Water in Nominally Anhydrous Minerals 9781501509476, 9780939950744

Citation preview

REVIEWS IN MINERALOGY AND GEOCHEMISTRY Volume 6 2

2006

WATER IN NOMINALLY ANHYDROUS MINERALS EDITORS

Hans Keppler

Joseph R. Smyth

Universität Bayreuth Bayreuth, Germany

University Colorado Boulder, Colorado

C O V E R P H O T O G R A P H : Thin section of a garnet lherzolite mantle xenolith from Pali-Aike, Patagonia. The almost colorless grains are olivine, orthopyroxene is brownish-green, clinopyroxene bright green and garnet is red. Grain size is about 1 mm. Photograph courtesy of Sylvie Demouchy.

Series Editor: J odi J.

Rosso

GEOCHEMICAL SOCIETY MINERALOGICAL SOCIETY OF AMERICA

SHORT COURSE SERIES DEDICATION Dr. William C. Luth has had a long and distinguished career in research, education and in the government. He was a leader in experimental petrology and in training graduate students at Stanford University. His efforts at Sandia National Laboratory and at the Department of Energy's headquarters resulted in the initiation and long-term support of many of the cutting edge research projects whose results form the foundations of these short courses. Bill's broad interest in understanding fundamental geochemical processes and their applications to national problems is a continuous thread through both his university and government career. He retired in 1996, but his efforts to foster excellent basic research, and to promote the development of advanced analytical capabilities gave a unique focus to the basic research portfolio in Geosciences at the Department of Energy. He has been, and continues to be, a friend and mentor to many of us. It is appropriate to celebrate his career in education and government service with this series of courses.

Reviews in Mineralogy and Geochemistry, Volume 62 Water in Nominally Anhydrous Minerals ISSN ISBN

1529-6466

0-939950-74-X

COPYRIGHT 2 0 0 6 THE M I N E R A L O G I C A L S O C I E T Y OF A M E R I C A 3 6 3 5 CONCORDE PARKWAY, SUITE 5 0 0 CHANTILLY, VIRGINIA, 2 0 1 5 1 - 1 1 2 5 , 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.

"Journey to the Center of the Earth": In 1864, the French writer Jules Verne published his novel "Voyage au Centre de la Terre" (Journey to the Center of the Earth). In this novel, Otto Lidenbrock, a German professor of mineralogy (!) discovers a secret handwriting inside an old manuscript. The handwriting appears to show a way to get to the center of the Earth by climbing down the conduit of an extinct volcano on Iceland. Together with his nephew and a guide from Iceland, Lidenbrock follows this trail. On his way to the center of the Earth, he discovers, among many other things, a gigantic ocean in the Earth's interior. This scene is depicted in the woodcut illustration overleaf, which is taken from the first complete edition of the works of Jules Verne in French. The idea of a major water reservoir in the Earth's interior has therefore been anticipated already more than one century ago. Research in the last decades has entirely confirmed this idea, aside from some rather minor details, which are described in this book.

WATER in NOMINALLY ANHYDROUS MINERALS 62

Reviews in Mineralogy and Geochemistry

62

FROM THE SERIES EDITOR The review chapters in this volume were the basis for a four day short course on Water in Nominally Anhydrous Minerals held in Verbania, Lago Maggiore, Italy (October 1-4, 2006). The editors Hans Keppler and Joe Smyth have done an excellent job organizing this volume and the associated short course. Meeting deadlines (often ahead of schedule!) and keeping track of so many authors can be a thankless job at times but I truly appreciate all their hard work. Hans' "friendly reminder" e-mails certainly kept us all on task and his eye for detail (small and large) made my job much more enjoyable! I extend my sincere thanks to him for his efforts! Any supplemental material and errata (if any) can be found at the MSA website www. minsocam. org. Todi T. P-osso, Series Editor West Richland, Washington August 2006

PREFACE Earth is a water planet. Oceans of liquid water dominate the surface processes of the planet. On the surface, water controls weathering as well as transport and deposition of sediments. Liquid water is necessary for life. In the interior, water fluxes melting and controls the solidstate viscosity of the convecting mantle and so controls volcanism and tectonics. Oceans cover more than 70% of the surface but make up only about 0.025% of the planet's mass. Hydrogen is the most abundant element in the cosmos, but in the bulk Earth, it is one of the most poorly constrained chemical compositional variables. Almost all of the nominally anhydrous minerals that compose the Earth's crust and mantle can incorporate measurable amounts of hydrogen. Because these are minerals that contain oxygen as the principal anion, the major incorporation mechanism is as hydroxyl, OH", and the chemical component is equivalent to water, H 2 0. Although the hydrogen proton can be considered a monovalent cation, it does not occupy same structural position as a typical cation in a mineral structure, but rather forms a hydrogen bond with the oxygens on the edge of the coordination polyhedron. The amount incorporated is thus quite sensitive to pressure and the amount of H that can be incorporated in these phases generally increases with pressure and sometimes with temperature. Hydrogen solubility in nominally anhydrous minerals is thus much more sensitive to temperature and pressure than that of other elements. Because the mass of rock in the mantle is so large relative to ocean mass, the amount that is incorporated the nominally anhydrous phases of the interior may constitute the largest reservoir of water in the planet. Understanding the behavior and chemistry of hydrogen in minerals at the atomic scale is thus central to understanding the geology of the planet. There have been significant recent advances in the detection, measurement, and location of H in the nominally anhydrous silicate and oxide minerals that compose the planet. There have also been advances in experimental methods for measurement of H diffusion and the effects of H on the phase 1529-6466/06/0062-0000505.00

DOI: 10.2138/rmg.2006.62.0

Water in Nominally Anhydrous Minerals - Preface boundaries and physical properties whereby the presence of H in the interior may be inferred from seismic or other geophysical studies. It is the objective of this volume to consolidate these advances with reviews of recent research in the geochemistry and mineral physics of hydrogen in the principal mineral phases of the Earth's crust and mantle. The chapters We begin with a review of analytical methods for measuring and calibrating water contents in nominally anhydrous minerals by George Rossman. While infrared spectroscopy is still the most sensitive and most convenient method for detecting water in minerals, it is not intrinsically quantitative but requires calibration by some other, independent analytical method, such as nuclear reaction analysis, hydrogen manometry, or SIMS. A particular advantage of infrared spectroscopy, however, is the fact that it does not only probe the concentration, but also the structure of hydrous species in a mineral and in many cases the precise location of a proton in a mineral structure can be worked out based on infrared spectra alone. The methods and principles behind this are reviewed by Eugen Libowitzky and Anton Beran, with many illustrative examples. Compared to infrared spectroscopy, NMR is much less used in studying hydrogen in minerals, mostly due to its lower sensitivity, the requirement of samples free of paramagnetic ions such as Fe2+ and because of the more complicated instrumentation required for NMR measurements. However, NMR could be very useful under some circumstances. It could detect any hydrogen species in a sample, including such species as H2 that would be invisible with infrared. Potential applications of NMR to the study of hydrogen in minerals are reviewed by Simon Kohn. While structural models of "water" in minerals have already been deduced from infrared spectra several decades ago, in recent years atomistic modeling has become a powerful tool for predicting potential sites for hydrogen in minerals. The review by Kate Wright gives an overview over both quantum mechanical methods and classical methods based on interatomic potentials. Joseph Smyth then summarizes the crystal chemistry of hydrogen in high-pressure silicate and oxide minerals. As a general rule, the incorporation of hydrogen is not controlled by the size of potential sites in the crystal lattice; rather, the protons will preferentially attach to oxygen atoms that are electrostatically underbonded, such as the non-silicate oxygen atoms in some high-pressure phases. Moreover, heterovalent substitutions, e.g., the substitution of Al3+ for Si4+, can have a major effect on the incorporation of hydrogen. Data on water in natural minerals from crust and mantle are compiled and discussed in three reviews by Elisabeth Johnson, Henrik Skogby and by Anton Beran and Eugen Libowitzky. Among the major mantle minerals, clinopyroxenes usually retain the highest water contents, followed by orthopyroxenes and olivine, while the water contents in garnets are generally low. Most of these water contents need to be considered as minimum values, as many of the mantle xenoliths may have lost water during ascent. However, there are some cases where the correlation between the water contents and other geochemical parameters suggest that the measured water concentrations reflect the true original water content in the mantle. The basic thermodynamics as well as experimental data on water solubility and partitioning are reviewed by Hans Keppler and Nathalie Bolfan Casanova. Water solubility in minerals depends in a complicated way on pressure, temperature, water fugacity and bulk composition. For example, water solubility in the same mineral can increase or decrease with temperature, depending on the pressure of the experiments. Nevertheless, the pressure and temperature dependence of water solubility can be described by a rather simple thermodynamic formalism and for most minerals of the upper mantle, the relevant thermodynamic parameters are known. The highest water solubilities are reached in the minerals wadsleyite and ringwoodite stable in the transition zone, while the minerals of the lower mantle are probably mostly dry. The rather limited experimental data on water partitioning between silicate melts and minerals are reviewed by Simon Kohn and Kevin Grant. One important observation here is that comparing vi

Water in Nominally Anhydrous Minerals - Preface the compatibility of hydrogen with that of some rare earth element is misleading, as such correlations are always limited to a small range of pressure and temperature for a given mineral. The stabilities of hydrous phases in the peridotite mantle and in subducted slabs are reviewed by Daniel Frost and by Tatsuhiko Kawamoto. While most of the water in the mantle is certainly stored in the nominally anhydrous minerals, hydrous phases can be important storage sites of water in certain environments. Amphibole and phlogopite require a significant metasomatic enrichment of Na and K in order to be stabilized in the upper mantle, but serpentine may be an important carrier of water in cold subducted slabs. The diffusion of hydrogen in minerals is reviewed by Jannick Ingrin and Marc Blanchard. An important general observation here is that natural minerals usually do not loose hydrogen as water, but as H 2 generated by redox reaction of OH with Fe2+. Moreover, diffusion coefficients of different mantle minerals can vary by orders of magnitude, often with significant anisotropy. While some minerals in a mantle xenolith may therefore have lost virtually all of their water during ascent, other minerals may still preserve the original water content and in general, the apparent partition coefficients of water between the minerals of the same xenolith can be totally out of equilibrium. Accordingly, it would be highly desirable to directly deduce the water content in the mantle from geophysical data. One strategy, based on seismic velocities and therefore ultimately on the effect of water on the equation of state of minerals, is outlined by Steve Jacobsen. The dissolution of water in minerals usually increases the number of cation vacancies, yielding reduced bulk and shear moduli and seismic velocities. Particularly, the effect on shear velocities is strong and probably larger than the effect expected from local temperature variations. Accordingly, the v s /v P ratio could be a sensitive indicator of mantle hydration. A more general approach towards remote sensing of hydrogen in the Earth's mantle, including effects of seismic anisotropy due to lattice preferred orientation and the use of electrical conductivity data is presented by Shun-ichiro Karato. Probably the most important effect of water on geodynamics is related to the fact that even traces of water dramatically reduce the mechanical strength of rocks during deformation. The physics behind this effect is discussed by David Kohlstedt. Interestingly, it appears that the main mechanism behind "hydrolytic weakening" is related to the effect of water on the concentration and mobility of Si vacancies, rather than to the protons themselves. Water may have major effects on the location of mantle discontinuities, as reviewed by Eiji Ohtani and Konstantin Litasov. Most of these effects can be rationalized as being due to the expansion of the stability fields of those phases (e.g., wadsleyite) that preferentially incorporate water. Together with other geophysical data, the changes in the depths of discontinuities are a promising tool for the remote sensing of water contents in the mantle. The global effects of water on the evolution of our planet are reviewed in the last two chapters by Bernard Marty, Reika Yokochi and Klaus Regenauer-Lieb. By combining hydrogen und nitrogen isotope data, Marty and Yokochi demonstrate convincingly that most of the Earth's water very likely originated from a chondritic source. Water may have had a profound effect on the early evolution of our planet, since a water-rich dense atmosphere could have favored melting by a thermal blanketing effect. However, Marty and Yokochi also show very clearly that it is pretty much impossible to derive reliable estimates of the Earth's present-day water content from cosmochemical arguments, since many factors affecting the loss of water during and after accretion are poorly constrained or not constrained at all. In the last chapter, Klaus Regenauer-Lieb investigates the effect of water on the style of global tectonics. He demonstrates that plate tectonics as we know it is only possible if the water content of the mantle is above a threshold value. The different tectonic style observed on Mars and Venus may therefore be directly related to differences in mantle water content. Earth is the water planet - not just because of its oceans, but also because of its tectonic evolution.

Water in Nominally

Anhydrous

Minerals

- Preface

Acknowledgments This volume and the accompanying short course in Verbania were made possible by generous support from the Mineralogical Society of America, the Geochemical Society, the United State Department of Energy, the German Mineralogical Society and Bayerisches Geoinstitut. The Verbania short course is the first MSA/GS short course ever held in Italy. We are very grateful for the generosity and the international spirit of the supporting institutions, which made this project possible. The preparation of the short course benefited enormously from the permanent advice by Alex Speer. Finally, we would like to thank Jodi Rosso for the efficient and professional handling of the manuscript and for her patience with authors and editors who ignore deadlines. August 2006 Hans Keppler Bayreuth, Germany

DMÒ •

Deutsche Deutsc Mineralogische Minerale Gesellschaft Gesellscha

Joseph R. Smyth Boulder, Colorado, USA

BGI(

f:. 0

Bayerisch« Geoinstitut

viii

WATER in NOMINALLY ANHYDROUS MINERALS 62

Reviews in Mineralogy and. Geochemistry

62

TABLE OF CONTENTS 1

Analytical Methods for Measuring Water in Nominally Anhydrous Minerals George R. Rossman

INTRODUCTION ANALYTICAL METHODS Early infrared studies Quantitative IR methods Mineral specific calibrations Thermogravimetric methods P 2 0 5 cell coulometry Hydrogen extraction with uranium reduction methods Nuclear methods for hydrogen determination Nuclear magnetic resonance Secondary ion mass spectrometry (SIMS) PREVIOUS REVIEWS OF METHODS SURFACE WATER CURRENT STATUS OF CALIBRATIONS GLASSES ACKNOWLEDGMENTS REFERENCES

2

1 2 2 3 8 9 9 10 11 16 18 20 21 23 23 24 24

The Structure of Hydrous Species in Nominally Anhydrous Minerals: Information from Polarized IR Spectroscopy Eugen Libowitzky and Anton Beran

INTRODUCTION The importance of hydrous species in NAMs Why use IR spectroscopy? History CONCEPTS OF INFRARED SPECTROSCOPY Introduction to IR spectroscopy Sample requirements ix

29 29 30 30 31 31 32

Water in Nominally Anhydrous Minerals - Table of Contents Experimental equipment QUANTITATIVE DATA FROM INFRARED SPECTROSCOPY The distance - frequency correlation of hydrogen bonds The spatial orientation of hydrous species Total absorbance: a first step towards quantitative water analysis CONCEPTS OF STRUCTURAL MODELS FROM INFRARED DATA Charge balance and substitution Electrostatic considerations on defect geometry Space requirements: ideal and distorted models Influence on band energies from cation substitution Discrimination among hydrous defects Deuteration I XAMl'l I S Vesuvianite: orientation and hydrogen bonding of hydroxyl groups Hydrogarnet substitution - the (OH)44~ cluster Water molecules in structural cavities: beryl and cordierite OH substitution in topaz OH incorporation in diopside OH defects in perovskite OH traces in corundum ACKNOWLEDGMENTS REFERENCES

3

32 33 33 34 35 36 36 37 38 38 39 40 40 40 41 43 44 44 47 48 49 49

Structural Studies of OH in NominallyAnhydrous Minerals Using NMR Simon C. Kohti

INTRODUCTION 53 PRINCIPLES OF SOI ID STATE NMR 54 Positions of 'H MAS NMR resonances 55 Widths of 'H MAS NMR resonances 55 Intensity of 'H MAS NMR resonances 56 APPLICATION OF II MAS NMR TO NOMINALLY ANHYDROUS MINERALS 58 Attractive features of 'H MAS NMR for studies of NAMs 58 Problems and difficulties in applying 'H MAS NMR to NAMs 58 'H MAS NMR studies of orthopyroxene 59 'H MAS NMR studies of clinopyroxene 60 'H MAS NMR studies of olivine 60 'H MAS NMR studies of garnet 61 'H MAS NMR studies of Si0 2 polymorphs 61 'H MAS NMR studies of feldspars and other aluminosilicate framework minerals... 61 'H MAS NMR studies of wadsleyite 62 NON-SPINNING II NMR EXPERIMENTS 62 STUDIES OF OTHER NUCLEI IN NAMS 63 PROSPECTS FOR FUTURE DEVELOPMENT OF NMR FOR STUDIES OF NAMS 65 ACKNOWLEDGMENTS 65 REFERENCES 65 x

Water in Nominally

Anhydrous

Minerals

- Table of Contents

in Nominally Anhydrous Minerals Kate Wright INTRODUCTION POINT 1)11 I ( I S IN MINERALS THEORETICAL BACKGROUND Quantum mechanical methods Classical methods Treatment of defects OH DEFECTS IN MANTLE SILICATES The Mg 2 Si0 4 polymorphs Pyroxene GENERAL REMARKS AND FUTURE DIRECTIONS ACKNOWLEDGMENTS REFERENCES

5

67 67 69 69 70 71 72 73 79 80 81 81

Hydrogen in High Pressure Silicate and Oxide Mineral Structures Joseph R. Smyth

INTRODUCTION GEOCHEMISTRY OF H CRYSTAL CHEMISTRY OF H NOMINALLY HYDROUS HIGH-PRESSURE SILICATE PHASES Brucite Serpentine Talc True micas Chlorite Amphiboles Lawsonite Epidote Humite Clinohumite Chondrite Phase A Phase B Superhydrous Phase B Phase D Phase E Phase Pi Topaz-OH Phase Egg K-cymrite xi

85 85 86 87 90 90 90 91 92 92 92 93 94 94 94 95 96 96 96 97 97 98 98 98

Water in Nominally Anhydrous Minerals - Table of Contents NOMINALLY ANHYDROUS HIGH-PRESSURE SILICATE AND OXIDE PHASES Periclase-wiistite Corundum Coesite Stishovite and rutile Pyroxenes Akimotoite Garnet Olivine Wadsleyite Wadsleyite II Ringwoodite Anhydrous phase B Kyanite Perovskite Post-perovskite Zircon Titanite CONCLUSIONS ACKNOWLEDGMENT REFERENCES

6

99 99 99 101 101 102 103 104 104 105 106 106 107 107 108 108 109 110 110 110 110

Water in Nominally Anhydrous Crustal Minerals: Speciation, Concentration, and Geologic Significance Elizabeth A. Johnson

INTRODUCTION Importance of nominally anhydrous minerals in the crust Scope and goals of this chapter HYDROUS SPECIES AND CONCENTRATIONS IN CRUSTAL MINERALS Quartz and coesite Feldspars and nepheline Pyroxenes Garnets Al 2 Si0 5 polymorphs Rutile and cassiterite Zircon and titanite Cordierite and beryl UNDERSTANDING GEOLOGIC SYSTI MS Thermodynamic properties Physical properties The water budget of the Earth SUMMARY AND FUTURE POSSIBILITIES ACKNOWLEDGMENTS REFERENCES APPENDIX Studies of hydrogen in quartz xii

117 117 117 118 118 119 122 123 124 126 128 129 130 130 134 136 136 137 137 142 143

Water in Nominally Anhydrous Minerals

- Table of Contents

Geological studies of H 2 0 and C 0 2 in cordierite Hydrous species in feldspars Structural hydroxyl concentrations in crustal and mantle pyroxenes Structural hydroxyl concentrations in crustal garnets Structural hydroxyl concentrations in kyanite Structural hydroxyl concentrations in sillimanite Structural hydroxyl concentration in andalusite Structural hydroxyl concentrations in rutile Structural hydroxyl concentrations in cassiterite

7

143 144 146 148 152 152 153 153 154

Water in Natural Mantle Minerals I: Pyroxenes Henrik Skogby

INTRODUCTION OH ABSORPTION BANDS IN IR SPECTRA Diopside Augite Omphacite Orthopyroxene Absorption from inclusions CORRELATIONS OF OH AND SAMPLE CHEMISTRY WATER CONCENTRATION IN MANTLE PYROXENES IMPLICATIONS FOR WATER IN THE UPPER MANTLE ACKNOWLEDGMENTS REFERENCES

8

155 156 156 156 157 157 158 159 162 164 165 166

Water in Natural Mantle Minerals II: Olivine, Garnet and Accessory Minerals Anton Beran and Eugen

INTRODUCTION OLIVINE Basic structure and possible sites of hydrogen incorporation Defect types in mantle-related olivines from different localities Calibration approaches and summary of hydrogen contents GARNET Structural and spectral features Calibration and hydrogen content \( ( I SSORY MINERALS Kyanite Rutile Coesite Spinel Zircon ACKNOWLEDGMENTS REFERENCES xiii

Libowitzky 169 170 170 171 176 176 176 180 182 182 184 185 186 186 188 188

Water in Nominally Anhydrous Minerals - Table of Contents 7

Thermodynamics of Water Solubility and Partitioning Hans Keppler and Nathalie

Bolfan-Casanova

INTRODUCTION BASIC THERMODYNAMICS OF WATER SOLUBILITY AND PARTITIONING The meaning of the term "water solubility" Thermodynamics of water solubility Relationship between water solubility and partitioning EXPERIMENTAL STRATEGIES FOR MEASURING WATER SOLUBILITY AND WATER PARTITION ( O i l 1 l( U N I S Annealing experiments Crystallization experiments WATER IN UPPER MANTLE MINERALS Water solubility in and the A1 content of orthopyroxenes as "geohygrometer" Water solubility in olivine Water solubility in garnet Water solubility in clinopyroxene Water partitioning among upper mantle minerals Water storage capacity of the upper mantle and the origin of the Earth's asthenosphere Water recycling by subducted slabs WATER IN TRANSITION ZONE MINERALS Water solubility in wadsleyite and water partitioning between wadsleyite and olivine Partitioning of water between wadsleyite and ringwoodite Partition coefficients of water between other high-pressure phases WATER IN MINERALS OF THE LOWER MANTLE Water in ferropericlase Water in magnesium silicate perovskite The distribution of water at the 660 km discontinuity THE EQUILIBRIUM DISTRIBUTION OF WATER IN THE EARTH'S INTERIOR ACKNOWLEDGMENTS REFERENCES

1 0

193 194 194 195 198 199 199 200 201 201 205 211 211 212 214 215 216 216 219 220 222 222 223 225 226 227 227

The Partitioning of Water Between Nominally Anhydrous Minerals and Silicate Melts Simon C. Kohn and Kevin J. Grant

INTRODUCTION PARTITIONING OF WATER BETWEEN NAMS AND MELTS; METHODOLOGY AND APPROACH Experimental studies of water partitioning between NAMs and melts SUMMARY, IMPLICATIONS AND FUTURE RESEARCH ACKNOWLEDGMENTS REFERENCES xiv

231 233 233 238 239 239

Water in Nominally

Anhydrous

Minerals

- Table of Contents

Daniel J. Frost INTRODUCTION MANTLE METASOMATISM EVIDENCE 1 ROM MANTLE XI NOI I I IIS Peridotite massifs and xenoliths from alkaline basalts Xenoliths from kimberlites Mantle amphibole mineralogy Mantle mica mineralogy EXPERIMENTAL STUDIES ON THE STABILITY OF KNOWN MANTLE HYDROUS MINERALS Pargasitic amphiboles Apatite Phlogopite K-richterite EXPERIMENTAL STUDIES ON THE STABILITY OF POTENTIAL HIGH PRESSURE HYDROUS MANTLE MINERALS Phase X Humite and dense hydrous magnesium silicate phases THE STABILITY OF HYDROUS PHASES IN ULTRAMAFIC LITHOSPHERE AND THE CONVECTING MANTLE ACKNOWLEDGMENTS REFERENCES

1 2

243 244 246 247 247 248 250 251 251 255 256 259 260 261 262 262 265 266

Hydrous Phases and Water Transport in the Subducting Slab Tatsuhiko Kawamoto

INTRODUCTION LOW-PRESSURE HYDROUS MINERALS AND HIGH-PRESSURE HYDROUS PHASES STABILITY OF HYDROUS PHASES IN DOWNGOING PERIDOTITE STABILITY OF HYDROUS PHASES IN DOWNGOING BASALT AND SEDIMENT PRESSURE - TEMPERATURE CONDITIONS AND DEHYDRATION REACTIONS IN THE SUBDUCTING SLAB COMPOSITION AND DIHEDRAL ANGLES OF AQUEOUS FLUIDS IN MANTLE PERIDOTITE SECOND CRITICAL ENDPOINT BETWEEN MAGMAS AND AQUEOUS FLUID: IMPLICATIONS FOR SLAB-DERIVED COMPONENT CONCLUDING REMARKS ACKNOWLEDGMENT REFERENCES xv

273 273 277 279 280 281 .282 285 286 286

Water in Nominally

I «3

Anhydrous

Minerals

- Table of Contents

Diffusion of Hydrogen in Minerals Jannick Ingrin and Marc Blanchard

INTRODUCTION BASIC CONCEPTS OF DIFFUSION IN MINERALS EXPERIMENTAL METHODS MEASUREMENT TECHNIQUES Infrared spectroscopy Mass spectrometry Thermogravimetry Nuclear reaction analysis Liquid scintillation counting Proton-proton scattering Theoretical techniques DETECTION OF H DIFFUSION THROUGH ISOTOPE EXCHANGE Anhydrous minerals Hydrous minerals EXTRACTION/INCORPORATION REACTIONS IN ANHYDROUS MINERALS Olivine Diopside Enstatite Garnets Quartz Feldspars CONCLUSION AND FUTURE DIRECTIONS ACKNOWLEDGMENTS REFERENCES

1 4

291 291 292 294 295 296 297 297 298 298 298 299 299 304 307 307 310 313 314 316 316 317 318 318

Effect of Water on the Equation of State of Nominally Anhydrous Minerals Steven D. Jacobsen

INTRODUCTION ELASTIC PROPERTIES OF NOMINALLY ANHYDROUS MINERALS IN THE UPPER MANTLE Olivine Humite-group minerals along the forsterite-brucite join Garnet Grossular-hydrogrossular Pyroxene ELASTIC PROPERTIES OF NOMINALLY ANHYDROUS MINERALS IN THE TRANSITION ZONE Wadsleyite Wadsleyite-II Ringwoodite DENSE HYDROUS MAGNESIUM SILICATES xvi

321 322 322 323 323 326 326 328 328 330 330 332

Water in Nominally Anhydrous Minerals

- Table of Contents

Phase A Phase-B group minerals Phase D Phase E WATER-ELASTICITY SYSTI M ATK S CALCULATED HYDROUS VELOCITIES IN THE UPPER MANTLE A N D TRANSITION ZONE CONCLUSIONS A N D FUTURE RESEARCH OPPORTUNITIES ACKNOWLEDGMENTS REFERENCES

1 5

335 338 338 338

Remote Sensing of Hydrogen in Earth's Mantle Shun-ichiro

INTRODUCTION GEOPHYSICAL OBSERVATIONS Electrical conductivity Seismic wave velocities Seismic wave attenuation Seismic anisotropy Topography of discontinuity Sharpness of discontinuities PHYSICAL BASIS FOR INFERRING HYDROGEN CONTENT FROM GEOPHYSICAL OBSERVATIONS Electrical conductivity Seismic properties Partial melting? SOME EXAMPLES Water content in the transition zone Distribution of hydrogen in the upper mantle Hydrogen in the lower mantle SUMMARY AND OUTLOOK ACKNOWLEDGMENTS REFERENCES

1 6

332 332 332 334 334

Karato 343 344 344 346 346 347 348 349 349 349 351 362 363 363 366 369 370 370 371

The Role of Water in High-Temperature Rock Deformation David L.

INTRODUCTION BACKGROUND MODELS OF CLIMB-CONTROLLED CREEP THE CASE FOR OLIVINE DISLOCATION CLIMB DIFFUSION DEPENDENCE OF CREEP RATE ON WATER FUGACITY xvii

Kohlstedt 377 379 379 381 383 384 386

Water in Nominally Anhydrous Minerals - Table of Contents CONCLUDING REMARKS ACKNOWLEDGMENTS REFERENCES APPENDIX Charge neutrality Flux equations for a semi-conducting silicate

1 7

388 390 390 394 394 395

The Effect of Water on Mantle Phase Transitions Eiji Ohtani and K. D. Litasov

INTRODUCTION RECENT PROGRESS ON PRESSURE SCALES FOR THE DETERMINATION OF PHASE BOUNDARIES IN MANTLE MINERALS EFFECT OF WATER ON PHASE TRANSFORMATION Dry and wet phase boundaries in the olivine-wadsleyite transformation Wadsleyite-ringwoodite transformation Post-spinel transformation Post-garnet transformation in basalt (MORB) EFFECT OF WATER ON PHASE TRANSFORMATION KINETICS Olivine-wadsleyite phase transformation kinetics Post-spinel and post garnet phase transformation kinetics IMPLICATION FOR SEISMIC DISCONTINUITIES AND PHASE TRANSFORMATION BOUNDARIES UNDER DRY AND WET CONDITIONS 410 km seismic discontinuity and olivine-wadsleyite phase boundary The 660 km seismic discontinuity and the post-spinel transformation: average depth and topography of the 660 km seismic discontinuity The density relation of basalt and peridotite near the 660 km discontinuity Seismic reflectors: the possible existence of fluid in the lower mantle CONCLUDING REMARKS ACKNOWLEDGMENTS REFERENCES

1 8

397 398 400 400 401 401 404 406 406 408 409 409 410 412 413 415 415 415

Water in the Early Earth Bernard Marty and Reika Yokochi

INTRODUCTION ISOTOPIC CONSTRAINTS ON THE ORIGIN OF TERRESTRIAL WATER Hydrogen isotopic ratios Nitrogen and carbon isotopic ratios Noble gas isotopic ratios Other isotopic constraints POTENTIAL WATER CONTRIBUTORS Contribution of water-rich planetary embryos Asteroid contribution Constraints on water delivery by asteroidal material from the terrestrial highly xviii

421 422 422 424 424 427 428 429 429

Water in Nominally

Anhydrous

Minerals

- Table of Contents

siderophile element budget The case of interplanetary dust as a source of terrestrial water PROCESSES OF WATER INCORPORATION IN EARTH Solar nebula Impact degassing Impact erosion Post-accretional role of a proto-atmosphere in the Early Earth's evolution A summary of volatile delivery processes and of their inherent uncertainties Cooling of the primordial Earth I Hi: WATER ( Y( I i: IN I Hi: HADE AN WATER CONTENT OF THE ARCHEAN MANTLE FROM THE COMPOSITION OF KOMATIITES CONCLUSIONS ACKNOWLEDGMENTS REFERENCES

1 9

430 432 435 435 437 438 438 439 439 440 443 444 444 445

Water and Geodynamics Klaus Regenauer-Lieb

INTRODUCTION WATER IN m i : i rniosi'iii Ri : Water and the rigidity of (oceanic) plates Water and the nucleation of (new) plate boundaries Water and the evolution of plate boundaries Water and the stored energy potential Y Water and the dissipated energy potential ® Solid versus fluid dynamic modeling setups Application to subduction initiation WATER IN IIIi: CONVECTING MANTLE DISCUSSION AND CONCLUSIONS REFERENCES APPENDIX: THERMOMECHANICAL APPROACH

451 452 452 454 455 455 457 461 462 465 465 467 471

Additional Volume Content: COLOR COLOR COLOR COLOR

PLATE I PLATE 2 PLATE 3 PLATE 4

475 476 477 478

xix

1

Reviews in Mineralogy & Geochemistry Vol. 62, pp. 1-28, 2006 Copyright © Mineralogical Society of America

Analytical Methods for Measuring Water in Nominally Anhydrous Minerals George R. Rossman Division of Geological and Planetary Sciences California Institute of Technology Pasadena, California, 91125-2500, U.S.A. e-mail:

[email protected]

INTRODUCTION Decades of work have shown that trace- to minor-amounts of hydrous components commonly occur in minerals whose chemical formula would be normally written without any hydrogen, namely, the nominally anhydrous minerals (NAMs). When the concentrations of the hydrous components are several tenths of a percent by weight or higher, a variety of analytical methods such as weight loss on heating, X-ray cell parameters, X-ray structure refinement, Karl-Fischer titrations, or even careful electron microprobe analyses can be used to establish their concentrations (e.g., Aines and Rossman 1991). However, for most NAMs, accurate determinations with these common analytical methods prove difficult if not impossible. For this reason, infrared (IR) spectroscopy has become, and remains, the most widely used method to detect and analyze hydrous components (OH or H 2 0) in minerals and glasses because it is both highly sensitive and can be done rapidly with a commonly available, modestly priced instrument and at dimensions of just a few tens of micrometers. A change in the electric dipole occurs when the OH bond in either water and hydroxy 1 ions vibrate. This motion has a resonance coupling with electromagnetic radiation generally in the 3500 cm - 1 region of the infrared spectrum. In addition, bending motions of the water molecule, and overtones and combination of these motions produce absorption in the infrared. Under favorable conditions, namely a sharp band in a single orientation, just a few nanometers equivalent thickness of a hydroxyl species such as an amphibole can be detected in an otherwise anhydrous mineral such a pyroxene (Skogby et al. 1990). Routinely, detection limits of a few to tens of ppm wt of H 2 0 in a mineral can be detected and often quantitatively determined. The overtone and combination modes of OH and H 2 0 behave in predictable fashion in minerals (Rossman 1975) so that the two species can usually be separated from each other. Infrared spectra, however easily obtained, are not rigorously self-calibrating, so independent methods of analysis have been necessary to calibrate the spectroscopic work. A couple general correlations of IR band intensity with the absorption energy have proven useful, if approximate. Various absolute hydrogen extraction methods have proven highly useful for purpose of rigorous calibration. More recently, nuclear methods that rely upon specific resonant reactions with the hydrogen nucleus or nuclear scattering specific to hydrogen have gained importance and have provided critical absolute calibrations of the infrared spectra. Secondary Ion Mass Spectroscopy (SIMS) for hydrogen is still in the early stages of development but once calibrated, and with established protocols, should play an ever-expanding role in the future. NanoSIMS promises to bring hydrogen analyses to ever finer spatial dimensions but will require significant effort before it can be regarded as an accurate analytical technique for small concentrations of hydrogen. The purpose of this chapter is to review the various methods that have been used to analyze hydrous components in the NAMs. 1529-6466/06/0062-0001505.00

DOI: 10.2138/rmg.2006.62.1

2

Rossman ANALYTICAL METHODS

Early infrared studies Much of the early interest in OH in minerals came from the study of synthetic minerals used in the electronics industry. Quartz, in particular, was an important phase used for frequency control in telecommunications and radio circuits. Consequently, much effort was directed towards the understanding of factors that influenced the efficiency and cost of these devices. Water in quartz was one of the most important factors. The OH bond is dipolar with a partial negative charge on the oxide ion and a partial positive charge on the hydrogen ion. Thus, the vibrations of the OH bond coupled well to infrared radiation and infrared spectroscopy quickly became the tool of choice to study OH in both natural and synthetic minerals. An important early study was conducted by Kats and Haven (1960) who used deuteration to demonstrate which bands in the complex quartz spectrum in the 3000 to 4000 cm -1 region originated from ! H as opposed to overtone or combination bands of the quartz vibrational spectrum that appeared in the same region. Once the OH vibrations were positively identified, Kats (1962) performed a comprehensive study of OH in quartz and identified which of the sharp band absorptions in the 3000-3600 cm -1 region are due to O-H stretching vibrations. Kats further showed that most of the absorptions are primarily due to the presence of Al3+ substitution for Si4+ with charge compensating cations (such as H + , Li+, Na+) in defects in the crystal. Other studies were taking place at Bell Labs in the United States where elastic properties and dielectric loss in synthetic quartz was related to H defects (King et al. 1960; Dodd and Fraser 1965,1967). In these studies, the relationship between infrared absorption, and hydroxyl and water defects in quartz was also being established. During these times, Brunner et al. (1961) concluded that H enters defects in clear, natural quartz in the form of OH ions and estimated the amount of H as 1018 per cc (corresponding to about 15 ppm H 2 0 wt). These early estimates showed that small amounts of hydrous components could have a large impact on the physical properties of the host phase. As work on synthetic quartz progressed, studies of quartz also used natural samples and ultimately, the results were reported in the mineralogical literature through the work of Dodd and Fraser (1965). Simultaneously, interest in the low concentrations of water in ring silicate minerals was generated by infrared (Schreyer and Yoder 1964; Wood and Nassau 1967; Farrell and Newnham 1967) and NMR (proton nuclear magnetic resonance) (Pare and Ducros 1964; Sugitani et al. 1966) studies of beryl that demonstrated that water molecules occur in the c-axis channels. The NMR work showed that the water molecules were in motion and the IR studies showed that the water molecule existed in two independent crystallographic orientations in the crystal. In Austria, in the late 1960's and early 1970's, Beran and Zemann obtained the IR spectra of a number of minerals such as titanite, kyanite, axinite, titanium oxides, cassiterite (Beran 1970a,b,c,d; Beran and Zemann 1969a,b, 1971) and demonstrated that they had structurally bound, crystallographically oriented OH groups. These studied demonstrated that polarized infrared radiation could establish the orientation of the OH groups in minerals and demonstrated that trace amounts of hydroxyl occur broadly in a number of nominally anhydrous minerals. A couple of significant motivations to develop quantitative understanding of the H-content of nominally anhydrous minerals appeared in the early 1970's. Martin and Donnay (1972) suggested that hydrogen may be stored as OH groups in minerals in the deep earth, and Wilkins and Sabine (1973) initiated a broad effort to determine the amount of hydrous components in a variety of minerals by combining infrared absorption with independent water analysis (P 2 0 5 electrolytic coulometry). Although we now recognize that many of the analyses of Wilkins and Sabine included alteration products and water in micro-inclusions, they did set the quantitative stage for further detailed studies.

