PHYSICS AND CHEMISTRY OF THE DEEP EARTH

Physics and Chemistryof the Deep Earth

Edited by

Shun-ichiro KaratoDepartment of Geology and Geophysics

Yale University, New HavenCT, USA

A John Wiley & Sons, Ltd., Publication

This edition first published 2013 © 2013 by John Wiley & Sons, Ltd.

Wiley-Blackwell is an imprint of John Wiley & Sons, formed by the merger of Wiley’s global Scientific, Technicaland Medical business with Blackwell Publishing.

Registered office: John Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial offices: 9600 Garsington Road, Oxford, OX4 2DQ, UKThe Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK111 River Street, Hoboken, NJ 07030-5774, USA

For details of our global editorial offices, for customer services and for information about how to apply forpermission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell.

The right of the author to be identified as the author of this work has been asserted in accordance with the UKCopyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, inany form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted bythe UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand namesand product names used in this book are trade names, service marks, trademarks or registered trademarks of theirrespective owners. The publisher is not associated with any product or vendor mentioned in this book.

Limit of Liability/Disclaimer of Warranty: While the publisher and author(s) have used their best efforts inpreparing this book, they make no representations or warranties with respect to the accuracy or completeness ofthe contents of this book and specifically disclaim any implied warranties of merchantability or fitness for aparticular purpose. It is sold on the understanding that the publisher is not engaged in rendering professionalservices and neither the publisher nor the author shall be liable for damages arising herefrom. If professional adviceor other expert assistance is required, the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Physics and chemistry of the deep Earth / Shun-ichiro Karato.pages cm

Includes bibliographical references and index.ISBN 978-0-470-65914-4 (cloth)

1. Geophysics. 2. Geochemistry. 3. Earth – Core. I. Karato, Shun-ichiro, 1949-QE501.K325 2013551.1′2 – dc23

2012045123

A catalogue record for this book is available from the British Library.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not beavailable in electronic books.

Cover image: © iStockphoto.com/Thomas VogelCover design by Design Deluxe

Set in 9/11.5pt Trump Mediaeval by Laserwords Private Limited, Chennai, India

1 2013

Contents

Contributors, viiPreface, ix

PART 1 MATERIALS’ PROPERTIES, 1

1 Volatiles under High Pressure, 3Hans Keppler

2 Earth’s Mantle Melting in the Presence ofC–O–H–Bearing Fluid, 38Konstantin D. Litasov, Anton Shatskiy, andEiji Ohtani

3 Elasticity, Anelasticity, and Viscosity of aPartially Molten Rock, 66Yasuko Takei

4 Rheological Properties of Minerals andRocks, 94Shun-ichiro Karato

5 Electrical Conductivity of Minerals andRocks, 145Shun-ichiro Karato and Duojun Wang

PART 2 COMPOSITIONAL MODELS, 183

6 Chemical Composition of the Earth’s LowerMantle: Constraints from Elasticity, 185Motohiko Murakami

7 Ab Initio Mineralogical Model of the Earth’sLower Mantle, 213Taku Tsuchiya and Kenji Kawai

8 Chemical and Physical Properties andThermal State of the Core, 244Eiji Ohtani

9 Composition and Internal Dynamics ofSuper-Earths, 271Diana Valencia

PART 3 GEOPHYSICAL OBSERVATIONSAND MODELS OF MATERIALCIRCULATION, 295

10 Seismic Observations of MantleDiscontinuities and Their Mineralogical andDynamical Interpretation, 297Arwen Deuss, Jennifer Andrews, andElizabeth Day

11 Global Imaging of the Earth’s Deep Interior:Seismic Constraints on (An)isotropy,Density and Attenuation, 324Jeannot Trampert and Andreas Fichtner

12 Mantle Mixing: Processes andModeling, 351Peter E. van Keken

13 Fluid Processes in Subduction Zones andWater Transport to the Deep Mantle, 372Hikaru Iwamori and Tomoeki Nakakuki

Index, 393

Colour plate section can be found betweenpages 214–215

Contributors

JENNIFER ANDREWS Bullard Laboratory, Cam-bridge University, Cambridge, UK

ELIZABETH DAY Bullard Laboratory, CambridgeUniversity, Cambridge, UK

ARWEN DEUSS Bullard Laboratory, CambridgeUniversity, Cambridge, UK

ANDREAS FICHTNER Department of Earth Sci-ences, Utrecht University, Utrecht, The Nether-land

HIKARU IWAMORI Department of Earth andPlanetary Sciences, Tokyo Institute of Technol-ogy, Tokyo, Japan

SHUN-ICHIRO KARATO Department of Geologyand Geophysics, Yale University, New Haven,CT, USA

KENJI KAWAI Department of Earth and Plan-etary Sciences, Tokyo Institute of Technology,Tokyo, Japan

HANS KEPPLER Byerisched Geoinstitut, Univer-sitat Bayreuth, Bayreuth, Germany

KONSTANTIN LITASOV Department of Earth andPlanetary Materials Science, Graduate School ofScience, Tohoku University, Sendai, Japan

MOTOHIKO MURAKAMI Department of Earthand Planetary Materials Science, GraduateSchool of Science, Tohoku University, Sendai,Japan

TOMOEKI NAKAKUKI Department of Earth andPlanetary Systems Science, Hiroshima Univer-sity, Hiroshima, Japan

EIJ I OHTANI Department of Earth and PlanetaryMaterials Science, Graduate School of Science,Tohoku University, Sendai, Japan

ANTON SHATSKIY Department of Earth andPlanetary Materials Science, Graduate School ofScience, Tohoku University, Sendai, Japan

YASUKO TAKEI Earthquake Research Institute,University of Tokyo, Tokyo, Japan

JEANNOT TRAMPERT Department of Earth Sci-ences, Utrecht University, Utrecht, The Nether-land

viii Contributors

TAKU TSUCHIYA Geodynamic Research Center,Ehime University, Matsuyama, Ehime, Japan

DIANA VALENCIA Department of Earth,Atmospheric and Planetary Sciences,Massachusetts Institute of Technology,Cambridge, MA, USA

PETER VAN KEKEN Department of Earth andEnvironmental Sciences, University of Michigan,Ann Arbor, MI, USA

DUOJUN WANG Graduate University of ChineseAcademy of Sciences, College of Earth Sciences,Beijing, China

Preface

Earth’s deep interior is largely inaccessible. Thedeepest hole that human beings have drilled isonly to ∼11 km (Kola peninsula in Russia) whichis less than 0.2 % of the radius of Earth. Somevolcanoes carry rock samples from the deep in-terior, but a majority of these rocks come fromless than ∼200 km depth. Although some frag-ments of deep rocks (deeper than 300 km) arediscovered, the total amount of these rocks ismuch less than the lunar samples collected duringthe Apollo mission. Most of geological activitiesthat we daily face occur in the shallow por-tions of Earth. Devastating earthquakes occur inthe crust or in the shallow upper mantle (lessthan ∼50 km depth), and the surface lithosphere(‘‘plates’’) whose relative motion controls mostof near surface geological activities has less than∼100 km thickness. So why do we worry about‘‘deep Earth’’?

In a sense, the importance of deep processesto understand the surface processes controlled byplate tectonics is obvious. Although plate motionappears to be nearly two-dimensional, the geom-etry of plate motion is in fact three-dimensional:Plates are created at mid-ocean ridges and theysink into the deep mantle at ocean trenches,sometimes to the bottom of the mantle. Platemotion that we see on the surface is part ofthe three-dimensional material circulation in thedeep mantle. High-resolution seismological stud-ies show evidence of intense interaction betweensinking plates and the deep mantle, particularlythe mid-mantle (transition zone) where minerals

undergo a series of phase transformations. Cir-culating materials of the mantle sometimes goto the bottom (the core–mantle boundary) wherechemical interaction between these two distinctmaterials occurs. Deep material circulation isassociated with a range of chemical processesincluding partial melting and dehydration and/orrehydration. These processes define the chemicalcompositions of various regions, and the mate-rial circulation modifies the materials’ properties,which in turn control the processes of materialscirculation.