Analytical

Methods

3

Another major impetus to the study of water in the nominally anhydrous minerals came from the studies of the rheological properties of, first, quartz (Griggs and Blacic 1965; Kirby and McCormick 1979), and then olivine (Mackwell et al. 1985). To study how water weakens minerals, it was necessary to know both the chemical species of the hydrous components that enter nominally anhydrous minerals, and to know their absolute concentrations. Quantitative IR methods The determination of the concentration of OH or H 2 0 in an "anhydrous" mineral depends upon accurate measurement of the infrared spectrum and ultimately on an independent calibration. Infrared spectra are intrinsically not self-calibrating. A number attempts have been made to develop generic calibrations. These often may be good as an initial estimate of the water concentration, but, for many systems, have been shown to be inadequate for precise work. Thus, mineral-specific calibrations have been developed. Once such calibrations are established and properly published, they can be used by other labs worldwide, even if an inhouse standard is not available. The well-established Beer-Lambert law is used to determine the concentration of hydrous species in a mineral from the infrared spectra: Absorbance = £ x c x t

(1)

This relates Absorbance (A), the band height in the region of interest (corrected for baseline), c, the concentration of hydrous species expressed in moles of H 2 0 per liter of mineral, and t, the thickness of the path (in cm) through which the measurement is made where £ is a mineral-specific calibration factor. In the classical chemical applications, the sample is in solution, so only one measurement is made. In the case of anisotropic solids, it is necessary to make the measurement in multiple directions (Libowitzky and Rossman 1996). Typically, linearly polarized light would be used and measurements would be made along the three principal extinction directions, X, Y, and Z. In this case, the intensities would be summed so A becomes AX + AY + Az (where Ax is the absorbance obtained with light polarized in the X direction, etc.). This approach tends to work best with phases that have one or a small number of narrow bands in the OH region. It also requires knowledge of the density of the mineral to convert from moles per liter to weight percent (or ppm) water. For most minerals, it is usually more useful to use a modified version of the Beer-Lambert law that uses integrated band areas rather than band heights. Band heights can vary depending on both the quality of the polarizer in the instrument and on the spectroscopic resolution of the instrument whereas band areas are less dependent on these parameters. The band height measured by the Absorbance is replaced by the total integrated area of bands in the region of interest Absorbancetotai (also written as Abstotai or Atotal). The concentration, c, remains expressed as moles of H 2 0 per liter of mineral. In this case, the absorption coefficient, £, is replaced by the integral molar absorption coefficient, I, in units of L/(mol-cm2). When c is expressed as ppm H 2 0 by weight, the absorption coefficient becomes the integral specific absorption coefficient (/', ppm_1-cm~2). The absorption coefficient for each species of hydrogen is found by determining the concentration, c, by an independent, absolute method and measuring Abstotai from polarized IR spectra in the three principal optical directions (X, Y, and Z) for the mineral of interest. For an orthorhombic mineral such as olivine: (2) Here, the equation specifies measuring the integrated area of an orthorhombic crystal with light polarized in the Ella, Ellb, and Ellc directions between the appropriate wavenumber limits of the OH bands, v l and v2. For lower symmetry crystals (monoclinic, and triclinic)

4

Rossman

Abstotai = \Absx+ \AbsY+ jAbs z , and for a uniaxial crystal (hexagonal or tetragonal) Abs,otai = 2 J'Abs±c+ \Absc (e.g., Libowitzky and Rossman 1996). To be comparable to measurements on lower symmetry crystals, an isotropic crystal would need to have Atotal = 3¡Abs a . Paterson's method. If the absorption frequency and intensity of a unit concentration of OH were a constant, then a single calibration of the OH spectrum would be all that is needed to conduct quantitative analysis with IR spectroscopy. Unfortunately, that is not the case. First of all, while the fundamental stretching vibration of a free (gaseous) hydroxide ion occurs at 3555.59 cm - 1 (Lutz 1995), the OH stretching frequency in a mineral commonly can occur over a range of several hundred wavenumbers and can vary by nearly 2000 cm -1 . A variety of studies (Nakamoto et al. 1955; Bellamy and Owen 1969; Novak 1974) showed that for a variety of chemical elements, the stretching frequency of an X-H bond in an X - H - Y hydrogen bonded system is a function of the X-Y distance. This includes O-H bonds. These authors derived empirical fits to experimental data that mathematically expressed this relationship. The second observation of interest is that the infrared absorption intensity of a unit concentration of OH in a solid is obviously not constant. Paterson (1982) confirmed that the strength of the OH absorption in the 3600 to 3000 cm - 1 region was frequency dependent. From the calibrations available for various substances, he presented a single empirical calibration line that related the OH intensity to band position that could be applied as a first approximation for determining the amount of OH in a variety of substances such as silicate glasses, quartz, and various forms of water. This was the first generic calibration specifically designed for the study of hydrous components in minerals and glasses. Paterson demonstrated that the intensity of an OH band (normalized to a unit concentration of H 2 0) increases when the band occurs at lower wavenumbers (stronger hydrogen bonding). This trend has been used by a number of authors to estimate the OH content of various minerals. Subsequent work has shown that determinations based on Paterson's trends are a reasonable first estimate, but that accurate determinations do require mineral-specific calibrations. Paterson's method first assumes that if a crystal is being measured, it is in a known crystallographic orientation. To determine the concentration of hydroxyl groups in the sample, the integrated absorbance is determined by integration of the infrared spectra over the region dominated by the stretching vibrations due to O-H bonds, typically from approximately 3750 to 3000 c m 4 . The integral molar absorption coefficient (I) is scaled to reflect the higher intrinsic intensities of bands at lower wavenumbers (stronger H-bonds) through the equation: I = y xl50 x (3780 - v)

(3)

where v is the wavenumber and gamma (y) is a factor to take account of the anisotropy of the crystal based on an assumption that O-H bonds are oriented in a single direction. The OH concentration is then calculated from a Beer-Lambert law relationship:

Concentration™, =

(150 xy)

f

——dv

(3780 - v )

(4)

assuming that the data are scaled for 1 cm sample thickness. Although uncertainties in this calibration were thought by Paterson to be about 30%, it has been widely adopted, partly in the hope that it would eliminate the need for more involved polarized light observations with multiple crystallographic directions. However, the studies of Libowitzky and Rossman (1997) and Bell et al. (2003) show that it can result in non-systematic underestimates of hydrogen concentrations. Examples of mineral specific calibrations that fall far from the trend are documented, particularly those that involve nominally anhydrous minerals with low concentrations of OH. As examples, the pyrope analyzed by Bell et al. (1995) departs from the Paterson trend by nearly a factor of three, the nuclear reaction analysis

Analytical

5

Methods

of olivine by Bell et al. (2003) departs by more than a factor of two (Fig. 1) and the SIMS analysis of both olivine and orthopyroxene (Koga et al. 2003) show that the Paterson trend also underestimates their OH concentrations. Libowitzky and Rossman's revision. Libowitzky and Rossman (1997) presented an updated version of the correlation of Paterson (1982). They measured polarized IR absorption data from single crystal minerals that contained stoichiometric water contents in the form of either OH or H 2 0. These data were used to construct a calibration curve for the intensity of the infrared absorption as a function of the band energy. Specifically, integrated molar absorption coefficient, £i (in units of cm - 2 per moleH2o/liter), was evaluated as function of the mean wavenumber of the OH stretching band (in units of cm -1 ). The result in Figure 2 shows that an increase in the hydrogen bonding leads to a decrease in the energy of the OH stretching energy which, in turn, is associated with an increase in the intensity of absorption. The form of the correlation is Ei = 246.6 x (3752 - v)

(5)

where v is the mean wavenumber of the OH stretching band. The results in Figure 2 show that the revised calibration produces £; values about threequarters of those of Paterson (1982). Measurements of minerals with stoichiometric OH are difficult to obtain. Their OH intensities are so high that crystals must be prepared very thin (perhaps as thin as 2 )im). Such preparations are difficult to near impossible; and when successful, the determination of their thickness to a high degree of accuracy is difficult.

_ 250 CM CO

Figure 1. Comparison of the results of the calibration developed by Bell et al. (2003) using Nuclear Reaction Analysis and the OH analysis method of Paterson (1982), as applied to polarized (solid circle) olivine spectra. Modified after Fig. 6 of Belletal. (2003).

1:1 line

2 200 150 (0

Q.

GRR 1012-2

100

0C-J 1 50

KLV-23 GRR 1695-2

E Q.

50

100

150

200

250

ppm H 2 0, Bell & Rossman (2003)

Figure 2. The correlation of the integrated molar absorption coefficient of OH stretching vs. wavenumber. Circles are experimental data points for stoichiometric minerals. The correlation of Paterson (1982) is shown for comparison. This means that if all things are equal, the Paterson trend underestimates the OH content. From Libowitzky and Rossman (1997). Wavenumber (cm"1)

6

Rossman

In a related effort, Libowitzky (1999) evaluated correlations specific to minerals between the frequency of the O-H stretching vibration and the length of the oxygen-oxygen distance and the H - O distances in the O-H—O hydrogen bond. Effectively, the shorter these distances are, the lower becomes the energy of the O-H stretch. Because the intensity of the OH band is related to the energy of the vibration (Libowitzky and Rossman 1997), such correlations provide some degree of a predictive estimate about the intensity of an OH absorption that arises from a particular site in a crystal. Use of unoriented grains. Asimow et al. (2006) present a method that allows multiple, randomly oriented grains of a mineral to be used to determine the total absorbance. In their method, the spectra of oriented sample of the phase of interest must already exist. Then, the spectra of three different randomly oriented crystals are measured, and the orientations of the grains are determined via methods such as electron backscatter diffraction (EBSD) or from the silicate overtone bands in the infrared spectra. They demonstrated that such methods result in angular errors of typically only 6 degrees and provide a surprising good determination of the OH content of the phase. Polarizer considerations. A linear polarizer must be used in the infrared beam of conventional spectrometers to obtain the total absorbance of anisotropic crystals. Commonly, the polarizers are made of a fine, parallel wire grid deposited on an infrared-transparent substrate such as CaF 2 or KRS5 (a thallium bromide iodide). These polarizers have wide acceptance angles and are readily available, but have only moderate polarization ratios. Crystal polarizers of a design similar to calcite polarizers used in the visible wavelength region are also available, but often have a narrow range of wavelengths over which they function. Lithium iodate covers a wide wavelength range and has a very high polarization ratio, but is hydroscopic and no longer readily available. Libowitzky and Rossman (1996) discussed the principles of quantitative absorbance measurements of anisotropic crystals and paid particular attention to the influence of the quality of the polarizers upon the results. First, they showed that the use of unpolarized radiation with an anisotropic crystal could not produce quantitatively accurate results. The Beer-Lambert law demands that the height of an absorption band will scale with the thickness of the sample. Figure 3 demonstrates how the spectrum taken with linearly polarized radiation follows the law. It also shows that unpolarized spectra do not scale according to the law. This means that unpolarized spectra should not be used to calibrate the infrared spectrum of OH

3

Figure 3. Comparison of the intensity of a carbonate overtone band in the calcite spectrum taken with well-polarized and unpolarized radiation. The experiment that used different thickness of calcite to test the Beer Lambert law shows that unpolarized spectra are not appropriate to quantitatively measure anisotropic crystals. From Libowitzky and Rossman (1997).

5

Thickness, m m

Analytical

7

Methods

in an anisotropic standard, and cannot be used to accurately determine the concentration of OH in an anisotropic unknown. The more highly anisotropic the sample is, the more problematic this issue will become. Libowitzky and Rossman also showed that the intensity of an absorption band of an anisotropic crystal is highly dependent upon the polarization ratio of the polarizers (Fig. 4) which means that if band heights are used to calibrate the infrared spectra, results can vary significantly from lab to lab if the appropriate in-lab standards are not available. Baselines issues. Figure 5 shows that strongly rising, non-linear baselines may be an intrinsic part of the spectrum in the OH region. These baselines commonly arise from Fe 2+ and may arise from silicate overtones in thick samples. A major, subjective source of uncertainty in IR measurements of OH in minerals remains the choice of the baseline.

Figure 4. The intensity of absorbance depends on the quality of the polarizer used for the measurement. Here, the spectrum (E ± c) of three bands in the calcite spectrum was obtained with a high efficiency polarizer (LilO,), a lower efficiency wire-grid polarizer (gold wire on AgBr), and without polarization. Modified after Figure 6 of Libowitzky and Rossman (1996).

4600

4400

4200

4000

3B00

3600

340C

1

Wavenumber {cm )

Figure 5. Infrared spectra of a clinopyroxene that show the baseline remaining after the crystal is fully dehydrated. Contributions from ferrous iron cause the rising baseline towards the long wavenumber side. From Bell et al. 1995, Figure 1.

(1) O c to -Q O CO -Q
-!mpurity replacement of an atom or ion at specific sites in the crystal. They are important, ( lb, £ £ since they are the means by which atomic migration takes place and can influence color, electrical conductivity and reactivity. In general, we define three types of point defect: vacancies, where an ion is removed Interstitial from its normal lattice site; interstitials, where an ion is present at a non-lattice site; and impurities, where dopants are present either at lattice or interstitial sites (Fig. 1). In ionic and semi-ionic crystals, these point Figure 1. Schematic representation of point defects defects will typically be charged species in a crystal of composition MX. and so must occur in balanced defect populations to maintain charge neutrality. Within pure crystals, defects made up of balanced populations of cations and anions are termed Schottky defects, while those composed of a vacancy and interstitial of the same species are known as Frenkel defects. In the strictest sense, Schottky disorder requires that charge neutrality and stoichiometry be maintained, however, the term is fairly loosely applied in the literature and we will use Schottky to refer to any charge neutral group of vacancies. The formation energy of a Schottky defect (ESch) is the sum of the individual vacancy energies (Ex) plus the lattice energy (U) of the phase removed:

• ••tA«

ESch=Evl+EV2+

Ev„+U

(1)

For a Frenkel defect, the energy is simply the sum of the corresponding vacancy and interstitial energies. Point defects occur in all crystals at temperatures above 0 K and, in pure crystals, there will be a finite population of these intrinsic defects in thermodynamic equilibrium with the system. The change in free energy (AG) associated with the introduction of a defect is expressed in the usual way as: AG = AH -TAS

(2)

AH is the enthalpy, associated with changes in nearest neighbor interactions, and AS the entropy increase due to the introduction of disorder into an otherwise perfect crystal. The entropy term includes vibrational disorder in the atoms around the defect as well as configurational terms related to way in which the defects are distributed within the crystal. The equilibrium concentration of defects in a stoichiometric material of composition MX (e.g., MgO, NaCl) can be approximated by the following: "def =

, , -&H/2RT

N e

m

(3)

where N is the number of sites, AH is the enthalpy required to form the defect, R the universal gas constant, and T temperature. The full derivation of this formula can be found in Putnis (1992). Populations of intrinsic defects are generally small; for a typical alkali halide at room temperature less than 1 site per million will be vacant (Tilley 1987). Normally, we would expect one type of point defect to dominate, and this will be the one with the lowest value of AH. Generally speaking, Schottky defects tend to dominate in close packed solids, while Frenkel defects are more common in framework and layered structure materials.

Atomistic Models of OH in

NAM's

69

In many crystalline materials, non-intrinsic vacancies and interstitials can be created in response to the presence of impurities and are thus termed extrinsic. These impurities may occur as neutral species (e.g., Mn 2+ replacing Mg 2+ ), or as charged species (e.g., Al 3+ replacing Si4+) that must be charge balanced by another impurity (coupled substitution) or by an accompanying vacancy or interstitial. Since our defects are all charged species, we might expect them to interact strongly and form discrete defect clusters with significant binding energies. When defects form, the ions around them must relax to accommodate the new configuration, and in some cases, the energy will be lower for a cluster than for the same isolated defects. At this stage it's a good idea to introduce the notation used for point defects, so that they can be easily identified and written down. The Kröger-Vink (Kröger 1972) notation is widely used to describe point defects. Vacancies (V) are defined in terms of species (subscript) and charge (superscript), where the charge may be neutral (*), positive (') or negative (')• For example, V^describes a metal (Me) vacancy with an effective 2~ charge. A Me2+ interstitial is written Me", while a neutral impurity (A) at a. Me site is given as AxMe. In the case of hydrogen, we are generally referring to an (OH)~ group replacing an O 2 - , which is written (OH)"0-

THEORETICAL BACKGROUND This section presents a brief introduction to the different computational approaches used for the study of defects. Technical details of the different theories are not included since there are many excellent texts available that cover computational methods (e.g., Foresman and Frisch 1997; Leach 2001; Gale and Röhl 2003; Griffiths 2005) and readers are referred to these for a more detailed and rigorous treatment. Computer simulation methods aim to determine the energy of a solid as a function of the interaction of all particles within that system, with varying degrees of approximation depending on the level of theory used. Simulations can be static, where the system is essentially at 0 K, or dynamic, where the free energy is calculated by molecular dynamics (MD) or lattice dynamics techniques. This can be obtained either by quantum mechanics or by atomistic techniques, based on classical mechanics, in which the details of the electronic structure are subsumed into a series of effective interactions that depend only on the nuclear positions.

Quantum mechanical methods Within quantum mechanical (QM) theory, both the electrons and nuclei are explicitly considered and their interactions are generally calculated using either Hartree-Fock (HF) or density functional theory (DFT). In both cases the Born-Oppenheimer approximation, that the motion of electrons can be separated from that of the nuclei, is assumed to hold true. However, the difficulty arises when trying to calculate interactions between electrons, since the potential experienced by one electron depends on the position of all other electrons in the system. These exchange interactions, between electrons of like spin, and correlation interactions, between electrons of opposite spin, are treated in different ways depending on the approach used. HF theory calculates the exchange energy explicitly but ignores correlation, although post-HF techniques, such as Moller-Plesset (Leininger et al. 2000) and Coupled Cluster theory, can overcome this to some extent at the price of significantly increased computational cost. DFT, while being in principle an exact theory, in practice has to approximate both the exchange and correlation potentials, using either the Local Density Approximation (LDA) or Generalized Gradient Approximation (GGA). Hybrid functionals, such as B3LYP (Becke 1993), that combine GGA or LDA with exact HF exchange, are also available. The wave function is generally described by a linear combination of basis functions that can be atom centered Gaussian type functions, or plane waves. In many cases, only the valence electrons need

70

Wright

to be explicitly considered, as it is these that are responsible for bonding. The electrostatic potential due to the frozen core electrons and the nucleus are commonly represented by a pseudopotential, and can lead to substantial computational savings, particularly when used in conjunction with plane waves. Quantum mechanical calculations, are by their very nature, the most accurate and reliable approach, although they require significant computational resources. Early studies using these techniques were limited to the use of clusters of atoms representing a solid, or very small unit cells. Recent developments in both hardware and software mean that it is now possible to calculate the properties of complex mineral phases using these methods and thus their use is increasing. DFT is by far most commonly used technique within the Earth Sciences, with particular success being enjoyed using the planewave, pseudopotential codes such as CASTEP (Segall et al. 2002) and VASP (Kresse and Furthmuller 1996a,b). DFT does, however, have its limitations; band gaps are typically underestimated by about 50%, and while LDA overestimates binding, GGA underestimates it leading to calculated cell parameters that are normally 1-2% too large. In addition, long-range van de Waals interactions are not well described, so that layered structures can prove difficult to model accurately. Classical methods Classical atomistic, or molecular mechanics (MM), simulation techniques employ interatomic potential functions to describe the total energy of the system in terms of atomic positions. These potentials include long-range electrostatic effects as well as short-range interactions produced by the overlap of nearest neighbor electron clouds. Terms to describe oxygen ion polarizibility and directionality of bonding are also available. The effective potential parameters are derived either by fitting to experimental data (structure, elastic constants, dielectric constants, etc.), or by fitting to potential energy surfaces generated by high level QM calculations. The equilibrium positions of the ions are then evaluated by minimizing the lattice energy until all forces acting on the crystal are close or equal to zero. The majority of defect calculations carried out using these methods are performed at 0 K and 0 GPa and so the energies obtained are enthalpies rather than free energies of defect formation, although free energies can be obtained by the use of lattice dynamical techniques. A comprehensive overview of interatomic potential methods can be found in Gale and Rohl (2003). Interactions between closed shell ionic species are well modeled by standard two-body potentials of the Born-Meyer type but bonded molecules, such as (OH), need to be treated differently. The hydroxyl molecule is generally described using a Morse potential of the form:

(4) where D corresponds to the dissociation energy of the bond, r0 is the equilibrium bond length and a , in combination with D, is related to the vibrational frequency of the stretching mode. Traditionally, studies of hydrogen defects in minerals use the Morse potential parameters originally derived from HF calculations on NaOH (Saul et al. 1985). These have been used extensively in the study of hydroxyl in zeolites (e.g., Schroder et al. 1992), and in a whole range of other minerals including muscovite (Collins and Catlow 1990) and goethite (Steele 2002). In the Saul et al. model, polarizibility of the hydroxyl oxygen ion is not included, and the ions have fractional charges (qO = -1.412, qW = +0.412) such that the (OH) unit has - 1 overall charge. Despite its success in calculating defect structures and energies, the Saul et al. potential does not give O-H vibrational frequencies that agree with experiment. This is a reflection on the fact that HF calculations were used in the original fitting procedure. Gatzemeier and Wright (2006) modified the a parameter by fitting it to O-H stretching frequencies in forsterite obtained using QM/MM methods (Braithwaite et al. 2003). Other,

Atomistic Models of OH in

NAM's

71

more sophisticated, Morse potentials have been developed for OH, such as that by Baram and Parker (1996). The model explicitly includes oxygen polarizibility via the use of a shell model (Dick and Overhauser 1956), where the ion is divided into a core containing all of the mass, and a shell, coupled by an harmonic spring. Other potential forms used to model water and hydroxyl groups include simple LennardJones type models as well as more sophisticated potentials, such as that of Stillinger-David (Stillinger and David 1980) that allow the dissociation of the water molecule. However, these models have not been used in the study of hydrogen in NAMs and so will not be discussed further. Treatment of defects The choice of theoretical method to use for defect calculations depends on the level of accuracy required and the actual quantity to be calculated. This, along with the limitations imposed by available computational resources, determine which level of theory to use. Classical MM methods have been extremely successful at predicting defect behavior in a whole range of solids from complex ionic materials, such as zeolites (Schroder et al. 1992), clays (Cygan et al. 2004) and carbonates (Austen et al. 2005), to battery materials (Islam et al. 2005) and semi-conductors (Wright and Gale 2004). They need minimal computer time and memory so that large numbers of possible configurations can be easily sampled. However, the quality of the results will only be as good as the potential parameters available for the system under consideration. Interatomic potential methods do have their limitations, as they are generally unable to model bond breaking and bond forming processes, although there are some exceptions to this such as the reactive empirical bond order (REBO) type models (e.g., Brenner et al. 2002). Equally importantly, they cannot be used to assess the influence of defects on those properties that explicitly depend on the electronic structure, such as JahnTeller distortions associated with transition metal ions. QM methods usually give a much more accurate description of a system, but use far more resources. There are essentially two approaches to calculating the structure and energetics of defects in solids, namely the supercell (SC) and cluster methods. The supercell approach, illustrated in Figure 2, has the defect within in a large supercell and the system is modeled in any code, QM (e.g., CASTEP, VASP) or classical (e.g., GULP, PARAPOCS), using periodic boundary conditions. The cell should be m m m fe fe m large enough that the defect does not interact with those in the periodic images as this i fe fe will introduce an additional component into fe i. fe fe ; fe fe the total energy obtained. Defect-defect interactions can be corrected for in terms of electrostatic multipoles, and, in the case of charged defects, a neutralizing background needs to be applied (Leslie and Gillan 1985). fe fe fe fe fe' fe Although this first consideration is not an issue for force field calculations, which can fe fe fe fe fe fe handle cells containing thousands of atoms, fe fe fe it can be a problem when performing QM calculations. For periodic DFT codes (Pfe fe fe' fe fe fe DFT) such as CASTEP and VASP, the CPU time required to run a calculation goes up in Figure 2. Illustration of a supercell used in defect a non-linear fashion as the number of atoms calculations. The cell defined by a dark line is increases. With the advances in parallel the unit cell which is periodically repeated in the supercell. computing, and increases in efficiency, large-

fe fe • fe fe•fe fe fe

72

Wright

scale simulations of cells containing hundreds of atoms are possible, although most calculations use much less than this. The cluster method involves cutting out a fragment of a crystal that has the defect at its centre, and embedding it in some representation of the bulk material (Fig. 3). The embedding approach is most commonly implemented in MM calculations, where the polarization caused by introducing the defect is handled using the formulation of Mott and Littleton (Mott and Littleton, 1938). In this approach the crystal is divided into two concentric spherical regions (Fig. 3). In region 1, which contains the defect at its centre, an explicit atomistic simulation is carried out to adjust Figure 3. The embedded cluster has a central area containing the defect that is embedded in a representation of the bulk the coordinates of all ions within material. the region until they are at positions at which no net forces act on them, i.e. they are relaxed around the defect. In region 2, the effects of the defect are relatively weak and the relaxation can be calculated essentially as the polarization response to the effective charge of the defect. In practice an interface region between regions 1 and 2, referred to as 2A, is normally used. The resulting defect energy is a measure of the perturbation by the defect of the static lattice energy of the crystal. As with supercells, size matters, and region 1 should be large enough that the defect energy is converged with respect to region size. The Mott-Littleton method is implemented in codes such as GULP (Gale and Rohl 2003). Although the cluster approach works well within classical calculations, its use in quantum mechanical simulations is more problematic, primarily due to edge effects and the limited number of atoms that can be included in the cluster. Hybrid, so called QM/MM embedded cluster methods (see for example Braithwaite et al. 2002) overcome these problems, by having a quantum region surrounded by a classical one. The central region contains the defect of interest and is treated at the quantum mechanical level of theory, using either HF or DFT. This QM region is normally terminated with atoms described by effective core pseudopotentials and is embedded in a large (>50 A) array of point charges which represent the potential due to the bulk crystal that acts on the embedded cluster. Between the two, is a sphere of classical atoms that are described by interatomic potentials. The embedded cluster approach overcomes many of the problems associated with studying charged defects using periodic supercell methods, including the problem of the energy term produced by defect - defect interactions.

OH DEFECTS IN MANTLE SILICATES Water can be incorporated into NAMs via a number of different mechanisms, depending on the chemistry and defect structure of the host mineral. Equation (5) describes the formation of a hydrogarnet defect, where two molecules of water interact with a silicon ion to form a silicon vacancy charge balanced by four (OH) groups, and a unit of Si0 2 : 2 H 2 0 + Si* + 40Q

[Vsi- 4 ( O H ) 0 f + Si0 2

(5)

Atomistic Models of OH in

NAM's

73

The energy of the above reaction is found by summing together the energies of the different terms, including the self-energy of the water molecule, calculated using the same level of theory. In some MM calculations, this self-energy is substituted by a proton transfer term representing the energy of the H 2 0 + O 2 - —> 2(OH)~ reaction (see Wright et al. 1994 for details). Other reaction pathways (Eqns. 6-8) involve the creation of other vacancies, or reactions with impurities and their energetics are calculated in a similar manner. In Equation (6), water is incorporated via the formation of V ^ , two (OH) groups and a unit of metal oxide. H 2 0 + Me*Me + 20*0

[VMe 2(OH) 0 f + MeO

(6)

Similar reactions can occur for metal cations of different charges, with corresponding numbers of (OH) and different oxide products. Reactions, such as those in Equations (7) and (8), involve interactions of water with impurity cations and vacancies on the oxygen sub-lattice. Within clinopyroxenes, Al^ and Na^, substitutions can be charge balanced by the inclusion of (OH) as: H 2 0 + 2Al^ + 20Q + V " ^ 2 [ A l s i - ( O H ) 0 f + 0*0 H 2 0 + 2 N a ^ + V " + OS

2[Na Mc ( O H ) 0 f

(7) (8)

The identity of the charge compensating defects in NAMs had been the subject of considerable debate in the literature, as has the extent to which these defects bind with the hydrogen. This is the sort of problem that can readily be addressed by computational methods, as the relative stabilities of known defect configurations can be assessed by calculating their formation energies, both bound and unbound. In addition, the O-H stretching frequency can be determined for each configuration and compared with experiment. In this way, the results obtained from the calculations can be used to both constrain models, and to aid in the interpretation of experimental data. In the following sections we consider the literature on hydrogen defects in the major mineral phases of the Earth's upper mantle and transition zone; i.e., the Mg 2 Si0 4 polymorphs and the clinopyroxenes diopside and jadeite. Hydrogen defects in a number of other important nominally anhydrous minerals, have been studied using computational methods including garnets (Wright et al. 1994; Nobes et al. 2000), quartz (Lin et al. 1994; Purton et al. 1992) and feldspar (Wright et al. 1996) although these will not be covered here. The Mg 2 Si0 4 polymorphs Forsterite. There is considerable experimental evidence for the presence of hydrogen in all three of the Mg 2 Si0 4 polymorphs, as discussed in various chapters in this volume. Of the three, forsterite is by far the most well studied both experimentally (Kohlstedt et al. 1996; Matveev et al. 2001; Demouchy and Mackwell 2003) and computationally (see Table 1 for references). Olivine [(Fe,Mg) 2 Si0 4 ] is the dominant mineral in the Earth's upper mantle and thus will exert a major control on the rheological behavior. Natural samples show levels of hydrogen in the range 1 to 140 ppm H 2 0, where hydrogen is expected to be incorporated into the olivine lattice in association with both silicon and magnesium vacancies (e.g., Kohlstedt et al. 1996; Kohn 1996). Concentrations of OH in natural samples appear to show some correlation with the geological setting and composition suggesting that PIT history, as well as local stoichiometry, can affect the uptake of hydrogen. There is some evidence (Bell and Rossman 1992) suggesting that iron rich olivines contain a greater proportion of hydrogen than those with low iron content; however, this relationship has not been quantified in any way. Calculations, based on both QM and classical methods (Table 1), have been used to investigate the relative stability of hydrogen defects at different positions in the forsterite

74

Wright Table 1. Summary of calculations carried out on the M g 2 S i 0 4 polymorphs. All calculations with the exception of those marked with *, were carried out at 0 K and 0 GPa. Mineral

Method

Reference

Forsterite

P-DFT P-DFT QM/MM MM MM MM

Haiber et al. (1997) Brodholt and Refson (2000) Braithwaite et al. (2003) Wright and Catlow (1994) de Leeuw et al. (2000) Walker et al. (2006)

Wadsleyite

P-DFT MM MM MM

Haiber et al. (1997)* Wright and Catlow (1996) Parker et al. (2004)* Walker et al. (2006)

Ringwoodite

MM P-DFT

Blanchard et al. (2005) Haiber et al. (1997)

lattice and to calculate energies of the reactions in Equations (5) and (6). Forsterite has an orthorhombic unit cell with isolated Si0 4 tetrahedra separated by magnesium ions octahedrally coordinated by oxygens. There are two symmetry inequivalent magnesium positions, and three different oxygen sites, as shown in Figure 4. Looking at the structure of forsterite, we can identify a number of possible environments for hydrogen: (i) interstitial hydrogen bound to any one of the three oxygen sites but isolated from any cation vacancies; (ii) hydrogen bound to oxygen adjacent to either Ml or Ml vacancies; and (iii) hydrogen bound to oxygen adjacent to silicon vacancies. Of the three oxygen sites, all calculations agree that 0 3 is the most easily protonated, and that the Ml vacancy has a lower formation energy that M2. The most favorable defect configuration involving magnesium vacancies has hydrogen bound to two 02 oxygens around the vacant Mgl site, as shown in Figure 5 a. The third possibility, of hydrogen surrounding a silicon vacancy is shown in Figure 5b. Calculated binding energies (Brodholt and Refson 2000; Braithwaite et al. 2003; Walker et al. 2006) for [VMg-2(OH)0]j: and [Vsi-4(OH)0]j: and associated clusters, given in Table 2, are substantial, and hence there is a strong driving force for hydroxyl groups to combine with cation vacancies. Haiber et al. (1997) have suggested that under mantle conditions entropy would cancel out any vacancy-hydrogen binding and therefore only isolated interstitial hydrogen defects would be present. However, the magnitude of the binding energies is sufficient to overcome the activation energy for hydrogen diffusion in olivine, estimated as 130 kJ-mol-1 (Mackwell and Kohlstedt 1990), so that cation vacancies will act as a 'sink' for unassociated hydroxyl species. Brodholt and Refson (2000) suggest that reactions with water could actively promote the formation of metal vacancies, particularly silicon vacancies, leading to a form of hydrolytic weakening. Calculated energies for the dissolution reactions in Equations (5) and (6) are given in Table 3 and show considerable variation depending on the methodology used. The P-DFT and QM/MM calculations compare well, with both methods showing that reactions of water with silicon vacancies will be exothermic. The error on the QM/MM values comes from uncertainties in the value for the lattice energy of oxide phases produced on formation of the defect, which by necessity had to be calculated using a periodic QM code. Calculated values from MM calculations are much higher in energy than either of the QM values, although reaction with

Atomistic Models of OH in

NAM's

75

Figure 4. The unit cell of forsterite (Mg2Si04) viewed along the [100] direction.

(b)

Figure 5. Structure of hydrogen defect complexes in forsterite produced from the data of Braithwaite et al. (2003). (a) [VMg-2(OH)0]v cluster, and (b) hydrogarnet defect [Vsi-4(OH)0]V cluster.

Table 2. Defect binding energies in forsterite calculated using periodic DFT (Brodholt and Refson 2000) and QM/MM DFT (Braithwaite et al. 2003). Binding energy (kj-mol 1) Reaction

P-DFT

QM/MM DFT

1

-239

-245

2

-157

-203

3

-546

-519

-155

-202

4

(3
a =ß*

Augite

a = ß,y = 0

y>a = ß

y>a = ß

Omphacite

a = ß, y = 0 *

y>a = ß

y>a = ß

— —

3600 - 3610

3410 - 3560

3060 - 3300

Orthopyroxene

a = ß, y = 0 *

y>a ~ß

y>a>ß

* Band not always observed.

close to 3630 cm - 1 polarized in the a and P directions, and two bands close to 3530 and 3460 with the pleochroism y > a = P (Fig. 3). Compared to diopside, the 3355 cm - 1 band is absent, and the bands are normally somewhat broader and slightly shifted to lower wavenumbers. Omphacite IR spectra of omphacite are similar to those of augite, with one a - and P-polarized band around 3630 cm - 1 and two bands around 3530 and 3460 cm - 1 with their strongest absorption in the y direction (Fig. 4). However, omphacite spectra differ from augite spectra in that the 3460 cm - 1 band in omphacite is normally much stronger than the other bands, and that the 3630 cm - 1 bands is sometimes absent (Koch-Miiller et al. 2004; Katayama et al. 2005).

< D O cto -Q O CO .Q
a~ P occur at 3560, 3510 and 3410 c n r 1 (Fig. 5). A third group occur at the relatively low wavenumbers 3300, 3210 and 3060 cnr 1 , with the pleochroic scheme y > a > P (e.g., Bell et al. 1995; Peslier et al. 2002).

C.Ì c tc -9 o!/)
3400 cm -1 ) are caused by OH associated with Al.

WATER CONCENTRATION IN MANTLE PYROXENES Ortho-and clinopyroxenes have been shown to carry the largest amounts of water among the major upper mantle minerals, and it is evident that pyroxenes play important roles both in providing a repository for water in the upper mantle and in mantle water recycling processes. A fairly large number of studies have addressed the specific amounts of OH in pyroxenes from different mantle occurrences, including kimberlites (e.g., Bell et al. 1992, 2004), peridotites of different types (e.g., Skogby et al. 1990; Peslier et al. 2002) as well as eclogites from both mantle and crustal environments (Smyth et al. 1991; Katayama and Nakashima 2003; Koch-Miiller at al. 2004; Katayama et al. 2005). Most studies have relied on IR spectroscopy for quantification of water concentrations. The calculation procedures involved to translate spectral absorption parameters to absolute water concentration data require calibration by independent water analysis methods, which have continuously been improved over time. An overview of suitable hydrogen analysis methods and their application in calibrations of OH absorbances in IR spectra is given in Rossman (2006). As also the IR measurements can be performed in different ways, for instance concerning polarization, background subtraction, and use of linear or integrated intensities, published concentration data are not always directly

Water in Natural

Mantle

Minerals

I:

163

Pyroxenes

comparable. A summary of published data on contents in mantle pyroxenes are given in Table 2, including information on analytical procedures. The OH concentrations vary as a function of mineral species and composition, and in some cases also in relation to the type of mantle environment. However, strong variations are also observed for samples within each mineral group from similar environments. Calcic clinopyroxenes (diopside and augite) from peridotites vary in concentration from 140 to 740 ppm H 2 0 , whereas orthopyroxene show lower concentrations in the range 40 to 530 ppm H 2 0 . In general, orthopyroxenes hold about half the amount of OH being present in co-existing clinopyroxenes. The calcic clinopyroxenes do not appear to show significant correlation with type of the environment in general. However, Demouchy (2004) observed an increasing trend in water concentrations in diopside when going from spinel lherzolite via garnet/spinel lherzolite to garnet lherzolite. Similarly, Bell et al. (1992) observed the highest OH concentration for orthopyroxene in coarse-grained samples from garnet peridotites, but noted that these samples may have been affected by metasomatic reactions in the mantle. The highest levels of water concentrations among pyroxenes have been recorded for omphacite and sodic clinopyroxene (Smyth et al. 1991; Katayama and Nakashima 2003; Koch-Miiller at al. 2004; Katayama et al. 2005). Reported concentrations (Table 2) range Table 2. Summary of observed OH concentrations in mantle pyroxenes. Geological occurrence

Mineral

#of samples

OH conc. (wt-ppm H 2 0 )

Analysis method*

Ref.

kimberlite and alkali basalt xenoliths, mantle eclogite

calcic cpx omphacite opx

5 1 3

200 - 530 640 60- 260

1

[1]

mantle eclogite

omphacite

11

130 - 970

1

[2]

kimberlite and alkali basalt xenoliths, mantle eclogite

calcic cpx omphacite opx

7 2 10

150 -590 470- 1080 50- 460

2

[3]

basalt xenoliths

cpx opx

3 3

388 -492 174 -212

2

[4]

spinel peridotite xenoliths

cpx opx

15 16

140 -528 39--265

3

[5]

crustal eclogite

omphacite

6

230 - 870

4

[6][7]

kimberlite megacrysts

cpx opx

9 3

195 -620 215--263

2

[8]

peridotite xenoliths

diopside opx

4 5

150 -420 70- 310

5

[9]

mantle and crustal eclogite

cpx

8

31 --514 61 - 872

6 2

[10]

* Analysis based on: 1) IR, based on a linear molar absorption coefficient e 0 H = 150 L/(molcm), Skogby et al. (1990); 2) IR, calibrations of Bell et al. 1995, with integral molar absorption coefficients I cpx = 38300, I o p x = 80600 L/(mol cm 2 ); 3) IR, calibrations of Bell et al. 1995, using "specific" integral absorption coefficients I' c p x = 7.09, r o p x = 14.84 p p m - 1 cm - 2 ); 4) SIMS data, unpolarized IR data using calibration of Bell et al. 1995 (see above) indicate that the OH concentrations are twice as high as those listed here; 5) Unpolarized IR data, based on the calibration of Paterson (1982), with the general integral molar absorption coefficient defined as I H = 150 (3780 - v); 6) IR, calibration of Libowitzky and Rossman (1997), with the general integral molar absorption coefficient defined as I H2 o = 246.6 (3753 - v). References: [1] Skogby et al. (1990); [2] Smyth et al. (1991); [3] Bell et al. (1992); [4] Ingrin and Beran (2001); [5] Peslier et al. (2002); [6] Katayama and Nakashima (2003); [7] Katayama et al. (2005); [8] Bell et al. (2004); [9] Demouchy (2004); [10] Koch-Müller et al. (2004)

164

Skogby

from 30 to 1080 ppm H20, although it 1200 should be noted that a mineral specific Clinopyroxenes IR calibration for omphacite is lacking, o M 1000 and the reported concentrations based I™ on IR data should be viewed with some Q. 800 caution. This is exemplified by recent Q. SIMS data on omphacite from a crustal 600 eclogite occurrence (Katayama et al. 2005) oc that indicated that the concentration based o o 400 on IR absorbances was overestimated by is a factor of two. Additional uncertainties 15 200 regarding the high water concentrations observed in omphacite are the observations o of nano-inclusions of hydrous phases 1.0 2.0 3.0 4.0 5.0 6.0 7.0 related to high-wavenumber bands obPressure (GPa) served by Koch-Müller et al. (2004). A Figure 14. Pressure dependence of water contents in strong dependence of water concentration omphacite in eclogites from the Kokchetav massif, and crystallization depth was observed by Kazachstan. Data from Katayama et al. (2005). Katayama et al. 2005 (Fig. 14), who related the pressure-dependence to increased stability of the Ca-Eskola component. However, Koch-Müller et al. (2004) observed a reversed situation with unusually low concentrations (31 ppm) in diamond-bearing eclogite xenoliths and the highest concentrations (437-514 ppm) in lower-pressure grospydites and granulites. They interpreted the unusually low water concentrations for the samples from the high-pressure occurrence as being caused by low water activity during crystallization, or alternatively by hydrogen loss during uplift.

IMPLICATIONS FOR WATER IN THE UPPER MANTLE A fundamental question regarding OH contents in mantle pyroxenes, as well as in other mantle-derived NAMs, is whether the water concentrations recorded in xenolith samples are representative for the conditions in the upper mantle, or if the original hydrogen contents have been reset during different types of ascent processes. The substantial amounts of kinetic data for hydrogen diffusion in pyroxenes that now are available (e.g., Hercule and Ingrin 1999; Carpenter Woods et al. 2000; Stalder and Skogby 2003; Ingrin and Blanchard 2006) indicate that major resetting is indeed possible, also during relatively fast ascent processes (Ingrin and Skogby 2000). On the other hand, several lines of evidence indicate that OH concentrations representative for mantle conditions to large extents are preserved in pyroxenes and other mantle NAMs. A major argument for preservation of mantle OH concentrations comes from the observations of systematic correlations between sample chemistry and OH content. In a study of OH concentration of pyroxenes from spinel peridotites from the sub-arc mantle wedge, Peslier et al. (2002) found that the OH contents were correlated with sample composition, but also with the chemistry of associated spinels and whole-rock xenolith data. These parameters can be expected to be completely independent of xenolith transport processes, which indicate that the OH contents were neither reset to large extents. An important parameter for OH incorporation in peridotite pyroxenes appears to be the redox conditions. This was demonstrated in the study by Peslier et al. (2002) who observed a clear negative correlation between OH concentration and oxygen fugacities estimated from spinel compositions in a series of samples collected from different regions of the sub-arc

Water in Natural Mantle Minerals I: Pyroxenes

165

mantle wedge, which is known to be more oxidized than other parts of the upper mantle. The low OH concentrations observed in the more oxidized environments were interpreted to be caused by redox dehydration reactions associated with metasomatism and partial melting, which led to a loss of more than half of the initial pyroxene water contents. Estimates of the water budget in subduction zones indicated, however, that the water released from NAMs in the sub-arc mantle wedge only account for a minor proportion of ca 5% of the total water in this environment. Negative correlations between oxygen fugacity and OH in pyroxene have also been observed in experimental studies (e.g., Skogby 1994). Support for preservation of mantle OH concentrations in pyroxenes is put forward by the study of Bell et al. (2004) on different NAMs from a suite of megacrysts from the Monastery kimberlite. They observed that the OH contents in both clino- and orthopyroxenes followed trends with other elements in Fe-enrichment and Ca content, reflecting igneous differentiation. In accordance with the study of Peslier et al. (2002), they concluded that such systematic behavior would unlikely be observed if OH contents were fully reset at crustal environments at later stages of ascent processes. Furthermore, they noted that the relatively high OH concentrations recorded for the Monastery olivines (54-262 ppm H 2 0 ) according to experimental studies require equilibration pressures corresponding to mantle conditions, and are not compatible to resetting in crustal environments. Even if arguments exist for preservation of original mantle water contents in pyroxenes found in xenoliths, the available kinetic data regarding dehydrogenation reactions (cf. Ingrin and Blanchard 2006) indicate that hydrogen loss may be significant. The fastest reaction in this respect involves concomitant oxidation of Fe 2+ according to the redox reaction: Fe 2+ + OH" = Fe 3+ + O 2 " + Yi H 2 This reaction has been shown to be considerable faster than dehydrogenation reactions involving more rigorous structural changes (e.g., cation diffusion, resetting of defect chemistry). Significant dehydrogenation following this reaction will lead to enhanced Fe 3+ / Fe 2+ ratios, similar to what have been observed for some mantle-derived amphiboles (e.g., Dyar et al. 1992). The ferric iron contents of mantle pyroxenes can hence be used to estimate the maximum amounts of hydrogen that may have been lost via the redox reaction. By adopting this approach, Ingrin and Skogby (2000) found that the ferric iron levels observed in mantle pyroxenes correspond to a maximum loss of 900 ppm H 2 0 for clinopyroxene and 570 ppm H 2 0 for orthopyroxene. Apart from providing a host phase for mineral-bound water in the upper mantle, pyroxenes appear to play an important role for water recycling related to subduction zones. The high OH concentrations recorded in omphacitic pyroxenes from both crustal and mantle occurrences indicate that they provide an efficient means for water transport down to the deeper levels of the upper mantle, beyond the stability fields of the hydrous minerals. In a recent study, Katayama et al. (2005) showed that the water contents of omphacite from the Kokchetav massif, which has been subducted to a depth of 180 km, contain up to 870 ppm H 2 0 and that water contents systematically increase with pressure (Fig. 14). They suggested that omphacite, together with garnet and rutile, in subducted crust may carry water down towards the transition zone at depths around 400 km, where the eclogite minerals are expected to transform to majoritic garnet and stishovite.

ACKNOWLEDGMENTS E. Libowitzky, A. Beran and H. Keppler are thanked for providing constructive reviews of this manuscript.

166

Skogby REFERENCES

Andrut M, Brandstätter F, Beran A (2003) Trace hydrogen zoning in diopside. Mineral Petrol 78:231-241 Bell DR, Ihinger PD, Rossman GR (1995) Quantitative analysis of trace OH in garnet and pyroxenes. Am Mineral 80:465-474 Bell DR, Rossman GR (1992) Water in Earth's mantle: The role of nominally anhydrous minerals. Science 255:1391-1397 Bell DR, Rossman GR, Moore RO (2004) Abundance and partitioning of OH in a high-pressure magmatic system: Megacrysts from the Monastery kimberlite, South Africa. J Petrol 11:1539-1564 Beran A (1976) Messung des Ultrarot-Pleochroismus von Mineralen. XIV. Der Pleochroismus der OHStreckfrequenz in Diopsid. Tscherm Min Petr Mitt 23:79-85 Beran A (1981) Infrarotspektroskopische Untersuchungen über den OH-Gehalt von Pyroxenen und Cyaniten aus Gesteinen. Forts Mineral 59:16-18 Beran A, Zemann J (1986) The pleochroism of a gem-quality enstatite in the region of the OH stretching frequency, with a stereochemical interpretation. Tscherm Min Petr Mitt 35:19-25 Bolfan-Casanova N (2005) Water in the Earth's mantle. Mineral Mag 69:229-257 Bromiley GD, Keppler H (2004) An experimental investigation of hydroxyl solubility in jadeite and Na-rich clinopyroxenes. Contrib Mineral Petrol 147:189-200 Carpenter Woods S, Mackwell S, Dyar MD (2000) Hydrogen in diopside: Diffusion profiles. Am Mineral 85: 480-487 Demouchy S (2004) Water in the Earth's interior: Thermodynamics and kinetics of hydrogen incorporation in olivine and wadsleyite. PhD Dissertation, University of Bayreuth, Germany Dyar MD, McGuire AV, Mackwell S J (1992) Fe3+/H+ and D/H in kaersutites - Misleading indicators of mantle source fugacities. Geology 20:565-568 GasparikT (1986) Experimental study of subsolidus phase relations and mixing properties of clinopyroxene in the silica-saturated system Ca0-Mg0-Al 2 0 3 -Si0 2 . Am Mineral 71:686-693 Hercule S, Ingrin J (1999) Hydrogen in diopside: diffusion, extraction-incorporation, and solubility. Am Mineral 84:1577-1588 Ingrin J, Beran A (2001) Hydrogen content in spinel lherzolite xenoliths coming from different depths (Kilbourne Hole, NM). Terra abstract, X, 458 Ingrin J, Blanchard M (2006) Diffusion of hydrogen in minerals. Rev Mineral Geochem 62:291-320 Ingrin J, Latrous K, Doukhan JC, Doukhan N (1989) Water in diopside: an electron microscopy and infrared spectroscopy study. Eur J Mineral 1:327-341 Ingrin J, Skogby H (2000) Hydrogen in nominally anhydrous upper-mantle minerals: concentrations levels and implications. Eur J Mineral 12:543-570 Johnson EA (2006) Water in nominally anhydrous crustal minerals: speciation, concentration, and geologic significance. Rev Mineral Geochem 62:117-154 Johnson EA, Rossman GR, Dyar MD, and Valley JW (2002) Correlation between OH concentration and oxygen isotope diffusion rate in diopsides from the Adirondack Mountains, New York. Am Mineral 87: 899-908 Katayama I, Nakashima S (2003) Hydroxyl incorporation from deep subducted crust: Evidence for H 2 0 transport into the mantle. Am Mineral 88:229-234 Katayama I, Nakashima S, Yurimoto H (2005) Water content in natural eclogite and implication for water transport into the deep upper mantle. Lithos 86:245-259 Koch-Müller M, Matsyuk SS, Wirth R (2004) Hydroxyl in omphacites and omphacitic clinopyroxenes of upper mantle to lower crustal origin beneath the Siberian platform. Am Mineral 89:921-931 Libowitzky E, Beran A (2006) The structure of hydrous species in nominally anhydrous minerals : information from polarized IR spectroscopy. Rev Mineral Geochem 62:29-52 Libowitzky E, Rossman GR (1997) An IR absorption calibration for water in minerals. Am Mineral 82:11111115 Martin RF, Donnay G (1972) Hydroxyl in the mantle. Am Mineral 57:554-570 McCormick T (1986) Crystal-chemical aspects of nonstoichiometric pyroxenes. Am Mineral 71:1434-1440 Paterson MS (1982) The determination of hydroxyl by infrared absorption in quartz, silicate glasses and similar materials. Bull Minéral 105:20-29 Peslier AH, Luhr JF, Post J (2002) Low water contents in pyroxenes from spinel-peridotites of the oxidized, sub-arc mantle wedge. Earth Plan Sc Lett 201:69-86 Rauch M, Keppler H (2002) Water solubility in orthopyroxene. Contrib Mineral Petrol 143:525-536 Rossman GR (1996) Studies of OH in nominally anhydrous minerals. Phys Chem Minerals 23: 299-304 Rossman GR (2006) Analytical methods for measuring water in nominally anhydrous minerals. Rev Mineral Geochem 62:1-28 Runciman WA, Sengupta D, Marshall M (1973) The polarized spectra of iron in silicates. I. Enstatite. Am Mineral 58:444-450

Water in Natural Mantle Minerals I: Pyroxenes

167

Skogby H (1994) OH incorporation in synthetic clinopyroxene. Am Mineral 79:240-249 Skogby H, Rossman GR (1989) OH~ in pyroxene: An experimental study of incorporation mechanisms and stability. Am Mineral 74:1059-1069 Skogby H, Rossman GR (1991) The intensity of amphibole OH bands in the infrared absorption spectrum. PhysChemMin 18:64-68 Skogby H, Bell DR, Rossman GR (1990) Hydroxide in pyroxene: Variations in the natural environment. Am Mineral 75:764-774 Stalder R (2004) Influence of Fe, Cr and A1 on hydrogen incorporation in orthopyroxene. Eur J Mineral 16: 703-711 Stalder R, Skogby H (2003) Hydrogen diffusion in natural and synthetic orthopyroxene. Phys Chem Minerals 30:12-19 Smyth JR, Bell DR, Rossman GR (1991) Incorporation of hydroxyl in upper-mantle clinopyroxenes. Nature 351:732-735 Veblen DR, Buseck PR (1981) Hydrous pyriboles and sheet silicates in pyroxenes and uralites: Intergrown microstructures and reaction mechanisms. Am Mineral 66:1107-1134 Wilkins RW, Sabine W (1973) Water content of some nominally anhydrous silicates. Am Mineral 58:508-516

8

Reviews in Mineralogy & Geochemistry Vol. 62, pp. 169-191,2006 Copyright © Mineralogical Society of America

Water in Natural Mantle Minerals II: Olivine, Garnet and Accessory Minerals Anton Beran and Eugen Libowitzky Institut fiir Mineralogie und Kristallographie Universität Wien - Geozentrum Althanstraße 14, A-1090 Wien, Austria anton. [email protected]

engen, libowitzky @ univie. ac. at

INTRODUCTION Hydrogen traces change the physical properties of mantle minerals to an extent that is far out of proportion to its low concentration. These properties include mechanical strength, melting behavior, diffusion rate, electrical conductivity, viscosity and rheology. Besides minerals of the pyroxene group (as discussed by Skogby 2006, this volume), the close-packed mineral structures of olivine, garnet and some accessory minerals offer important storage sites for hydrogen traces in the Earth's mantle. However, the water content stored in olivine and mantle garnet is quite low compared to that in pyroxenes. Based on chemical considerations, but also on information obtained from infrared (IR) data, Martin and Donnay (1972) proposed the existence of hydroxyl in nominally anhydrous minerals (NAMS) occurring in the upper mantle, especially in pyroxene and olivine. The review articles of Bell and Rossman (1992a), Skogby (1999), and Ingrin and Skogby (2000) reveal a wide range of water contents for mantle-derived pyroxenes, olivines and garnets, which are derived from IR spectroscopic data. Fourier transform infrared (FTIR) spectroscopy provides an extremely sensitive method for detecting trace hydrogen bonded to oxygen in the structures of various NAMS (Beran 1999; Beran and Libowitzky 2003; Libowitzky and Beran 2004). As this method is not self-calibrating, attempts have been made to calibrate the IR spectra with independent absolute methods. These methods include hydrogen manometry and measurement of thermally released water in an electrolytic cell or by Karl Fischer titration. ! H Magic-Angle-Spinning Nuclear Magnetic Resonance (MAS NMR), Secondary Ion Mass Spectrometry (SIMS) and Nuclear Reaction Analysis (NRA) are encouraging but experimentally demanding and expensive methods (as discussed by Rossman 2006, this volume). To overcome these problems, approximations such as that proposed by Paterson (1982) and refined by Libowitzky and Rossman (1997), attempt to provide a way to deal with the general dependence of the molar absorption coefficient on OH band positions in a more accurate way. The basis of the quantitative determination of the water content is the Beer-Lambert's law. IR absorbances A (A = log IQ/I) are directly related by the molar absorption coefficient e to the concentration c of OH groups and to the thickness t of the sample: A = e • c • t. In optically anisotropic crystals normally the sum of absorbances measured with polarized radiation in principal axis directions of the optical indicatrix in oriented crystal sections results in accurate values which can be used for the determination of the water content (Libowitzky and Rossman 1996). In optically isotropic (cubic) crystals, e.g., garnets, the absorbance values from (un)polarized spectra must be multiplied by 3 to account for all three spatial directions (compare with Paterson's 1982 orientation factor y = 1/3). Concentration values may be best obtained from integral absorbances A; (in cm-1). The integral absorption coefficient cti = A Jt (in cm -2 ; t measured in cm) is then expressed by c^ = s^c, where £; is the integral molar absorption coefficient in L-moH-cm-2. When £; is determined for a specific structural matrix, e.g., 1529-6466/06/0062-0008505.00

DOI: 10.2138/rmg.2006.62.8

170

Beran & Libowiizky

for olivine or garnet of specific composition, then for all other olivines or garnets of the same composition and same type of OH defect the water content can be quantitatively determined by the relation c (in wt% H 2 0) = (cti • 1.8)/(£j • D), where D is the density of the mineral in g-cm~3 (Beran et al. 1993; Libowitzky and Beran 2004). Asimow et al. (2006) reported a method to derive accurate water contents from polarized measurements of randomly oriented grains. OLIVINE Basic structure and possible sites of hydrogen incorporation The structure of olivine is best described as an approximately hexagonal close-packing of oxygen atoms with one half of the distorted octahedral interstices occupied by (Mg,Fe) atoms and one eighth of the tetrahedral interstices occupied by Si. One formula unit (Mg,Fe) 2 Si0 4 contains two crystallographically different (Mg,Fe) sites, i.e., Ml on a center of symmetry, M2 in a mirror plane. Si is also placed in that mirror plane. Two of the three different oxygen positions, Ol and 02, are localized in a mirror plane, while the third, 03, occupies a general position. All oxygen atoms are coordinated by three (Mg,Fe) and one Si atom in a distorted tetrahedron. Thus, any possible structural OH defect is part of a full/vacant coordination tetrahedron and octahedron. A first model for OH positions in olivine based on polarized IR spectroscopic measurements was proposed by Beran and Putnis (1983) for gem-quality crystals of hydrothermal origin from Zabargad, Egypt. This olivine is characterized by pleochroic absorption bands at ~3590, 3570, 3520, and 3230 cm -1 . It was suggested that [0(0H) 3 ] and [0 2 (0H) 2 ] tetrahedra with a specific combination of hydrogen positions occur as structural elements, assuming that vacancies are on Si sites. If M2 site vacancies were assumed, [Si0 3 (0H)] and [Si0 2 (0H) 2 ] tetrahedra occur as structural elements. Libowitzky and Beran (1995) presented a polarized IR study of a colorless near-endmember forsterite, revealing OH stretching bands (Fig. 1) predominantly in the high-energy wavenumber

Wavenumber (cm"1) Figure 1. Polarized IR absorption spectra of an oriented forsterite crystal from a skarn deposit in Pamir, Tadzikistan, representing a crustal occurrence (Libowitzky and Beran 1995). The sharp, strongly pleochroic OH absorption bands are restricted to the high-energy region of band group I, comprising the range 36503450 crrr 1 . Band group II covers the 3450-3200 crrr 1 region (according to Bai and Kohlstedt 1993).

Water in Mantle Minerals:

Olivine, Garnet, & Accessory Minerals

region (band group I according to Bai and Kohlstedt 1993). Sharp, strongly pleochroic band doublets centered at 3674/3624, 3647/3598, 3640/3592 cm -1 are assigned to OH dipoles oriented parallel to [100]. An OH band doublet at 3570/3535 cm -1 shows both, a strong absorption parallel to [100] and a strong component parallel to [001]. Under the assumption of vacancies at Si and (Mg,Fe) sites, the Ol site represents the most favorable position for OH defects pointing to a vacant Si site. 0 3 is proposed as donor oxygen of OH dipoles lying near the 03-01 tetrahedral edge or roughly pointing to a vacant Ml site. In this model also 0 2 can act as donor oxygen of an OH group oriented along the 02-03 edge of a vacant Ml octahedron (Fig. 2).

171

[M2]

03

a Figure 2. Schematic diagram of a part of the olivine structure with a Si vacancy showing possible OH orientations derived from the pleochroic behavior of the Pamir forsterite spectra presented in Figure 1 (modified after Libowitzky and Beran 1995).

In a recent polarized FTIR spectroscopic study of synthetic pure forsterite by Lemaire et al. (2004) the proposed OH incorporation model assuming vacant Si, Ml, and Ml sites is essentially confirmed. OH bands at 3613, 3580, 3566, 3555, and 3480 cm -1 are assigned to OH groups compensating Si vacancies. Bands at 3600, 3220, and 3160 cm -1 are enhanced in samples with higher silica activity, suggesting that these bands are related to Ml (3160 cm -1 ) and Ml vacancies (3600 and 3220 cm-1). From the crystal chemical approach it is interesting to note that the predominant alignment of OH groups parallel to [100] was also observed by IR spectroscopic studies of Bauerhansl and Beran (1997) in the olivine-type mineral chrysoberyl, Al 2 Be0 4 . An intense discussion of the cation vacancy type, i.e., D Mg vs. n s i , related to hydrogen solubility was initiated on the basis of experimental results by Bai and Kohlstedt (1992, 1993). The authors carried out annealing experiments on olivine crystals from San Carlos, Arizona, with samples left unbuffered, samples buffered with orthopyroxene, and samples buffered with magnesiowustite. IR spectra from the annealed samples revealed two distinct groups of OH bands, group I bands occurred in the 3650-3450 cm -1 region, group II bands in the 34503200 cm -1 range. The origin of these band groups—Si vacancy and/or Ml, Ml vacancies related—and their relation to P, T, silica activity, and oxygen fugacity, is still a matter of debate (Kohlstedt et al. 1996; Matveev et al. 2001, 2005; Lemaire et al. 2004; Zhao et al. 2004; Berry et al. 2005; Mosenfelder et al. 2006, and contributions in this volume, e.g., Keppler and BolfanCasanova 2006). Defect types in mantle-related olivines from different localities Based on TEM observations and IR spectroscopic investigations, Kitamura et al. (1987) described planar OH bearing defects in olivine from a kimberlite in Buell Park, Arizona. The structure of the defects resembles that of an OH bearing monolayer within the olivine structure as it exists in humite group minerals. Bands present at 3571, 3524, 3402, and possibly 3319 cm -1 can be assigned to titanian clinohumite. Polarized IR spectra of an olivine crystal from this locality, which contains about 50 wt ppm H 2 0, have been reported by Mosenfelder et al. (2006). The strongest bands are at 3613, 3598, 3579, and 3567 cm -1 . The spectra of this olivine sample differ from that reported by Kitamura et al. (1987) by lack of bands due to planar titanian clinohumite defects. Two modes of hydrogen incorporation in mantle olivine from Yakutia, Siberia, were suggested by Khisina et al. (2001): Intrinsic hydrogen in form of ordered OH bearing point defects in "hydrous olivine" and extrinsic hydrogen contained in exsolutions of hydrous minerals in form of "large" inclusions. Bands observed in the olivine

Beran & Libowiizky

172

spectra at 3704 and 3683 cm - 1 can be referred to serpentine and a band at 3677 c m - 1 to talc. The bands observed at 3591 and 3660 cm - 1 match those of the "10 A-phase." Lamellar and hexagon-like inclusions of several ten nm in size of Mg-vacant hydrous olivine have been described in a combined FTIR/TEM study by Khisina and Wirth (2002). Typical OH band positions of mantle-derived olivines, including the band assignment for possible planar OH bearing defects are summarized in Table 1. Considering especially the Ti-clinohumite defects, synthetic olivines, crystallized experimentally under upper mantle conditions, that reproduce the common and intense OH bands at 3572 and 3525 cm - 1 were reported by Berry et al. (2005). According to these authors the bands arise from OH point defects associated with traces of Ti. It should also be noted that bands at 3355 and 3325 cm - 1 are assigned to Fe 3+ related OH defects. A polarized IR study of naturally occurring olivines from 17 different localities by Miller Table 1. OH band positions (in cm ') and band assignments for possible planar OH-bearing defects for mantle olivines from different localities and occurrences. Locality 1

2

3675 3637

3610 3598 3571

3524

3623 3615 3602 3592 3576 3565 3542 3527

3 3683 3676 3660 3637 3630 3624 3613 3597 3591 3572 3540 3526

3499 3481 3455

3500 3482 3458

3413 3400 3375

3414 3401 3374 3355 3330

4 3704 3688 3677 3660 3640 3623 3599 3591 3572 3562 3540 3526

Band

5

3639 3630 3624 3612 3597 3591 3572 3567 3541 3525 3512

6

7

assignment serpentine serpentine talc 10 A-phase serpentine

3599 3572 3562 3545 3525

3481 3458

3573

10 A-phase Ti-clinohumite

3525

Ti-clinohumite

3486 3451

3402

Ti-clinohumite 3369 3331

3354 3331 Ti-clinohumite

3319 3230

3298 3225

Wavenumber values with deviations of ± 2.5 cm - 1 are listed within one line. 1 - kimberlite xenolith, Buell Park, Arizona (Kitamura et al. 1987) 2 - kimberlite xenolith, Monastery, South Africa (Miller et al. 1987) 3 - kimberlite xenolith, Obnazennaya, Yakutia (Kishina et al. 2001) 4 - kimberlite xenolith, Udachnaya, Yakutia (Kishina et al. 2001) 5 - kimberlite xenolith, Udachnaya, Yakutia (Koch-Miiller et al. 2006) 6 - spinel peridotite, Ichinomegata, Japan (Kurosawa et al. 1997) 7 - garnet peridotite, Wesselton, South Africa (Kurosawa et al. 1997)

Water in Mantle

Minerals:

Olivine,

Garnet, & Accessory

Minerals

173

et al. (1987) clearly demonstrated that olivines from kimberlite occurrences contain the highest hydrogen contents at a concentration level of 37-138 wt ppm H 2 0 (if a factor of 2.3 is applied to adjust Paterson's 1982 approximation to the recent calibration of Bell et al. 2003 - see the following chapter). The IR spectra are essentially characterized by bands in the 3600-3500 cm - 1 high-energy and 3400-3300 cm - 1 low-energy region. The authors also noted that over 30 distinct OH absorption bands have been identified in olivine from Monastery kimberlite, South Africa, and that the majority of these bands are inclined towards [100]. Representative polarized IR absorption spectra of olivines from South African occurrences, including olivine from Monastery, are shown in Figure 3. The presence of serpentine and talc has been determined by their characteristic OH absorption bands at 3685 and 3678 cm -1 , respectively (see above). Prominent OH bands at 3572 and 3525 cm -1 are attributed to humite group minerals (see above). Relatively uniform spectra of olivines from the Monastery kimberlite have been

Figure 3. Representative polarized IR absorption spectra in the OH stretching frequency region of olivines from kimberlitic xenoliths of South African occurrences, indicating a preferred orientation of the OH defects parallel to a (modified after Miller et al. 1987 and Bell et al. 2003).

174

Beran &

Libounizky

reported by Bell et al. (2004a). The olivine megacrysts represent the crystallization product of a kimberlite-like magma at pressures of about 5 GPa and temperatures of 1400-1100 °C. Spectra of olivines from the magnesian "main silicate trend" group (~13 wt% FeO) differ slightly from spectra of high-Fe olivines (17-19 wt% FeO). Both groups show a main band centered at 3572 cm - 1 with strong polarization parallel to [100]. The high-Fe olivines are characterized by an enhanced intensity of bands at the high-energy side of the main band, and by a reduced intensity of the 3526 cm - 1 band, relative to the main band. With respect to the OH defect assignment it is important to note that the Ti content of high-Fe olivines is significantly lower than that of low-Fe olivines. The olivines display H 2 0 contents in the range 45-262 wt ppm. Olivine and clinopyroxene water contents appear to increase with differentiation of the host magma, consistent with an enrichment of water in the residual melt during fractional crystallization. Inter-mineral distribution coefficients for OH between olivine and clinopyroxene are thus constant. However, the presence of strong, titanium related OH defect bands is evident. Matsyuk and Langer (2004) published a comprehensive IR study of Yakutian upper mantle material and proposed a new nomenclature for the hydrous component in olivines. Selected IR absorption spectra, also illustrating the presence of group II bands, are shown in Figure 4. Hydroxyl groups in the form of non-intrinsic separate inclusions (NSI) were discerned from isolated local defects (ILD) or condensed extended defects (CED) intrinsic to the olivine

25

1

1

i

1

20

E c

® O

it a) o o c o 8

•Q TO

15

AJ

:

i

l

i

l

i

l

i E II a

1

"KrS 10

| W \ AA

TO a) c

l

/ A

-... AJftlUv :

A

"•

V

.

> — B a z o v a y a - 3

1 3700

Udachnaya

— ; — 3600

3500

3400

3300

3200

Wavenumber (cm 1 Figure 4. Selected IR absorption spectra, polarized parallel to a, of olivines from kimberlitic xenoliths of occurrences from the Siberian Platform, Yakutia, showing OH bands in both group I and group II regions ( modified after Matsyuk and Langer 2004).

Water in Mantle

Minerals:

Olivine,

structure. As the two latter types cannot be simply distinguished by IR spectroscopy and as they are presumably interconnected by condensation reactions, it was proposed to symbolize the intrinsic defects as ILD/CED. NSI frequently comprise serpentine and talc with OH bands in the 3704-3657 cm - 1 range (see above); Mg-edenite and Mg-pargasite occur rarely, showing bands at 3711-3709 cm -1 . OH stretching bands strongly polarized along [100] are a significant feature of ILD/CED. Bands in the 3570-3510 c n r 1 region are intensity-correlated and are assigned to Sidepleted "Ti-clinohumite-like" defects (for OHclinohumite and OH-chondrodite see Liu et al. 2003). Bands in the 3500-3300 c n r 1 low-energy region and in the 3640-3580 c n r 1 high-energy region are suggested to originate from OH in different types of (Mg,Fe)-depleted defects. The complex nature of the strongly polarized OH bands in the group I region is demonstrated in Figure 5. The study of Matsyuk and Langer (2004) is based on a total of 335 olivine crystal grains extracted from 174 different specimen of Yakutian upper mantle material representing all rock types occurring in kimberlites of the Siberian platform. Though there are indications that the occurrence of individual defect types is related to the genetic peculiarities of their host rocks, straight-forward and simple correlations do not exist. It is important to note that, according to Matsyuk and Langer (2004), olivine included in diamond does not contain water, detectable by IR spectroscopy, neither as NSI nor as ILD/CED. Among the rock-forming olivines, those of the ilmenite bearing specimens are highest in total water content, except for olivines from peridotites with primary phlogopite. Values of the absolute water content range from 4 to about 350 wt ppm H 2 0 with an average around 140 wt ppm.

Garnet, & Accessory

Minerals

175

E c 'o it d) o o c o o t/> •Q TO TO d) C

3700

3600

3500

Wavenumber (cm1) Figure 5. Polarized IR absorption spectra of olivines from kimberlitic xenoliths of Yakutian occurrences, showing the complex nature of the OH bands around 3570 crrr 1 . The preferred orientation of the OH defects parallel to a is clearly indicated (modified after Matsyuk and Langer 2004).

Kurosawa et al. (1997) determined the concentration of hydrogen and other trace elements in olivines from mantle xenoliths by a combined SIMS and IR spectroscopic study. The H 2 0 contents are in the range of 13 to 60 wt ppm. In contrast to the observations made by Matsyuk and Langer (2004), the SIMS analyses of olivine included in diamond yielded similar hydrogen concentrations. Therefore, the authors concluded that the hydrogen content of xenolithic olivines does not equilibrate with water in the host magma during transport from the mantle to the surface. Comparing olivines from spinel peridotites with those from garnet peridotites, the presence of additional absorption bands in the band group II region (3354, 3331 cm -1 ) is a significant feature of spinel peridotitic olivines. In garnet peridotitic olivines a positive correlation of hydrogen with the trivalent cation content (Al+Cr) was observed by

176

Beran & Libowiizky

Kurosawa et al. (1997), thus indicating the incorporation of hydrogen into mantle olivines by a coupled substitution mechanism. Light elements for a suite of ten mantle-derived olivine crystals have been measured by EMP, SIMS and FTIR spectroscopy (Kent and Rossman 2002). Li, B and H 2 0 concentrations are in the range of 0.9-7.8, 0.01-67, and 0.8-61 wt ppm, respectively. Although Li, B and H 2 0 contents vary substantially, their cation proportions are not strongly correlated, arguing against coupled substitutions. More than 20 strongly polarized OH bands in the 3730-3330 cm - 1 range have been reported by Koch-Miiller et al. (2006) for olivine crystals from the Udachnaya kimberlite. Bands in the 3730-3670 cm - 1 region were assigned to inclusions of serpentine, talc and "10 A-phase." All other bands were believed to be intrinsic to the olivine structure and it was proposed that the corresponding OH point defects are associated with vacant Si and vacant M l sites, i.e., Ol-H defects, aligned strongly parallel to [100]. The absolute water contents range from 49 to 392 wt ppm. Calibration approaches and summary of hydrogen contents Though there exist older approaches to calibrate the hydrogen content in olivine, e.g., by the general diagram of Paterson (1982), only the two most recent calibration studies that end up with comparable values are reported here. The hydrogen contents of three natural olivines determined by 15N nuclear reaction analysis (NRA), i.e., 140, 220, and 16 wt ppm H 2 0, were used by Bell et al. (2003) to calibrate the IR spectroscopic data for the quantitative hydrogen analysis of olivines. The OH defect concentration expressed as wt ppm H 2 0 is 0.188 times the total integral absorbance of the fundamental OH stretching bands, rigorously applicable to samples dominated by OH absorptions in the high-wavenumber range 3650-3450 cm -1 . This equals a value of the integral molar absorption coefficient £¡ = 28450 + 1830 L-moH-cm -2 . A comparison of the OH concentrations determined with the Bell et al. (2003) calibration and those derived from the general Paterson (1982) trend indicates that by using polarized radiation, the Bell et al. (2003) calibration yields OH concentrations that are higher by a factor of 2.3. Based on SIMS analyses of four olivine samples from the Udachnaya kimberlite, KochMiiller et al. (2006) calculated the integral molar absorption coefficient to 37500 + 5000 L-moH-cm -2 . This value is by a factor of about 1.3 slightly higher than that determined by Bell et al. (2003). Table 2 summarizes the OH concentration values of mantle olivines from different studies and geological occurrences, based on the calibration of Bell et al. (2003). In general, the scatter of data is relatively limited and ranges from a few wt ppm to maximum values of about 400 wt ppm H 2 0. The mean values are roughly in the region between 100-200 wt ppm H 2 0. Considering the wide variation of localities the rather homogeneous hydrogen contents of olivines are amazing. In comparison to water contents of further important mantle minerals the sequence pyroxene > olivine > garnet can be observed. On the other hand, the water contents of these natural mantle minerals are comparatively low in the light of high-P/T phases, such as wadsleyite and ringwoodite.