In order to understand deep Earth, a multi-disciplinary approach is essential. First, we needto know the behavior of materials under theextreme conditions of deep Earth (and of deepinterior of other planets). Drastic changes in prop-erties of materials occur under the deep plane-tary conditions including phase transformations(changes in crystal structures and melting). Resis-tance to plastic flow also changes with pressureand temperature as well as with water content.Secondly, we must develop methods to infer deepEarth structures from the surface observations.Thirdly, given some observations, we need to de-velop a model (or models) to interpret them in theframework of physical/chemical models.

In this book, a collection of papers coveringthese three areas is presented. The book is dividedinto three parts. The first part (Keppler, Litasovet al., Takei, Karato, Karato and Wang) includespapers on materials properties that form the basis

x Preface

for developing models and interpreting geophys-ical/geochemical observations. The second part(Murakami, Tsuchiya and Kawai, Ohtani, Va-lencia) contains papers on the composition ofdeep Earth and planets including the models ofthe mantle and core of Earth as well as modelsof super-Earths (Earth-like planets orbiting starsother than the Sun). And finally the third part(Deuss et al., Trampert and Fichtner, van Keken,Iwamori) provides several papers that summa-rize seismological and geochemical observationspertinent to deep mantle materials circulationand geodynamic models of materials circulationwhere geophysical/geochemical observations and

mineral physics data are integrated. All of thesepapers contain reviews of the related area tohelp readers understand the current status ofthese areas.

I thank all the authors and reviewers and edi-tors of Wiley-Blackwell who made it possible toprepare this volume. I hope that this volume willhelp readers to develop their own understandingof this exciting area of research and to play a rolein the future of deep Earth and planet studies.

Shun-ichiro KaratoNew Haven, Connecticut

Part 1

Materials’ Properties

1 Volatiles under High Pressure

H A N S K E P P L E RBayerisches Geoinstitut, Universitat Bayreuth, Bayreuth, Germany

Summary

Hydrogen and carbon are the two most importantvolatile elements in the Earth’s interior, yet theirbehavior is very different. Hydrogen is soluble inmantle minerals as OH point defects and theseminerals constitute a water reservoir comparablein size to the oceans. The distribution of water inthe Earth’s interior is primarily controlled by thepartitioning between minerals, melts and fluids.Most of the water is probably concentrated in theminerals wadsleyite and ringwoodite in the tran-sition zone of the mantle. Carbon, on the otherhand, is nearly completely insoluble in the sili-cates of the mantle and therefore forms a separatephase. Stable carbon-bearing phases are likely car-bonates in the upper mantle and diamond orcarbides in the deeper mantle. Already minuteamounts of water and carbon in its oxidized form(as carbonate or CO2) greatly reduce the solidus ofmantle peridotite. Melting in subduction zones istriggered by water and both water and CO2 con-tribute to the melting below mid-ocean ridges andin the seismic low-velocity zone. Redox melt-ing may occur when oxygen fugacity increasesupon upwelling of reduced deep mantle, convert-ing reduced carbon species to carbonate or CO2that strongly depress solidus temperatures. Thelarge contrast of water storage capacity betweentransition zone minerals and the mineral assem-blages of the upper and lower mantle implies that

Physics and Chemistry of the Deep Earth, First Edition. Edited by Shun-ichiro Karato.© 2013 John Wiley & Sons, Ltd. Published 2013 by John Wiley & Sons, Ltd.

melt may form near the 440 and 660 km seismicdiscontinuities. Water and carbon have been ex-changed during the Earth’s history between thesurface and the mantle with typical mantle res-idence times in the order of billions of years.However, the initial distribution of volatiles be-tween these reservoirs at the beginning of theEarth’s history is not well known. Nitrogen, noblegases, sulfur and halogens are also continuouslyexchanged between mantle, oceans and atmo-sphere, but the details of these element fluxes arenot well constrained.

1.1 Introduction: What Are Volatiles and WhyAre They Important?

Volatiles are chemical elements and com-pounds that tend to enter the gas phase inhigh-temperature magmatic and metamorphicprocesses. Accordingly, one can get some ideaabout the types of volatiles occurring in theEarth’s interior by looking at compositions ofvolcanic gases. Table 1.1 compiles some typicalvolcanic gas analyses. As is obvious from thistable, water and carbon dioxide are the twomost abundant volatiles and they are also mostimportant for the dynamics of the Earth’s interior(e.g. Bercovici & Karato, 2003; Mierdel et al.,2007; Dasgupta & Hirschmann, 2010). Other,less abundant volatiles are sulfur and halogen

4 hans keppler

Table 1.1 Composition of volcanic gases (in mol%).

Mt. St. Helens Kilauea Kilauea Etna1980 1918 1983 2000

H2O 91.6 37.1 79.8 92H2 0.85 0.49 0.90CO2 6.94 48.9 3.15 7.3CO 0.06 1.51 0.06SO2 0.21 11.84 14.9 1.0H2S 0.35 0.04 0.62HCl 0.08 0.1 0.1HF 0.19 0.07

Source: Data from Symonds et al. (1994) except for Etna (fromAllard et al., 2005).

compounds, particularly SO2, H2S, HCl, and HF.Noble gases are only trace constituents of vol-canic gases, but they carry important informationon the origins and history of the reservoirs theyare coming from (Graham, 2002; Hilton et al.,2002). Nitrogen is a particular case. Volcanicgas analyses sometimes include nitrogen, butit is often very difficult to distinguish primarynitrogen from contamination by air duringthe sampling process. The most conclusiveevidence for the importance of nitrogen as avolatile component in the Earth’s interior isthe occurrence of N2-filled fluid inclusions ineclogites and granulites (Andersen et al., 1993).Ammonium (NH4

+) appears to be a commonconstituent in metamorphic micas, which maytherefore recycle nitrogen into the mantle insubduction zones (Sadofsky & Bebout, 2000).

Generally, the composition of fluids trapped asfluid inclusions in magmatic and metamorphicrocks of the Earth’s crust is similar to volcanicgases. Water and carbon dioxide prevail; hydrousfluid inclusions often contain abundant dissolvedsalts. Methane (CH4) containing inclusions arealso sometimes found, particularly in low-grademetamorphic rocks of sedimentary origin and insediments containing organic matter (Roedder,1984). Fluid inclusions in diamonds are an impor-tant window to fluid compositions in the man-tle. Observed types include CO2-rich inclusions,carbonatitic compositions, water-rich inclusions

with often very high silicate content, and highlysaline brines (Navon et al., 1988; Schrauder &Navon, 1994; Izraeli et al., 2001). Methane andhydrocarbon-bearing inclusions have also been re-ported from xenoliths in kimberlites (Tomilenkoet al., 2009).

Although volatiles are only minor or trace con-stituents of the Earth’s interior, they controlmany aspects of the evolution of our planet. Thisis for several reasons: (1) Volatiles, particularlywater and carbon dioxide, strongly reduce melt-ing temperatures; melting in subduction zones,in the seismic low velocity zone and in deeperparts of the mantle cannot be understood with-out considering the effect of water and carbondioxide (e.g. Tuttle & Bowen, 1958; Kushiro, 1969;Kushiro, 1972; Tatsumi, 1989; Mierdel et al.,2007; Hirschmann, 2010). (2) Even trace amountof water dissolved in major mantle minerals suchas olivine can reduce their mechanical strengthand therefore the viscosity of the mantle by or-ders of magnitude (Mackwell et al., 1985; Karato& Jung, 1998; Kohlstedt, 2006). Mantle convec-tion and all associated phenomena, such as platemovements on the Earth’s surface, are there-fore intimately linked to the storage of waterin the mantle. (3) Hydrous fluids and carbon-atite melts only occur in trace amounts in theEarth’s interior. Nevertheless they are respon-sible for chemical transport processes on localand on global scales (e.g. Tatsumi, 1989; Iwamoriet al., 2010). (4) The formation and evolutionof the oceans and of the atmosphere is directlylinked to the outgassing of the mantle and tothe recycling (‘‘ingassing’’) of volatiles into themantle (e.g. McGovern & Schubert, 1989; Rupkeet al., 2006; Karato, 2011).

1.2 Earth’s Volatile Budget

The Earth very likely formed by accretion ofchondritic material that resembles the bulkcomposition of the solar system. In principle,it should therefore be possible to estimate theEarth’s volatile budget by considering the volatilecontent of chondritic meteorites (e.g. Morbidelli

Volatiles under High Pressure 5

et al., 2002; Albarede, 2009). Unfortunately,there is a large variation in the contents of water,carbon and other volatiles between the differentkinds of chondritic meteorites and the Earth,likely formed by accretion of a mixture of thesedifferent materials, the precise fractions beingpoorly constrained. Moreover, during accretion,massive loss of volatiles to space likely occurredcaused by impacts. This volatile loss has tobe accounted for, which introduces another,considerable uncertainty.

Estimating the volatile content of the bulkmantle or of the bulk silicate Earth (crust +mantle) from cosmochemical arguments is evenmore difficult, since the iron–nickel alloy ofthe Earth’s core very likely sequestered at leastsome fraction of the available volatiles. Evidencefor this comes from the occurrence of sulfides(troilite, FeS), carbides (cohenite, Fe3C) and ni-trides (osbornite, TiN) as minerals in iron mete-orites and from various experimental studies thatshow that under appropriate conditions, carbon,sulfur, nitrogen and hydrogen are quite soluble inmolten iron (Fukai, 1984; Wood, 1993; Okuchi,1997; Adler & Williams, 2005; Terasaki et al.,2011). Another line of evidence is the densitydeficit of the Earth’s outer core (Birch, 1952),which requires the presence of some light ele-ments in the iron nickel melt. While most presentmodels suggest that Si and/or O account for mostof the density deficit, a significant contributionfrom other volatiles is possible. The recent modelby Rubie et al. (2011) yields 8 wt % Si, 2 wt % Sand 0.5 wt % O as light elements in the core. Thelow oxygen content appears to be consistent withshock wave data on melts in the Fe–S–O system(Huang et al., 2011).

The timing of volatile acquisition on the Earthis another poorly constrained variable. One typeof models assumes that volatiles were acquiredduring the main phase of accretion, while anotherview holds that volatiles, in particular water weredelivered to the Earth very late (Albarede, 2009),possibly during the formation of a ‘‘late veneer’’ ofchondritic materials or perhaps by comets. How-ever, both the D/H and 15N/14N isotope ratios ofterrestrial reservoirs are close to the chondritic

values, while they are much lower than thoseobserved in comets. This limits the cometarycontribution to the terrestrial water and nitro-gen budget to a few percent at most (Marty &Yokochi, 2006).

Recent models of the Earth’s formation (e.g.Rubie et al., 2011) suggest that during accretion,initially very volatile depleted chondritic mate-rial accreted, which possibly became more waterand volatile-rich towards the end of accretion,but still before core formation. Such models areconsistent with the observed depletion of mod-erately volatile elements (e.g. Na, K, Zn) on theEarth relative to CI chondrites; these elementsmay have failed to condense in chondritic ma-terial that formed close to the sun. Numericalmodels of early solar system evolution suggestthat at later stages of accretion, stronger radialmixing in the solar system occurred, so thatwater and volatile-rich material from the coldouter part of the solar system entered the growingplanet (Morbidelli et al., 2002). Taking all of theavailable evidence together, it is plausible thatthe Earth after complete accretion contained 1–5ocean masses of water (Jambon & Zimmermann,1990; Hirschmann, 2006). A major depletion ofhydrogen and other light elements by loss tospace during later Earth history can be ruled out,because the expected depletions of light isotopesresulting from such a distillation process are notobserved on the Earth.

Evidence on the present-day volatile content ofthe Earth’s mantle comes from direct studies ofmantle samples, particularly xenoliths, from mea-surements of water contents in basalts, which arepartial melts formed in the shallow part of the up-per mantle and from remote sensing by seismicmethods and magnetotelluric studies of electricalconductivity. While the first two methods mayprovide constraints on all volatiles, remote sens-ing techniques are primarily sensitive to water(Karato 2006).

Pyroxenes in mantle xenoliths that wererapidly transported to the surface contain from<100 to about 1000 ppm of water (Skogby, 2006);olivines may be nearly anhydrous but some-times contain up to 300 ppm of water (Beran &

6 hans keppler

Libowitzky, 2006). These observations show thatthe upper mantle is by no means completely dry(Bell & Rossman, 1992). However, estimatingmantle abundances of water and other volatilesfrom such data is difficult, because samples oftenhave lost water on their way to the surface; insome cases, this water loss is evident in diffusionprofiles that may be used to constrain ascentrates (Demouchy et al., 2006; Peslier & Luhr,2006; Peslier et al., 2008). Moreover, many ofthese xenoliths come from alkali basalts orkimberlites. The source region of these magmasmay be more enriched in volatiles than thenormal mantle.

Mid-ocean ridge basalts (MORB) tap a volatile-depleted reservoir that is believed to representmost of the upper mantle. Ocean island basalts(OIB) appear to come from a less depleted, likelydeeper source. Probably the best constraints onvolatile abundances in the mantle come fromMORB and OIB samples that have been quenchedto a glass by contact with sea water at the bottomof the ocean (e.g. Saal et al., 2002; Dixon et al.,2002); the fast quenching rate and the confiningpressure probably suppressed volatile loss. In prin-ciple, one can calculate from observed volatileconcentrations in quenched glasses the volatilecontent in the source, if the degree of meltingand the mineral/melt partition coefficients of thevolatiles are known. Such calculations, however,are subject to considerable uncertainties. A muchmore reliable and widely used method is basedon the ratio of volatiles to certain incompatibletrace elements, such as H2O/Ce and CO2/Nb(Saal et al., 2002). These ratios are nearly con-stant in MORB glasses over a large range ofH2O and CO2 contents that represent differentdegrees of melting and crystal fractionation. Thismeans that the bulk mineral/melt partition co-efficient of H2O is similar to that of Ce and thebulk mineral/melt partition coefficient of CO2is similar to Nb. For equal bulk partition coeffi-cients, the H2O/Ce ratio and the CO2/Nb ratiomust be the same in the mantle source and inthe basalt, independent of the degree of melting.Therefore, measured H2O/Ce ratios and CO2/Nbratios of the basalts can be used together with the

quite well-constrained Ce and Nb contents of themantle to estimate the water and carbon dioxidecontent in the MORB and OIB sources. Using thismethod, Saal et al. (2002) estimated the volatilecontents of the MORB-source upper mantle tobe 142 ± 85 ppm H2O (by weight), 72 ± 19 ppmCO2, 146 ± 35 ppm S, 1 ± 0.5 ppm Cl and 250 ± 50ppm F. In general, estimates of the water contentin the depleted MORB source using similar meth-ods yield values of 100–250 ppm by weight forH2O. In particular, the work by Michael (1995)suggests some regional variability of the MORBsource water content. Much higher volatile con-centrations with up to about 1000 ppm of H2Ohave been obtained for the OIB source region (e.g.Dixon et al., 1997; Hauri, 2002). The CO2 contentin the OIB source may range from 120 to 1830 ppmCO2 (Hirschmann & Dasgupta, 2009). If one as-sumes that the MORB source is representativefor most of the mantle and the OIB source con-tributes a maximum of 40% to the total mantle,these numbers would translate to a total mantlecarbon budget in the order of (1–12) . 1023 g of C(Dasgupta & Hirschmann, 2010). A similar cal-culation assuming 142 ppm H2O in the MORBsource and 1000 ppm H2O in the OIB sourcewould give a bulk water reservoir in the man-tle of 2 . 1024 g, i.e. about 1.4 ocean masses. Theuncertainty in this estimate is, however, quitesignificant and the number given is likely to beonly an upper limit of the actual water content.