GARNET Structural and spectral features The structure of garnet group minerals is built up by alternating corner-sharing Me3+Oe octahedra and Si0 4 tetrahedra, forming chains parallel to the three axes of the cubic unit cell. The resulting framework contains pseudo-cubic cavities which incorporate the larger Me2+ cations

Water in Mantle

Minerals:

Olivine,

Garnet,

& Accessory

Minerals

177

Table 2. Observed OH defect concentrations in mantle olivines. Mean values in parentheses. Geological occurrence; , ... Locality

., „ . No. ot samples

OH concentration ,. . . . ,,, (in wt ppm H 2 0)

,, „ References

Kimberlite, xenoliths; South Africa

3

37 - 138 (71)

Kimberlite, xenoliths; South Africa

2

140-220(180)

Bell et al. (2003)

Kimberlite, megacrysts; Monastery, South Africa

29

45 - 262 (159)

Bell et al. (2004a)

Kimberlite, xenoliths; Siberian platform

36

4 - 350 (140)

Matsyuk and Langer (2004)

Kimberlite, xenoliths; Udachnaya, Yakutia

9

49 - 392 (239)

Koch-Muller et al. (2006)

Kimberlite; Buell Park, Arizona

1

-50

Mosenfelder et al. (2006)

Miller et al. (1987)

Concentration values are derived from IR spectroscopic data, calibrated against H 2 0 values obtained from

15

N

Nuclear Reaction Analysis (Bell et al. 2003).

in eight-fold dodecahedral coordination. The S i 0 4 tetrahedra are distorted by an amount that depends on the size of the Me2+ cation in the distorted pseudo-cubes, which share two opposite edges with two tetrahedra. However, the relatively rigid S i 0 4 tetrahedra can accommodate to varying Me2+ cation sizes by a rotation which increases the size of the Me2+ sites and therefore the shared Me3+06 octahedral edges as well. The oxygen atoms, representing possible docking sites of hydrogen, occupy only one general crystallographic site and are coordinated by one Si, one Me3+ and two Me2+ cations in the form of an almost ideal tetrahedron. One of the well-established OH defect types is the hydrogarnet substitution (see also Libowitzky and Beran 2006, this volume), where (Si0 4 ) is (partially) replaced by (OH) 4 . This substitution mode is generally observed in OH rich samples of the grossular (more often with an andradite component)-hydrogrossular series. As originally observed by Cohen-Addad et al. (1967) and by Kobayashi and Shoji (1983) the IR spectroscopic characteristics of hydro grossulars with more than five wt% H 2 0 are two broad overlapping absorption bands centered around 3600 and 3660 cm - 1 (Rossman and Aines 1991). However, these spectroscopic characteristics were generally not observed in grossular containing less than 0.3 wt% H 2 0 . The great variability of the IR spectra of grossular with low H 2 0 content suggests that the hydrogarnet substitution is not the only means of incorporating OH groups. In addition to tetrahedral sites, OH defects apparently exist in multiple other environments. Because of the optically isotropic behavior of garnets and their widely varying OH stretching frequencies, usually in the 3700-3500 cm - 1 region, possible sites of hydrogen incorporation can only scarcely be assigned. The anisotropic OH stretching vibrational behavior of non-cubic natural garnet crystals with compositions close to the uvarovite-grossular binary was investigated by Andrut et al. (2002). The IR absorption behavior complies with orthorhombic, monoclinic, and triclinic crystal symmetry, respectively. According to the individual pleochroic behavior of ten nonisotropic bands, six different pleochroic patterns, i.e., four band doublets and two single bands, are distinguished. For band doublets at 3559/3540, 3572/3565, and 3595/3588 cm" 1 as well as for a single band at 3618 cm - 1 , models for a structural OH incorporation based on the classical (OH) 4 hydrogarnet substitution are proposed. In contrast, for the band doublet at 3652/3602 cm - 1 and the single band at 3640 cm - 1 , OH defect incorporation is explained by assuming the

178

Beran & Libowiizky

presence of vacancies on octahedral and dodecahedral cation positions, leading to [Si0 3 (0H)] tetrahedral groups. It is concluded that in garnets containing only H traces the [Si0 3 (0H)] substitution mode plays an essential role as OH defect incorporation mechanism. Anisotropy of OH bands for a birefringent grossular from Asbestos, Quebec, has also been reported by Rossman and Aines (1986). Based on the presence of absorption bands around 3685, 3570, and 3530 cm - 1 of hydrothermally grown Ti-bearing pyropes, an [(0H) 3 0] substitution was proposed by Khomenko et al. (1994) to compensate for the higher valence of Ti4+ at the Al 3+ site. Distance-least-squares calculations were used by Lager et al. (1989) to simulate the effect of the hydrogarnet substitution on the grossular structure. Those garnets in which the shared octahedral edge is longer than the unshared can incorporate more OH. The application of these observations to other garnet compositions suggests that mantle garnets, rich in pyrope component, may contain only very limited amounts of water. Due to the presence of a relatively sharp absorption band near 3600 cm - 1 in synthetic pyrope, Ackermann et al. (1983) proposed the hydrogarnet substitution as a possible location for water in the mantle. According to Geiger et al. (1991) OH defects in synthetic pyropes, grown from oxides are also incorporated into the structure as a hydrogarnet component, showing a characteristic absorption band at 3629 cm - 1 . The IR spectra of pyropes, grown from a gel starting material display several absorption bands, indicating that OH substitution is not governed solely by the hydrogarnet substitution. The splitting of the 3629 cm - 1 band at 79 K into two narrow bands centered around 3636 and 3618 cm - 1 has been confirmed by Geiger et al. (2000) for pyrope single-crystals doped with transition elements (Co, Cr, Ni, Ti, V). The spectrum of the Ti-bearing pyrope essentially corresponds to that observed by Khomenko et al. (1994). Geiger et al. (2000) suggested that due to the presence of four OH stretching bands in Ti-bearing pyrope (3686, 3630, 3568, and 3527 cm -1 ), additional mechanisms of OH substitution occur. In the IR spectrum of a V 4+ -bearing pyrope the same number of bands is observed, suggesting that higher charged cations cause additional OH substitutions and increased OH concentrations in garnet. Withers et al. (1998) found that under identical conditions of high pressure and temperature the OH content of pyrope is similar to that of grossular; at P = 3 GPa and T= 1000 °C the H 2 0 values amount to 0.04 wt% for pyrope and to 0.02 wt% for grossular. Both garnets, pyrope and grossular, are characterized by a single band centered at 3630 and 3622 cm -1 , respectively, being much sharper in grossular than in pyrope. The IR spectra of some natural pyropes appear to be different from those of synthetic samples. Observed OH band positions of garnets from mantle occurrences are summarized in Table 3. The OH spectrum of a nearly endmember natural pyrope from high-grade blueschists of Dora Maira, Western Alps, was originally described by Rossman et al. (1989). The spectrum does not resemble that of any other natural pyrope. As shown in Figure 6, the spectrum consists of four narrow bands at 3661, 3651, 3641, and 3602 cm -1 , forming a triplet and a single band system. If the calibration of Bell et al. (1995) is applied (see below), the estimated H 2 0 content is about 58 wt ppm. From high-temperature and high-pressure IR spectra, Lu and Keppler (1997) concluded that the absorption features arise from almost free OH groups in sites with different compressibility and thermal expansivity. The intensity of the high-energy triplet increases with increasing pressure (up to 10 GPa), while the intensity of the single band at 3602 cm - 1 decreases significantly. However, Dora Maira pyrope may not be fully representative for mantle garnets. Representative IR absorption spectra in the OH stretching frequency region of garnets from kimberlitic xenoliths are presented in Figure 7. The IR spectra of garnets rich in pyrope component from mantle-derived xenoliths of the Colorado Plateau show a broad absorption band centered around 3570 cm - 1 with an additional weak but broad absorption around 3650 cm - 1 (Aines and Rossman 1984a,b). The presence of broad OH bands centered around 3570 and 3670 cm - 1 for megacrysts of pyrope from ultramafic diatremes of the Colorado Plateau containing 22-112 wt ppm H 2 0 was reported by Wang et al. (1996). The authors stated that pyrope crystals from the mantle may dehydrogenate during ascent and that caution should be exercised in using the OH content of natural pyrope to infer conditions of the source region.

Water in Mantle

Minerals:

Olivine,

Garnet,

& Accessory

Minerals

179

Table 3. OH band positions (in cm ') of mantle garnets from different localities and occurrences.

Locality

3661 3651 3641

Remarks

3650

3650

usually weak 3630

3630

3602 3590 3570

3570 3512

3570 3512

usually strong Ti-related band

Wavenumber values with deviations of ± 4 c m - 1 are listed within one line. 1 - blueschist. Dora Maira. Italy (Lu and Keppler 1997 ) 2 - kimberlite, megacryst. Kaalvallei. South Africa (Bell and Rossman 1992b ) 3 - kimberlite, megacryst. Lace. South Africa ( Bell and Rossman 1992b) 4 - kimberlite, megacryst. Monastery. South Africa ( Bell et al. 2004a) 5 - kimberlite, xenolith. Udachnaya. Yakutia ( Snyder et al. 1995 ) 6 - eclogite. Rietfontein. South Africa (Bell and Rossman 1992b) 7 - grospydite. Zagadochnaya. Yakutia (Beran et al. 1993 )

• 1 . . . .

3700

I • . . . - . . . .

3650

3600

I .•

3550

.

l_i_

3500

W a v e n u m b e r (erri1) Figure 6. Room- and high-temperature IR spectra of the OH stretching vibrational region of a pyrope single-crystal from high-grade blueschists of Dora Maira, Italy (modified after Lu and Keppler 1997).

Two broad bands centered in the same wavenumber region were reported by Bell and Rossman (1992b) for pyrope-rich garnet samples from different localities of the subcontinental mantle of southern Africa (Fig. 7). A usually weak absorption centered around 3510 c m - 1 is described as a typical feature of Ti-bearing mantle garnets (Table 3). A significant absorption at 3570 c m - 1 for pyrope-rich garnet from the Liaoning-50 kimberlite, N E China was

180

Beran & LibowUzky i

1

1

1

1—

Megacryst garnet

Peridotite garnet 5 Louwrensia

Kimberley

4

Lace

3 Eclogite garnet

Monastery

2 Monastery

Vctor Kaalvallei

4000

3800

3600

3400

3200 1

Wavenumber (cm )

3000 4 0 0 0

3800

3600

3400

3200

3000

1

Wavenumber (cm )

Figure 7. Representative IR absorption spectra in the OH stretching frequency region of garnets from kimberlitic xenoliths of occurrences in southern Africa. The broad absorption rising towards higher wavenumbers is due to an electronic transition of Fe 2+ (modified after Bell and Rossman 1992b).

reported by Langer et al. (1993). Beran et al. (1993) reported a single broad absorption centered around 3630 cm - 1 for a garnet of an eclogitic mantle xenolith from the Zagadochnaya kimberlite, Yakutia. Absorptions near 3590 and 3650 cm - 1 are dominant in the spectra of pyrope-rich garnets from the Udachnaya kimberlite, Yakutia (Snyder et al. 1995; Table 3). A suite of 200 garnet single-crystals extracted from 150 mantle xenoliths from kimberlites of the Siberian platform was studied in the OH vibrational range by Matsyuk et al. (1998). Representative OH spectra show either one or a combination of two major bands in the 3660-3645, 3585-3560, and 3525-3515 cm - 1 range. Bands in the latter region occur only in Ti-rich (>0.4 wt% TiOo) garnets. Bands at 3570 and 3512 have also been reported for pyrope garnets from the Monastery kimberlite, South Africa, by Bell et al. (2004a) (Figure 7). The presence of amphibole exsolution lamellae discovered by TEM in garnets from peridotites from northern Tibet (Song et al. 2005) confirms the role of garnet as an important reservoir of water in the mantle. Calibration and hydrogen content Several attempts have been made to calibrate the amount of the hydrous component of garnets derived from IR spectroscopic investigations. Water contents were calculated by Aines and Rossman (1984b) by calibrating the integrated IR absorbances against the water content measured on the basis of P2O5 cell coulometry. 15N Nuclear Reaction Analysis (NRA) has first been performed by Rossman et al. (1988) for the determination of the hydrogen content of a series of almandine, pyrope, and spessartine garnets. It was stated that pyropes contain so little hydrogen ( 1. The second is not naturally observed but would be normal subalkaline lherzolite rocks at pressures above 6-7 GPa, as discussed later. Konzett et al. (1997) and Konzett and Fei (2000) have shown that the K/Na ratio of K-richterite increases with pressure in both peralkaline and subalkaline bulk compositions. The K/Na ratio of experimentally produced K-richterite reflects the K/Na ratio of the bulk composition, which, as Konzett et al. (1997) point out, is quite different to natural MARID rocks. K-richterites in MARIDs have a generally narrow range of K/Na ratio even though the bulk rocks have a far more variable range. As Konzett et al. (1997) reasoned, this means that MARID rocks themselves cannot represent the entire liquid from which the rocks crystallized, i.e., MARID rocks are cumulates. The results of Konzett et al. (1997) and Konzett and Fei (2000) demonstrate that in peralkaline bulk compositions, where diopside is stable but garnet is only present at pressures >8 GPa, K-richterite stability is very close to that of the pure phase. Only the thermal maxima is reduced by approximately 100 °C compared to the pure phase stability determined by Tr0nnes (2002). The vast majority of mantle peridotite rocks, on the other hand, are subalkaline (i.e., Na 2 0+K 2 0)/Al 2 0 3 > 1). K-richterite is only stable in subalkaline lherzolitic bulk compositions above 6-7 GPa as a result of the reaction: 0.5K 2 Mg 6 Al 2 Si 6 O 20 (OH) 4 + CaMgSi 2 O e + NaAlSi 2 O e + Mg 2 Si 2 O e = phlogopite

in cpx

opx

KNaCaMg 5 Si s 0 2 2 (0H) 2 + Mg 3 Al 2 Si 3 0 12 K-richterite

garnet

(7)

260

Frost

Di+Cen +Wad+X+fl

Di+Cen +Wad+fl "

o 900

1000 1100

|

K-richterite stable or

I

I unstable in peridot ite

1200 1300 1400 1500

1600

Temperature (°C) Figure 11. The stability field of pure KNaCaMg 5 Si 8 022(0H)2 K-richterite as bracketed by Tr0nnes (2002) is shown by curve (A). All of the indicated named products are with respect to curve (A) with Wad being wadite-structured K 2 Si 4 09, X is Phase X, en and Cen are enstatite and clinoenstatite, Di is diopside and St is stishovite. Curves (B) and (C) are stability fields of K-richterite determined by Foley (1991) and Gilbert and Briggs (1974), respectively. The closed and open symbols indicate the presence and absence of K-richterite in a synthetic KNCMASH subalkaline peridotite assemblage, as determined by Konzett and Ulmer (1999) and Konzett and Fei (2000). In these experiments the coexisting assemblage always contained olivine, garnet, clinopyroxene and enstatite or clinoenstatite. The breakdown products at high pressure also contain Phase X and below curve (D) K-richterite breaks down to a phlogopite-bearing assemblage. Curve (E) shows the high pressure stability of KK-richterite K 2 CaMg 5 Si 8 022(0H)2 as determined by Inoue et al. (1998).

Equation (7), the Na present equivalent of Equation (6) proposed by Sudo and Tatsumi (1990), demonstrates that the breakdown of minor amounts of phlogopite in the presence of pyroxenes containing some jadeite component can produce the observed natural mantle Krichterite without any fluid release. The identical K/OH-ratio of phlogopite and K-richterite (but not KK-richterite) is the fundamental requirement for such a fluid-free reaction (Konzett and Ulmer 1999). Equation (7) and the general lack of K-richterite in garnet-bearing mantle xenoliths (e.g., Erlank et al. 1987), indicate that the majority of such xenoliths equilibrated at depths shallower than 200 km.

EXPERIMENTAL STUDIES ON THE STABILITY OF POTENTIAL HIGH PRESSURE HYDROUS MANTLE MINERALS Whereas most mantle xenoliths originate from the upper 200 km of the mantle or up to approximately 7 GPa, experimental studies at higher pressures have identified a number of other hydrous minerals that are potentially stable in the deeper parts of the mantle, although mainly in subduction zones. These phases generally don't have mineral names and are simply referred to by letters e.g., A, B, superhydrous B, D, E and X. Their stability in subduction zones is covered in this volume by Kawamoto (2006). The criterion for evaluating their presence in the ambient mantle is their compatibility with typical mantle minerals and their high temperature stability,

Stability

of Hydrous

Mantle

Phases

261

which can only be assessed through experimental studies. Here I only consider phases with upper thermal stability limits that are close to an average mantle adiabat. Phase X The enigmatically named Phase X has been observed in several studies as a high-pressure product of the decomposition of K-richterite. Phase X has a variable composition with reported K 2 0 contents of between 10 and 19 wt%. In the KCMSH system Inoue et al. (1998) reported Phase X with the approximate composition K4Mg s Si s 0 2 5(0H) 2 whereas Tr0nnes (2002) reported the composition K 3 7 Mg 74 Al 06 Si s O 2 5(OH)2 in the KMASH system. In addition to Phase X with the formula Kj.54Mgj.93Sij.g9O7Hj.04, Yang et al. (2001) synthesized and solved the structures of sodic Phase X, Naj 16K0.0iMgj 93Al0.i4Sij g 9 0 7 Hj .04, and the anhydrous end members Kj g5Mg2.06Si2.01O7 and Naj 7gMgj 93Al0.j3Si2.02O7. Phase X is composed of layers of brucite-like MgO e octahedra linked by Si 2 0 7 tetrahedral dimers and K cations (Yang et al. 2001 ; Mancini et al. 2002). Yang et al. (2001) proposed the general formula A2_rM2Si207Hr where A can be K and/or Na, M can be Mg or Al and x = 0-1. An increase in the K content of Phase X is therefore coupled to a decrease in the H content. The only measurement of the H 2 0 content of Phase X, performed using SIMS, yielded a value of 1.7±0.1 wt% H 2 0 (Inoue et al. 1998), which is significantly below the theoretical maximum of 3.51 wt%. No studies have been performed on the stability of any pure Phase X composition; however, Konzett and Fei (2000) have examined the stability of Phase X in a subalkaline KNCMASH analogue peridotite composition. Phase X coexists with a typical mantle assemblage of olivine/wadsleyite, clinopyroxene and garnet between 14 and 20 GPa and at temperatures up to 1600 °C. Phase X, therefore, has the highest thermal stability of any yet investigated nominally hydrous silicate. As shown in Figure 12 the high temperature stability of Phase X is for the main part undetermined. Konzett and Fei (2000) showed that the reactions that produce Phase X from K-richterite in a mantle peridotite composition release fluid because the K/H ratio of Phase X is higher than that of K-richterite. These results also show that the K content and K/Na ratio of Phase X both increase with pressure, which implies a decrease in the H 2 0 content of Phase X with pressure. The change in K/Na ratio occurs as Na is partitioned into coexisting garnet with increasing pressure. Between 20 and 22 GPa Phase X breaks down to an assemblage containing K-hollandite (KAlSi 3 O s ).

0

0

K-Hollandite bearing - *

Phase X out

• -

t

P h a s e X bearing

- •• •

• •

•

®

Ü O K-richterite bearing

1000 1100 1200 1300 1400 1500 1600 1700 1800

Temperature (°C)

Figure 12. The closed and open symbols indicate the presence and absence of phase-X in a synthetic KNCMASH subalkaline peridotite assemblage, as determined Konzett and Fei (2000). In these experiments the coexisting assemblage was that expected for a peridotite composition at the indicated conditions, i.e., olivine or high-pressure polymorphs, garnet and Ca-perovskite at and above 20 GPa. Filled rectangles show conditions where Luth (1997) observed Phase X in a KCMASH bulk composition. At high pressures Phase X breaksdown to an assemblage containing K-hollandite (KAlSi,0 8 ).

262

Frost

Humite and dense hydrous magnesium silicate phases A number of high pressure experimental studies have shown that the humite minerals chondrodite and clinohumite and the dense hydrous magnesium silicate phases A, superhydrous B, D and E can coexist with ultramafic assemblages at various conditions above 6 GPa and below 1200 °C (Kanzaki 1991; Kawamoto et al. 1995; Ohtani et al. 1995; Frost and Fei 1998; Irifune et al. 1998). The stability fields of these phases are significantly below reasonable average mantle adiabats and they are therefore only expected to be stable in the cooler regions of subduction zones, provided that significant H 2 0 is in fact present within such regions at pressures above 6 GPa. Although the stability fields have been examined in natural systems (Luth 1995; Kawamoto et al. 1995; Frost 1999; Kawamoto 2004) there remains some question as to whether the strong partitioning of some element by a particular hydrous phase may cause some increase in thermal stability. In addition the large amounts of H 2 0 added in some bulk compositions may result in the breakdown of hydrous phases at a lower temperature than we might expect in the mantle as a result of excessive melting. Experiments in relatively 1OW-H20 bulk compositions show, however, that the presence of Al and Fe in phases A and E, superhydrous phases B and D has a limited effect on stability relations in comparison to the MSH system (Luth 1995; Frost 1999). Humite minerals have a preference for Ti and F. Titanian clinohumite is a common accessory mineral in metamorphosed ultrabasic rocks and occurs in serpentinites and kimberlites (López Sánchez-Vizcaíno et al. 2005). The stability of titanian clinohumite is below 1000 °C at 8 GPa although the pure fluorine clinohumite end-member is stable to over 1400 °C at 3 GPa (Weiss 1997; Ulmer and Trommsdorf 1999). The experiments of Kawamoto (2004) contained Ti and showed clinohumite and chondrodite stability to be limited to below 1100 °C at 11 GPa. Phase D is the highest-pressure dense hydrous magnesium silicate and its stability in the lower mantle is ultimately controlled by the reaction, MgSi 2 0 4 (0H) 2 Phase D

+

MgO periclase

=

2MgSi0 3 MgSi-perovskite

+

H20

(8)

Liquid

The slope of this reaction is not clear however. From a Schreinemakers analysis of the existing experimental data Komabayashi et al. (2004) reported a negative Clapyron slope at approximately 25 GPa with a maximum thermal stability for phase D of 1100°C. Laser heated diamond cell experiments of Shieh et al. (1998) indicate that this reaction leads to the breakdown of phase D at 44 GPa at temperatures between 1000 and 1400 °C, which would be consistent with a convex shape of the reaction boundary of Equation (8), like many other dehydration reactions. Phase D may therefore be stable at temperatures higher than 1100 C at pressures between 25 and 44 GPa but it is probably unlikely that these temperatures approach that of the mantle adiabat. In natural systems phase D contains significant amounts of Al and ferric and ferrous Fe but not in quantities higher than coexisting silicate perovskite, so they have little effect on the thermal stability of phase D (Frost 1999; Frost unpublished data).

THE STABILITY OF HYDROUS PHASES IN ULTRAMAFIC LITHOSPHERE AND THE CONVECTING MANTLE In considering the significance of hydrous minerals in the mantle it is not only of interest to define stability fields, but it is also important to assess the proportion of hydrous minerals that may exist at particular conditions, identify how much of the mantle's water budget they may account for and examine further factors, such as H 2 0 activity, that may affect their stability. Changing redox conditions as a function of depth in the upper mantle and transition zone may also control fluid speciation and H 2 0 activity, which, in turn, may affect the stability of the hydrous phases.

Stability

of Hydrous

Mantle

263

Phases

Figure 13 shows the stability fields of the major mantle hydrous phases derived from the previously described experimental studies. The experimental data employed are from studies where hydrous phases formed in equilibrium with typical ultramafic mantle assemblages at H 2 0 undersaturated conditions. A mantle adiabat with a potential temperature of 1600 K (i.e., the temperature at the surface when extrapolated through the melting region) is shown with branching geotherms for Achean cratonic and oceanic lithosphere. A water saturated peridotite solidus interpolated from the data of Mysen and Boettcher (1975) to 4 GPa and Kamamoto (2004) >4 GPa is shown. The solidus is not followed into the region of dense hydrous magnesium silicate stability because huge amounts of H 2 0 are required to produce a melt at these conditions and the solvus between fluid and melt may anyway disappear. Figure 13 indicates that the only hydrous mineral to be stable along an average mantle adiabat (AMA) is Phase X, which could be present in the mantle between depths of 400 and 600 km. The Archean lithospheric geotherm (ACL), which branches off the average mantle adiabat at temperatures approaching 1400 °C, misses the stability field of K-richterite but enters the phlogopite stability field at pressures of approximately 6.8 GPa at 1280 °C. In Figure 13 the data on phlogopite and K-richterite are taken from experiments in Fe-free systems (Konzett and Ulmer 1999; Konzett and Fei 2000). Preliminary experiments seem to indicate that Fe destabilizes these hydrous phases further (Konzett and Ulmer 1999) and the extent of the stability fields in Figure 13 may, therefore, be slightly overestimated. It is important to reiterate that in nature K-richterite occurs in mantle xenoliths of peralkline rocks where the K-richterite stability field extends to much lower pressures (Konzett et al. 1997) than in normal subalkaline

700

25

Q.

900

1000 1100 1200 1300 1400 1500 1600 1700

Temperature (°C) Figure 13. Stability fields of hydrous minerals in mantle of peridotite composition at H 2 0-undersaturated conditions. Data are combined from Figures 5,9.11 and 12. The grey shaded region shows where the dense hydrous magnesium silicate phases A, E, super hydrous phase B (sB) and D are stable from Kawamoto (2005). Thick grey curves show the peridotite solidus under H 2 0 saturated and dry conditions. An average mantle adiabat (AMA) and geotherms for Archean cratonic lithosphere (ACL) and 100 million year old oceanic lithosphere (OL) are shown by thin black lines.

264

Frost

peridotite compositions that are depicted in Figure 13. It seems that the only environment where K-richterite could exist in subalkaline mantle rocks is in a subduction zone. The oceanic lithosphere geotherm (OL) passes into the stability field of phlogopite and pargasitic amphibole below 3 GPa. The proportion of hydrous minerals that form along typical geotherms will depend on the Na and K content of the mantle, which in turn will depend on the degree of depletion and metasomatism. Using a primitive mantle composition as a benchmark, it is possible to appreciate the degree of metasomatic enrichment necessary for significant hydrous minerals to form in the lithosphere. If we consider a typical primitive mantle N a 2 0 content of 0.3 wt% then from the experiments of Niida and Green (1999) we can calculate that along an oceanic geotherm at approximately 70 km the lherzolitic assemblage could contain 9 wt% pargasitic amphibole which would accommodate approximately 1500 ppm H 2 0 , assuming stoichiometric amphibole OH contents. At only 50 km this rises to approximately 25 wt% pargasite which would host 4000 ppm H 2 0 in the bulk. Significant amounts of amphibole can, therefore, form in lithospheric mantle with typical Na contents, mostly at the expense of clinopyroxene, by adding relatively small amounts of H 2 0 alone. Primitive mantle K contents, on the other hand, are generally 10 times lower than corresponding Na contents. Therefore, along an Archean craton lithospheric geotherm at approximately 150 km depth 0.03 wt% K 2 0 in the bulk rock will allow a maximum of just 0.2 wt% phlogopite to form, which will host 90 ppm H 2 0 , using data from Konzett and Ulmer (1999). In comparison, metasomatized garnet phlogopite peridotite rocks (GPP) reported by Erlank et al. (1987) have average bulk K 2 0 contents of 0.16% which would result in 1.4 wt% phlogopite forming at 150 km with a bulk H 2 0 content of 600 ppm. Erlank et al. (1987) classified GPP rocks as the least metasomatized, whereas PKP rocks, which are considered to be the most metasomatized, have average K 2 0 contents of approximately 1%. The presence of significant phlogopite in some mantle xenoliths means, therefore, that there are processes that occur in the mantle that strongly concentrate K while having much smaller effects on other major elements and in particular Na. One possibility is that such high K-bearing liquids are produced by the breakdown of the white mica phengite in subducting lithosphere (Schmidt et al. 2004). If this is the only explanation then all K-rich metasomatism of the lithosphere must be related to subduction. Another possibility is that high-pressure metasomatic fluids or melts are K-rich because Na becomes compatible in clinopyroxene during melting at high pressures and low temperatures (Blundy et al. 1995). Clinopyroxene/melt partition coefficients for K, on the other hand, are normally 2 orders of magnitude below those of Na. The compositions of low-fraction hydrous melts or fluids at pressures above 3 GPa are poorly constrained but further study may provide important insights into metasomatic agents in the lithosphere. Phase X is the only hydrous mineral that could be stable along an average mantle adiabat in the convecting mantle, at least given the available experimental data that extend to lower mantle conditions (>660 km). Assuming a primitive mantle bulk composition we can calculate how much Phase X could form at the top of the transition zone (410 km) and how much of the convecting mantle's water budget it could account for at these depths. Using the data of Konzett and Fei (2000) who used a K 2 0 enriched KLB-1 peridotite composition the K 2 0 content of Phase X expected in the transition zone is 14.5 wt% and the stoichiometric H 2 0 content is approximately 3 wt%. If the bulk rock contains 0.03 wt% K 2 0 then 0.1 wt% Phase X can form with the proportion of H 2 0 hosted by Phase X being just 30 ppm. In addition to the low K content of the primitive mantle, at these conditions high Ca-clinopyroxene also contains as much as 0.1% K 2 0 and given the uncertainty on some of the values it is quite possible that all K 2 0 and H 2 0 may be accommodated by the nominally anhydrous assemblage. It is of course also possible that K is inhomogeneously distributed in the convecting mantle in a similar way to that found in the lithosphere, resulting in regions with higher proportions of Phase X. If the bulk of the transition zone has a primitive mantle composition, however, then the formation of these regions must leave the remaining transition zone depleted in K and the amount of H 2 0 stored

Stability

of Hydrous

Mantle

Phases

265

by Phase X over the bulk of the transition zone cannot be much greater than 30 ppm. It seems clear, therefore, that nominally anhydrous minerals and melts, or possibly fluids at reducing conditions, must host the majority of hydrogen stored in the ambient convecting mantle. As shown in Figure 4 and explained previously for pargasitic amphibole, hydrous phases display the highest thermal stability at fluid-absent conditions. Figure 13 should, therefore, depict the maximum thermal stability within ultramafic bulk compositions with respect to water activity. Lower H 2 0 activities or H 2 0-saturated conditions should lead to lower hydrous mineral stability fields. One of the problems of relating experimental studies of hydrous mineral stability to natural mantle mineral assemblages is that we generally have only circumstantial evidence for the nature of the metasomatic or igneous melt/fluid phase from which the minerals formed. The activity of H 2 0 at the conditions of formation are, therefore, poorly constrained. For this reason the previously described methodology of Popp et al. (1995) to determine H 2 0 activity through the use of Equation (2) is particularly attractive. Above subduction zones for example where high concentrations of H 2 0 may enter the mantle wedge the maximum thermal stability of hydrous minerals such as pargasite or phlogopite may be closer to that of H 2 0-saturated conditions, shown for pargasite in Figures 4 and 5, which may be a few hundred degrees below those in Figure 13. Another poorly constrained factor is that the oxygen fugacity of the mantle may decrease with depth causing C-O-H fluids to become richer in CH 4 and lowering the activity of H 2 0 (Woermann and Rosenhauer 1985; Wood et al. 1990). Several studies have argued for a lowering of mantle fo2 with depth as a result of the pressure effect on the ferric-ferrous equilibria that likely define mantle fo2 and due to changes in the solubility of ferric iron in major mantle minerals (Wood et al. 1990; Gudmundsson and Wood 1995; O'Neill et al. 1993; Ballhaus and Frost 1994; Frost et al. 2004). Several oxygen thermobarometry studies on garnet peridotite xenoliths have observed a decrease in fo2 with depth from values of around FMQ-1 (one log unit below the fayalite-magnetite-quartz oxygen buffer) close to the spinel peridotite field at 80 km depth down to FMQ-4 at approximately 200 km (McCammon et al. 2001; Woodland and Koch 2003; McCammon and Kopylova 2004). O'Neill et al. (1993) argued that oxygen fugacities in the transition zone may be close to the iron-wiistite buffer (IW i.e., ~FMQ-5). At these conditions a C-O-H fluid may contain over 50% CH 4 and up to 5% H 2 although values are uncertain as equations of states for reduced gas phases are poorly constrained at these conditions (Holloway 1987; Belonoshko and Saxena 1992). H 2 contents may also increase depending on the activity of carbon and components such as H 2 S could also be relevant. Although there is very little experimental data on their behavior, reduced fluid phases may be more mobile in the mantle as their components likely have lower solubilities in minerals and melts and the solubilities of silicate components in these fluids may be low. As previously discussed, Taylor and Green (1988) observed an increase in the fluid-saturated peridotite solidus between 1.0 and 3.5 GPa at low fo2 (~FMQ-4) where CH 4 became a major fluid component. This occurred because CH 4 lowered the H 2 0 activity in the fluid, which lowered the H 2 0 solubility in the coexisting silicate melt. The presence of a reduced fluid phase with a low H 2 0 activity in the mantle may affect hydrous phase stability and may also lower the solubility of hydroxyl in nominally anhydrous phases. The mobility and low density of a reduced fluid phase in the deeper convecting mantle may help to redistribute hydrogen and might even tend to focus H 2 0 in the upper more oxidized regions of the upper mantle.

ACKNOWLEDGMENTS I am tremendously grateful to Jurgen Konzett, Reidar Tr0nnes and Alan Woodland for lengthy discussions and for making numerous comments on an earlier version of the manuscript. I also appreciate the comments and corrections of Hans Keppler and John Winter.

266

Frost

REFERENCES Adam J, Green TH, Sie SH (1993) Proton microprobe determined partitioning of Rb, Sr, Ba, Y, Nb and Ta between experimentally produced amphiboles and silicate melts with variable F content. Chem Geol 109: 29-49 Adam J, Green TH, Sie SH, Ryan CG (1997) Trace element partitioning between aqueous fluids, silicate melts and minerals. Eur J Mineral 9:569-584 Arai S, Matsukage K, Isobe E, Vysotskiy S (1997) Concentration of incompatible elements in oceanic mantle: effect of melt/wall interaction in stagnant or failed melt conduits within peridotite. Geochim Cosmochim Acta 61:671-675 Agrinier P, Mevel C, Bosch D, Javoy M (1993) Metasomatic hydrous fluids in amphibole peridotites from Zabargad island (Red Sea). Earth Planet Sci Lett 120:187-205 Baker MB, Wyllie PJ (1992) High-pressure apatite solubility in carbonate-rich liquids: implications for mantle metasomatism. Geochim Cosmochim Acta 56:3409-3422 Bailey DK (1982) Mantle metasomatism-continuing chemical change within the Earth. Nature 296:525-530 Ballhaus C, Frost BR (1994) The generation of oxidized C0 2 -bearing basaltic melts from reduced CH4-bearing upper mantle sources. Geochim Cosmochim Acta 58:4931-4940 Belonoshko AB, Saxena SK (1992) A unified equation of state for fluids of C-H-O-N-S-Ar composition and their mixtures up to very high-temperatures and pressures. Geochim Cosmochim Acta 56:3611-3626 Bureau H, Keppler H (1999) Complete miscibility between silicate melts and hydrous fluids in the upper mantle: experimental evidence and geochemical implications. Earth Planet Sci Lett 165:187-196 Blundy JD, Falloon TJ, Wood BJ, Dalton JA (1995) Sodium partitioning between clinopyroxene and silicate melts. J Geophys Res 100:15501-15515 Bodinier J-L, Dupuy C, Dostal J (1988) Geochemistry and pedogenesis of Eastern Pyrenean peridotites. Geochim Cosmochim Acta 52:2893-2907 Bose K, Ganguly J (1995) Experimental and theoretical studies of the stabilities of talc, antigorite and phase A at high pressures with applications to subduction processes. Earth Planet Sci Lett 136:109-121 Boyd FR (1959) Hydrothermal investigations of amphiboles. In: Researches in Geochemistry Vol 1. Abelson PH (ed) Wiley, p 377-396 Boyd FR, Mertzman SA (1987) Composition and structure of the Kaapvaal lithosphere, South Africa. In: Magmatic Processes: Physiochemical Principles. Mysen B (ed) Geochemical Society Special Publication 1, p 13-24 Boyd FR (1990) Mantle metasomatism: evidence from a MARID-harzburgite compound xenolith. Carnegie Institution Washington Yearbook 90:18-23 Brandon AD, Draper DS (1996) Constraints on the origin of the oxidation state of mantle overlying subduction zones: An example from Simcoe, Washington, USA. Geochim Cosmochim Acta 60:1739-1749 Bryndzia LT, Wood BJ (1990) Oxygen thermobarometry of abyssal spinel peridotites: the redox state and the C-O-H volatile composition of the Earth's sub-oceanic upper mantle. Am J Sci 290:1093-1116 Bureau H, Keppler H (1999) Complete miscibility between silicate melts and hydrous fluids in the upper mantle: experimental evidence and geochemical implications. Earth Planet Sci Lett 165:187-196 Chazot G, Menzies MA, Harte B (1996) Determination of partition coefficients between apatite, clinopyroxene, amphibole, and melt in natural spinel lherzolites from Yemen: implications for wet melting of the lithospheric mantle. Geochim Cosmochim Acta 60:423-437 Cohen RE, O'Nions RK, Dawson JB (1984) Isotope geochemistry of xenoliths from East Africa: implications for development of mantle reservoirs and their interaction. Earth Planet Sci Lett 68:209-220 Conceicao RV, Green DH (2004) Derivation of potassic (shoshonitic) magmas by decompression melting of phlogopite+pargasite lherzolite. Lithos 72:209-229 Dawson JB, Smith JV (1977) The MARID (mica-amphibole-rutile-ilmenite-diopside) suite of xenoliths in kimberlite. Geochim Cosmochim Acta 41:309-323 Dawson JB, Smith JV (1982) Upper-mantle amphiboles: a review. Mineral Mag 45:35-46 Dawson JB (1984) Contrasting types of upper-mantle metasomatism? In: Kimberlites II: The mantle and CrustMantle Relationships. Kornprobst J (ed) Elsevier, pp 289-294 Dyar MD, Mackwell SJ, McGuire AV, Cross, LR, Robertson JD (1993) Crystal chemistry of Fe3+ and H + in mantle kaersutite: implications for mantle metasomatism. Am Mineral 78:968-979 Edgar AD, Vukadinovic D (1992) Implications of experimental petrology to the evolution of ultrapotassic rocks. Lithos 28:205-220 Eggler DH (1978) The effect of C0 2 upon partial melting of peridotite in the system Na 2 0-Ca0-Al 2 0 3 -Mg0Si0 2 -C0 2 to 35 kb, with an analysis of melting in a peridotite-H 2 0-C0 2 system. Am J Sci 278:305-343 Elkins-Tanton LT, Grove TL (2003) Evidence for deep melting of hydrous metasomatized mantle: Pliocene high-potassium magmas from the Sierra Nevadas. J Geophys Res 108:2350, doi:10.1029/2002JB002168