Water has a strong effect on the physical prop-erties, particularly density, seismic velocities andelectrical conductivity of mantle minerals (Jacob-sen, 2006; Karato, 2006). In addition, water maychange the depth and the width of seismic dis-continuities (e.g. Frost & Dolejs, 2007), because itstabilizes phases that can incorporate significantamounts of water as OH point defects in theirstructure. These effects may be used for a re-mote sensing of the water content in parts of themantle that are not accessible to direct sampling.The dissolution of water as OH point defectsin minerals generally reduces their density andboth P and S wave seismic velocities (Jacobsen,2006). This is mostly due to the formation ofcation vacancies that usually – but not always

Volatiles under High Pressure 7

(e.g. Gavrilenko et al., 2010) – is associated withthe dissolution of water as OH point defects insilicates. As such, the effect of increasing watercontents is qualitatively similar to the effect of in-creasing temperature. However, the ratio of P andS wave velocities vp/vs appears to be particularlysensitive to water as water in minerals affectsS wave velocities much more than P wave ve-locities. The vp/vs ratio may therefore be able todistinguish the effects of water from temperature,although the effects may be subtle (Karato, 2011).

A property that is particularly sensitive towater is electrical conductivity, which may begreatly enhanced by proton conduction (Karato,1990) (see also Chapter 5 of this book). Therefore,numerous studies on the effect of water onminerals such as wadslyeite and ringwooditehave recently been carried out, since these twomain constituents of the Earth’s transition zoneare able to dissolve more than 2–3 wt % of water(Smyth, 1987; Kohlstedt et al., 1996). Althoughthere are some significant discrepancies betweenthe available studies (Huang et al., 2005; Yoshinoet al., 2008), it appears that a fully hydrated transi-tion zone containing several wt % of water can beruled out. The data may, however, be compatiblewith water concentrations up to 1000 or 2000 ppmby weight (Huang et al., 2005; Dai & Karato,2009b). Such concentrations, if confirmed, wouldimply that the transition zone is the mostimportant reservoir of water in the mantle,containing roughly 0.5 ocean masses of water.

1.3 Water

1.3.1 Hydrous minerals in the mantle

According to available phase equilibria studies(e.g. Gasparik, 2003), the bulk of the Earth’s man-tle is made up by nominally anhydrous minerals,i.e. silicates and oxides that do not contain anyH2O or OH in their formula. Mantle xenoliths,however, while they mostly consist of olivine,pyroxenes, and garnet or spinel, sometimes docontain some hydrous minerals such as amphi-boles or phlogopite-rich micas (Nixon, 1987),indicating that these phases may be stable inthe mantle under some circumstances.

Experimental data on the stability range ofhydrous mantle minerals (Frost, 2006) are com-piled in Figure 1.1. At low temperatures andhigh pressures, a variety of dense hydrous magne-sium silicates (DHMS) are stable, including phaseA, D, E, and superhydrous phase B. However,Figure 1.1 also shows clearly that the stabil-ity range of these phases is far away from anormal mantle adiabat; they may perhaps existin cold subducted slabs, but not in the nor-mal mantle. Of all the DHMS phases, phase Dwith composition MgSi2O4(OH)2 is stable at thehighest pressures. A new aluminum-rich versionof phase D has recently been described (Boffa-Ballaran et al., 2010). All other hydrous phasesshown in Figure 1.1 also contain some Na and/orK. Both alkali elements occur in normal mantleperidotite only at a minor or trace element level(typically about 0.3 wt % Na2O and 0.03 wt %K2O). The occurrence of these phases is therefore

D

E

A

SB

DHMS

Phase X out

Phase X

K-richterite

K-richterite in

Phlogopite

Na-amphibole

Phlogopite out

AM

A

OL

100

0

15

10

20

25

5

1000900 1100 1200 1300 1400 1500 17001600

200

300

400

600

700

500

Dep

th(k

m)

Temperature(°C)

Pre

ssur

e (G

Pa)

Fig. 1.1 Stability range of hydrous minerals in Earth’smantle. AMA is an average mantle adiabat, OL is ageotherm for a 100 million year old oceaniclithosphere. The thick black line gives the tentativelocation of the water-saturated peridotite solidus.Modified after Frost (2006).

8 hans keppler

not only limited by their maximum P,T stabilityfield, but also by the availability of Na and K. Atnormal mantle abundances, most, if not all, Naand K may be dissolved in clinopyroxene (Harlow,1997; Gasparik, 2003; Perchuk et al., 2002; Har-low & Davies, 2004), so that these hydrous phaseswill not form, even in the pressure range wherethey may be stable for suitable bulk composi-tions. The only hydrous phase that is stable alongan average mantle adiabat is phase X, a silicatewith an unusual layer structure containing Si2O7groups. The formula of phase X may be writ-ten as (K, Na)2−x(Mg, Al)2Si2O7Hx, where x = 0–1(Yang et al., 2001). The oceanic geotherm passesthrough the thermodynamic stability fields ofphlogopite and Na-amphibole. The sodic amphi-boles found in mantle samples are usually rich inpargasite NaCa2Mg4Al3Si6O22(OH)2 component,often with a significant content of Ti (kaersutites).However, due to the low abundance of alkalis inthe normal mantle, their occurrence in mantlexenoliths is probably related to unusual chem-ical environments affected by subduction zoneprocesses or mantle metasomatism. The sameapplies for phlogopite KMg3(OH)2AlSi3O10.

Hydrous phases are stable in some parts ofsubduction zones. The subducted slab containshydrous minerals in the sediment layer. In addi-tion, the basaltic layer and parts of the underlyingperidotite may have been hydrated by contactwith seawater to some degree. Anomalies in heatflow near mid-ocean ridges suggest that the up-permost 2–5 km may be hydrated (Fehn et al.,1983). However, much deeper hydration may oc-cur by the development of permeable fracturesrelated to the bending of the slab when it entersthe subduction zone (Faccenda et al., 2009). For along time, it was believed that amphibole in thebasaltic layer is the major carrier of water into themantle and that the volcanic front in island arcsis located above the zone of amphibole decompo-sition. More recent work (Poli & Schmidt, 1995;Schmidt & Poli, 1998), however, suggests thatseveral phases in the sedimentary, basaltic andultramafic parts of the slab are involved in trans-porting water, including amphibole, lawsonite,phengite and serpentine. Due to the formation of

solid solutions in multicomponent systems, eachphase decomposes over a range of pressures andtemperatures, and the decomposition reactions ofthese phases overlap. Water released by decompo-sition of hydrous phases in the slab likely causesserpentinization of the shallow and cool parts ofthe mantle wedge above the slab. Under somecircumstances, particularly for the subduction ofold oceanic lithosphere along a cool geotherm,some of the serpentine in the ultramafic part ofthe subducted lithosphere may escape decompo-sition and may be able to transport water into thedeep mantle (Rupke et al., 2004).

1.3.2 Water in nominally anhydrous minerals

Already more than 50 years ago, it was noticedthat chemical analyses of nominally anhydrousminerals occasionally suggest the presence oftraces of water (Brunner et al., 1961; Griggs &Blacic, 1965; Wilkins & Sabine, 1973; Martin &Donnay, 1972). However, since water is an ubiqui-tous contaminant that may occur as mechanicalimpurities in samples, the significance of theseobservations was uncertain. This has changed bythe application of infrared spectroscopy to thestudy of water in nominally anhydrous minerals,a method that was pioneered by the groups ofJosef Zemann and Anton Beran in Vienna and byGeorge Rossman at Caltech (e.g. Beran, 1976; Be-ran & Zemann, 1986; Bell & Rossman, 1992).Figure 1.2 shows polarized infrared spectra ofan olivine crystal from the upper mantle. Allof the absorption bands in the range between3000 to 3700 cm-1 in such spectra correspond toOH groups (or, very rarely in a few minerals, tomolecular H2O). The intensity of absorption ob-viously depends on the polarization, i.e. on theorientation of the electrical field vector of polar-ized light relative to the crystal structure. Thisdemonstrates that these samples contain ‘‘water’’in the form of OH defects incorporated into thecrystal lattice. If the bands were due to somemechanical impurities, a dependence of infraredabsorption on the orientation of the crystal latticewould not be expected.

Infrared spectra are exceedingly sensitive totraces of water in minerals and can detect OH

Volatiles under High Pressure 9

8

6

4

2

0

Abs

orba

nce

4000 3800 3600 3400 3200 3000 2800

Wavenumber (cm−1)

(001)

(100)

(010)

Fig. 1.2 Polarized infrared spectra of awater-containing olivine. The figure shows threespectra measured with the electrical field vectorparallel to the a, b, and c axis of the crystal. Spectracourtesy of Xiaozhi Yang.

concentrations down to the ppb level, if suf-ficiently large single crystals are available. Inprinciple, quantitative water contents can alsobe measured using the Lambert Beer law

E = logI0/I = ε c d, (1.1)

where E is extinction or absorbance, I0 is in-frared intensity before the sample, I is infraredintensity after the sample, ε is the extinctioncoefficient, c is concentration (of water) and dis sample thickness. Therefore, if the extinc-tion coefficient of water in a mineral is known,water contents down to the ppb level can beaccurately determined by a simple infrared mea-surement. Unfortunately, ε varies by orders ofmagnitude from mineral to mineral and ε mayeven be different for the same mineral, if OHgroups are incorporated on different sites or by adifferent substitution mechanism, which resultsin a different type of infrared spectrum. Extinc-tion coefficients therefore have to be calibratedfor individual minerals by comparison with anindependent and absolute measure of water con-centration, such as water extraction combinedwith hydrogen manometry or nuclear reaction

analysis (Rossman, 2006). Approximate numbersfor water contents may also be obtained usingempirical correlations between extinction coeffi-cients and OH stretching frequencies (Paterson,1982; Libowitzky & Rossman, 1997). Only inrecent years has secondary ion mass spectrome-try (SIMS) been developed sufficiently to be usedto routinely measure water contents in mineralsdown to the ppm level (e.g. Koga et al., 2003;Mosenfelder et al., 2011). Unlike infrared spec-troscopy, however, SIMS measurements do notyield any information on water speciation and,as such, they cannot directly distinguish betweenOH groups in the crystal lattice and mechanicalimpurities, such as water on grain boundaries,dislocations or fluid inclusions.

The recognition that nearly all nominally an-hydrous minerals from the upper mantle containtraces of water (e.g. Bell & Rossman, 1992; Milleret al., 1987; Matsyuk & Langer, 2004) has stimu-lated experimental studies of water solubility inthese minerals. The suggestion by Smyth (1987)that wadsleyite could be a major host of water inthe transition zone of the mantle, capable of stor-ing several ocean volumes of water has furtherstimulated this work and it is now generally ac-cepted that most of the water in the mantle occursas OH point defects in nominally anhydrous min-erals. Hydrous minerals, fluids or water-bearingsilicate melts only form under special circum-stances and at specific locations in the mantle.While the absolute concentrations of water inmantle minerals are low, due to the very largemass of the Earth’s mantle they constitute a wa-ter reservoir probably comparable in size to theoceans, with a potential storage capacity severaltimes larger than the ocean mass. In other words,most parts of the upper mantle are undersatu-rated with water, so that the actual OH contentsin minerals are far below their saturation level.

Water is dissolved in silicates by protonationof oxygen atoms, charge balanced by Mg2+ vacan-cies, Si4+ vacancies and by coupled substitutions,such as Al3+ + H+ for Si4+ or Al3+ + H+ for2 Mg2+. Evidence for these substitution mech-anisms comes from infrared spectra and fromexperimental observations. For example, certain

10 hans keppler

bands in the spectrum of olivine are only seenat low Si activities that may favor the forma-tion of Si4+ vacancies, while other bands areenhanced at low Mg2+ activities and may there-fore be associated with Mg vacancies (Matveevet al., 2001). The occurrence of Mg vacancies inhydrous olivine was directly confirmed by sin-gle crystal X-ray diffraction studies (Smyth et al.,2006). As a general rule, when the protonationof oxygen atoms is associated with cation vacan-cies, the protons are not located on the vacantcation site. Rather, their location is controlledby hydrogen bonding interactions with neigh-boring oxygen atoms. In the hydrogarnet (OH)4defect which corresponds to a fully protonatedSiO4 tetrahedron with a vacant Si site, the pro-tons are located on the outside of the tetrahedronabove the O-O edges (Lager et al., 2005). Rauchand Keppler (2002) demonstrated that certain OHbands in the infrared spectra of orthopyroxeneonly occur in the presence of Al and are thereforeassigned to coupled substitutions of H+ and Al3+.Similar studies have now also been carried outfor olivine and several bands have been identi-fied that are related to various trivalent cations

and also to titanoclinohumite-like point defects(Berry et al., 2005; 2007). Table 1.2 compiles themain substitution mechanisms for water in nom-inally anhydrous mantle minerals.

Water solubility, i.e. the amount of water dis-solved in a mineral in equilibrium with a fluidphase consisting of pure water, can be describedby the equation (Keppler & Bolfan-Casanova,2006):

cwater = A fnH2O exp

(−�H1bar + �VsolidP

RT

),

(1.2)where A is a constant which essentially containsthe entropy of reaction and n is an exponentrelated to the dissolution mechanism of OH:n = 0.5 for isolated OH groups, n = 1 for OHpairs (or molecular water), n = 2 for the hydrog-arnet defect. �H1bar is the reaction enthalpy at1 bar and �Vsolid is the change in the volumeof the solid phases upon incorporation of water.Experimentally derived parameters for Equation(1.2) applied to the water solubility in a varietyof nominally anhydrous minerals are compiled inTable 1.3.

Table 1.2 Hydrogen-bearing defects in nominally anhydrous minerals.

Substitution mechanism Mineral Reference

------------- Isolated protons -------------H+ + Al3+ ↔ 2 Mg2+ orthopyroxene Mierdel et al. (2007)H+ + Al3+ ↔ Si4+ orthopyroxene Rauch & Keppler (2002)(Fe3, Cr3+ , Sc3+ , REE3+) + H+ ↔ 2 Mg2+ olivine Berry et al. (2007)H+ + B3+ ↔ Al3+ B-rich olivine Sykes et al. (1994)H+ + Li+ ↔ Mg2+ possibly in pyrope Lu & Keppler (1997)

--------------- Proton pairs ---------------2 H+ ↔ Mg2+ olivine Smyth et al. (2006)

enstatite Rauch & Keppler (2002)wadsleyite Smyth (1987)ringwoodite Smyth et al. (2003)

possibly MgSiO3 perovskite Ross et al. (2003)2 H+ + Mg2+ ↔ Si4+ possibly ringwoodite Kudoh et al. (2000)Ti4+ + 2 H+ ↔ Mg2+ + Si4+ olivine Berry et al. (2005)interstitial H2O feldspars Johnson & Rossman (2004)

--------- Cluster of four protons ----------4 H+ ↔ Si4+ garnet Lager et al. (2005)

possibly in olivine Matveev et al. (2001)

Volatiles under High Pressure 11

Table 1.3 Thermodynamic models for water solubility in minerals.

Mineral A n �Vsolid �H1bar Reference(ppm/barn) (cm3/mol) (kJ/mol)

Olivine 0.0066 1 10.6 – Kohlstedt et al. (1996)0.0147a 1 10.2 – Mosenfelder et al. (2006)0.54 1 10.0 50 Zhao et al. (2004)b

MgSiO3 enstatite 0.0135 1 12.1 −4.56 Mierdel & Keppler (2004)Aluminous enstatitec 0.042 0.5 11.3 −79.7 Mierdel et al. (2007)Jadeite 7.144 0.5 8.02 – Bromiley & Keppler (2004)Cr-diopsided 2.15 0.5 7.43 – Bromiley et al. (2004)Pyrope 0.679 0.5 5.71 – Lu & Keppler (1997)Ferropericlase 0.0004 0.5 4.0 – Bolfan-Casanova et al. (2002)

NotesTabulated parameters refer to Equation (1.2). Where no value for �H1bar is given, the enthalpy term is missing because the temperaturedependence of water solubility was not calibrated and the equations are strictly valid only at the temperatures they were calibrated.aRecalculated from a value of A = 2.45 H/106Si/MPa which in the original publication is probably misprinted as A = 2.45 H/106 Si/GPa.bThe equation by Zhao et al. (2004) also includes a term exp (αxFa/RT), where a is 97 kJ/mol and xFa is the molar fraction of fayalite.cThis equation gives the water solubility couples to Al of an Al-saturated enstatite. In order to get the total water solubility in Al-saturatedenstatite, the water solubility in pure MgSiO3 according to Mierdel et al. (2007) has to be added.dThese data may reflect metastable equilibria.