Stability of Hydrous Mantle

Phases

267

Erlank AJ, Waters FG, Hawkesworth CJ, Haggerty SE, Allsopp HL, Rickard RS, Menzies M (1987) Evidence for mantle metasomatism in peridotite nodules from the Kimberley pipes, South Africa. In: Mantle Metasomatism. Menzies MA, Hawkesworth CJ (eds). Academic Press, p 313-361 Fabries J, Lorand J-P, Guiraud M (2001) Petrogenesis of the amphibole-rich veins from the Lherz orogenic lherzolite massif (Eastern Pyrenees, France): a case study for the origin of the orthopyroxene-bearing amphibole pyroxenites in the lithospheric mantle. Contrib Mineral Petrol 140:383-403 Field SW, Haggerty, SE, Erlank AJ (1989) Subcontinental metasomatism in the region of Jagersfontein, South Africa. In: Kimberlites and Related Rocks. Geol Soc Aust Spec Publ 14. Ross J (ed). Geological Society of Australia, p 771-784 Fleet ME (2003) Micas. The Geological Society Foley S (1991) High-pressure stability of the fluor- and hydroxyl-endmembers of pargasite and K-richterite. Geochim Cosmochim Acta 55:2689-2694 Foley S (1992) Vein-plus-wall-rock melting mechanisms in the lithosphere and the origin of potassic alkaline magmas. Lithos 28:435-453 Frey FA, Prinz M (1978) Ultramafic inclusions from San Carlos, Arizona: petrological and geochemical data bearing on their petrogenesis. Earth Planet Sci Lett 38:129-176 Frost DJ, Fei Y (1998). Stability of phase D at high pressure and high temperature. J Geophys Res 103B:74637474 Frost DJ (1999) The stability of dense hydrous magnesium silicates in Earth's transition zone and lower mantle. In Mantle Petrology: Field Observations and High-Pressure Experimentation. Fei Y, Bertka CM, Mysen BO(ed) Geochemical Society, Special Publication No. 6, p 283-295 Frost DJ, Liebske C, Langenhorst F, McCammon CA, Tronnes RG, Rubie DC (2004) Experimental evidence for the existence of iron-rich metal in the Earth's lower mantle. Nature 428:409-412 Gilbert MC, Briggs DF (1974) Comparison of the stabilities of OH- and F-potassic richterites- a preliminary report. Trans Am Geophys Union 55:480-481 Gilbert MC, Helz RT, Popp RK, Spear FS (1982) Experimental studies of amphibole stability. Rev Mineral 9B: 229-353 Green DH (1973) Experimental melting studies on a model upper mantle composition at high pressure under water-saturated and water-undersaturated conditions. Earth Planet Sci Lett 19:37-53 Gregoire M, Lorand JP, O'Reilly SY, Cottin JY (2000) Armalcolite-bearing, Ti-rich metasomatic assemblages in harzburgitic xenoliths from the Kerguelen Islands: Implications for the oceanic mantle budget of high-field strength elements. Geochim Cosmochim Acta 64:673-694 Gregoire M, Bell DR, Le Roex AP (2003) Garnet lherzolites from the Kaapvaal craton (South Africa): trace element evidence for a metasomatic history. J Petrol 44:629-657 Gudmundsson G, Wood BJ (1995) Experimental tests of garnet peridotite oxygen barometry. Contrib Mineral Petrol 119:56-67 Halliday AN, Lee D-C, Tommasini S, Davies GR, Paslick CR, Fitton JG, James DE (1995) Incompatible trace elements in OIB and MORB and source enrichment in the sub-oceanic mantle. Earth Planet Sci Lett 133: 379-395 Harte B (1993) Mantle peridotites and processes- the kimberlite sample. In: Continental Basalts and Mantle Xenoliths. Hawkesworth CJ, Norry MJ (eds) Shiva, p 46-91 Hauri EH, Shimizu N, Dieu JJ, Hart SR (1993) Evidence for hotspot-related carbonatite metasomatism in the oceanic upper-mantle. Nature 365:221-227 Hawkesworth CJ, Erlank AJ, Marsh JS, Menzies MA, Van Calsteren P (1983) Evolution of the continental lithosphere: evidence from volcanics and xenoliths in southern Africa. In: Continental Basalts and Mantle Xenoliths. Hawkesworth MJ, Norry MJ (eds). Shiva, p 111-138 Hoal KEO, Hoal BG, Erlank AJ, Shimizu N (1994) Metasomatism of the mantle lithosphere recorded by rare earth elements in garnets. Earth Planet Sci Lett 126:303-313 Holloway JR (1973) The system pargasite H 2 0-C0 2 : a model for melting of a hydrous mineral with a mixedvolatile fluid. I. Experimental results to 8 kbar. Geochim Cosmochim Acta 37:651-666 Holloway JR (1987) Igneous fluids. Rev Mineral 17:211-233 Hiibner SJ, Papike JJ (1970) Synthesis and crystal chemistry of sodium-potassium richterite (Na,K)NaCaMg5 Si 8 0 22 (0H,F) 2 ): a model for amphiboles. Am Mineral 55:1973-1992 Huckenholz HG, Gilbert MC, Kunzmann T (1992) Stability and phase relations of calcic amphiboles crystallized from magnesio-hastingsite compositions in the 1 to 45 kbar pressure range. N Jahr Miner Abh 164:229-268 Inoue T, Irifune T, Yurimoto H, Miyagi I (1998) Decomposition of K-amphibole at high pressures and implications for subduction zone volcanism. Phys Earth Planet Int 107:221-231 Ionov DA, Hofmann AW (1995) Nb-Ta-rich mantle amphiboles and micas: implications for subduction-related metasomatic trace element fractionations. Earth Planet Sci Lett 131:341-356 Ionov D (1998) Trace element composition of mantle-derived carbonates and coexisting phases in peridotite xenoliths from alkaline basalts. J Petrol 39:1931-1941

268

Frost

IrifuneT, Kubo T, Isshiki M, Yamasaki Y (1998) Phase transformations in serpentine and transportation of water into the lower mantle. Geophys Res Lett 22:117-120 Jenkins DM (1983) Stability and composition relations of calcic amphiboles in ultramafic rocks. Contrib Mineral Petrol 83:375-384 Jones AP, Smith JV, Dawson JB (1982) Mantle metasomatism in 14 veined peridotites from Bultfontein Mine, South Africa. J Geol 90:435-453 Jones AP (1989) Upper mantle enrichment by kimberlitic or carbonatitic magmatism. In: Carbonates: Genesis and Evolutions. Bell K (ed) Unwin Hyman, p 448-463 Kanzaki M (1991) Stability of hydrous magnesium silicates in the mantle transition zone. Phys Earth Planet Inter 66:307-312 Kawamoto T, Leinenweber K, Hervig RL, Holloway JR (1995) Stability of hydrous minerals in H 2 0-saturated KLB-1 peridotite up to 15 GPa. In: Volátiles in the Earth and Solar System. Farley KA (ed) American Institute of Physics, p 229-239 Kawamoto T (2004) Hydrous phase stability and partial melt chemistry in H 2 0-saturated KLB-1 peridotite up to the uppermost lower mantle. Phys Earth Planet Int 143-144:387-395 Kawamoto T (2006) Hydrous phases and water transport in the subducting slab. Rev Mineral Geochem 62: 273-289 King PL, Hervig RL, Holloway JR, Vennemann TW, Righter K (1999) Oxy-substitution and dehydrogenation in mantle-derived amphibole megacrysts. Geochim Cosmochim Acta 63:3635-3651 Kinny PD, Compston JW, Bristow JW, Williams IS (1989) Archaean mantle xenocrysts in a Permian kimberlite: two generations of kimberlitic zircon in Jwaneng DK2, southern Botswana. In: Kimberlites and related rocks. Geol Soc Aust Spec Publ 14. Ross J (ed). Geological Society of Australia, p 833-842 Kinny PD, Meyer HOA (1994) Zircon from the mantle: a new way to date old diamonds. J Geol 102:475-481 Komabayashi T, Omori S, Maruyama S (2004) Petrogenetic grid in the system Mg0-Si0 2 -H 2 0 up to 30 GPa, 1600 degrees C: Applications to hydrous peridotite subducting into the Earth's deep interior. J Geophys Res 109:Art. No. B03206 Konzett J, Sweeney RJ, Thompson AB, Ulmer P (1997) Potassium amphibole stability in the upper mantle: an experimental study in a peralkaline KNCMASH system to 8.5 GPa. J Petrol 38:537-568 Konzett J, Ulmer P (1999) The stability of hydrous potassic phases in lherzolitic mantle- an experimental study to 9.5 GPa in simplified and natural bulk compositions. J Petrol 40:629-652 Konzett J, Fei Y (2000) Transport and storage of potassium in the Earth's upper mantle and transition zone: an experimental study to 23 GPa in simplified and natural bulk compositions. J Petrol 41:583-603 Kushiro I (1970) Stability of amphibole and phlogopite in the upper mantle. Carnegie Inst Wash Year Book 68: 245-247 Kushiro I (1987) A petrological model of the mantle wedge and lower crust in the Japanese island arcs. In: Magmatic Processes: Physiochemical Principles. Geochem Soc Spec Publ 1. Mysen BO (ed) The Geochemical Society, p 165-181 Lambert IB, Wyllie PJ (1968) Stability of hornblende and a model for the low velocity zone. Nature 215:12401241 LaTourrette T, Hervig RL, Holloway JR (1995) Trace element partitioning between amphibole, phlogopite and basanite melt. Earth Planet Sci Lett 135:13-50 Leake BE, Woolley AR, Arps CES, Birch WD, Gilbert MC, Grice JD, Hawthorne FC, Kato A, Kisch HJ, Krivovichev VG, Linthout K, Laird J, Mandarino JA, Maresch WV, Nickel EH, Rock NMS, Schumacher JC, Smith DC, Stephenson NCN, Ungaretti L, Whittaker EJW, Guo YZ (1997) Nomenclature of amphiboles: Report of the subcommittee on amphiboles of the International Mineralogical Association, commission on new minerals and mineral names. Am Mineral 82:1019-1037 López Sánchez-Vizcaíno V, Trommsdorff V, Gómez-Pugnaire MT, Garrido CJ, Müntener O, Connolly JAD (2005) Petrology of titanian clinohumite and olivine at the high-pressure breakdown of antigorite serpentinite to chlorite harzburgite (Almirez Massif, S. Spain). Contrib Mineral Petrol 149:627-646 Luth RW (1995) Is phase A relevant to the Earth's mantle? Geochim Cosmochim Acta 59:679-682 Luth RW (1997) Experimental study of the system phlogopite-diopside from 3.5 to 17 GPa. Am Mineral 82: 1198-1209 Lykins RW, Jenkins DM (1992) Experimental determination of pargasite stability relations in the presence of orthopyroxene. Contrib Mineral Petrol 112:405-413 McCammon CA, Griffin WL, Shee SR, O'Neill HSC (2001) Oxidation during metasomatism in ultramafic xenoliths from Wesselton kimberlite, South Africa. Implications for the survival of diamond. Contrib Mineral Petrol 141:287-296. McCammon C, Kopylova MG (2004) A redox profile of the Slave mantle and oxygen fugacity control in the cratonic mantle. Contrib Mineral Petrol 148:55-68 Mclnnes BIA, Gregoire M, Binns RA, Herzig PM, Hannington MD (2001) Hydrous metasomatism of oceanic sub-arc mantle, Lihir, Papua New Guinea: petrology and geochemistry of fluid-metasomatized mantle wedge xenoliths. Earth Planet Sci Lett 188:169-183

Stability of Hydrous Mantle

Phases

269

McNeil AM, Edgar AD (1987) Sodium-rich metasomatism in the upper mantle: implications of experiments on the pyrolite-Na 2 0-rich fluid system at 950 °C, 20 kbar. Geochim Cosmochim Acta 51:2285-2294 Mancini F, Harlow GE, Cahill C (2002) The crystal structure and cation ordering of p h a s e - X - i K ^ J ^ M g ^ [Al,Cr]n)2Si207H2X: A potential K- and H-bearing phase in the mantle. Am Mineral 87:302-306 Matson DW, Muenow DW, Garcia MO (1986) Volatile contents of phlogopite micas from South African kimberlite. Contrib Mineral Petrol 93:399-408 Mengel K, Green DH (1989) Stability of amphibole and phlogopite in metasomatized peridotite under watersaturated and water-undersaturated conditions. In: Kimberlites and related rocks. Geol Soc Aust Spec Publ 14. Ross J (ed) Geological Society of Australia, p 571-581 Menzies MA, Murthy VR (1980) Nd and Sr isotope geochemistry isotope geochemistry of hydrous mantle nodules and their host alkali basalts: implications for local heterogeneities in metasomatically veined mantle. Earth Planet Sci Lett 46:323-334 Menzies MA, Rogers N, Tindle A, Hawkesworth CJ (1987) Metasomatic enrichment processes in lithospheric peridotites, an effect of asthenospheric-lithospheric interaction. In: Mantle Metasomatism. Menzies MA, Hawkesworth CJ (eds) Academic Press, p 313-361 Merrill RB, Wyllie PJ (1975) Kaersutite and kaersutite eclogite from Kakanui, New Zealand- water-excess and water-deficient melting to 30 kilobars. Geol Soc Am Bull 86:555-570 Michael PJ (1988) The concentration, behavior and storage of H 2 0 in the sub-oceanic upper mantle: implications for mantle metasomatism. Geochim Cosmochim Acta 52:555-566 Millhollen GL, Irving AJ, Wyllie PJ (1974) Melting interval of peridotite with 5.7 percent water to 30 kilobars. J Geol 82:575-587 Miyagi I, Matsubaya O, Nakashima S (1998) Change in D/H ratio, water content and color during dehydration of hornblende. Geochem J 32:33-48 Modreski PJ, Boettcher AL (1972) Stability of phlogopite + enstatite at high pressure: A model for micas in the interior of the Earth. Am J Sci 272:852-869 Murayama JK, Nakai S, Kato M, Kumazawa M. (1986) A dense polymorph of Ca 3 (P0 4 ) 2: a high pressure phase of apatite decomposition and its geochemical significance. Phys Earth Planet Int 44:293-303 Mysen BO, Boettcher AL (1975) Melting of a hydrous mantle: I Phase relations of natural peridotite at high pressures and temperatures with controlled activities of water, carbon dioxide, and hydrogen. J Petrol 16: 520-548 Mysen BO, Virgo D, Popp RK, Bertka CM (1998) The role of H 2 0 in Martian magmatic systems. Am Mineral 83:942-946 Niida K, Green DH (1999) Stability and chemical composition of pargasitic amphibole in MORB pyrolite under upper mantle conditions. Contrib Mineral Petrol 135:18-40 Obata M (1980) The Ronda peridotite: garnet-, spinel- and plagioclase-lherzolite facies and the P-T trajectories of a high-temperature mantle intrusion. J Petrol 21:533-572 Obata M, Morten L (1987) Transformation of spinel lherzolite to garnet lherzolite in ultramafic lenses of the Austridic crystalline complex, northern Italy. J Petrol 28:599-623 O'Hara MJ, Mercy ELP (1963) Petrology and pedogenesis of some garnetiferous peridotites. Trans Royal Soc Ed 65:251-314 O'Neill HStC, Rubie DC, Canil, D, Geiger CA, Ross CR, Seifert F, Woodland AB (1993) Ferric iron in the upper mantle and in transition zone assemblages: Implications for relative oxygen fugacities in the mantle. Geophys Monograph 74IUGG 14:73-89 O'Reilly SY, Griffin WL (2000) Apatite in the mantle: implications for metasomatic processes and high heat production in Phanerozoic mantle. Lithos 53:217-232 O'Reilly SY, Griffin WL (1988) Mantle metasomatism beneath western Victoria, Australia: I. Metasomatic processes in Cr-diopside lherzolites. Geochim Cosmochim Acta 52:433-447 Ohtani, E, Shibata T, Kubo T, Kato T (1995) Stability of hydrous phases in the transition zone and the upper most part of the lower mantle. Geophys Res Lett 22:2553-2556 Popp RK, Virgo D, Yoder HS, Hoering TC, Phillips MW (1995) An experimental study of phase equilibria and Fe oxy-component in kaersutitic amphibole: implications for the fm and amo in the upper mantle. Am Mineral 80:534-548 Righter K, Dyar MD, Delaney JS, Vennemann TW, Hervig RL, King PL (2002) Correlations of octahedral cations with OH~, O 2- , Cl~ and F~ in biotite from volcanic rocks and xenoliths. Am Mineral 87:142-153 Robinson P, Spear FS, Schumacher JC (1982) Phase relations of metamorphic amphiboles: natural occurrence and theory. Rev Mineral 9B: 1-227 Roden MF, Murthy R (1985) Mantle metasomatism. Ann Rev Earth Planet Sci 13:269-296 Ryabchikov ID, Boettcher AL (1980) Experimental evidence at high pressure for potassic metasomatism in the mantle of the Earth. Am Mineral 65:915-919 Sato K, Katsura T, Ito E (1997) Phase relations of natural phlogopite with and without enstatite up to 8 GPa: implication for mantle metasomatism. Earth Planet Sci Lett 146:511-526

270

Frost

Scambelluri M, Hermann J, Morten L, Rampone E (2006) Melt- versus fluid-induced metasomatism in spinel to garnet wedge peridotites (Ulten Zone, Eastern Italian Alps): clues from trace element and Li abundances. Contrib Mineral Petrol 151:372-394 Schmidt MW, Poli S (1998) Experimentally based water budgets for dehydrating slabs and consequences for arc magma generation. Earth Planet Sci Lett 163:361-379 Schmidt MW, Vielzeuf D, Auzanneau E (2004) Melting and dissolution of subducting crust at high pressures: the key role of white mica. Earth Planet Sci Lett 228:65-84 Schneider ME, Eggler DH (1986) Fluids in equilibrium with peridotite minerals: implications for mantle metasomatism. Geochim Cosmochim Acta 50:711-724 Schrauder M, Navon O (1994) Hydrous and carbonatitic mantle fluids in fibrous diamonds from Jwaneng Botswana. Geochim Cosmochim Acta 58:761-771 Seyler M, Mattson PH (1989) Petrology and thermal evolution of the Tinaquillo peridotite (Venezuela). Journal Geophys Res 94:7629-7660 Shaw CSJ, Eyzaguirre J (2000) Origin of megacrysts in the mafic alkaline lavas of the West Eifel volcanic field, Germany. Lithos 50:75-95 Shen AH, Keppler H (1997) Direct observation of complete miscibility in the albite-H 2 0 system. Nature 385: 710-712 Shieh SR, Mao HK, Hemley RJ, Ming LC (1998) Decomposition of phase D in the lower mantle and the fate of dense hydrous silicates in subducting slabs. Earth Planet Sci Lett 159:13-23 Shimizu N (1975) Rare earth elements in garnets and clinopyroxenes from garnet lherzolite nodules in kimberlites. Earth Planet Sci Lett 25:26-32 Sweeney RJ, Thompson AB, Ulmer P (1993) Phase relations of a natural MARID composition and implications for MARID genesis, lithospheric melting and mantle metasomatism. Contrib Mineral Petrol 115:225-241 Sudo A, Tatsumi Y (1990) Phlogopite and K-amphibole in the upper mantle: implication for magma genesis in subduction zones. Geophys Res Lett 17:29-32 Tatsumi Y, Hamilton DL, Nesbitt RW (1986) Chemical characteristics of fluid phase released from a subducted lithosphere and origin of arc magmas: evidence from high pressure experiments and natural rocks. J Volcanol Geotherm Res 29:293-310 Taylor WR, Green DH (1988) Measurement of reduced peridotite-C-O-H solidus and implications for redox melting of the mantle. Nature 332:349-352 Thibault Y, Edgar AD, Lloyd EE (1992) Experimental investigation of melts from a carbonated phlogopite lherzolite: implications for metasomatism in the continental lithospheric mantle. Am Mineral 77:784-794 Tiepolo M, Vannucci R, Oberti R, Foley S, Bottazzi P, Zanetti A (2000) Nb and Ta incorportation and fractionation in titanian pargasite and kaersutite: crystal-chemical constraints and implications for natural systems. Earth Planet Sci Lett 176:185-201 Thompson AB (1992) Water in the Earth's upper mantle. Nature 358:295-302 Thompson JB, Laird J, Thompson AB (1981) Reactions in amphibolite, greenschist and Blueschist. J Petrol 23: 1-27 Tr0nnes RG, Edgar AD, Arima M (1985) A high pressure-high temperature study of T i 0 2 solubility in Mg-rich phlogopite: implications to phlogopite chemistry. Geochim Cosmochim Acta 49:2323-2329 Tr0nnes RG (2002) Stability range and decomposition of potassic richterite and phlogopite end members at 5-15 GPa. Mineral Petrol 74:129-148 Ulmer P, Trommsdorff V (1999) Phase relations of hydrous mantle subducted to 300 km. In Mantle Petrology: Field Observations and High-pressure Experimentation. Fei Y, Bertka CM, Mysen BO (ed) Geochemical Society, Special Publication No. 6, p 259-281 Ulmer P (2001) Partial melting in the mantle wedge - the role of H 2 0 in the genesis of mantle-derived 'arcrelated' magmas. Phys Earth Planet Int 127:215-232 van Achterbergh E, Griffin WL, Stiefenhofer J (2001) Metasomatism in mantle xenoliths from the Letlhakane kimberlites: estimation of element fluxes. Contrib Mineral Petrol 141:397-414 Wagner C, Deloule E, Mokhtari A (1996) Richterite-bearing peridotites and MARID-type inclusions in lavas from North Eastern Morocco: mineralogy and D/H isotopic studies. Contrib Mineral Petrol 124:406-421 Wallace ME, Green DH (1988) An experimental determination of primary carbonatite magma composition. Nature 335:343-346 Wallace ME, Green DH (1991) The effect of bulk rock composition on the stability of amphibole in the upper mantle: implications for solidus positions and mantle metasomatism. Mineral Petrol 44:1-19 Wallace PJ (1998) Water and partial melting in mantle plumes: Inferences from the dissolved H 2 0 concentrations of Hawaiian basaltic magmas. Geophys Res Lett 25:3639-3642 Waters FG (1987) A suggested origin of MARID xenoliths in kimberlites by high pressure crystallization of an ultrapotassic rock such as lamproite. Contrib Mineral Petrol 95:523-533 Watson B E (1980) Apatite and phosphorus in mantle source regions: an experimental study of apatite/melt equilibria at pressures to 25 kbar. Earth Planet Sci Lett 51:322-335

Stability of Hydrous Mantle

Phases

271

Weiss M (1997) Clinohumites: a field and experimental study. ETH Dissertation No. 12202, 168 pp. Wilkinson JFG, Le Maitre RW (1987) Upper mantle amphiboles and micas and Ti0 2 , K 2 0 and P 2 O s abundances and 100Mg/(Mg+Fe2+) ratios of common basalts and andesites: implications for modal mantle metasomatism and undepleted mantle compositions. J Petrol 28:37-73 Wilshire HG, Shervais JW (1975) Al-augite and Cr-diopside ultramafic xenoliths in basaltic rocks from Western United States. Phys Chem Earth 9:257-276 Witt-Eickschen G, Seek HA, Reys CH (1993) Multiple enrichment processes and their relationships in the subcrustal lithosphere beneath the Eifel (Germany). J Petrol 34:1-22 Witt-Eickschen G, Kramm U (1998) Evidence for the multiple stage evolution of the subcontinental lithospheric mantle beneath the Eifel (Germany) from pyroxenite and composite pyroxenite/peridotite xenoliths. Contrib Mineral Petrol 131:258-272 Woermann E, Rosenhauer M (1985) Fluid phases and the redox state of the earths mantle. Fortschr Mineral 63: 263-349 Wood BJ, Bryndzia LT, Johnson KE (1990) Mantle oxidation state and its relationship to tectonic environment and fluid speciation. Science 248:337-345 Woodland AB, Kornprobst J, McPherson E, Bodinier J-L, Menzies MA (1996) Metasomatic interactions in the lithospheric mantle: petrologic evidence from the Lherz massif, French Pyrenees. Chem Geol 134:83-112 Woodland AB, Koch M (2003) Variation in oxygen fugacity with depth in the upper mantle beneath the Kaapvaal craton, South Africa. Earth Planet Sci Lett 214:295-310 Wyllie PJ (1978) Mantle fluid compositions buffered in peridotite-C0 2 -H 2 0 by carbonates, amphibole, and phlogopite. J Geol 86:687-713 Yang HX, Konzett J, Prewitt CT (2001) Crystal structure of Phase X, a high pressure alkali-rich hydrous silicate and its anhydrous equivalent. Am Mineral 86:1483-1488 Yang H-J, Frey FA, Clague DA (2003) Constraints on the source components of lavas forming the Hawaiian North Arch and the Honolulu volcanics. J Petrol 44:603-627 Yaxley GM, Green DH, Kamenetsky V (1998) Carbonatite metasomatism in the southeastern Australian lithosphere. J Petrol 39:1917-1930 Yoder HS, Kushiro I (1969) Melting of a hydrous phase: phlogopite. Am J Sci 267A: 558-582 Young ED, Virgo D, Popp RK (1997) Eliminating closure in mineral formulae with specific application to amphiboles. Am Mineral 82:790-806 Zanetti A, Mazzucchelli M, Rivalenti G, Vannucci R (1999) The Finero phlogopite-peridotite massif: an example of subduction-related metasomatism. Contrib Mineral Petrol 134:107-122

12

Reviews in Mineralogy & Geochemistry Vol. 62, pp. 273-289, 2006 Copyright © Mineralogical Society of America

Hydrous Phases and Water Transport in the Subducting Slab Tatsuhiko Kawamoto Institute for Geothermal Sciences Graduate School of Science Kyoto University Beppu 874-0903, Japan e-mail: kawamoto @bep.

vgs.kyoto-u.ac.jp

INTRODUCTION Arc volcanoes are typically located 90-180 km above the surface of downgoing slabs, as shown by Wadati-Benioff deep seismic foci (Gill 1981; Tatsumi 1989). The intimate relationship between the dip angles of the subducting slab and the locations of volcanic arcs indicates that subduction zone magmatism is triggered by material input from the subducting slab (Tatsumi and Eggins 1995). The slab-derived components are thought to be aqueous fluids or H 2 0-rich partial melts of subducted oceanic crust. Therefore, knowledge of the stability of hydrous phases and the chemical and physical properties of aqueous fluids in downgoing slabs is essential to understand the material transport in subduction zones. In this section, I will review the stability of hydrous phases in downgoing peridotite, basalt and sediment systems, and the chemical and the wetting properties of aqueous fluids. Recent experimental studies indicate that 3-4 GPa, equivalent to 90-120 km depth, is a key pressure, where (1) the chemical compositions of silicate components dissolved in aqueous fluids equilibrated with mantle minerals approach the composition of mantle peridotite itself (Stalder et al. 2001; Mibe et al. 2002; Kawamoto et al. 2004), (2) the dihedral angle between olivine and aqueous fluids starts becoming smaller than 60° (Watson et al. 1990; Mibe et al. 1998, 1999), and (3) the immisciblity gap between peridotitic melts and aqueous fluids disappears and consequently hydrous minerals liberate supercritical aqueous fluids (Mibe et al. 2004a, 2006). The similarity between these pressures and the depths of downgoing slab underneath volcanic fronts, where the maximum numbers of volcanoes are formed, 124 ± 38 km (Gill 1981) or 112 ± 19 km (Tatsumi 1986), suggests that subduction zone magmatism can be triggered by the input of supercritical fluids from the downgoing peridotite and basalt.

LOW-PRESSURE HYDROUS MINERALS AND HIGH-PRESSURE HYDROUS PHASES Many hydrous crystalline phases are stable in peridotite, basalt and sediment systems over a wide range of pressure. Their chemical formulae and H 2 0 contents are summarized together with those of nominally anhydrous minerals in Table 1. Some hydrous phases have been found only in high-pressure and high-temperature experimental products and have not yet been found in nature: dense hydrous magnesium silicates (DHMS) or alphabet phases (Ringwood and Major 1967), phase Egg (Eggleton et al. 1978), phase Pi (Wunder et al. 1993a), topaz-OH (Wunder et al. 1993b), and 8-AlOOH (Suzuki et al. 2000). Although phase D, F, and G were originally suggested as different phases, these phases seem to be identical (Frost 1999; Ohtani et al. 2001). The chemical compositions of DHMS are plotted in Figure 1 with the estimated 1529-6466/06/0062-0012505.00

DOI: 10.213 8/rmg.2006.62.12

Kawamoto

274

H20

MgC

Si02

Figure 1. Compositions of hydrous minerals and dense magnesium hydrous silicates stable in peridotite system plotted with compositions of silicates dissolved into aqueous fluids coexisting with forsterite and enstatiteat 1100 °C at 1-10 GPa estimated by Zhang and Frantz (2000) andMibe etal. (2002) in theMgOS i 0 2 - H 2 0 system. Phase D, E, antigorite, and 10 A phase are non-stoichiometric phases. Humite is located between chondrodite and clinohumite. Abbreviations are in Table 1.

chemistry of aqueous fluids equilibrated with forsterite + enstatite in the M g 0 - S i 0 2 - H 2 0 system (Mibe et al. 2002). The hydrous crystalline phases can be divided into three major groups with respect to their stability range (Fig. 2): (1) low-pressure hydrous minerals such as chlorite (clinochlore), talc, and amphibole (the relevant end members are listed in Table 1), which are commonly observed in metamorphic rocks, (2) high-pressure hydrous phases such as DHMS (Fig. 1), K-richterite, topaz-OH, and phase Egg, and (3) middle-pressure hydrous minerals such as phlogopite, antigorite, Mg-sursassite and 10 A phase in peridotite, lawsonite in basalt, and phengite in sediment. The last group is stable between 5 and 7 GPa, and may be important for delivering H 2 0 from low-pressure hydrous minerals to high-pressure hydrous phases (Fig. 2). Liu (1987) recognized that phase A, a DHMS, can accommodate much more water than amphibole or phlogopite. Therefore he emphasized the important reaction forsterite + H 2 0 = phase A + enstatite, and he described this reaction boundary as a "water-line," implying that a region deeper than the water-line can be a H 2 0 reservoir in the mantle. In Figure 2, the water-line is shown by the low-pressure stability of DHMS. Kawamoto et al. (1996) identified the presence of a "choke point" in a down going slab. A choke point represents a pressure and temperature condition along a PT path where low-pressure and middle-pressure hydrous minerals get dehydrated at certain pressure conditions and cannot deliver H 2 0 to high-pressure hydrous phases (Fig. 2). The choke point curve, the curve connecting the array of choke points, represents the high-pressure and high-temperature stability limit of the lowpressure and middle-pressure hydrous minerals. In the M g 0 - S i 0 2 - H 2 0 system, the invariant point composed of antigorite, phase A, enstatite, forsterite and H 2 0 represents the lowest temperature and highest pressure of the choke point. In recent literature, this point is at around 6.2 GPa and 620 °C (Iwamori 2004), and at 5.1 GPa and 550 °C (Komabayashi et al. 2005). In Figure 2, based on the KLB-1 peridotite data, T i 0 2 stabilizes chondrodite and clinohumite. Therefore, in the peridotite systems, the PT1 conditions where antigorite meets chondrodite and clinohumite represent the lowest temperature and highest pressure choke point. In the MgO-

Hydrous

Phases & Water Transport in the Subducting

400

600

Slab

275

Temperature (°Q 800 1000 1200 1400 1600

2 4 6

100

200

8

p 1 £

12

14 16 18

500

20 600 22 24 700 26 Figure 2. Pressure and temperature diagram showing stability of hydrous minerals/phases in peridotite (Kawamoto 2004a) with some hydrous phases in basalt/sediment systems. The wet solidus is from Kawamoto and Holloway (1997). Since a second critical endpoint between peridotite melt and aqueous fluids is located at around 3.8 GPa (Mibe et al. 2004a; 2006), the wet solidus is drawn by dashed line at pressures higher than 4 GPa. Stability of lawsonite in basalt is indicated by solid dots and stability boundaries among phengite, topaz-OH, and phase Egg is drawn by open dots, respectively. The stabilities of Par, Chi, Talc, Atg, Phi, K-rich, Lws, Top, Eg are after Schmidt and Poli (1998), Pawley (2003), Ulmer and Trommsdorff (1995), Sudo and Tatsumi (1990), and Ono (1998); phase boundaries among Ol, Ol + Wd, Wd, and Wd + Rg - (Mg 0 9 Fe 0 1 ) 2 Si0 4 and Rg - (Mg 0 9 Fe 0 1 ) 2 Si0 4 and Mg-perovskite (Mg-Pv) + magnesium wiistite (Mw) in dry conditions are after Katsura and Ito (1989), and Ito and Takahashi (1989), respectively. The phase boundary of Hy- wd and Hy-rg (dashed line) is at higher pressure than under dry conditions. The 60° isopleths of the dihedral angle in garnet-garnet-fluid (gt-fl) and olivine-olivine-fluid (ol-fl) are also shown (thick gray line). The data of the dihedral angle are compiled in Figure 5. HT and LT represent PT paths of high-temperature and low-temperature subducting slab surface, respectively (Peacock and Wang 1999). Abbreviations are in Table 1.

276

Kawamoto

Table 1. Formula of hydrous minerals/phases and nominally anhydrous minerals in metamorphic basalts, metamorphic sediments, and peridotite (after Wunder and Schreyer 1992, Pawley and Wood 1995, Mysen et al. 1998, Ono 1999, Forneris and Holloway 2003). Name

Symbols

(Amphibole groups) Tremolite

Formula

wt% H2C

Trm

Ca2Mg5Sis022(0H)2

2.2

Pargasite Barroisite Glaucophane

Par Bar Gin

Na 2 Ca 3 Mg 8 FeAl 3 Si 1 3 044(0H)4 NaCaMg 3 Al 2 Si 7 A10 2 2 (0H) 2 Na 2 Mg 3 Al 2 Si s 0 2 2 (0H) 2

2.2 2.3 2.3

K-richterite

K-ric

K L 9 Ca L 1 Mg 5 Si 7 . 9 Al 0 . 1 O 2 2 (OH) 2

2.1

(Peridotite system) Chlorite Talc

Chi Tic

(Mg 5 Al)(AlSi 3 )O 10 (OH) 8 Mg 6 Si s O 2 0 (OH) 4

13 4.8

Serpentine Antigorite Clinohumite

Serp Atg Chm

Mg3Si205(0H)4 Mg 4 8 Si 3 4 O s 5 (OH) 6 2 M g 9 S i 4 0 1 6 ( 0 H ) 2 , Ti 0 5 Mg 8 5 S i 4 0 1 7 ( 0 H )

13 12.3

Humite Chondrodite Norbergite

Hm Chn Nor

Mg7Si3012(0H)2 Mg 5 Si 2 O s (OH) 2 , Ti 0 5 Mg 4 5 Si 2 O 9 (OH) Mg 3 S10 4 (0H) 2

Phase A Brucite Phase B

A Br B

Mg7Si208(0H)6 Mg(OH) 2 Mg24Sis03s(0H)4

Superhydrous B Anhydrous B Phase E

sB AhyB E

Mg 1 0 Si 3 O 1 4 (OH) 4 Mg14Si5024 Mg 2 . 2 7 Si L 2 6 H 2 4 0 6

11.4

Phase D/F/G Anthophyllite

D Ant

MgSi2H206 Mg7Si8022(0H)2

10.1 2.3

Talc 10 A phase Mg-sursassite

Talc 10 Â MgS

Mg 3 Si 4 O 1 0 (OH) 2 Mg 3 Si 4 O 1 0 (OH) 2 XK20 Mg 5 Al 5 Si 6 0 2 1 (0H) 7

Hydrous wadsleyite Hydrous ringwoodite

Hy-wd Hy-rg

Mg L 7 5 SiO 4 (OH) 0 . 5 Mg L 7 5 SiO 4 (OH) 0 . 5

Zo / Czo Sta

Ca 2 Al 3 Si 3 0 1 2 (0H) (Mg,Fe) 2 (Al,Fe) 9 Si 4 0 2 2 (0,0H) 2

Apatite Sphene Phlogopite

Ap Spn Phi

Ca 5 (P0 4 ) 3 (0H,F,Cl) CaTiSi04(0,0H,F) KMg 2 Si 3 AlO 1 0 (OH) 2

1.8 1.5 4.8

Phase Egg Topaz-OH Phase Pi

Eg Top Pi

AlSi03(0H) Al2Si04(0H)2 Al3Si207(0H)3

7.5 10.0 9.0

Lawsonite Chloritoid

Lws Cid

CaAl2Si207(0H)2 H 2 0 (Mg, Fe) 2 (Al,Fe) 4 Si 2 O 1 0 (OH) 4

11.5 8

Phengite Ö-AIOOH

Phe S-Al

K(Al 2 - x Mg x )(Si 3+x Al 1 _ x )O 10 (OH,F) 2 AlOOH

4.6 15

(Basalt and sediment systems) Zoisite/clinozoisite Staurolite

(Nominally anhydrous minerals) Olivine/Wadsleyite/Ringwoodite

Ol/Wd/Rg

Mg 2 S10 4

Clinopyroxene Ca-perovskite Orthopyroxene/ Majorité/ Akimotoite/ Perovskite Quartz/ Coesite/ Stishovite

Cpx Ca-pv Opx/Mj/ Ak/Pv Qz / Coe / St

(Na,Ca)(Mg,Al)Si 2 0 6 CaSi03 MgSi03

Spinel Garnet

Sp Gt

MgAl204 (Fe,Mg,Ca) 3 Al 2 Si 3 0 1 2

Si02

2 . 9 - 1.4 3.75 5.3-2.6 9.0 11.8 30.9 2.4 1.6

4.75 7 . 6 - 13 7.2 3.3 3.3

2 2

Hydrous

Phases & Water Transport

in the Subducting

Slab

Til

AI2O3-SÌO2-H2O system, Mg-sursassite (Gottschalk et al. 2000), which was previously called MgMgAl-pumpellyite (for example, Domanik and Holloway 1996), is stabilized at higher temperature than this invariant point (Fig. 3; Bromiley and Pawley 2003), and its presence therefore increases the temperature of the choke point. The transition zone (410-660 km depth) is also characterized by the high H 2 0 storage capacity of hydrous wadsleyite and hydrous ringwoodite (Fig. 3; Smyth 1987; Inoue et al. 1995; Kawamoto et al. 1996; Kohlstedt et al. 1996; Kudoh et al. 1996; Smyth and Kawamoto 1997; Smyth et al. 1997; Demouchy et al. 2005). Therefore, the transition zone could play a significant role as a large H 2 0 -reservoir formed by crystallization of hydrous wadsleyite and ringwoodite from a hydrous magma ocean. Kawamoto and Holloway (1997) measured the partition coefficient of H 2 0 between hydrous wadsleyite/ringwoodite and hydrous partial melts of peridotite, and suggested the possible existence of a hydrous transition zone in the early history of the Earth. Upwelling from such a hydrous reservoir could generate partial melting at 410 km and produce komatiitic magmas. Through partial melting of a hydrous transition zone, in this hypothesis, the transition zone has been getting drier during the geological time, because the choke point prevents H 2 0 from subducting into the transition zone. Therefore the present transition zone has much less ability to produce komatiite magmas. This hypothesis thus explains why komatiites were produced mainly in the Archean period.