Since water fugacity increases with pressure,one would normally expect that at higher pres-sures, more water is dissolved in minerals. How-ever, the term �Vsolid in Equation (1.2) is usuallypositive and therefore counteracts the effect ofincreasing water fugacity. For this reason, watersolubility in minerals usually first increases withpressure, and then decreases again. Depending onthe sign and magnitude of �H, water solubilitymay either increase or decrease with temperature.

The partition coefficient of water between twophases α and β can be described as the ratio of thewater solubilities in the two phases (Keppler &Bolfan-Casanova, 2006):

Dα/βwater = cα

water

cβwater

= Aα

Aβ

fnα−nβ

H2O

× exp

⎛⎜⎜⎜⎜⎜⎝−

(�H1bar

α − �H1barβ

)+

(�Vsolidα − �Vsolid

β )P

RT

⎞⎟⎟⎟⎟⎟⎠ .

(1.3)

An important consequence of Equation (1.3) isthat the partition coefficient may depend on waterfugacity at given P and T if the two exponents nfor the two phases are not equal. In other words, insuch a case the partition coefficient may vary withbulk water content (e.g. Dai & Karato, 2009a).

Water storage in the upper mantle is largelycontrolled by olivine and orthopyroxene. Watersolubility in garnet appears to be relatively low(Lu & Keppler, 1997) and no trace of water has everbeen detected in the MgAl2O4 spinel phase of theupper mantle. Clinopyroxenes from mantle xeno-liths may contain twice more water than coexist-ing orthopyroxenes (Skogby, 2006), but the verylow modal abundance of clinopyroxene impliesthat it is not a major repository of water in themantle. Clinoyroxene (omphacite) may, however,be very important for recycling water deep intothe lower mantle in subducting slabs; this idea isconsistent with the observation that the highestwater contents from xenolith samples are usuallyfound in eclogitic omphacites (Skogby, 2006).

Water solubility in olivine has been extensivelystudied (Kohlstedt et al., 1996; Zhao et al., 2004;

12 hans keppler

Mosenfelder et al., 2005; Withers et al., 2011).While earlier studies (e.g. Kohlstedt et al., 1996)calculated water contents from infrared spectrausing the Paterson (1982) calibration, later worksuggested that this calibration underestimates thewater contents of olivine by about a factor ofthree (Bell et al., 2003). If this factor is taken intoaccount, there is very good mutual consistencybetween the available studies. Water solubility in-creases continuously with pressure; Mosenfelderet al. (2005) observed 6399 ppm H2O by weightat 12 GPa and 1100 ◦C. In addition, water solubil-ity also increases with temperature (Zhao et al.,2004; Smyth et al., 2006), but at very high temper-atures, observed water contents start to decreaseagain. This effect is likely related to a decrease ofwater activity in the coexisting fluid phase, notto an intrinsic reduction of water solubility inolivine. Smyth et al. (2006) found a maximum of8900 ppm water in olivine at 12 GPa and 1250 ◦C.

Water solubility in orthopyroxene is very dif-ferent from olivine. This is because in orthopy-roxene, water solubility mostly involves coupledsubstitutions with Al3+, such as Al3+ and H+

replacing Si4+ or Al3+ + H+ replacing 2 Mg2+.

Water solubility in pure MgSiO3 enstatite isrelatively low (Mierdel & Keppler, 2004), butincreases greatly with Al (Rauch & Keppler, 2002;Mierdel et al., 2007). If orthopyroxene coexistswith olivine and an Al-rich phase such as spinelor garnet, the Al-content of the orthopyroxeneis buffered. The water solubility in such an Al-saturated orthopyroxene decreases strongly withboth pressure and temperature. This is probablypartially due to the fact that the substitution ofthe larger Al3+ cation on the Si4+ site becomesunfavorable at high pressure. The contrastingbehavior of water solubility in olivine and inorthopyroxene produces a pronounced minimumin water solubility in the upper mantle, which co-incides with the depth of the seismic low velocityzone (Figure 1.3). The minimum in water solubil-ity implies that water partitions more stronglyinto melts and therefore likely stabilizes a smallfraction of partial melt in the seismic low velocitylayer (Mierdel et al., 2007).

At transition zone and lower mantle pressures,water solubility in minerals is not always easyto define, because the minerals usually coexistwith a very solute-rich fluid or with a hydrous

0

20

40

60

80

100

120

140

100

200

300

400

600 1000 1400 0 500 1000 1500 2000 2500 500 1000 1500 2000 2500 30000

Temperature (°C) Water solubility (ppm H2O) Water solubility (ppm H2O)

Depth (km

)P

ress

ure

(kba

r)

oceanic geothermcontinental geotherm

upper mantle adiabat

pureenstatile

AI-saturatedenstatite

olivine

60% olivine +40% AI-saturated enstatite

continental oceanic

LVZLVZ

Fig. 1.3 Water solubility in the upper mantle as a function of depth, for an oceanic and a continental geotherm.LVZ = seismic low-velocity zone. Modified after Mierdel et al. (2007). Reprinted with permission from AAAS.

Volatiles under High Pressure 13

silicate melt of unknown water activity. This isbecause in the deep mantle, silicate melts andaqueous fluids are completely miscible, so thatthe composition of a fluid phase coexisting withminerals changes continuously from fluid-like tomelt-like as a function of temperature (e.g. Shen& Keppler, 1997; Bureau & Keppler, 1999; Kesselet al., 2005). Only in situations where water activ-ity is buffered by phase equilibria (e.g. Demouchyet al., 2005) a thorough thermodynamic treatmentof the data is possible. In the deep mantle, the be-havior of water is therefore often best describedby the partition coefficient of water between solidphases and melts.

In the transition zone, most water is storedin wadsleyite and ringwoodite, the two high-pressure polymorphs of olivine. Compared tothese two phases, majorite-rich garnet appears todissolve less water (Bolfan-Casanova et al., 2000).The prediction by Smyth (1987) that wadsleyiteshould be able to dissolve up to 3.3 wt % ofwater was confirmed by later experimental work(Inoue et al., 1995: 3.1 wt % H2O measured bySIMS; Kohlstedt et al., 1996: 2.4 wt % measuredby FTIR). Polarized infrared spectra (Jacobsenet al., 2005) are consistent with the protonationof the electrostatically underbonded O1 site, asoriginally proposed by Smyth (1987). In contrastto wadsleyite, the observation by Kohlstedtet al. (1996) that spinel-structure (Mg, Fe)2SiO4ringwoodite also dissolves up to 2.6 wt % of H2Owas unexpected, in particular considering thataluminate spinel appears not to dissolve anywater. The dissolution of water in ringwooditemay be related to Mg vacancies and/or topartial disordering between Mg2+ and Si4+, withsome Mg2+ substituting for Si4+ and chargecompensation by two protons (Smyth et al., 2003;Kudoh et al., 2000). Several studies have shownthat the solubility of water in both wadsleyiteand ringwoodite decreases with temperatureabove approximately 1300 ◦C, presumably dueto reduced water activity in the coexisting meltphase. Litasov et al. (2011) estimate that themaximum water solubility in wadsleyite along anaverage mantle geotherm is about 0.4 wt %. The

pressure effect on water solubility in both phasesappears to be small. In equilibrium between wad-sleyite and ringwoodite, water usually appearsto partition preferentially into wadsleyite withDwadsleyite/ringwoodite of approximately 2 at 1450 ◦C;however, this partition coefficient may decreasewith temperature (Demouchy et al., 2005).Chen et al. (2002) report a partition coefficientDwadsleyite/olivine = 5 for directly coexisting phases,which is roughly in agreement with predictionsfrom solubility data. Deon et al. (2011) measuredDwadsleyite/olivine = 3.7 and Dwadsleyite/ringwoodite =2.5. Similar data were reported by Inoue et al.(2010) and by Litasov et al. (2011).