STABILITY OF HYDROUS PHASES IN DOWNGOING PERIDOTITE There are two potentially-hydrated peridotite layers in subduction zones. One is the harzburgite/lherzolite of the subducting lithospheric mantle, which is overlain by oceanic basaltic crust and sediments. The other is downdragged mantle at the base of the mantle wedge. To what extent the peridotite layers are hydrated remains uncertain. Along transform faults, serpentine minerals (antigorite, lizardite, chrysotile) can be formed by seawater alteration. However, the rest of the subducting lithospheric mantle may not be hydrated. The downdragged mantle peridotite at the base of the mantle wedge should be hydrated by aqueous fluids liberated by dehydration reactions of hydrous minerals in downgoing sediment and basalt layers. Nicholls and Ringwood (1973) suggested that subducting basalt will be almost dry beneath the fore-arc region. Sakuyama and Nesbitt (1986), therefore, suggested that downdragged peridotite in the mantle wedge will be hydrated through H 2 0 released by dehydration of the hydrous minerals in the basaltic layer and may carry H 2 0 beneath the volcanic arc. Iwamori (2004) compiled the stability of hydrous phases in the Mg0-Si0 2 -H 2 0, the Mg0-Al 2 0 3 -Si0 2 -H 2 0, and KLB-1 peridotite systems, and presented the distribution of maximum H 2 0 contents bound in mantle peridotite (Fig. 3). Komabayashi et al. (2004) also presented a similar stability diagram of hydrous phases based on Schreinemakers' net analysis. They noticed two main differences of hydrous phase stability between the peridotite system and simple systems: (1) the addition of A1 2 0 3 expands the stability field of phase E to the lower pressures and (2) the addition of Ti0 2 enhances the stability field of clinohumite and chondrodite (Fig. 2). The addition of fluorine is also found to expand the stability of clinohumite into a lower pressure range (Stalder and Ulmer 2001). According to Fumagalli et al. (2001), thelO A phase (Table 1) is reported to be stable in the peridotite system at 5.2 GPa and 680 °C. Fumagalli and Poli (2005) found that the 10 À phase has high A1 2 0 3 contents (about 10 wt%) and suggested that this phase is a mixed layer of chlorite and pure 10 A phase formed in the Mg0-Si0 2 -H 2 0 system. The stability field of this Al-rich 10 A phase is close to the stability of Mg-sursassite (Bromiley and Pawley 2002). These phases cover some regions of the choke point (Fig. 3), though the H 2 0 content contributed by Mg-sursassite and Al-rich 10 A phase to peridotite is limited to 0.7 (Iwamori 2004; Fig. 3) and 1 wt% (Fumagalli and Poli 2005), respectively.

278

Kawamoto

:

.

o

E

i

> CT" Ol > "H -H joj

i

>

^

r

t

: £ + -H E • \ + I - g 03 HH + . ™ m L II) î»

> > r

+ ï -P i E P + + O Q + + m m (/in

f f i

ï

i

p

\ > on

I O o + o 7+ > « û. û. • + + + m > a> Kl û . Û .

Z> Ì + +

00

r**

60° to 60

: >60 °

4

5

1 11

i1

1 11

i1

1 11

,0.--*ol - fl

60°. These would get dragged down in the induced flow, and this process would be repeated until the fluid reaches the zone of partial melting.

SECOND CRITICAL ENDPOINT BETWEEN MAGMAS AND AQUEOUS FLUID: IMPLICATIONS FOR SLAB-DERIVED COMPONENT Simple silicate melts and aqueous fluids can mix completely under certain FT conditions (Fig. 6A,B). At pressure conditions equivalent to the Earth's upper mantle, silicate melts and aqueous fluids cannot be distinguished from each other at the temperature-pressure conditions beyond a second critical endpoint, where a critical temperature meets its wet solidus (Kennedy et al. 1962; Paillat et al. 1992; Shen and Keppler 1997; Bureau and Keppler 1999). Following the visual demonstration of the complete mixing between albite melt and H 2 0 (Shen and Keppler 1997), Bureau and Keppler (1999) reported complete miscibility between aqueous fluids and K 2 0-bearing nepheline melt, pure jadeite melt, haplogranitic melt, Ca-bearing haplogranitic melt and dacite in the Si0 2 -Al 2 0 3 -Na 2 0-K 2 0-Ca0-Mg0 system. Sowerby and Keppler (2002) demonstrated complete miscibility between B 2 0 3 - F enriched albite melt or pegmatite and H 2 0

Hydrous

Phases & Water Transport

(A)

dry solidus

Tc

a) a.

fluid

/melt \ melt / +A fluid\ absent solidus / melt A wet solidus 1/ + H + 3 H o fluid < D 0) « +H

/ fluid / +melt

H,0

H

A

H,0

H

A

H,0

H

m -a < «D « c 60° to - — — exp %n=02n + \

- D (2n +1)2 7l2 t

(2n + l) 71 x 2L

(5)

294

Ingrin

&

Blanchard

where the origin is located at the mid-point of the slab. If the measurements provide the average concentration rather than the concentration profile like in mass-loss experiments on single-crystals or powder, then Equation (5) has to be integrated over the sample thickness. The solution is then

c{t)~ca

1

Cj - C0

(2n + l)'

- D (2n + l) 2 7ï2 t

• exp

(6)

4 L

For the same kind of geometry, if the sample is not thin enough to assume unidirectional diffusion, Equation (5) becomes C(x,y,z,t)

-C0

C - C

(-ir

(V)

n 3 i ^ T o i S i 2 1 +1 ){2m + 1 X 2 « +1) -7Ï2 t D A 2 l L

exp

+

rf I M

\

(21 +1) 71 x 2L

Î

cos

(2m

2 w

+ 1

+ l ) 71

y

)

|

gz(2W

\

Î

cos

2a

+

l) 2

(2W + 1)tiz 2b

This corresponds to the case of a 2a x 2b x 2L parallelepiped where the diffusion coefficients along the three crystallographic directions are different. As before, the expression of the average concentration is obtained by integrating this equation over the volume analyzed. By integrating over the whole volume (2a x 2b x 2L), the average concentration is expressed as follows: 1 Cl

CQ

rc exp

i 5 „ o , o ( 2 / + l) (2m + \)(2n -7l2 t

+

(8)

+ \)

, P y ( 2 m + 1)

|

D z ( 2 n + l) 2

The corresponding solutions for a spherical geometry (radius, R) are, respectively, — - — - — - = i + — y ^—

ex

P

-Dn2n2

F /

^ i i k ^ - ^ e x p C0

t

^w 71 r ^ v

(9)

y

„ ? ? \ J2

(10)

MEASUREMENT TECHNIQUES In this section, we describe briefly the techniques of analysis that have been reported in the literature for measuring hydrogen diffusion (infrared spectroscopy, mass spectrometry, nuclear reaction analysis, thermogravimetry, scintillation counting) as well as some other new techniques that have not been used for diffusion measurements but are very promising like proton-proton scattering. More details on analytical methods used for measuring water in minerals can be found in this volume (Rossman 2006). We end this section with a short review of the theoretical simulations that contribute to our understanding of the diffusion mechanisms.

Diffusion

of Hydrogen

in

Minerals

295

Infrared spectroscopy Infrared spectroscopy is the most frequently used method to measure hydrogen diffusion. The vibrational modes of the OH dipole within the sample interact with the infrared beam and give rise to absorption bands. The concentration of OH is directly related to the intensity of the bands, and the concentration can be determined if the spectra are measured accurately; IR spectra of anisotropic minerals should be measured in polarized mode (cf. Libowitzky and Rossman 1996a) and the relation between absorption and concentration must be calibrated against an independent hydrogen analysis method. The position in wavenumber of the absorption band depends on the strength of the hydrogen bond, bond geometry and neighbors. Therefore polarized spectra also provide information about the structure of OH (Libowitzky and Beran 2006). Advantages of the IR technique are very high sensitivity (< ppm H 2 0), its ability to distinguish hydroxyl ions from adsorbed and intrinsic water molecules, and the ability to distinguish OH appearing in inclusions of hydrous phases from OH structurally in the parent phase (e.g., amphibole lamellae in clinopyroxene, Ingrin et al. 1989; Skogby and Rossman 1989). For diffusion experiments only the relative change of hydrogen content is necessary and no independent calibration is required. This technique can be used for measurements of the average concentration of hydrogen and deuterium in mass-loss experiments or measurements of the diffusion profiles. When the spectrometer is equipped with a microscope, measurements with a spatial resolution better than 50 )im can be easily achieved. Sample preparation, for profile measurements is described in Figure 1 and an example of a profile measurement is shown in Figure 2 for a garnet sample that has undergone partial hydrogen extraction in air.

Figure 1. Sample preparation for the measurement of diffusion profiles in a single-crystal by infrared spectroscopy.

296

Ingrin

&

Blanchard

distance (mm) Figure 2. Infrared spectra recorded through a single-crystal of grossular partially dehydrogenated (a) and corresponding fits of the normalized H concentration profile assuming constant diffusion coefficients (b). (after Kurka 2005).

Mass spectrometry Mass spectrometry techniques offer efficient means of measuring hydrogen isotope exchange (Graham 1981). The main disadvantage is that it is a destructive technique and has relatively poor spatial resolution. The sample is ionized. The ions of differing masses are separated and their relative abundances are recorded by measuring the intensities of ion flux (Graham et al. 1980). Secondary ion mass spectrometry (SIMS) represents an improved method by measuring directly hydrogen-deuterium exchange profiles over a few microns in length (Vennemann et al. 1996). This technique involves bombarding the sample surface with 160~ primary ion beam.

Diffusion

of Hydrogen

in

Minerals

297

The secondary ions emitted from the sample are then measured with the mass spectrometer. Because the primary beam erodes the surface, a depth profile can be obtained. Vennemann et al. (1996) and Suman et al. (2000) report measurements of H-D diffusion profiles in hornblende, epidote and pargasite done by this method. Thermogravimetry A novel thermogravimetric method has been developed to study H 2 0 - D 2 0 exchange in lawsonite (Marion et al. 2001). The mass difference between hydrogen and deuterium is sufficient to monitor the global weight change due to H-D exchange with a thermobalance. This simple experimental setup has proved to be efficient and accurate for hydrous minerals but cannot apply to minerals with low water content such as nominally anhydrous minerals. Nuclear reaction analysis Nuclear reaction analysis allows measurements of hydrogen or deuterium diffusion profiles (Dersch et al. 1997). To detect the hydrogen, the surface sample is irradiated with 15N with variable energies in order to produce the following nuclear reaction under resonant conditions: 15

N + ! H h> 12C + a + y

The number of y-rays emitted at any incident energy is proportional to the hydrogen concentration at the respective depth. To contribute to the nuclear reaction, the 15N nuclei have to slow down in the sample to reach the resonance energy (6.385 MeV). Thus each initial kinetic energy corresponds to a depth in the sample where the reaction occurs. The hydrogen diffusion profile is obtained by measuring the yield of the characteristic reaction y-rays versus the beam energy. To probe the deuterium, a 3 He beam with a fixed energy is used. The depth of the deuterium atom in the sample is known from the energy of the proton emitted by the reaction: 3

He + D H> 4 He + p

Ion-beam analysis displays a depth resolution of few nanometers decreasing slightly with the depth while the size of the beam is of the order of a millimeter. This technique has been used to investigate hydrogen diffusion in quartz (e.g., Dersch et al. 1997; Dersch and Rauch 1999). Figure 3 shows an example of a deuterium concentration profile measured in quartz.

13.0

Energy (MeV) 13.5

14.0

14.5

200 300 400 Depth (nm)

_L_i

800

850

Channel

500

u.

600

900

Figure 3. Proton spectrum from the analysis of a deuterated quartz sample with a 700 keV 3 He beam. The inset shows the depth profile determined from the spectrum. [Used with kind permission of SpringerVerlag from Dersch and Rauch 1999).

298

Ingrin & Blanchard

Liquid scintillation counting Tritium is sometimes used as a tracer of hydrogen diffusion (e.g., Shaffer et al. 1974). This hydrogen isotope, 3 H, is radioactive and its concentration is measured by liquid scintillation counting. Tritium transforms by beta decay into stable helium (half life = 12.3 years). This analysis technique involves the detection of beta decay within the sample via capture of beta emissions in a scintillation cocktail containing organic solvents and solutes, fluor for instance. Beta particles emitted from the sample excite the solvent molecules, which in turn transfer the energy to the solute. The excited solute molecules dissipate the energy by emitting photons, which can be detected via a photomultiplier tube within a scintillation counter. The low-energy beta particles emitted from tritium have a small penetration length (~1 |im). This makes liquid scintillation counting a surface technique and depth resolution is ~1 jim. Tritium concentration profiles can be determined by performing successive steps of grinding and analysis. After each grinding and polishing event, the single crystal is immersed into the cocktail with its back face masked. Analyzing the activity of the removed material can be used to check the results provided by this procedure. Proton-proton scattering A new analytical procedure has recently been developed to measure hydrogen depth profiles by elastic proton-proton scattering (Wegden et al. 2005). A beam of 2.8 MeV protons (5-10 |im in diameter) in normal incidence is scanned over the sample. The scattered proton and recoiled target proton are detected coincidentally in the forward direction with an annular surface barrier detector. The summed energy of every detected proton-proton pair and the difference in their energy is used in an indirect approach to determine the depth location for every hydrogen event. The depth resolution is on the order of a micrometer. The major advantages of this method are the high detection cross section and the lowest possible irradiation damage effects compared to other ion-beam techniques (e.g., nuclear reaction analysis). However, sample preparation represents this technique's main disadvantage. The sample must be thinner than 10-15 |im so that proton pairs from the entrance surface can travel through the whole sample thickness. Thus, diffusion profiles can only be determined to a limited depth. This technique has been tested for several nominally anhydrous minerals with hydrogen concentrations of 10 to 100 ppm H 2 0 (Wegden et al. 2005) but has not been applied yet to diffusion studies. Theoretical techniques Many theoretical studies of the migration of protons (H+), neutral hydrogen atoms (H°) or hydrogen molecules (H2) in various materials are reported in the literature. Beyond providing important information on the microscopic mechanisms involved in hydrogen diffusion, these studies have industrial applications such as, for example, the trapping-detrapping of incident hydrogen in fusion devices, hydrogen storage in fuel cells, hydrogen separation processes in molecular sieves and the use of perovskite-type oxides with high proton conductivity as a separator material in electrochemical cells. The modeling techniques used are numerous and follow the technological progress in computational resources. Among them, we distinguish two main methods of determining diffusion laws. Trajectories of the diffusing particle can be obtained at any temperature in molecular dynamics simulations by integrating Newton's equations of motion. The diffusion coefficients are then computed according to the Einstein relation for random walk:

where r(f) is the position of the particle at time t and D is the diffusion coefficient. As in experimental studies, the diffusion law (activation energy and pre-exponential factor) is determined by plotting the diffusion coefficients on an Arrhenius graph. This method implies long simula-

Diffusion

of Hydrogen

in

Minerals

299

tions (several nanoseconds) recording many hopping events in order to achieve a satisfactory statistics. Therefore only classical molecular dynamics employing an empirical atomistic model can be used to simulate the interatomic interactions. The second method is based on the equation: 1 ? D = - v r exp 2

kT

where v is the attempt frequency, I is the hopping distance between energy minima, Q is the activation energy, k is the Boltzmann constant and T is the temperature. The hopping distance, I, is given by the crystal geometry. The activation energy corresponds to the difference between the energy of the system when hydrogen is located at the most energetically favorable position and when hydrogen is at the saddle point connecting the two energy minima. It is then known from the calculation of the potential energy surface. Any static calculations can provide this information. The simulation techniques can be based either on empirical interatomic potentials or on first-principles solving the Schrodinger equations with some approximations. On the other hand, the attempt frequency is determined by molecular dynamics. The attempt frequency is much faster than the rate of hopping events, and molecular dynamics calculations can be much shorter than in the first method (several picoseconds). Quantum molecular dynamics calculations are much more computationally expensive and can be performed following the classical molecular dynamics calculation. In this case, the ground state energy for each atomic configuration is calculated using first-principles techniques.

DETECTION OF H DIFFUSION THROUGH ISOTOPE EXCHANGE Isotopic exchange is a basic tool to estimate the diffusion coefficients of hydrogen in mineral structures by measurement of deuterium (D) or tritium (T). Such isotopic diffusion experiments also provide a direct measurement of the kinetics of isotope exchange in natural rocks and an understanding of the conditions that preserve the observed H-D isotope disequilibrium in nature.

Anhydrous minerals Rutile. Except for rutile there are few data for true isotopic exchange in single oxide minerals. In rutile, data have been collected for the three isotopes, H, D and T (Johnson et al. 1975; Cathcart et al. 1979). Exchanges have been done under equilibrium conditions; the hydrous content of the sample was in equilibrium with the enriched H 2 0, D 2 0 or T 2 0 gas. The diffusion laws obtained in the two crystallographic directions a, and c, for H-D and H-T exchange were recalculated using York's least-squares fit method, assuming an uncertainty of 5 K in T and 0.15 in logD (York 1966; Table 1): D

TiO, a = D O e X P

D

TiO,c= D O e X P

D

D

TiO, a = O

eX

DJio,c=D0exv

P

-121 ± 6 k J m o r 1 RT -56 ± 2 kJmor1 ^ RT -106 ± 3 k J m o r 1 RT -70 ± 2 kJmor1 ^ RT

with logD 0 = —4.55 ± 0.29 , with logD 0 = - 6 . 7 8 ±0.12 , with logD 0 = - 5 . 8 4 ±0.13 with logD 0 = - 6 . 2 5 ±0.10

300

Ingrin & Blanchard

Table 1. Isotope diffusion data. Mineral / range of proton solubility

Sample orientation

Diffusing Trange species (K)

Rutile Ti0 2 ; ~ 17 ppm H 2 0

Single xtal // a

H-D exch.

887-994

7xl0-3

Rutile Ti0 2 ; ~ 17-45 ppm H 2 0

Single xtal // c

H-D exch.

623-973

7xl0-3

Rutile Ti0 2 ; ~ 3 ppm H 2 0

Single xtal // a

H-T exch.

773-1183

4x10-"

Rutile Ti0 2 ; ~ 3 ppm H 2 0

Single xtal // c

H-T exch.

527-973

4x10-"

Quartz-a; ~ 30 ppm H 2 0

Single xtal // c

H-D exch.

673-893

2.5; H 2 0 / D 2 0 vapor

Quartz-ß; ~ 30 ppm H 2 0

Single xtal // c

H-D exch.

893-1273

2.5; H 2 0 / D 2 0 vapor

Quartz-ß; ~ 13 ppm H 2 0

Single xtal // c

H-D exch.

973-1173

1100; D 2 0 fluid

Quartz-ß; ~ 13 ppm H 2 0

Single xtal J_ c

H-D exch. 1073-1173

1100; D 2 0 fluid

Diopside Fe/(Fe+Mg) = 0.036; ~ 10-40 ppm H 2 0

Single xtal // c and a*

H-D exch.

873-1139

0.1; 90%Ar -t- 10% D 2

Diopside Fe/(Fe+Mg) = 0.036; ~ 10-40 ppm H 2 0

Single xtal // b

H-D exch.

973-1173

0.1; 90%Ar -t- 10% D 2

Forsterite with 0.25 wt% Fe; (Libowitzky and Beran 1995) < 100 ppm H 2 0

Single xtal // c

H-D exch.

973-1423

0.1; 90%Ar -t- 10% D 2

y-Spinel (synthetic, Mg 2 Ge0 4 );

Single xtal

H-D exch.

873-973

0.1; 90%Ar -t- 10% D 2

Pyrope (Gr 3 Almi 5 Py 8 i); ~ 13-36 ppm H 2 0

Single xtal

H-D exch.

973-1223

0.1; 90%Ar -t- 10% D 2

Grossular (Gr 84 Andi 4 Py 2 ); 220 ppm H 2 0

Single xtal

H-D exch. 1073-1323

0.1; 90%Ar -t- 10% D 2

Grossular (Gr 73 And 23 Py 2 ); 1400 ppm H 2 0

Single xtal

H-D exch.

973-1223

0.1; 90%Ar -t- 10% D 2

Andradite (GrjAnd,,); 1500 ppm H 2 0

Single xtal

H-D exch. 1073-1173

0.1; 90%Ar -t- 10% D 2

Ilvaite; ~ CaFe 2 Fe 34 -Si 2 O s (OH)

powder

H-D exch.

623-923

5-20; H 2 0 / D 2 0 fluid

Zoisite; ~ C a j F e o j A l ^ S i j O ^ O H )

powder

H-D exch.f

623-923

200 or 400; H 2 0 / D 2 0 fluid

Epidote; ~ Ca 2 Fe 0 !) Al 2 . 1 Si3O 12 (OH)

powder

H-D exch.f

523-923

200 or 400; H 2 0 / D 2 0 fluid

Epidote; ~ Ca 2 FeAl 2 Si 3 0 1 2 (0H)

Single xtal // b

H-D exch.

473-873

200; 9 9 % D 2 0

P

(MPa)

Nominally a n h y d r o u s minerals

Hydrous minerals

Epidote

?

H-D exch.

423-673

Lawsonite; C a A l 2 S i 2 0 7 ( 0 H ) 2 H 2 0 |

powder

H-D exch.

648-698

0.1;Ar + D 2 0

Tourmaline; 14.27% FeO, 1.93% MgO, 31.34% AI 2 O 3

powder

H-D exch.

723-1073

15-25; H 2 0 / D 2 0 fluid

Hornblende; ~ (NaK) Ca 2 (Mg 2 . 4 Fe 1 . 8 Al 0 . 8 )(Si 6 . 5 Al L5 )O 22 (OH) 2

powder

H-D exch.f

623-823

200 to 800; H 2 0 / D 2 0 fluid

Kaersutite; ( \ a ü l K ü -1( ( 'a| i,\a ü , i (Mg 2 . 6 Fe 2t 1 Fe 3t c ,. 5 Ti 0 . 5 Al 0 . 3 )(Si 5 . 9 Al 2 . 1 )O 22 (OH) 2

Single xtal // b

H-D exch.

873-1173

0.1; 90%Ar -t- 10% D 2

Tremolite; ~ Ca 2 (Mg 4 8 Fe„. 2 .)Si 8 0 2 2 (0H) 2

powder

H-D exch.

623-1073

200 to 400; H 2 0 / D 2 0 fluid

Actinolite; ~ Ca 2 (Mg 4 Fe!)Si 8 0 2 2 (0H) 2

powder

H-D exch.f

673-943

200; H 2 0 / D 2 0 fluid

Chlorite; ~ (Mg a7 Al a3 Fe) 12 Si5.5Al 2 .5O 2 0(OH) 16

powder

H-D exch.f 773-973

200 or 500; H 2 0 / D 2 0 fluid

Muscovite; ~ K 2 Al 4 Si 6 Al 2 O 20 (OH,F) 4

powder

H-D exch.f 723-1023

200 or 400; H 2 0 / D 2 0 fluid

Notes: $ Data from two different samples, f possible H-D leak through the capsule References: [1] Johnson al. (1975); [2] Cathcart et al. (1979); [3] Kats et al. (1962); [4] Kronenberg et al. (1986); [5] Hercule and Ingrin (1999); [6] Ingrin unpublished data; [7] Hertweck and Ingrin (2005a, 2005b); [8] Blanchard and Ingrin (2004a); [9] Kurka et al. (2005); [10] Kurka (2005); [11] Yaqian and Jibao (1993); [12] Graham (1981);

Diffusion

o2

of Hydrogen

in

Minerals

301

log D0 (mV1)

Comments

H -4.55±0.29 D -4.72±0.30

Sequential IR measurement

[1]

56±2

H -6.78±0.12 D -6.92±0.12

Sequential IR measurement

[1]

10-»-10- 6

106±3

-5.84±0.13

Liquid-scintillation counting; No effect of p 0 2

[2]

io- 39 -io- 17

70±2

-6.25±0.10

Liquid-scintillation counting; No effect of p 0 2

[2]

unbuffered

69±12

-8.95±0.83

Sequential IR measurement

[3]

unbuffered

169±15

-3.66±0.67

Sequential IR measurement

[3]

unbuffered

215±92

-0.45±4.45

Bulk IR after a single annealing, only 3 data

[4]

unbuffered

156

-3.45

Bulk IR after a single annealing, only 2 data

[4]

10-M-10-19

149±16

-3.4±0.8

Sequential IR measurement

[5]

io- 23 -io- 18

143±33

-5.0±1.7

Sequential IR measurement

[5]

io- 23 -io- 13

134 ±7

-7.5±0.3

Sequential IR measurement

[6]

io- 23 -io- 13

140±34

-5.8±1.9

Sequential IR measurement

[7]

io- 23 -io- 18

140±38

-5.8±1.9

Sequential IR measurement

[8]

io- 21 -io- 16

102±45

-7.6

Sequential IR measurement

io- 23 -io- 18

185±28

-3.8±1.3

Sequential and profile IR measurement

[9] [10]

10-21-10-19

70

-8.9

Sequential and profile IR measurement Only 2 data points

[10]

unbuffered

115-119

-7.0 to -7.4

Bulk analysis by mass spectrometer. 1D-2D diffusion model (overestimation of D). Grain size ~ 50 u m

[11]

unbuffered

100-103

-7.8 to -8.4

Bulk analysis by mass spectrometer. 1D-2D diffusion model, grain size ~ 68-75 (jm

[12]

unbuffered

52 to 58 (7>723 K) 128 (T723 K) - 4 to - 3 (TvMe we have D e g « 3 D v . For the same reasons as for Relation (13), only the effective diffusivities are reported in Table 2. The diffusivity measured from H-D exchange in natural forsterite along c is lower than the effective diffusivity of hydrogen uptake in synthetic forsterite (Fig. 11). More data need to be collected to confirm this for synthetic forsterite and the other crystallographic direction to learn whether the assumption D H » ^>vMe is valid. A summary of the effective diffusivities measured in San Carlos olivine is presented in Figure 12. H-D exchange experiments have not been reported in San Carlos olivine, but experience shows that when we try to do it at room pressure, the experiment fails due to the concurrent extraction of hydrogen following Reaction (11). Thus we expect that the redox-exchange reaction is limited by the diffusion of hydrogen DH with Dh » D H , at least for diffusion along the slowest directions, b and c (Fig. 12). Despite the conclusion of Mackwell and Kohlstedt (1990) suggesting that diffusion is anisotropic with different kinetics along b and c directions, it is not clear yet from the latest data of Kohlstedt and Mackwell (1998) that the diffusion rates of hydrogen along these two directions are really different (Fig. 12). The assumption that DH » DyMe is justified for directions a and b but not for c; the diffusivity of hydrogen along c is comparable to the effective diffusivity of hydrogen uptake attributed to Reaction (12) (see Fig. 12). Metal vacancy diffusivities along a, and b directions are not expected to be significantly different considering the data of effective diffusivities plotted in Figure 12.

Diopside Extensive hydrogen extraction/incorporation experiments have been performed on natural diopsides with different iron contents (Ingrin et al. 1995; Hercule 1996; Hercule and Ingrin

Diffusion

of Hydrogen

in

Minerals

311

T(K) 1423 -10

1173

1073

1

1

1

973

873

1

1

o

-ii Fo

¿So

H-VMe

gv

-12

-

ÙO

o

'Fo

H-D

-16

0.6

0.7

0.8

0.9

1

1.1

1.2

103/T(K_1) Figure 11. Comparison of hydrogen uptake effective diffusivities in synthetic forsterite along the three crystallographic directions (Demouchy and Mackwell 2003) and H-D diffusivity along c in natural forsterite.

T(K) 1173

1073

1

1

1

\

K -

Olivine

H-h

A

£ D Ù0

1

1

s? o//a

s i Olivine -12

\

H-VMe

^

\

o

^

He

// c // b

\ //a

-15 0.6

0.7

0.8

0.9

1

1.1

1.2

1 0 3 / T (K"1) Figure 12. Effective diffusivities of hydrogen uptake along the three crystallographic directions of San Carlos olivine from Kohlstedt and Mackwell (1998). Lines with data points are for redox-exchange reactions (open circles: // a; full circles: // b; triangles: // c) and lines without data points are for exchange involving metal vacancies.

312

Ingrin

&

Blanchard

1999; Carpenter-Wood et al. 2000). These results can be compared with the diffusivities determined from H-D exchange experiments (Fig. 13). For low iron content (iron molar fraction, XFe = 0.036) the effective diffusivity of H extraction and uptake is isotropic and much lower than hydrogen diffusivities. Effective diffusivity of hydrogen extraction increases with iron content and then shows the same anisotropy as hydrogen diffusivity (D a = Dc » Db, Jaipur diopside with XFe = 0.07; Fig. 13). For the richest iron content (XFe = 0.07 and 0.126), the extraction diffusivities reach values comparable to hydrogen diffusivity. Studies of hydrogen incorporation in iron-rich synthetic diopside single crystals by infrared, Mossbauer- and optical spectroscopies have shown that hydrogen uptake is accompanied by iron reduction from Fe 3+ to Fe 2+ (Skogby 1994; Forneris and Skogby 2004), suggesting that hydrogen uptake is controlled by the redox-exchange Reaction (11) much as shown for olivine. It is thus possible to analyze the effective diffusivities of uptake in diopside from Equation (13) if we assume that Dh increases with iron content. However, the concentration of hydrogen point defects XH in Relation (13) is supposed to be close to the concentration of polarons Xh. The general solution is: f

D

'eff

=

^

h!

H

h

o r

X, D H

eff

'

A

Y

HDh

(15)

XH

T(K) 1173

X

-ii

Fe

X ,-11.5

-12

X

Fe

Fe

1073

=0.126 =0.05

=0.036

8

O D // b -13

-13.5

-14 0.9

1

10 3 /T(K _1 ) Figure 13. Effective diffusivities of hydrogen extraction and uptake for diopside single crystals with different iron contents compared to hydrogen diffusivities determined from H-D exchange experiments. Solid lines: hydrogen diffusivities from Hercule and Ingrin (1999). Dashed lines and data points: extraction/ incorporation experiments (diamonds: Jaipur diopside, extraction // a and c; Carpenter-Wood et al. 2000; full triangles: Jaipur diopside, extraction // b; Carpenter-Wood et al. 2000; empty circles: Russian diopside, extraction/incorporation // a*, b and c; Ingrin et al. 1995; Hercule and Ingrin 1999; long dashed line: Baikal diopside, extraction // c; Hercule 1996; short dashed line: Malacacheta diopside, extraction // ~ [212]; Hercule 1996).

Diffusion

of Hydrogen

in

313

Minerals

The assumption XH = XH can be discussed for olivine but it is certainly wrong for diopside. Figure 14 shows the results of the values of DH deduced from Equation (15) assuming that DH is given by the diffusion laws obtained from H-D exchange and XH is arbitrarily fixed equal to 10% of the iron content (XH/XH is equal to 6, 17, 36 and 85 respectively for diopsides of increasing iron content). At very low XFe, (XFe = 0.036), DH » XJDiJXn and Dh ~ D^l where DH is isotropic. We assume that polaron diffusivity is isotropic for other iron concentrations too. For XFe = 0.05, along the c-direction, we still have DH » XhDhIX]H and Dh ~ D^ 18. At XFe = 0.07, XiPiJX^ is comparable to DH along c and a directions and is much greater than DH along b. Thus DH is close to DH along b. For XFe = 0.126, the effective diffusivities were not measured along the main crystallographic axes but the data confirm that XHDF/XH » DH; thus DEG ~ DH (Fig. 14). The only data that reports uptake involving another reaction than the redox-exchange shows a diffusivity only slightly smaller than the diffusivity of polarons (XFe = 0.036; the square symbol in Fig. 14; Guillaumou et al. 1999; Ingrin and Skogby 2000). Enstatite Stalder and Skogby (2003) published a study on hydrogen exchange in orthopyroxene. They show that the extraction of hydrogen in enstatite single crystals from a Kilbourne Hole xenolith is two orders of magnitude faster than the extraction from end-member synthetic enstatite at 973 K. In end-member enstatite the effective diffusivity is isotropic and the activation enthalpy is high (295 ± 55 kJ-mol -1 ; Table 2). Values of the effective diffusivity of extraction/incorporation in Kilbourne Hole enstatite are of the same order as the diffusivity of the iron-poor diopside. The low ferric iron content of this enstatite (Fe3+/Fetotal = 3.5%) suggests that the exchange is controlled by the mobility of polarons (DH » XHDH/XH). In end-member enstatite, diffusion is slower probably because the exchange is controlled by Reaction (12) and the diffusivity of metal vacancies is much slower than hydrogen diffusivity

T(K) 1423

0.6

0.7

1173

0.8

1073

0.9

973

1

873

1.1

1.2

10 3 /T(K _1 ) Figure 14. Evolution of polaron diffusivities determined f r o m equation (15), D ^ , for various iron contents: grey lines. Full square: effective diffusivity of a reaction involving metal vacancies (Guillaumou et al. 1999: Ingrin and Skogby 2000). Other symbols are the same as in Figure 13.

314

Ingrin & Blanchard

(.XhDh/XVMe » Dv ). Stalder and Skogby (2003) report that a second reaction is involved in the uptake of hydrogen in Kilbourne Hole enstatite with the same effective diffusivity as for end-member enstatite suggesting that the same is true for mantle enstatite.

Garnets Water-related defects in garnet have been an important subject of study for decades but it is only recently that the kinetics of hydrogen extraction have been studied (Wang et al. 1996; Blanchard and Ingrin 2004b; Kurka 2005; Kurka et al. 2005). The activation enthalpy of hydrogen extraction in garnets is ~260 kJ-mol -1 (with a range of 180 - 329 kJ-mol -1 ), much higher than the enthalpies measured in olivine and diopside for redox-exchange reactions (107 - 1 8 1 kJ-mol -1 ; Table 2) and much higher than the average activation enthalpy of hydrogen diffusion determined from H-D exchange (130 kJ-mol -1 ; Figs. 7 , 1 5 ) . The extraction kinetics in mantle pyropes can be described by a single set of diffusion laws with an activation enthalpy of 253 kJ-mol -1 (Wang et al. 1996; Fig. 15). In this type of pyrope the extraction diffusivity is only slightly higher than the hydrogen diffusivity (Blanchard and Ingrin 2004a). For Dora Maira pyrope, which has a lower concentration of ferric iron than mantle pyrope, the extraction diffusivity is one to two orders of magnitude slower than for mantle pyropes. In that case, extraction diffusivities are comparable to or slightly lower than the hydrogen diffusivity (Fig. 15). Two general features of hydrogen diffusion in garnet, which were not observed in other silicates like olivine or pyroxenes, are the difference of hydrogen extraction kinetics between OH bands (groups of OH bands at 3650 and 3600 cm - 1 in Dora Maira pyrope and 3620 and 3560 cm - 1 in andradite; Table 2, Figs. 15, 16) and the diffusivity dependence on oxygen partial pressure (Blanchard and Ingrin 2004b; Kurka 2005; Fig. 15). In Dora Maira pyrope, the diffusivity decreases by two orders of magnitude for a p 0 2 drop of 15 to 20 orders (Fig. 15).

T(K) 1273

1173

1073

103/T (K-1) Figure 15. Effective diffusivities of hydrogen extraction in pyrope. Solid lines: mantle pyropes (Wang et al. 1996); solid lines with data points: Dora Maira pyrope for two different groups of OH bands (OH 3650 , OH 3 6 0 0 ; in air: big symbols; in reducing atmosphere: small symbols; Blanchard and Ingrin 2004b); dashed line: average law of H-D exchange in garnet.

Diffusion of Hydrogen in Minerals

315

In calcium-rich garnets, the diffusivities of hydrogen extraction roughly increase with ferric iron content (Table 2, Fig. 16). The kinetics dependence on iron concentration, like for diopside, suggests that a redox-exchange reaction controls the hydrogen extraction. The implication of Reaction (11) is also confirmed by the simultaneous decrease of Fe 2 + concentration measured in grossular by Mossbauer during its hydrogen extraction (sample Gr73And23Py2, Table 2; Forneris and Skogby 2004). However, complex kinetics and high activation enthalpy («260 kJ-mol -1 ) of hydrogen loss from garnet, close to the enthalpies found for reactions involving the migration of metal vacancies in olivine and enstatite, are difficult to interpret using Relations (13) and (15). In their experiments, Wang et al. (1996) could not fit the diffusion data of mantle garnets by a solution of Fick's law with D independent of the concentration. Doremus (2002) explains this behavior, assuming that the mobile species is molecular water, reacting with the oxygen sublattice to produce OH species like in glass. However, this effect was not observed in other garnets; a diffusion profile in grossular is fitted with a constant diffusivity in Figure 2. Moreover, the departure from a constant diffusion coefficient can be explained without assuming concentration dependence. The hydrogen species giving rise to different OH bands of garnet have different kinetics, and they must be considered separately in order to determine a diffusion law (Blanchard and Ingrin 2004b; Kurka 2005). In addition, the spatial resolution of infrared profiles is highly dependent on sample thickness and its measure close to the edge of the crystal is sometimes tricky. A dependence of D with hydrogen concentration for a single OH band closely resembles a relation given by Equation (15) where the effective diffusivity is a function of the ratio Xh and X H , as Xh and X H vary all along the exchange. The analysis of

T(K) 1323

1223

1123

1073

973

10 3 /T (K" 1 ) Figure 16. Diffusivity of hydrogen extraction in calcium-rich garnets. Data from Kurka et al. (2005) and Kurka (2005): "Andxx" gives the concentration of the garnet in the andradite component, And99 x x x x corresponds to the two types of OH bands present in the endmember andradite sample. Dashed line: average law of H-D exchange in garnet.

316

Ingrin

&

Blanchard

these results in terms of a specific reaction model cannot be done before the evolution during the exchange of both species Fe 3+ and H is determined. Quartz Very few diffusion experiments of extraction or incorporation of hydrogen have been done in quartz. Only the study of Kronenberg et al. (1986) reports hydrothermal experiments where the mobility of hydrogen, and not water, was involved during uptake. The diffusivity of uptake is close to the diffusivities for H-D exchange and the diffusion seems to be isotropic (Kats 1962; Kronenberg et al. 1986; Fig. 17). An experiment of tritium uptake under low water pressure by Shaffer (1974) gives values lower by a factor of 104 to 105 (Fig. 17). Shaffer noticed that the diffusivity increases with the decrease of water concentration in the sample. This behavior is contrary to that expected for water exchange; however, the values Shaffer reports are very close to data for water and oxygen diffusion in quartz (Cordier et al. 1988) suggesting that tritium uptake in Shaffer's experiments was due to the mobility of water. Feldspars To our knowledge, there is no experimental study providing a diffusion law for H-D exchange in feldspar and only two studies determining hydrogen extraction/incorporation diffusivities. Kronenberg et al. (1996) measured hydrogen diffusivity from extraction experiments in air from adularia feldspars that contain H 2 0 structural defects. Despite a debate as to whether the mobile defect is really the hydrogen atom or the water molecule (see Doremus 1998; Kronenberg et al. 1998), the results are interesting because the measured kinetics are very fast and may provide a lower limit for H-D exchange kinetics in feldspar. These results are plotted in Figure 18 with the diffusion law for H-D exchange in quartz from Kats (1962); the two laws are very close. Kronenberg et al. (1996) performed one extraction experiment in a direction parallel to b suggesting that the diffusion is probably isotropic. They also performed a few uptake experiments at 1000 MPa H 2 0 but no value of diffusivity was proposed. T(K) 1273

1173

1073

973

10 3 /T (K"1) Figure 17. Diffusivity of hydrogen extraction in quartz. Empty circles are for hydrogen uptake experiments done parallel to c, full circle for direction perpendicular to c.

Diffusion

of Hydrogen

in

317

Minerals

T(K) 1273

1173

1073

973

873

773

-10

\

°

Feldspars

-15 07

0.8

0.9

1.1

3

1.2

1.3

1

10 /T (K- ) Figure 18. Diffusivity of hydrogen extraction in feldspars (symbols and solid lines) compared to the H-D exchange kinetics in quartz (dashed line).

Pegmatite minerals like adularia are H 2 0-bearing feldspars but plagioclases from volcanic origin contain essentially structural OH groups (Johnson and Rossman 2004). Extraction experiments in nitrogen at atmospheric pressure in andesine crystal by Johnson (2003) give much lower diffusivities and a higher activation enthalpy than are given by the data on adularia (Fig. 18). The activation term is close to values measured for garnet and the diffusivities compare favorably to the diffusivities measured in iron-poor grossular (Figs. 16, 18; Table 2). The iron concentration in andesine is three times lower than the hydrogen concentration. Thus a redoxexchange reaction like (11) cannot be the main mechanism of extraction (Johnson 2003).

CONCLUSION AND FUTURE DIRECTIONS We have deliberately not considered migration of water molecules in hydrous or anhydrous minerals; the review was mainly focused on hydrogen diffusion in anhydrous minerals. Like for oxides, there are very few minerals with absolutely no evidence of the presence of hydrogen, so it is possible that the review has omitted some few diffusion data. However, the amount of published diffusion data on anhydrous minerals is now large enough to underline some general behavior: •

The values of hydrogen diffusivities determined from H-D exchanges in anhydrous minerals spread on 3 log units and the activation enthalpy is generally between 100 to 200 kJ-mol -1 . The activation enthalpies for hydrous minerals are generally lower than those for anhydrous minerals (Fig. 10).

•

Fast hydrogen uptake is controlled by a redox-exchange reaction involving the Fe 3+ /Fe 2+ couple (11) in most of minerals containing iron like olivine, pyroxenes and garnets. It is theoretically possible to determine the effective diffusivity of hydrogen exchange from the diffusivity of hydrogen and the diffusivity of polarons (15).

318

Ingrin

&

Blanchard

•

In garnets and iron-free minerals the activation enthalpy of hydrogen uptake is frequently higher than 200 kJ-mol -1 (pure forsterite, pure enstatite, plagioclase). For slow hydrogen uptakes controlled by the diffusion of vacancies (12), the activation enthalpy is also higher than 200 kJ-mol -1 .

•

Anisotropy of diffusion is important in rutile, olivine, diopside, epidote and amphibole, but surprisingly very low in quartz and feldspar.

The recent development of the infrared microscopy and techniques like ion and nuclear microprobes will increase the number of studies reporting hydrogen profiles in minerals (Demouchy 2004; Kurka 2005; Peslier et al. 2006). It will facilitate the study of the anisotropy of diffusion in single crystals and the systematic use of anisotropic data in modeling. Such profiles can be used to estimate the kinetics of natural processes like for instance the rate of magma ascent (Demouchy 2004; Peslier et al. 2006). However, there are still many experimental data missing even for well-studied minerals like olivine or diopside. We need these data in order to correctly analyze diffusion results and to build quantitative models of hydrogen uptake. For instance, there are no data for isotopic diffusion in olivine, orthopyroxenes and feldspars. The knowledge of the kinetics of H-D exchange is essential to identify the expression of the effective diffusivity and to validate simplified equations like (12) or (13). Experiments have demonstrated the importance of redox-exchange reactions in silicates but there are few if any studies that follow the Fe3+/Fe2+ change during hydrogen extraction/incorporation. Future works on hydrogen diffusion in anhydrous minerals will systematically incorporate studies of the iron oxidation during dehydrogenation. It is only with these new data that we will be able to understand the mechanisms of hydrogen uptake in major silicate minerals.

ACKNOWLEDGMENTS We thank Andreas Kurka, Juliette Forneris and Henrik Skogby for the communication of their most recent results. We are also grateful for helpful reviews from Andreas Kronenberg and Henrik Skogby. This study was supported by the EU, through the Human Potential Program HPRM-CT-2000-0056.

REFERENCES Bates JB, Wang JC, Perkins RA (1979) Mechanisms for hydrogen diffusion in Ti0 2 . Phys Rev B 19:41304139 Blanchard M, Ingrin J (2004a) Kinetics of deuteration in pyrope. Eur J Mineral 16:567-576 Blanchard M, Ingrin J (2004b) Hydrogen diffusion in Dora Maira pyrope. Phys Chem Mineral 31:593-605 Bongiorno A, Colombo L, Cargnoni F (1997) Hydrogen diffusion in crystalline Si0 2 . Chem Phys Lett 264: 435-440 Brady JB (1993) Diffusion data for silicate minerals, glasses, and liquids. In: Handbook of Physical Constants. Vol 2. Arhens TH (ed) Am Geophysical Union, p 269-290 Bunson PE, Di Ventra M, Pantelides ST, Schrimpf RD, Galloway KF (1999) Ab initio calculations of H+ energetics in Si0 2 : Implications for transport. IEEE Trans Nuclear Sci 46:1568-1573 Carpenter-Wood S, Mackwell SJ, Dyar D (2000) Hydrogen in diopside: Diffusion profiles. Am Mineral 85: 480-487 Carslaw HS, Jaeger JC (1959) Conduction of Heat in Solids. Oxford University Press Cathcart JV, Perkins RA, Bates JB, Manley LC (1979) Tritium diffusion in rutile (Ti0 2 ). J Appl Phys 50:41104119 Chacko T, Riciputi LR, Cole DR, Horita J (1999) A new technique for determining equilibrium hydrogen isotope fractionation factors using the ion microprobe: Application to the epidote-water system. Geochim Cosmochim Acta 63:1-10 Cordier P, Boulogne B, Doukhan JC (1988) Water precipitation and diffusion in wet quartz and wet berlinite A1P04. Bull Minéral 111:113-137 Crank J (1975) The Mathematics of Diffusion. Oxford University Press

Diffusion

of Hydrogen

in

Minerals

319

Demouchy S (2004) Water in the Earth's interior: thermodynamics and kinetics of hydrogen incorporation in olivine and wadsleyite. PhD Dissertation, Universität Bayreuth, Germany Demouchy S, Mackwell SJ (2003) Water diffusion in synthetic iron-free forsterite. Phys Chem Mineral 30: 486-494 Dersch O, Rauch F (1999) Water uptake of quartz investigated by means of ion-beam analysis. Fres J Anal Chem 365:114-116 Dersch O, Zouine A, Rauch F, Ericson JE (1997) Investigation of water diffusion into quartz using ion beam analysis techniques. Fres J Anal Chem 358:217-219 Doremus (1998) Comment on "Stationary and mobile hydrogen defects in potassium feldspar" by AK Kronenberg, RAYund and GR Rossman. Geochim Cosmochim Acta 62:377-378 Doremus RH (2002) Diffusion of Reactive Molecules in Solids and Melts. Wiley Forneris JF, Skogby H (2004) Is hydrogen loss via iron oxidation an important mechanism in nominally anhydrous minerals? Goldschmidt J Conf Abstr 1.1.23, A34 Graham CM (1981) Experimental hydrogen isotope studies III: Diffusion of hydrogen in hydrous minerals, and stable isotope exchange in metamorphic rocks. Contrib Mineral Petrol 76:216-228 Graham CM, Harmon RS, Sheppard SMF (1984) Experimental hydrogen isotope studies: hydrogen isotope exchange between amphibole and water. Am Mineral 69:128-138 Graham CM, Sheppard SMF, Heaton THE (1980) Experimental hydrogen isotope studies-I. Systematics of hydrogen isotope fractionation in the systems epidote-H 2 0, zoisite-H 2 0 and A10(0H)-H 2 0. Geochim Cosmochim Acta 44:353-364 Graham CM, Viglino JA, Harmon RS (1987) Experimental study of hydrogen-isotope exchange between aluminous chlorite and water and hydrogen diffusion in chlorite. Am Mineral 72:566-579 Guillaumou N, Dumas P, Ingrin J, Carr GL, Williams JP (1999) Microanalysis of fluids in minerals in the micron scale range by synchrotron infrared microspectrometry. Internet J Vibrational Spec [www.ijvs.com] 3: 1-11

Hercule (1996) Cinétique et solubilité de l'hydrogène dans le diopside monocrystallin. PhD Dissertation, University paris XI, Orsay, France Hercule S, Ingrin J (1999) Hydrogen in diopside: Diffusion, kinetics of extraction-incorporation, and mobility. Am Mineral 84:1577-1587 Hertweck B, Ingrin J (2005a) Hydrogen incorporation in a ringwoodite analogue: Mg 2 Ge0 4 spinel. Mineral Mag 69:335-341 Hertweck B, Ingrin J (2005b) Hydrogen incorporation in ringwoodite analogue: Mg 2 Ge0 4 spinel. Geophys Res Abstr, EGU 7:2690 Ingrin J, Blanchard M (2000) Hydrogen mobility in single crystal kaersutite. EMPG VIII, J Confe Abstr 5:52 Ingrin J, Hercule S, Charton T (1995) Diffusion of hydrogen in diopside: Results of dehydrogenation experiments. J Geophys Res 100:15489-15499 Ingrin J, Latrous K, Doukhan JC, Doukhan N (1989) Water in diopside: an electron microscopy and infrared spectroscopy study. Eur J Mineral 1:327-341 Ingrin J, Pacaud L, Jaoul O (2001) Anisotopy of oxygen diffusion in diopside. Earth Planet Sei Lett 92:347361 Ingrin J, Skogby H (2000) Hydrogen in nominally anhydrous upper-mantle minerals: concentration levels and implications. Eur J Mineral 12:543-570 Jibao J, Yaqian Q (1997) Hydrogen isotope fractionation and hydrogen diffusion in the tourmaline-water system. Geochim Cosmochim Acta 61:4679-4688 Johnson EA (2003) Hydrogen in nominally anhydrous crustal minerals. PhD Dissertation, California Institute of Technology, Pasadena, USA Johnson EA, Rossman GR (2004) A survey of hydrous species and concentrations in igneous feldspars. Am Mineral 89:586-600 Johnson OW, Paek SH, Deford JW (1975) Diffusion of H and D in Ti0 2 : Suppression of internal fields by isotope exchange. J Appl Phys 46:1026-1033 Kats A, Haven Y, Stevels JM (1962) Hydroxyl groups in a-quartz. Phys Chem Glasses 3:69-76 Kohlstedt DL, Mackwell SJ (1998) Diffusion of hydrogen and intrinsic point defects in olivine. Z Phys Chem 207:147-162 Kronenberg AK, Kirby SH, Aines RD, Rossman GR (1986) Solubility and diffusional uptake of hydrogen in quartz at high water pressure: implications for hydrolytic weakening. J Geophys Res 91:12723-12744 Kronenberg AK, Yund RA, Rossman GR (1996) Stationary and mobile hydrogen defects in potassium feldspar. Geochim Cosmochim Acta 60:4075-4094 Kronenberg AK, Yund RA, Rossman GR (1998) Reply to the comment by Robert H. Doremus on "Stationary and mobile hydrogen defects in potassium feldspar." Geochim Cosmochim Acta 62:379-382 Kurka A (2005) Hydrogen in Ca-rich garnets: diffusion and stability of OH-defects. PhD Dissertation, University Paul Sabatier, Toulouse, France

320

Ingrin & Blanchard

Kurka A, Blanchard M, Ingrin J (2005) Kinetics of hydrogen extraction and deuteration in grossular. Mineral Mag 69:359-371 Libowitzky E, Beran A (2006) The structure of hydrous species in nominally anhydrous minerals: information from polarized IR spectroscopy. Rev Mineral Geochem 62:29-52 Libowitzky E, Rossman GR (1996a) Principles of quantitative absorbance measurements in anisotropic crystals. Phys Chem Mineral 23:319-327 Libowitzky E, Rossman GR (1996b) FTIR spectroscopy of lawsonite between 82 and 325K. Am Mineral 81: 1080-1091 Mackwell SJ, Kohlstedt DL (1990) Diffusion of hydrogen in olivine: Implications for water in the mantel. J Geophys Res 95:5079-5088 Marion S, Meyer H-W, Carpenter M, Norby T (2001) H 2 0 - D 2 0 exchange in lawsonite. Am Mineral 86:11661169 Norby T, Larring Y (1997) Concentration and transport of protons in oxides. Curr Opin Solid State Mater Sci 2:593-599 Peslier AH, Luhr JF (2006) Hydrogen loss from olivines in mantle xenoliths from Simcoe (USA) and Mexico: Mafic alkalic magma ascent rates and water budget of the sub-continental lithosphere. Earth Planet Sci Lett 242:302-319 Rossman GR (2006) Analytical methods for measuring water in nominally anhydrous minerals. Rev Mineral Geochem 62:1-28 Shaffer EW, Sang SL, Cooper AR, Heuer AH (1974) Diffusion of tritiated water in p-quartz. In: Geochemical Kinetics and Transport. Hofmann AW, Giletti BJ, Yoder H, Yund RA (eds) Carnegie Inst Wash Publ 634: 131-138 Skogby H (1994) OH incorporation in synthetic clinopyroxene. Am Mineral 79:240-249 Skogby H, Rossman GR (1989) O H - in pyroxene: An experimental study of incorporation mechanisms and stability. Am Mineral 74:1059-1069 Stalder R, Skogby H (2003) Hydrogen diffusion in natural and synthetic orthopyroxene. Phys Chem Mineral 30:12-19 Suman KD, Cole DR, Riciputi LR, Chacko T, Horita J (2000) Experimental determination of hydrogen diffusion rates in hydrous minerals using the ion microprobe. J Conf Abstr 5(2):340 Suzuoki T, Epstein S (1976) Hydrogen isotope fractionation between OH-bearing minerals and water. Geochim Cosmochim Acta 40:1229-1240 Vennemann TW, O'Neil JR, Deloule E, Chaussidon M (1996) Mechanism of hydrogen exchange between hydrous minerals and molecular hydrogen: Ion microprobe study of D/H exchange and calculations of hydrogen self-diffusion rates. Goldschmidt. J Conf Abstr 1(1):648 Wang L, Zhang Y, Essene E (1996) Diffusion of the hydrous component in pyrope. Am Mineral 81:706-718 Wegden M, Kristiansson P, Skogby H, Auzelyte V, Elfman M, Malmqvist KG, Nilsson, Pallon J, Shariff A (2005) Hydrogen depth profiling by p-p scattering in nominally anhydrous minerals. Nucl Instr Methods Phys Res B 231:524-529 Yaqian Q, Jibao G (1993) Study of hydrogen isotope equilibrium and kinetic fractionation in the ilvaite-water system. Geochim Cosmochim Acta 57:3073-3082 York D (1966) Least-squares fitting of a straight line. Can J Phys 44:1079-1086

14

Reviews in Mineralogy & Geochemistry Vol. 62, pp. 321-342, 2006 Copyright © Mineralogical Society of America

Effect of Water on the Equation of State of Nominally Anhydrous Minerals Steven D. Jacobsen Department of Geological Sciences Northwestern University Evanston, Illinois, 60208, U.S.A. e-mail:

[email protected]

INTRODUCTION It is possible that the majority of Earth's H 2 0 budget is present as hydroxyl (OH) structurally incorporated into the major nominally anhydrous minerals (NAMs) of the mantle (e.g., Martin and Donnay 1972). Ring wood (1966) thought as much as five times the surface H 2 0-mass could be present in the mantle, amounting to ~0.2 wt% H 2 0 if distributed throughout the entire mantle (Harris and Middlemost 1969). We know now the (Mg,Fe) 2 Si0 4 polymorphs of the upper mantle and transition zone can incorporate up to several weight percent of water in their structures (e.g., Smyth 1987; Inoue et al. 1995; Kohlstedt et al. 1996; Bolfan-Casanvoa et al. 2000; Mosenfelder et al. 2006). The possibility of a deep-Earth water cycle leads naturally to the question, is "water" in the mantle, whether regional or globally distributed, detectible seismically? In order to address this question, it is necessary to know, quantitatively, the effects of water (or more precisely structurally bound hydroxyl) on the elastic moduli of mantle minerals. Also required are pressure and temperature derivatives of the elastic moduli for more direct comparison with seismological observation. This chapter will review what is known about the elastic properties of "hydroxylated" NAMs from experimental studies. From a crystal chemical perspective, hydrated NAMs are defect structures because hydrogen is usually incorporated through charge balance by cation vacancies. Therefore, small variations in water content can have a dramatic effect on thermoelastic parameters, more so than any other major geochemical substitution such as iron or aluminum. For example, at one atmosphere the addition of ~1 wt% H 2 0 into ringwoodite has a similar effect on the shear modulus as raising the temperature by 8001000 °C (Wang et al. 2003a; Jacobsen et al. 2004). However, due to elevated pressure derivatives, the difference between anhydrous and hydrous velocities diminishes at higher pressures. In this chapter, elasticity-water systematics will be evaluated in order to make some predictions about the elastic properties of major phases for which shear velocities are not yet available. Seismologists require accurate thermoelastic parameters of hydrated NAMs in order to evaluate velocity anomalies in potentially hydrous regions of the mantle. However, many of the thermoelastic properties of hydrated NAMs have not yet been measured. For example, pressure derivatives of the bulk and shear moduli are needed to estimate both compressional and shear wave velocities (vP and v s ) at high pressure, but are currently available only for hydrous ringwoodite (Wang et al. 2003b; Jacobsen and Smyth 2006). Furthermore, the effects of OH on temperature derivatives of the elastic moduli are unknown for any of the major NAMs. Thus, in order to obtain a broader picture of the effects of water on elastic properties, some stoichiometrically hydrous phase will also be considered, such as humites and selected dense-hydrous magnesium silicates (DHMS). Finally, using a model set of thermoelastic parameters, monomineralic velocities are calculated for hydrous (Mg,Fe) 2 Si0 4 polymorphs along a mantle adiabat for simple comparison with global seismic velocity models. 1529-6466/06/0062-0014505.00

DOI: 10.213 8/rmg.2006.62.14

322

Jacobsen

The chapter is divided into sections based on mineral groups stable within the upper mantle and transition zone. The DHMS are treated separately. There are no elasticity data for OH-bearing silicate perovskite or magnesiowiistite in the lower mantle. Within each section the reader will find different types of equation of state data. Pressure-volume (P-V) studies with X-ray or neutron diffraction techniques are static in nature, providing variation of density (and sometimes crystal structure) with pressure, p(P). P-V data are usually fit to a third-order Birch-Murnaghan equation of state (e.g., Angel et al. 2000), parameterized in terms of initial volume (Vq), isothermal bulk modulus, KT = -V{dP/dV)T, and first pressure derivative K' = dK-jjdP. With a third-order truncation, the second derivative of KT is implied, but probably plays an important role in the variation of density of hydrated NAMs with pressure due to the initial high compressibility of cation vacancies. It is possible that hydrated NAMs become as dense, or even denser than their anhydrous end members due to elevated pressure derivatives, which may also be the case for hydrous melts (Agee 2005; Matsukage et al. 2005; Sakamaki et al. 2006). Volume-temperature studies with X-rays or neutrons provide coefficients of thermal expansion, such as the volume thermal expansivity, av = V~l{dV/dT)P, needed to calculate mineral density at high temperature. One problem facing thermal expansion studies of hydrated mantle NAMs is the relatively low temperature (~500 °C) that these phases dehydrate or decompose when not under confining pressure (Inoue et al. 2004). To date, data for OHbearing NAMs is restricted to P-V or V-T studies, which means the temperature derivatives of hydrated NAMs at high pressure are not known. Dynamic studies of mineral elastic properties are effectively one derivative ahead of static compression. These include Brillouin spectroscopy, resonant ultrasound spectroscopy, and ultrasonic interferometry (and various adaptations of each technique). With these methods, elastic moduli are obtained from measured sound velocities, resonance frequencies, or elastic wave travel times, and at ambient pressure both the adiabatic bulk modulus, Ks = -V(dP/dV)s, and shear modulus (G, sometimes written |i) are obtained. The compressional (P) and shear (S) velocities and moduli are related through the equations: (1) (2)

High-pressure Brillouin or ultrasonic studies determine Ks and G at each pressure, so pressure derivatives of the moduli (K s ' and G') are usually very accurate in comparison with static compression studies where KT' is a second-order fitting parameter. Similarly, resonance techniques at elevated temperature determine the moduli at each temperature, resulting in very accurate temperature derivatives.

ELASTIC PROPERTIES OF NOMINALLY ANHYDROUS MINERALS IN THE UPPER MANTLE Olivine Olivine, a-(Mg,Fe) 2 Si0 4 , is the most abundant mineral in the upper mantle. An understanding of the effects of water on the elastic properties of olivine is therefore of primary importance. Natural olivines from mantle xenoliths exhibit a wide range of water contents from

-0.02

-0.04 -3

-2

-1

0

dlnQ

2

3

Remote Sensing of Hydrogen in the Earth's

Mantle

355

which can be inverted for five unknowns. Here 81ogX¡ (I = 1...«) is a set of observed data such as (81ogVp 81ogVs Sloggp 1 8 log ¡2s1 81ogp) and 8Yj (j = \...m) is a set of unknowns such as (8T SlogXj.. .81ogX n 81;), and A¡¡ = S l o g X / d Y j is a matrix made of partial derivatives of seismological observations with respect to physical/chemical variables (e.g., SlogV P /dT). This matrix is made of elements that need to be evaluated based on mineral physics. Note that in general, one has a large number of unknowns particularly because there are a large number of elements to specify the chemical composition, and the solution is non-unique. Matsukage et al. (2005) analyzed the compositional data and elasticity of constituent minerals in mantle peridotites and concluded that in most cases the chemical variation of mantle peridotite can be specified in terms of a small number of parameters. In the simplest case of peridotites in the oceanic environment, a single parameter, i.e., Mg# (mole fraction of Mg relative to Fe) is enough to specify the compositional dependence of seismic wave velocities of peridotite. With this simplification (and the assumed null effects of partial melting), one can invert seismic anomalies in terms of anomalies in water content, temperature and major element chemistry if at least three data are obtained for each point. The details of the inversion scheme are described in Shito et al. (2006). The quality of such an inversion depends strongly on the quality of seismological data and of the mineral physics-derived partial derivatives. In general, the inversion is non-linear because the values of matrix elements depend on the values of unknowns such as hydrogen content and temperature. Seismic anisotropy is in most cases caused by the lattice-preferred orientation (LPO) of elastically anisotropic minerals such as olivine (e.g., Nicolas and Christensen 1987; Chapter 21 of Karato 2006a). The possible influence of hydrogen on LPO of olivine was suggested by Karato (1995). This hypothesis was proposed based on the experimental study by Mackwell et al. (1985) who showed that the effect of hydrogen to enhance plastic deformation of olivine is anisotropic: deformation by slip systems with b = [001] is more enhanced by hydrogen than deformation by b = [100] slip systems. Consequently, Karato (1995) postulated that at high water fugacity conditions, slip systems with b = [001] (e.g., [001](010), [001](100)) might become the dominant (easiest) slip system, and consequently the LPO will be different from that usually observed at low water fugacity conditions. This hypothesis has been tested by experimental studies in my lab (Jung and Karato 2001b; Katayama et al. 2004; Jung et al. 2006; Katayama and Karato 2006). Based on these results, a variety of olivine LPOs have been identified (see Fig. 7), and each LPO has its own anisotropic signature (Table 2). A fabric transition is a commonly observed phenomenon in a material where multiple slip systems operate that have relatively small contrast in strength (e.g., quartz, Lister 1979). A fabric transition will occur when the relative easiness of two slip systems change, i.e., - J - ^ C rm(P)V(r,P)'

w

- J - ^ C rm(P)V(r,P)'

w

(17)

where è t 2 is strain-rate with the slip system 1, 2, T is temperature, Tm(P) is melting temperature, a is stress, and [i(T,P) is shear modulus. Consequently, the conditions for a fabric transition will be given by

F

Tm(P) \i(T,P)

=0

(18)

A boundary between different fabric types is characterized by a hyper-surface in the space defined by three variables, [T/(Tm(P)), a/(\i(T,P)), Cw], Several points may be noted on the nature of fabric transitions. First, because the fabric boundary is defined by the relative easiness of two slip systems, strain-rate does not explicitly enter the equation for a fabric boundary. In other words, the fabric diagram determined for

Kar ato

356

Table 2. Seismological signature of various olivine LPOs corresponding to the horizontal shear.

fabric

fast S-wave polarization

vsnlvsv

A-type

parallel to flow

>1

B-type

normal to flow*

>1

C-type

parallel to flow

1 (weak)

* This relation holds also for the vertical shear.

B-type

C-type

E-type

Figure 7. Various olivine deformation fabrics found in the experimental studies by (Jung and Karato 2001b; Katayama et al. 2004; Jung et al. 2006); pole figures on the equal area projection on the lower hemisphere. See Plate 2 for color figure.

a certain range of strain-rates can be applied to slower strain-rates without any (explicit) problems. Second, I note that a conventional power-law formula as applied to olivine does not predict stress-induced fabric transformation. The most detailed study on dislocation creep in olivine single crystal is the work by Bai et al. (1991) who showed a highly complicated creep laws for olivine single crystals, but a common feature they reported is that the stress exponent is common to all slip systems, n~3.5. In this case, Equation (18) will not contain stress as a variable and one should not expect stress-induced fabric transformations.

Remote Sensing of Hydrogen in the Earth's Mantle

35 7

To see these points, let us consider the following simple power-law creep constitutive relationship, ¿1 2 — -^l 2 ^^P

RT

\

(19) y

where A 1 j 2 are the pre-exponential factors, Hl2 are the activation enthalpy, and n12 are the stress exponent for the 1,2 slip systems respectively. Equating e t = e 2 , one gets the conditions for the fabric boundary, viz., («L - w 2 ) l o g c =

Hl — H2 Al --log A, RT

(20)

This equation does not contain strain-rate, so the fabric boundary does not explicitly depend on strain-rate. Also if nx = n2 as Bai et al. (1991) showed, then the boundary will be given = 0 a n d would not depend on stress. The latter point is by [(i/j inconsistent with some of the experimental results including Carter and Ave Lallemant (1970) and the B- to C-type, B- to E-type or B- to A-type transition observed in our study (Jung and Karato 2001b; Katayama et al. 2004; Jung et al. 2006; Katayama and Karato 2006). Therefore there is a need to go beyond a simple power-law constitutive relation to interpret the observed fabric transitions. One way is to incorporate a subtle deviation from the power-law formula observed in some of the experimental studies at high stress levels. This power-law breakdown occurs beyond a certain stress (>100-200 MPa), and can be explained by the stress dependence of activation enthalpy, H*(a) e l 2 — /A 2 exp

-a I

300

-a cS -a

J 200

g-a

100

0.8

u «g

0.6

£

0.4

Pure NaCl

o >

0.2 0

200

400

600

800

1000

1200

Time, min. Figure 4. The time dependence of the width and volume fraction of a wadsleyite rim growing on a single crystal of San Carlos olivine under dry and wet conditions at 13.5 GPa and 1303 K. Details are given in Kubo et al. (1998). Water was added as brucite Mg(OH) 2 and its content was controlled by the ratio of NaCl and brucite surrounding the single crystal sample. (A) dry condition; (B) 500:1 mixture of NaCl and brucite; (C) 10:1 mixture (weight) of NaCl and brucite. The transformation is clearly enhanced by water.

1600 • 1500 •

28.2 GPa, 1283 K, P= 5.4GPa ( M g 2 S i 0 4 ) d= 10 m 31.4 GPa, 1473 K, P= 6.0 GPa ¿=3.4m (Purepyrope)

0.6

35"8 GPa, 1273 K,

0.4 0.2 H

1

28.1 G P a 1600 K (Natural pyrope)

P= 5.5 G P a

P = 3.5 GPa d = 1 2 m

Post-garnet transformation

1000

2000 3000 Time, seconds

4000

Figure 7. The rate of the post-spinel and post-garnet transformations after Kubo et al. (2002a,b). The solid curve indicates the rate of the post-spinel transformation of Mg 2 Si0 4 at 28.2 GPa and 1283 K. The thin dotted curves are the rate of the post-garnet transformation in Mg3Al 2 Si30 12 at 31.4 GPa and 1473 K and 30.8 GPa and 1273 K. A thick dotted curve is the rate of the post-garnet transformation in natural pyropic garnet (Mg 0724 Fe 0184 Ca 0 nl ) 3 (Al 0872 Cr 000 4 4 Ti 0 010 ) 2 Si 3064 O 12 . The rate of the post-spinel transformation is significantly faster than that of the post-garnet transformation. AP, overpressure (a pressure interval from the equilibrium boundary): d, grain size of the sample.

Effect of Water on Mantle

Phase

Transitions

409

the wet conditions (Sano et al. 2006), whereas the temperature has to be raised to 1973 K for the same reaction time scale under dry conditions (Litasov et al. 2004). Water enhances the post-garnet transformation kinetics similar to the post-spinel transformation kinetics, although the reaction rate in the post-garnet transformation is sluggish compared to the post-spinel phase transformation. Although our preliminary studies suggest that both post-spinel and postgarnet transformation kinetics are enhanced by water, we need more quantitative studies by in situ X-ray diffraction on the reaction under wet conditions to quantify the effect of water. IMPLICATION FOR SEISMIC DISCONTINUITIES AND PHASE TRANSFORMATION BOUNDARIES UNDER DRY AND WET CONDITIONS 410 km seismic discontinuity and olivine-wadsleyite phase boundary The 410 km seismic discontinuity is considered to be caused by the transformation from olivine to wadsleyite in peridotite mantle. The average depth of the discontinuity is 411 km (Gu et al. 1998) and 418 km (Flanagan and Shearer 1998, 1999) using stacking SS and PP precursors. Figure 8 summarizes the shifts of the phase boundaries in the mantle as a result of the addition of water. Katsura et al. (2004) estimated the mean temperature at the 410 km seismic discontinuity to be 1760 ± 45 K for pyrolite mantle using the average depth of

Depth, km

Figure 8. Phase boundaries in peridotite and basalt under dry and wet conditions. The stability fields of wasleyite and ringwoodite expand under wet conditions at low temperatures (Litasov et al. 2005a,b), whereas the post garnet transformation shifts to lower pressures under wet conditions (Litasov et al. 2004; Sano et al. 2006). The grey dotted curve is the average temperature of the mantle (Akaogi et al. 1989). The shaded area represents the temperature range of the slabs (Kirby et al. 1996). The phase boundaries of the post-spinel and post-garnet transformations cross at around 1000 K (A) under dry conditions and 1400 K (B) under wet conditions. Rw, ringwoodite; Sb, superhydrous phase B; Pe, periclase. The other abbreviations are the same as those given in Figures 1 and 3.

410

Ohtani & Litasov

the discontinuity. As discussed in the previous section, the existence of water in the mantle (~1760 K) may elevate the depth of the 410 km discontinuity by a few kilometers. Global topography of the 410 km seismic discontinuity indicates depth variations of approximately ±20 km (Flanagan and Shearer 1998). Local seismological studies indicate greater depth variations in some areas. For example, the maximum elevation of the discontinuity in subduction zones is around 60-70 km, which is consistent with a temperature anomaly of about 1000 K if we apply the Clapeyron slope of the olivine-wadsleyite phase boundary of +3-4 MPa/K (e.g., Collier et al. 2001). Such a temperature anomaly of 1000 K at the 410 km discontinuity is too large given that the normal temperature of the discontinuity is around 1760 K based on the geotherm estimated by some authors (e.g., Akaogi et al. 1989; Katsura et al. 2004). Since water and temperature have a similar effect on the phase boundary, the topography of the 410 km seismic discontinuity in some regions may be also explained by the presence of water in subduction zones, i.e., some fraction of the topography of the discontinuity may be explained by the effect of water in combination with the temperature effect as shown in Figure 8. The width of the 410 km discontinuity varies from 4 to 35 km (Shearer 2000). The phase relations in the olivine composition in peridotite indicate a width of the binary loop of coexistence of olivine and wasleyite is 25 km at 1473 K and 14 km at 1873 K (Katsura and Ito 1989). Frost (2003) reviewed recent experimental and thermodynamic data on the olivine-wadsleyite transformation and discussed the possibility of a 4-6 km width, i.e., the minimum of the range of the width of discontinuity. Frost (2003) found this width to be consistent with the olivinewadsleyite phase transformation in a peridotite composition containing garnet and pyroxenes, which makes the phase boundary sharper due to partitioning of elements between minerals. There are several factors that make the phase boundary broader. First, we can expect a broader transformation in colder, garnet-poor or FeO-rich regions in the mantle based on the equilibrium phase relations. Second, the reaction kinetics at low temperature also tends to broaden the width of the discontinuity since olivine and wadsleyite can coexist metastably under the conditions of a cold subducting slab (e.g., Rubie and Ross 1994; Kubo et al. 1998). Third, the sharpness of the discontinuity may be affected by the presence of water. Wood (1995) used a thermodynamic calculation to estimate that the presence of a small amount of water, i.e., about 100 ppm, can broaden the stability field of coexistence of olivine and wadsleyite by about 3 km. He argued that the sharpness of the 410 km discontinuity indicates that the transition zone is essentially dry. On the other hand, Smyth and Frost (2002) suggested that the hydrogen diffusion and gravitational stratification narrow the phase transformation interval between olivine and wadsleyite, which is an opposite effect from the argument by Wood (1995). The preliminary experiments on determination of the olivine-wadsleyite transformation in peridotite-3.0 wt% water system made by Litasov et al. (2006) suggested that coexistence loop of olivine and wadsleyite expands under the wet conditions, which is consistent with the estimation by Wood (1995). Nolet and Zielhuis (1994), Revenaguh and Sipkin, (1994), and Song et al. (2004) suggested water as a possible cause of anomalies in the deep upper mantle and the transition zone. Recent observations that the width of the 410 km seismic discontinuity beneath the Mediterranean is between 20 and 35 km (Van der Meijde et al. 2003) may be explained by H 2 0 contents in this region. The cold and wet nature of this area may be supported by the high electrical conductivity of the upper mantle in this region, which is consistent with 1000-1500 ppm H 2 0 in olivine (Tarits et al. 2004). The 660 km seismic discontinuity and the post-spinel transformation: average depth and topography of the 660 km seismic discontinuity The 660 km seismic discontinuity has been considered to be caused by the post-spinel transformation in peridotite (e.g., Ito and Takahashi 1989). Seismological studies show that the 660 km discontinuity is observed globally and the average depth of the discontinuity

Effect of Water on Mantle

Phase

Transitions

411

is 654 km (Gu and Dziewonski 2002). If we adopt this depth as a global average for the discontinuity, the discrepancy between the pressure of the post-spinel transformation (i.e., decomposition of ringwoodite) determined by experiments and that in seismic studies becomes small. The phase boundary determined experimentally is lower by 0.2-0.3 GPa (i.e., 6-10 km shallower) compared with the pressure of the 660 km discontinuity assuming the temperature of the discontinuity around 1800-2000 K, when we apply the pressure scales of Au and MgO (Tsuchiya 2003; Speziale et al. 2001), which is believed to be the most reliable at present (Figs. 2A,B and Fig. 8). The above small discrepancy may be accounted for by the effect of pressure on the thermocouple EMF, i.e., the temperatures of the high pressure in situ X-ray diffraction study may be underestimated, resulting in a shift of the phase boundary to higher pressure (e.g., Ohtani 1979). The alternative explanation for the discrepancy may be the effect of water on the phase boundary. We observed a change of the Clapeyron slope of the post-spinel phase boundary under hydrous conditions, but found no significant shift of the boundary at higher temperatures around 1773-1873 K as shown in the previous section. On the other hand the boundary appears to shift to higher pressure at lower temperatures; i.e., Higo et al. (2001) and Inoue et al. (2001) reported 0.2 GPa shift of the phase boundary at 1873 K, and Katsura et al. (2003) also suggested 0.6 GPa shift of the post-spinel transition boundary to the higher pressure in hydrous Mg 2 Si0 4 based on their preliminary results at 1663 K. Our data indicate that the post-spinel transformation boundary in hydrous pyrolite shifts to higher pressure by about 0.6 GPa relative to anhydrous pyrolite at 1473 K, whereas there is no obvious shift of this boundary at higher temperatures (1773-1873 K). The Clapeyron slope of the post-spinel transformation determined recently by in situ X-ray diffraction experiments is very gentle (about -0.5 MPa/K) under dry conditions (e.g., Katsura et al. 2003; Fei et al. 2004a; Litasov et al. 2005b). The gentle slope of the postspinel transformation boundary may require a large temperature difference to account for the topography of the 660 km seismic discontinuity (Figs. 2A,B, Fig. 8). The depth of the 660 km seismic discontinuity varies by about ±20 km; about +20 km beneath Northern Pacific Ocean, Atlantic Ocean, and South Africa, whereas it is - 2 0 km beneath subduction zones such as the western Pacific and South America (Flanagan and Shearer 1998; Gu and Dziewonski 2002). Lebedev et al. (2003) proposed smaller variations in the range of ±15 km. The depressions caused by the slabs have been studied by many authors; depression of 20~30 km beneath Tonga (Niu and Kawakatsu 1995) and up to about 50 km beneath Izu-Bonin (e.g., Collier et al. 2001). Elevation of the 660 km discontinuity by 10-20 km is also reported in areas related to hot plumes such as Hawaii (Li et al. 2000) and the South Pacific (Niu et al. 2000). If the topography of the discontinuity is caused only by the effect of temperature on the phase boundary, an elevation of 20 km corresponds to a temperature elevation of about 1300 K and a depression of 50 km corresponds to a temperature decrease of about 3000 K, which implies unusually if not impossibly large temperature variations. This indicates that the variation in depth of the 660 km discontinuity cannot be explained only by the temperature effect, but we need to introduce the other effects to explain the topography. A delay of the phase transformation due to kinetics, or the influence of minor components or volatiles may be additional explanations for the large variations in 660 topography. Kinetics of the post spinel transformation in Mg 2 Si0 4 (Kubo et al. 2002a) indicates that the transformation is very fast compared to the speed of the slab subduction and it completes in 104 years with 1 GPa of overpressure (i.e., a pressure interval from the equilibrium phase boundary) even at the temperatures of 1000 K (Fig. 7). Only for very cold slabs with temperatures below 1000 K, the post-spinel transformation reaction delays by 106-108 years, a meaningful duration to explain a large depression of the 660 km discontinuity. We can expect a delay of transformation and shift of the phase boundary by about 1 GPa (corresponding to 20-25 km) due to kinetics only in unusually cold slabs.