The two main phases of the lower mantle,ferropericlase (Mg, Fe)O and MgSiO3 perovskiteappear to dissolve very little water, suggestingthat the lower mantle may be largely dry(Bolfan-Casanova, 2005). The solubility of waterin ferropericlase was systematically studied byBolfan Casanova et al. (2002). A maximum watersolubility of about 20 ppm H2O by weight wasobserved at 25 GPa, 1200 ◦C and Re–ReO2 bufferconditions.

Both the work by Bolfan-Casanova et al. (2000)and by Litasov et al. (2003) suggest that pureMgSiO3 perovskite dissolves very little water,with maximum water solubilities in the order of afew or at most a few tens of ppm. Bolfan-Casanovaet al. (2003) observed hydrous (Mg, Fe)2SiO4 ring-woodite to coexist with nearly dry (Mg, Fe)SiO3perovskite, yielding a water partition coefficientDringwodite/perovskite = 1050.

There is some controversy surrounding wa-ter solubility in Al-bearing MgSiO3 perovskite.Murakami et al. (2002) reported about 0.2 wt %of water in magnesiumsilicate perovskite and0.4 wt % water in amorphized calcium silicateperovskite synthesized from a natural peridotitecomposition at 25 GPa and 1600–1650 ◦C. Litasovet al. (2003) found up to 1400 ppm H2O in per-ovskites containing up to 7.2 wt % Al2O3, withthe water content increasing with Al. However,in all these studies, the infrared spectra of the per-ovskites show one very broad band centered near3400 cm−1, sometimes with some superimposed

14 hans keppler

weak and sharp peaks. A number of observationssuggests that the broad band, which accountsfor most of the water found in aluminous per-ovskites, is due to some mechanical impurities(i.e. inclusions of hydrous phases) and that thetrue water solubility in aluminous perovskites isvery low. In particular, Bolfan-Casanova et al.(2003) showed that the broad infrared absorp-tion band is only observed in perovskite samplesthat appear milky under the microscope. Thesesamples also show Raman bands of superhydrousphase B and the infrared peaks of this phase co-incide with the main infrared bands reported foraluminous perovskite. If only perfectly clear partsof aluminous perovskite are analyzed by infraredspectroscopy, observed water contents are verylow and comparable to those observed for pureMgSiO3 perovskite.

1.3.3 Water in silicate melts

Water is highly soluble in silicate melts and theassociated melting point depression is essentialfor melting in subduction zones and at other lo-cations in crust and mantle (e.g. Tuttle & Bowen,1958; Kushiro, 1972). Water solubility increases

with pressure and at low pressures, the solubilityof often proportion to the square root of waterfugacity (McMillan, 1994). There are subtle differ-ences in water solubility between felsic and basicmelts and even for felsic melts (e.g. Dixon et al.,1995; Holtz et al., 1995; Shishkina et al., 2010),parameters such as the Na/K ratio can affect wa-ter solubility (Figure 1.4). In general, however,the solubility at given pressure and temperatureis broadly similar for a wide range of melt compo-sitions. Water solubility in melts can be measuredwith high precision, if melts can be quenched tobubble-free glasses, which may be analyzed by avariety of methods, including Karl-Fischer titra-tion, infrared spectroscopy or SIMS (e.g. Behrens,1995). However, for ultrabasic melts, for basicmelts with high water contents and for any water-saturated melt beyond about 1 GPa, quenching toa homogeneous glass is usually not possible anymore. Accordingly, while water solubility undersubvolcanic conditions in the crust is very wellknown, it is only poorly constrained for melts inthe deep mantle.

Because of the square root dependence of wa-ter solubility on water fugacity, it was believed

0

2

4

6

8

10

14

16

12

Wat

er s

olu

bili

ty (

wt.

%)

0 100 200 300 400 500 600 700 800 900

Pressure (MPa)

Tholeitic basalt, 1250 °C

Haplogranite, 800 °C

Fig. 1.4 Water solubility in haplogranitic melt at 800 ◦C (Holtz et al., 1995) and in tholeitic basalt at 1250 ◦C(Shishkina et al., 2010).

Volatiles under High Pressure 15

0−5

−3

−1

1

3

5

7

9

11

13

10 20 30 40 50 60 70 80 90 100

0 10 20 30 40 50 60 70 80 90 100

wt % H2O

log

vis

cosi

ty (

Pa.

s)

mol % H2O

albiteleucitepectolite

Fig. 1.5 Viscosity in thealbite – H2O system at800 ◦C and 1–2 GPa.Modified after Audetat andKeppler (2004). Reprintedwith permission fromAAAS.

that water dissolves in silicate melts primarily asOH groups (e.g. Burnham, 1975), according to thereaction:

Omelt + H2Ogas = 2 OHmelt, (1.4)

where ‘‘O’’ is some oxygen atom in the anhydrousmelt. Since even very small concentrations ofwater reduce the viscosity of silicate melts byorders of magnitude (Figure 1.5), a commonlyheld view is that water ‘‘depolymerizes’’ silicatemelts by reacting with bridging oxygen atoms thatconnect two neighboring SiO4 or AlO4 tetrahedra,schematically:

H2O + Si–O–Si = Si–OH + HO–Si. (1.5)

The simple model of water dissolution out-lined above was modified when near infraredspectroscopy showed that quenched hydroussilicate glasses – and by inference, also thecorresponding silicate melts – contain physicallydissolved H2O molecules in addition to OHgroups (Bartholomew et al., 1980; Stolper, 1982).

Water speciation in silicate melts may thereforebe described by the following equilibrium (asEquation (1.4):

Omelt + H2Omelt = 2 OHmelt,

where Omelt is some bridging oxygen atom in thesilicate melt. The equilibrium constant Kwm ofthis reaction

Kwm = a2OH

aOaH20, (1.6)

where a are activities, implies that the concen-tration of molecular water should increase withthe square of the OH concentration. This meansthat molecular H2O becomes abundant in theglass only at high water concentrations, whileat low bulk water content, water is essentiallydissolved only as OH groups. This is entirelyconsistent with the observed square root de-pendence of water solubility on water fugacityfor low water contents. However, water specia-tion in glasses is only ‘‘frozen in’’ at the glass

16 hans keppler

Fig. 1.6 Complete miscibility in the albite-H2O system, as seen in an externally heated diamond anvil cell at 1.45GPa and 763–766 ◦C. The optical contrast between a droplet of hydrous albite melt and the surrounding fluiddisappears as the compositions of the coexisting phases approach each other. After Shen & Keppler (1997).Reproduced with permission of Nature.

transformation temperature, which may be verylow for these water-rich systems. To obtain equi-librium constants for reaction (1.4), direct infraredspectroscopic measurements by water speciationat magmatic temperatures are necessary. Suchmeasurements were first carried out by Nowakand Behrens (1995) and by Shen and Keppler(1995). They show that equilibrium (1.4) shiftsto the right side with increasing temperature, sothat at typical magmatic temperatures of 1000 ◦Cor higher, OH groups dominate in the melts evenif they contain several wt % of water. At pressuresapproaching 100 GPa, hydrogen becomes bondedto several oxygen atoms and forms extendedstructures in hydrous silicate melts (Mookherjeeet al., 2008).