Ohtani & Litasov

412

The most plausible explanation for such a large elevation and depression of the discontinuity may be the presence of water, although the elevation may be smaller due to lower water solubility in ringwoodite in hot environments. The observed shift of the postspinel transformation boundary is relevant to the topography of the 660 discontinuity detected by the seismological studies (Fig. 8). The displacement of the post-spinel phase boundary by 0.5 GPa under the hydrous conditions corresponds to about 15 km in depth, which is a half of the observed depressions of the 660 km discontinuity, 30-40 km, in the hot subduction zones at a temperature about 1473 K. Thus, a large displacement of the 660 km discontinuity may be considered as evidence for existence of water in the transition zone (Litasov et al. 2005a,b). The density relation of basalt and peridotite near the 660 km discontinuity Ringwood (1994) drew attention to the importance of the density contrast between the basaltic and peridotite layers of a subducting slab. He argued that there is a density crossover between peridotite and basalt due to the pressure difference between the post-spinel transition in the peridotite layer and the post-garnet transition in the basaltic layer of the slabs. Figure 8 summarizes the phase relations of peridotite and basalt under dry and wet conditions and shows the range of temperatures expected for subducting slabs and an average mantle geotherm. Figure 9 shows the density of the peridotite layer and the basalt layer of the slab. Irifune and Ringwood (1993) suggested that the density crossover occurs in the pressure interval of about 2 GPa along the slab geotherm. The density crossover may lead to a separation of the basaltic crust from the peridotite body at a depth of about 660 km resulting in the formation of a garnetite-bearing layer at the base of the transition zone (Ringwood 1994). On the other hand, Hirose et al. (1999) showed that the transformation of the basaltic crust from garnetite to a perovskite lithology occurs near 720 km and they argued that the density crossover between

Depth, km 600

640

680

720

760

800

840

Figure 9. Density profiles of dry and wet basalt (MORB) and pyrolite demonstrating the density crossover near 660 km. Density calculations were carried out along a normal mantle geotherm for anhydrous systems (Akaogi et al. 1989) and a cold subduction geotherm for hydrous systems (Kirby et al. 1996) using a third order Birch-Murnaghan equation of state and the set of thermoelastic parameters given by Litasov and Ohtani (2005). A density crossover exists between the peridotite (thin dotted curve) and basalt (thin solid curve) in the range from 23 GPa to 27 GPa under dry conditions. The sluggish transformation rate of the postgarnet transformation expands the region of the density crossover under dry conditions. There is no density crossover between the peridotite (thick dotted curve) and basalt (thick solid curve) under wet conditions.

Effect of Water on Mantle

Phase

Transitions

413

basalt and peridotite layers may be too narrow for separation of the basaltic crust to accumulate at the base of the transition zone. Therefore, the basaltic crust may gravitationally sink into the lower mantle. However, the equilibrium phase relations may not be applicable to the phase transformation in real subducting slabs, because reaction kinetics may be an important factor in controlling the mineralogy and density of subducting slabs. Kubo et al. (2002b) studied the kinetics of the post-garnet transformation in basalt, and they showed a very slow reaction rate allowing metastable garnet to survive for a long time (the order of 10 Ma) after crossing the 660 km seismic discontinuity. Therefore, a wider pressure interval of the density crossover can be expected by taking into account the kinetics of the post-garnet transformation under dry conditions, suggesting a separation of the basaltic layer of the slabs to form a garnetite layer at the base of the transition zone (Fig. 9). Litasov et al. (2004) and Sano et al. (2006) demonstrated that at the low temperature of slabs the density crossover between the peridotite and basalt layers might be absent (Figs. 8 and 9) especially in water-rich subduction zones. The crossover of the post-spinel and postgarnet transformation boundaries locates below the temperature path of a cold subducting slab in anhydrous subduction (point A in Fig. 8), whereas it lies above the temperature path of a hot slab at around 24 GPa and 1500 K (point B in Fig. 8) in hydrous subduction. Therefore, there may be no density crossover between the basalt and peridotite layers of hydrated slabs following cold subduction geotherms. If slabs pass through the 660 km seismic discontinuity, penetration of the basaltic crust component into the lower mantle can occur without gravitational separation from the peridotite body of the slab, at least for hydrous subduction environments. Seismic reflectors: the possible existence of fluid in the lower mantle Kaneshima and Helfrich (1998) and Niu et al. (2003) reported seismic reflectors in the lower mantle. Niu et al. (2003) indicate that the physical properties of the reflector observed in the upper part of the lower mantle beneath the Mariana subduction zone show a decrease in shear wave velocity by 2-6% and an increase in density by 2-9% within the reflectors, whereas, no difference exists in P-wave velocity (

Figure 10. Differences (%) in density and vP between basalt (MORB) in the slab and surrounding mantle peridotite (A) and those in density and v s between basalt and peridotite (B). The density and velocity (vP and v s ) differences between the reflector and surrounding mantle observed seismologically (Niu et al. 2003) are also shown in these figures. Density, vP, and v s of the mantle minerals are calculated at conditions of 30 GPa and 1873 K by using the procedure and parameters given in Ohtani (2006). Open squares and circles represent the v P -density jump and v s -density jump, respectively. The shaded areas indicate that the density and velocity differences of the majorité bearing assemblage and perovskite bearing assemblage. The properties of the reflectors cannot be explained either by a stable perovskite bearing assemblage nor a metastable majorité bearing assemblage. Abbreviations in the Figure are the same as those given in Figures 3 and 8.

properties of the reflector which cannot be accounted for by either the perovskite lithology or metastable majorité lithology of basalt, i.e., by a complete or incomplete phase transformation from majorité to perovskite lithologies, although there is a large uncertainty in seismic velocity, especially vs in basalt. The detailed calculations and the error estimations of the vP and vs are given in Ohtani (2006). We need additional factors to reduce v s , although the density increase can be explained by the lower mantle lithology; i.e., the complete phase transformation from majorité lithology to perovskite lithology shows positive jumps in density, vP and vs relative to the surrounding mantle, whereas the metastable majorité lithology shows physical properties (p, v'p, and v's) that are smaller than the surrounding mantle. It may be possible to explain the drastic decrease in v s as an effect of fluid or melt films in the subducted oceanic lithosphère in the lower mantle (e.g., Williams and Garnero 1996). Hydrous phase D dehydrates at pressures around 40-50 GPa (Shieh et al. 1998). Therefore, the physical properties of the reflectors might be explained by fluid in the oceanic crust supplied from dehydration of this phase in the underlying hydrous peridotite layer of the slab penetrating into the lower mantle. Although the existence of fluid films in the oceanic crust of the slabs is a plausible mechanism for the change of the physical properties of reflectors, the existence of minor metallic iron formed by the garnet-perovskite transformation in the oceanic crust (Miyajima et al. 1999; Frost et al. 2004) could also cause the reduction of shear wave velocity and increase the density of the reflectors. In addition, several other explanations might also account for the unusual properties of the reflectors. Theoretical calculations, for example, have revealed the existence of an elastic anomaly associated with the phase transformation from stishovite to the post-stishovite CaCl2 phase (e.g., Stixrude 1998). A similar anomaly could result from the transformation in A1 2 0 3 bearing CaSi0 3 perovskite from the orthorhombic to cubic phase (Kurashina et al. 2004). The elastic anomalies associated with theses phase transformations, however, are not yet confirmed experimentally.

Effect of Water on Mantle

Phase

Transitions

415

CONCLUDING REMARKS In this chapter we have summarized recent advances in the study of the effects of water on major phase transformations in the Earth's mantle with implications for the topography of seismic discontinuities and mantle dynamics based on the results of quenched multianvil and in situ X-ray diffraction experiments at high pressure and temperature. Differences in water solubility between (a) wadsleyite and olivine and (b) ringwoodite and Mg-perovskite and ferropericlase may cause displacement of the phase transition boundaries, which are believed to be responsible for the 410 and 660 km discontinuities, respectively. Experimental results show that water may increase the pressure interval of the binary olivine-wadsleyite transformation loop, i.e., water expands the stability field of wadsleyite to lower pressures. This interval is ~0.3 GPa (or ~6 km) wide at 1700 K in the anhydrous pyrolite, whereas it may be ~1.2 GPa at 1500 K (or 33 km) in pyrolite with 2-3 wt% of H 2 0. The broadening of the 410 km discontinuity observed in some regions of the mantle is consistent with enrichment by water in the mantle. A significant shift of the boundary of the wadsleyite to ringwoodite phase transformation to the higher pressure through the addition of water to the peridotite system may also be responsible for variations in the depth of the 520 km discontinuity. However, the cause of a significant displacement of the wadsleyite/ringwoodite phase boundary is not clear. In situ X-ray diffraction study of the post-spinel transformation in hydrous pyrolite indicates that the phase boundary is shifted to higher pressures by 0.6 GPa relative to anhydrous pyrolite at 1473 K, whereas it shows no obvious shift at higher temperature of around 1873 K. The displacement of the post-spinel phase transformation boundary in hydrous pyrolite would correspond to a depth variation of about 15 km and may account for approximately half of 30-40 km depression observed for the 660 km discontinuity in subduction zones at temperatures of around 1473 K. This effect should be much stronger at lower temperatures. Since the Clapeyron slope of the post-spinel transformation boundary in anhydrous pyrolite is very gentle (about -0.5 MPa/K), we cannot explain depressions of the 660 km discontinuity as a result of temperature variations alone. Thus, a large depression of the 660 km discontinuity might be considered as direct evidence for the existence of water in the transition zone. In situ X-ray diffraction studies of the post-garnet transformation in anhydrous and hydrous basalt show that the phase boundary shifts to lower pressures by ~2 GPa upon the addition of water. This observation demonstrates that at temperatures of subducting slabs the density crossover between peridotite and basalt near 660 km might be absent under hydrous conditions. The basaltic component of the slab may therefore penetrate into the lower mantle under hydrous conditions and not separate from the peridotite body at the 660 km discontinuity.

ACKNOWLEDGMENTS We thank A. Suzuki, T. Kubo, H. Terasaki, K. Funakoshi, T. Kondo for collaboration during the experiments at SPring-8. We appreciate S.D. Jacobsen and an anonymous reviewer for improving the manuscript. This work was supported by the grants in Aid for Scientific Researches from the Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 14102009 and 16075202) to E. Ohtani. This work was conducted as a part of the 21th Century Center-of-Excellence program Advanced Science and Technology Center for the Dynamic Earth at Tohoku University.

REFERENCES Agee CB (1998) Phase transformations and seismic structure in the upper mantle and transition zone. Rev Mineral 37:165-203

416

Ohtani & Litasov

Akaogi M, Ito E, Navrotsky A (1989) Olivine-modified spinel-spinel transitions in the system Mg 2 Si0 4 Fe 2 Si0 4 : calorimetric measurements, thermochemical calculation, and geophysical application. J Geophys Res 94:15671-15685 Anderson OL, Issak DG, Yamamoto S (1989) Anharmonicity and the equation state for gold. J Appl Phys 6: 1534-1543 Bell DR, Rossman GR, Maldener J, Endisch D, Rauch F (2003) Hydroxide in olivine: A quantitative determination of the absolute amount and calibration of the IR spectrum. J Geophys Res 108, doi: 10.1029/2001JB000679 Benz HM, Vidale JE (1993) Sharpness of upper-mantle discontinuities determined from high-frequency reflections. Nature 365:147-150 Bina CR, Helffrich GR (1994) Phase transition Clapeyron slopes and transition zone seismic discontinuity topography. J Geophys Res 99:15853-15860 Bolfan-Casanova N, Keppler H, Rubie DC (2000) Water partitioning between nominally anhydrous minerals in the M g 0 - S i 0 2 - H 2 0 system up to 24 GPa: implications for the distribution of water in the Earth's mantle. Earth Planet Sci Lett 182:209-221 Bolfan-Casanova N, Mackwell S, Keppler H, McCammon C, Rubie DC (2002) Pressure dependence of H solubility in magnesiowiistite up to 25 GPa: Implications for the storage of water in the Earth's lower mantle. Geophys Res Lett 29, doi: 10.1029/2001GL014457 Bolfan-Casanova N, Keppler H, Rubie DC (2003) Water partitioning at 660 km depth and evidence for very low water solubility in magnesium silicate perovskite. Geophys Res Lett 30, doi: 10.1029/2003GL017182 Brown JM (1999) High pressure equation of state for NaCl, KCL, and CsCl. J Appl Phys 86:5801-5808 Chen J, Inoue T, Weidner DJ, Wu Y (1998) Strength and water weakening of mantle minerals, a , p, and y Mg 2 Si0 4 . Geophys Res Lett 25:575-578 Chen J, Inoue T, Yurimoto H, Weidner DJ (2002) Effect of water on olivine-wadsleyite phase boundary in the (Mg,Fe) SiO system. Geophys Res Lett 29:1875, doi:10.1029/2001 GL014429 ChudinovskiBi L, fioehler R (2001) High-pressure polymorphs of olivine and the 660 km seismic discontinuity. Nature 411:574-577 Chung JI, Kagi H (2002) High concentration of water in stishovite in the MORB system. Geophys Res Lett 29, doi: 10.1029/2002GL015579 Collier JD, Helffrich GR, Wood BJ (2001) Seismic discontinuities and subduction zones. Phys Earth Plane Inter 127:35-49 Demouchy S, Deloule E, Frost DJ, Keppler H (2005) Pressure and temperature dependence of water solubility in Fe-free wadsleyite. Am Mineral 90:1084-1091 Dziewonski AM, Anderson DL (1981) Preliminary Reference Earth Model. Phys Earth Planet Int 25:297-356 Fei Y, Van Orman J, Li J, van Westrenen W, Sanloup C, Minarik W, Hirose K, Komabayashi K, Walter M, Funakoshi K (2004a) Experimentally determined postspinel transformation boundary in Mg 2 Si0 4 using MgO as an internal pressure standard and its geophysical implications. J Geophys Res 109, doi: 10.1029/ 2003JB002562 Fei Y, Li J, Hirose K, Minarik W, Van Orman J, Sanloup C, van Westrenen W, Komabayashi T, Funakoshi K (2004b) A critical evaluation of pressure scales at high temperatures by in situ X-ray diffraction measurements. Phys Earth Planet Inter 143-144:515-526 Flanagan MP, Shearer PM (1998) Global mapping of topography on transition zone velocity discontinuities by stacking SS precursors. J Geophys Res 103:2673-2692 Flanagan MP, Shearer PM (1999) A map of topography on the 410 k m discontinuity from PP precursors. Geophys Res Lett 26:549-552 Frost DJ (2003) The structure and sharpness of (Mg,Fe) 2 Si0 4 phase transformations in the transition zone. Earth Planet Sci Lett 216:313-328 Frost DJ, Liebske C, Langenhorst F, McCammon C, Tronnes RG, Rubie DC (2004) Experimental evidence of iron-rich metal in the Earth's lower mantle. Nature 428:409-411 Fukao Y, Koyama T, Obayashi M, Utada H (2004) Trans-Pacific temperature field in the mantle transition zone derived from seismic and electromagnetic tomography. Earth Planet Sci Lett 217:425-434 Gu YJ, Dziewonski AM (2002) Global variability of transition zone thickness. J Geophys Res 107, doi: 10/1029/ 2001JB000489 Gu YJ, Dziewonski AM, Agee CB (1998) Global de-correlation of the topography of the transition zone discontinuity. Earth Planet Sci Lett 157:57-67 Hae R, Ohtani E, Kubo T, Koyama T, Utada H (2006) Hydrogen diffusivity in wadsleyite and water distribution in the mantle transition zone. Earth Planet Sci Lett 243:141-148 Higo Y, Inoue T, Irifune T, Yurimoto H (2001) Effect of water on the spinel post-spinel transformation in Mg 2 Si0 4 . Geophys Res Lett 28:3505-3508 Hirose K (2002) Phase transitions in pyrolitic mantle around 670 km depth: implications for upwelling of plumes from the lower mantle. J Geophys Res 107, doi: 10.1029/2001JB000597

Effect of Water on Mantle Phase

Transitions

417

Hirose K, Fei Y (2002) Subsolidus and melting phase relations of basaltic composition in the uppermost lower mantle. Geochim Cosmochim Acta 66:2099-2108 Hirose K, Fei Y, Ma Y, Mao HK (1999) The fate of subducted basaltic crust in the Earth's lower mantle. Nature 397:53-56 Hirth G, Kohlstedt DL (1996) Water in the oceanic upper mantle: Implications for rheology, melt extraction and the evolution of the lithosphere. Earth Planet Sci Lett 144:93-108 Hosoya T, Kubo T, Ohtani, E, Sano A, Funakoshi K (2005) Water controls the fields of metastable olivine in cold subducting slabs. Geophys Res Lett 32:L17305 doi:10.1029/2005GL023398 Inoue T (1994) Effect of water on melting phase relations and melt compositions in the system Mg 2 Si0 4 MgSi0 3 -H 2 0 up to 15 GPa. Phys Earth Planet Inter 85:237-263 Inoue T, Weidner DJ, Northrup PA, Parise JB (1998) Elastic properties of hydrous ringwoodite (y-phase) in Mg 2 Si0 4 . Earth Planet Sci Lett 160:107-113 Inoue T, Higo Y, UedaT, TanimotoY, IrifuneT (2001) The effect of water on the high-pressure phase boundaries in the system Mg 2 Si04-Fe 2 Si04. Transport of Materials in the Dynamic Earth, Kurayoshi, p 128-129 Inoue T, Tanimoto Y, Irifune T, Suzuki T, Fukui H, Ohtaka O (2004) Thermal expansion of wadsleyite, ringwoodite, hydrous wadsleyite and hydrous ringwoodite. Phys Earth Planet Inter 143-144:279-290 Irifune T, Ringwood AE (1993) Phase transformations in sbducted oceanic crust and buoyancy relationships at depths of 600-800 km in the mantle. Earth Planet Sci Lett 117:101-110 Irifune T, Nishiyama N, Kuroda K, Inoue T, Isshiki M, Utsumi W, Funakoshi K, Urakawa S, Uchida T, Katsura T, Ohtaka O (1998) The postspinel phase boundary in Mg 2 Si0 4 determined by in situ X-ray diffraction. Science 279:1698-1700 Ito E, Takahashi E (1989) Post-spinel transformation in the system Mg 2 Si0 4 -Fe 2 Si0 4 and some geophysical implications. J Geophys Res 94:10637-10646 Jacobsen SD, Smyth JR, Spetzler H, Holl CM, Frost DJ (2004) Sound velocities and elastic constants of ironbearing hydrous ringwoodite. Phys Earth Planet Inter 143-144:47-56 Jamieson JC, Fritz JN, Manghnani MH (1982) Pressure measurements at high temperature in X-ray diffraction studies: gold as a primary standard. In: High pressure research in geophysics. Akimoto S, Manghnani MH (ed) Center for Academic Publications Japan, Tokyo, p 27-48 Kaneshima S, Helffrich G (1998) Detection of lower mantle scatterers northeast of the Mariana subduction zone using short-period array data. J Geophys Res 103:4825-4838 Kaneshima S, Helffrich G (1999) Dipping lower-velocity layer in the mid-lower mantle: evidence for geochemicalheterogeneity. Science 283:1888-1891 Karato S (1989) Grain growth kinetics in olivine aggregates. Tectonophysics 168:255- 273 Karato S (1990) The role of hydrogen in the electrical conductivity of the upper mantle. Nature 347: 72-273 Katayama I, Hirose K, Yurimoto H, Nakashima S (2003) Water solubility in majoritic garnet in subduction oceanic crust. Geophys Res Lett 30, doi: 10.1029/2003GL018127 Katsura T, Ito E (1989) The system Mg 2 Si0 4 -Fe 2 Si0 4 at high pressures and temperatures: Precise determination of stabilities of olivine, modified spinel, and spinel. J Geophys Res 94:15663-15670 Katsura T, Yamada H, Shinmei T, Kubo A, Ono S, Kanzaki M, Yoneda A, Walter MJ, Ito E, Urakawa S, Funakosh Ki, Utsumi W (2003) Post-spinel transition in Mg 2 Si0 4 determined by high P-T in situ X-ray diffraction. Phys Earth Planet Inter 136:11-24 Katsura T, Yamada H, Nishikawa O, Song M, Kubo A, Shinmei T, Yokoshi S, Aizawa Y, Yoshino T, Walter MJ, Ito E, Funakoshi K (2004) Olivine-wadsleyite transition in the system (Mg,Fe) 2 Si0 4 . J Geophys Res 109: B02209, doi: 10.1029/2003JB002438 Kawamoto T (2004) Hydrous phase stability and partial melt chemistry in H 2 0-saturated KLB-1 peridotite up to the uppermost lower mantle conditions. Phys Earth Planet Inter 143-144:387-395 Kawamoto T, Matsukage K, Mibe K, Isshik M, Nishimura K, Ishimatsu N, Ono S (2004) Mg/Si ratios of aqueous fluids coexisting with forsterite and enstatite based on the phase relations in the Mg 2 Si0 4 -Si0 2 H 2 0 system. Am Mineral 89:1433-1437 Kerschhofer L, Rubie DC, Sharp TG, McConnell JDC, Dupas-Bruzek C (2000) Kinetics of intercrystalline olivine-ringwood transformation. Phys Earth Planet Inter 121:59-76 Kirby SH, Stein S, Okal EA, Rubie DC (1996) Metastable mantle phase transformations and deep earthquakes in subducting oceanic lithosphere. Rev Geophys 34:261-306 Koyama T, Shimizu H, Utada H, Ohtani E, Hae R (2006) Water contents in the mantle transition zone beneath the north Pacific derived from the electrical conductivity anomaly. In: Earth's Deep Water Cycle. Jacobsen S, van der Lee S (eds) Am Geophys Union Monogr (in press) Kohlstedt DL, Keppler H, Rubie DC (1996) Solubility of water in the a, p, and y phases of (Mg,Fe) 2 Si0 4 . Contrib Mineral Petrol 123:345-357 Koper KD, Wiens DA, Dorman LM, Hildebrand JA, Webb SC (1998) Modeling the Tonga slab: can travel time data resolve a metastable olivine wedge? J Geophys Res 103:30079-30100 Kubo A, Akaogi M (2000) Post-garnet transitions in the system Mg 4 Si 4 0 12 -Mg 3 Al 2 Si 3 0 12 up to 28 GPa: phase relations of garnet, ilmenite and perovskite. Phys Earth Planet Inter 121:85-102

418

Ohtani & Litasov

Kubo T, Ohtani E, Kato T, Shinme T, Fujino K (1998) Effect of water on the a - p transformation kinetics in San Carlos Olivine. Science 281:85-87 Kubo T, Ohtani E, Kato T, Urakawa S, Suzuki A, Kanbe Y, Funakoshi K, Utsumi W, Kikegawa T, Fujino K (2002a) Mechanism and kinetics of the post-spinel transformation in Mg 2 Si0 4 . Phys Earth Planet Inter 129:153-171 Kubo T, Ohtani E, Kondo T, Kato T, Toma M, Hosoya T, Sano A, Kikegawa T, Nagase T (2002b) Metastable garnet in oceanic crust at the top of the lower mantle. Nature 420:803-806 Kubo T, Ohtani, E Funakoshi K (2004) Nucleation and growth kinetics of the a - p transformation in Mg 2 Si0 4 determined by in situ synchrotron powder X-ray diffraction. Am Mineral 89:285-293 Kurashina T, Hirose K, Ono S, Sata N, Ohishi Y (2004) Phase transition in Al-bearing C a S i 0 3 perovskite: implications for seismic discontinuities in the lower mantle. Phys Earth and Planet Inter 145:67-74 Kurosawa M, Yurimoto H, Sueno S (1997) Patterns in the hydrogen and trace element compositions of mantle olivines. Phys Chem Mineral 24:385-395 Lebedev S, Chevrot S, van der Hilst RD (2003) Correlation between the shear-speed structure and thickness of the mantle transition zone. Phys Earth Planet Inter 136:25-40 Li X, Kind R, Priestley K, Sobolev SV, Tilmann F, Yuan X, Weber M (2000) Mapping the Hawaiian plume conduit with converted seismic waves. Nature 405:938-941 Litasov KD, Ohtani E (2003) Stability of hydrous phases in CMAS-pyrolite-H 2 0 system up to 25 GPa. Phys Chem Mineral 30:147-156 Litasov KD, Ohtani E, Langenhorst F, Yurimoto H, Kubo T, Kondo T (2003) Water solubility in Mg-perovskites and water storage capacity in the lower mantle. Earth Planet Sci Lett 211:189-203 Litasov KD, Ohtani E, Suzuki A, Kawazoe T, Funakoshi K (2004) Absence of density crossover between basalt and peridotite in the cold slabs passing through 660 k m discontinuity. Geophys Res Lett 31, doi:10.1029/ 2004GL021306 Litasov KD, Ohtani E (2005) Phase relations in hydrous MORB at 18-28 GPa: implications for heterogeneity of the lower mantle. Phys Earth Planet Inter 150:239-263 Litasov KD, Ohtani E, Sano A, Suzuki A, Funakoshi K (2005a) Wet subduction versus cold subduction. Geophys Res Lett 32:L13312, doi:10.1029/2005GL022921 Litasov KD, Ohtani E, Sano A, Suzuki A, Funakoshi K (2005b) In situ X-ray diffraction study of post-spinel transformation in a peridotite mantle: Implication to the 660 k m discontinuity. Earth Planet Sci Lett 238: 311-328 Litasov KD, Ohtani E, Sano A (2006) Influence of water on major phase transitions in the earth's mantle. In: Earth's Deep Water Cycle. Jacobsen S, van der Lee S (eds) Am Geophys Union Monogr (in press) Liu M, Kerschhofer L, Mosenfelder JL, Rubie DC (1998) The effect of strain energy on growth rates during the olivine-spinel transformation. J Geophys Res 103:23897-23909 Matsui M, Parker SC, Leslie M (2000) The M D simulation of the equation of state of MgO: application as a pressure calibration standard at high temperature and high pressure. Am Mineral 85:312-316 Matsukage KN, Jing Z, Karato S (2005) Density of hydrous silicate melt at the conditions of Earth's deep upper mantle. Nature 438:488-491, doi: 10.1038/nature04241 McCammon C (1997) Perovskite as a possible sink for ferric iron in the lower mantle. Nature 387:694-696 Mei S, Kohlstedt DL (2000) Influence of water on plastic deformation of olivine aggregates. 1. Diffusion creep regime. J Geophys Res 105:21457-21470 Miyajima N, Fujino K, Funamori N, Kodo T, Yagi T (1999) Garnet-perovskite transformation under conditions of the Earth's lower mantle: an analytical transmission electron microscopy study. Phys Earth Planet Inter 116:117-131 Morishima H, Kato T, Suto M, Ohtani E, Urakawa S, Utsumi W, Shimomura O, Kikegawa T (1994) The phase boundary between a - and p-Mg 2 Si0 4 determined by in situ X-ray observation. Science 265:1202-1203 Mosenfelder JL, Marton FC, Ross C R I I , Kershhofer R, Rubie DC (2001) Phys Earth Planet Inter 127:165-180 Nishiyama N, Irifune T, Inoue T, Ando J, Funakoshi K (2004) Precise determination of phase relations in pyrolite across the 660 k m seismic discontinuity by in situ X-ray diffraction and quench experiments. Phys Earth Planet Inter 143-144:185-199 Niu F, Kawakatsu H (1995) Direct evidence for the undulation of the 660 km discontinuity beneath Tonga: Comparison of Japan and California array data. Geophys Res Lett 22:531-534 Niu F, Kawakatsu H, Fukao Y (2003) A slightly dipping and strong seismic reflector at mid-mantle depth beneath the Mariana subduction zone. J Geophys Res 108, doi:10.1029/2002JB002384 Niu F, Inoue H, Suegetsu D, Kanjo K (2000) Seismic evidence for a thinner mantle transition zone beneath the South Pacific Superswell. Geophys Res Lett 27:1981-1984 Nolet G, Zielhuis, A (1994) Low S velocities under the Tornquist-Teisseyre zone: Evidence for water injection into the transition zone by subduction. J Geophys Res 99:15813-15821, doi: 10.1029/94JB00083 Oguri K, Funamor N, Uchida T, Miyajima N, Yagi T, Fujino K (2000) Post-garnet transition in a natural pyrope: a multi-anvil study based on in situ X-ray diffraction and transmission electron microscopy. Phys Earth Planet Inter 122:175-186

Effect of Water on Mantle Phase

Transitions

419

Ohtani E (1979) Melting relation of Fe 2 Si0 4 up to about 200 kbar. J Phys Earth 27:189-203 Ohtani E (2006) Recent progress in experimental mineral physics: phase relations of hydrous systems and the role of water in slab dynamics. In: Earth's deep mantle: Structure, Composition, and evolution. Geophys Monograph Series 160. van der Hilst RD, Bass J, Matas J, Trampert J (eds), Am Geophys Union, p 321334 Ohtani E, Touma M, Litasov K, Kubo T, Suzuki A (2001) Stability of hydrous phases and water storage capacity in the transitional zone and lower mantle. Phys Earth Planet Inter 124:105-117 Ohtani E, Litasov K, Hosoya T, Kubo T, Kondo T (2004) Water transport into the deep mantle and formation of a hydrous transition zone. Phys Earth Planet Inter 143-144:255-269 Paterson MS (1982) The determination of hydroxyl by infrared absorption in quartz, silicate glasses and similar materials. Bull Mineral 105:20-29 Revenaugh, J, Sipkin SA (1994) Seismic evidence for silicate melt atop the 410 k m mantle discontinuity. Nature 369:474-476 Ringwood AE (1994) Role of the transition zone and 660 k m discontinuity in mantle dynamics. Phys Earth Planet Inter 86:5-24 Rubie DC, Thompson AB (1985) Metamorphic Reactions: Kinetics, Textures, and Deformation. In: Advances in Physical Geochemistry. Vol 4. Thompson AB, Rubie DC (eds) Springer-Verlag, p 27-79 Rubie DC, Ross II CR (1994) Kinetics of the olivine-spinel transformation in subducting lithosphere: experimental constraints and implications for deep slab processes. Phys Earth Planet Inter 86:223-241 Sano A, Ohtani E, Litasov KD, Kubo T, Hosoya T, Funakoshi K, Kikegawa T (2006) Effect of water on garnetperovskite transformation in MORB and implications for penetrating slab into the lower mantle. Phys Earth Planet Inter (in press) Sakamaki T, Suzuki A, Ohtani E (2006) Stability of hydrous melt at the base of the Earth's upper mantle. Nature 439:192-194 Shearer PM (1996) Transition zone velocity gradients and the 520 km discontinuity. J Geophys Res 101:30533066 Shearer PM (2000) Upper mantle seismic discontinuities. In: Earth's Deep Interior: Mineral Physics and Tomography from the Atomic to the Global Scale. Geophysical Monograph 117. Karato S, Forte AM, Liebermann RC, Masters G, Stixrude L (eds) Am Geophys Union, p 115-131 Shieh SR, Mao HK, Ming JC (1998) Decomposition of phase D in the lower mantle and the fate of dense hydrous silicates in subducting slabs. Earth Planet Sci Lett 159:13-23 Shim SH, Duffy TS, Shen G (2001) The post-spinel transformation in Mg 2 Si0 4 and its relation to the 660 km seismic discontinuity. Nature 411:571-574 Shim SH, Duffy TS, Takemura K (2002) Equation of state of gold and its application to the phase boundaries near 660 k m depth in Earth's mantle. Earth Planet Sci Lett 203:729-739 Smyth J, Frost DJ (2002) The effect of water on the 410 k m discontinuity: An experimental study. Geophys Res Lett 29, doi: 10.1029/2001GL014418 Song TR, Helmberger DV, Grand SP (2004) Low-velocity zone atop the 410 k m seismic discontinuity in the northwestern United States. Nature 427:530-533 Speziale S, Zha CS, Duffy TS, Hemley RJ, Mao HK (2001) Quasi-hydrostatic compression of magnesium oxide to 52 GPa: implication for the pressure-volume-temperature equation of state. J Geophys Res 106: 515-528 Stixrude L (1998) Elastic constants and anisotropy of M g S i 0 3 perovskite, periclase, and S i 0 2 at high pressure. In: Core-Mantle Boundary Region. Geodynamics Series 28. Gurnis M, Wysession ME, Knittle E, Biffet BA (eds) Am Geophys Union, p 83-96 Suzuki A, Ohtani E, Morishima H, Kubo T, Kanbe Y, Kondo T, Okada T, Terasaki H, Kato T, Kikegawa T (2000) In situ determination of the phase boundary between wadsleyite and ringwoodite in Mg 2 Si0 4 . Geophys Res Lett 27:803-806 Tarits P, Hautot S, Perrier F (2004) Water in the mantle: Results from electrical conductivity beneath the French Alps. Geophys Res Lett 31, doi:10.1029/2003GL019277 TsuchiyaT (2003) First-principles prediction of the P-V-T equation of state of gold and the 660 k m discontinuity in Earth's mantle. J Geophys Res 108, doi:10.1029/2003JB002446 Turnbull D (1956) Phase changes, in Solid State Physics. Vol 3. Seitz F, Turnbull D (eds) Elsevier, p 225-306 Van der Meijde M, Marone F, Giardini D, van der Lee S (2003) Seismic evidence for water deep in the Earth's upper mantle. Science 300:1556-1558 Weidner DJ, Wang Y (2000) Phase transformations: Implications for mantle structure Earth's Deep Interior. In: Earth's Deep Interior: Mineral Physics and Tomography from the Atomic to the Global Scale. Geophysical Monograph Vol. 117. Karato S, Forte AM, Liebermann RC, Masters G, Stixrude L (eds) Am Geophys Union, p 215-235 Williams Q, Garnero EJ (1996) Seismic evidence for partial melting at the base of the lower mantle. Science 273:1528-1530 Wood BJ (1995) The effect of H 2 0 on the 410-kilometer seismic discontinuity. Science 268:74-76

18

Reviews in Mineralogy & Geochemistry Vol. 62, pp. 421-450, 2006 Copyright © Mineralogical Society of America

Water in the Early Earth Bernard Marty Centre de Recherches Pétrographiques et Géochimiques Centre National de la Recherche Scientifique 15 Rue Notre Dame des Pauvres 54220 Vandoeuvre lès Nancy Cedex France e-mail:

[email protected]

Reika Yokochi University of Illinois at Chicago Department of Earth and Environmental Sciences Chicago, Illinois, 60607, U.S.A. e-mail:

[email protected]

INTRODUCTION The origin of water is a long standing problem that has fascinated generations of philosophers and scientists since the dawn of humanity. It has to do with processes that took place in the nascent solar system, but, unfortunately, we lack record of what really happened during this dark age. On Earth, the tectonic activity has erased completely any record dating back from this period, and the oldest rocks preserved on Earth crystallized about 600 Ma after start of solar system condensation (ASSC). What we observe today does not represent necessarily what was present when the solar system formed, and it may well be possible that the true water ancestors could have had chemical and isotopic characteristics that are not observed in any reservoir of the present-day solar system. However, the extraterrestrial objects that escaped planetary differentiation (chondrites, comets, interplanetary dust) are probably the best available precursor candidates for the source of volatile elements in Earth. In this context, the nature of such potential contributions can be estimated in view of chemical and isotopic mass balance, taking into account astrophysical and/or thermodynamic constraints on planetary system formation. The Earth is not a water-rich body. Water at the Earth's surface (1.5 x 1024 g), mostly in the oceans, makes about 0.025% H 2 0 over the whole Earth (5.97 xlO 27 g). Most estimates for the mantle water content advocate a few ocean masses, so that the bulk water content of Earth may be