The commonly held view that water depoly-merizes silicate melts and glasses was challengedby Kohn et al. (1989) on the basis of NMRstudies of hydrous albite glasses that failed todetect clear signs of depolymerization. The workby Kohn et al. has fostered intense research inthe structural role of water in silicate meltsand glasses. However, several more recent stud-ies (e.g. Kummerlen et al., 1992; Malfait & Xue,2010), often using more sophisticated NMR meth-ods, appear to fully confirm the traditional viewthat water depolymerizes the structure of sili-cate melts and glasses by forming Si–OH andAl–OH groups, as indicated by Equation (1.5). Inaluminosilicate glasses, however, the effects ofdepolymerization are not easily detected due tocomplications arising from Al/Si disorder in theglass and melt structure.

Water speciation in silicate melts is essentialfor understanding the strong effect of water on thephysical properties of silicate melts; water doesnot only reduce viscosities by many orders of mag-nitude (e.g. Hess & Dingwell, 1996; Audetat &Keppler, 2005), it also greatly enhances electricalconductivity (e.g. Gaillard, 2004; Ni et al., 2011)and diffusivity (e.g. Nowak & Behrens, 1997).While the effect on viscosity is likely due to de-polymerization, i.e. due to the formation of OHgroups, molecular water appears to be the maindiffusing hydrous species. Water also reduces thedensity of silicate melts, but this effect appearsto be rather insensitive to speciation and can bedescribed by one nearly constant partial molarvolume of water (Richet & Polian, 1998; Bouhifdet al., 2001).

Since water solubility in silicate melts in-creases with pressure and since at the same time,the solubility of silicates in aqueous fluids alsoincreases, the miscibility gap between water andsilicate melts may ultimately disappear. Thiseffect was already considered by Niggli (1920).Kennedy et al. (1962) showed that in the systemSiO2–H2O the compositions of water-saturatedmelt and coexisting fluid approach each otherclose to 1 GPa. The first direct observation ofcomplete miscibility in a silicate-H2O systemwas reported by Shen and Keppler (1997), seeFigure 1.6.

If complete miscibility between melt and fluidoccurs, a water-saturated solidus cannot be de-fined any more. At low pressures, melting ina binary silicate-H2O system in the presence

Volatiles under High Pressure 17

H2O H2O H2OSiO2 SiO2 SiO2

critical point

critical pointmelt

meltmelt

+ fluid melt+ fluid

Qtz+

melt

Qtz+

melt

Qtz + fluid Qtz + fluid Qtz + fluid

fluid

fluid fluid

T T T

Fig. 1.7 Melting relationships in the SiO2–H2O system, schematic. Left: Melting below the critical curve; right:melting above the critical curve. After Keppler and Audetat (2005). Shown are the phase relationships for threedifferent pressures, with pressure increasing from the left to the right diagram. Reprinted with permission from theEuropean Mineralogical Union.

of an excess fluid phase occurs at a definedwater-saturated solidus temperature, at whichhydrous melt, fluid and solid silicate coexist.This is the situation on the left-hand side ofFigure 1.7. A miscibility gap between the meltand the fluid allows both phases to coexist. How-ever, with increasing pressure, the miscibility gapis getting smaller and finally disappears. In theabsence of a miscibility gap (Figure 1.7, right di-agram), a fluid of variable composition coexistswith a solid at any temperature. The fluid com-position is very water-rich at low temperature;with increasing temperature, the solubility of sil-icate in the fluid increases until it finally reachesthe composition and properties of a hydrous sil-icate melt. Due to the continuous change froma ‘‘fluid-like’’ to a ‘‘melt-like’’ phase, the solidustemperature cannot be defined any more. In anybinary silicate-H2O system, a critical curve canbe defined, which for a given pressure specifiesthe maximum temperature under which a meltand a fluid phase may coexist. The intersection ofthis line with the water-saturated solidus curvegives a critical endpoint at which the soliduscurve ceases. Beyond the pressure of this crit-ical endpoint, a water-saturated solidus cannotbe defined any more and melting phase rela-tionships resemble the situation shown on theright-hand side of Figure 1.7. Bureau and Keppler(1999) have mapped out critical curves for several

felsic compositions; they suggest that in thesesystems the water-saturated solidus terminatesat about 1.5–2 GPa. However, the location of thecritical endpoint in basic to ultrabasic systemsis not yet well constrained; it may occur some-where in the 4–6 GPa range (Mibe et al., 2004;Kessel et al., 2005).

1.3.4 Hydrous fluids

Hydrous fluids are important agents of masstransport in subduction zones (e.g. Manning,2004) and they are responsible for the formationof many economically important hydrothermalore deposits (Hedenquist & Lowenstern, 1994).The properties of water under typical subvolcanichydrothermal conditions, e.g. at 0.1 0.2 GPaand at 700–800 ◦C are very different from thewater known at ambient conditions (Eugster,1986; Franck, 1987). Liquid water under standardconditions is an extremely good solvent for ionicspecies; salts such as NaCl are highly solublein water and they are practically completelydissociated into Na+ and Cl− ions. Similarly,HCl dissolved in water is a very strong acidwith nearly complete dissolution into (hydrated)protons and Cl− anions. This is due to the highdielectric constant (e = 80) of water under stan-dard conditions. According to the Coulomb law,the attractive force between a pair of cation and

18 hans keppler

anion is reduced proportional to 1/e; therefore,in a solvent with e = 80, the attractive forcebetween the two ions is just about 1% of the forceexpected in vacuum. This effect is responsible forthe strong dissociation and the associated highsolubility of many ionic solutes in water. At anatomistic level, the high dielectric constant isrelated to the polar nature of the angular H2Omolecules, which forms oriented layers aroundions in such a way that the preferred orientationof the water molecules shields the electricalfields. However, at 0.1–0.2 GPa and 700–800 ◦C,the density of water (Burnham et al., 1969; Pitzer& Sterner, 1994) is reduced to 0.2–0.4 g/cm3

so that much less water molecules per volumeunit are available to shield electrical charges;moreover the high temperature counteracts theformation of oriented dipole layers. For thisreason, the dielectric constant of water underthese typical magmatic hydrothermal conditionsin the crust is only about 5, which is similarto a chlorinated hydrocarbon under standardconditions. Accordingly, the solvent behavior ofwater changes dramatically. Salts, such as NaCl,are not fully dissociated any more; rather they aredissolved as ion pairs or as multiple ion clustersand immiscibility between a water-rich vapor anda salt-rich brine may occur (Quist & Marshall,1968; Bodnar et al., 1985; Brodholt, 1998). HClis only weakly dissociated and only a weak acidunder these conditions (Frantz & Marshall, 1984).Since ionic substances are primarily dissolved asion pairs, solubilities of cations strongly dependon the availability of suitable counter ions thatallow the formation of stable ion pairs. This isthe reason why under magmatic-hydrothermalconditions, the solubility of ore metals, such asCu or Au, very much depends on the availabilityof halogens as ligands. The fluid/silicate meltpartition coefficients of Cu, for example, in-creases by orders of magnitude in the presence ofa few wt % of NaCl or HCl in the fluid (Candela& Holland, 1984; Keppler & Wyllie, 1991).

When pressure increases by several GPa to typ-ical upper mantle conditions, dielectric constantsand ionic dissociation increase again, withprofound changes in the behavior of fluids. The

changes of fluid properties are best rationalizedas a function of temperature and fluid density(Burnham et al., 1969; Pitzer & Sterner, 1984),rather than of temperature and pressure. Thisimplies that the most profound changes in fluidproperties actually occur in the 0–1 GPa range,with more subtle changes at higher pressures.

SiO2 is the most important solute in hydrousfluids of the upper mantle. The solubility of SiO2in water has been well calibrated in several stud-ies (e.g. Manning, 1984). The solution mechanismwas studied by in-situ Raman spectroscopy in ahydrothermal diamond anvil cell by Zotov andKeppler (2000; 2002). These data show that silicais initially dissolved as orthosilicic acid H4SiO4at low concentrations, which then polymerizesfirst to pyrosilicic acid dimers H6Si2O7 and thento higher polymers, as solubility increases withpressure. This increasing polymerization of silicain the fluid, together with the depolymerizationof silicate melts by water, provides an atomisticexplanation for the occurrence of complete misci-bility in silicate-H2O systems. Viscosities in suchsystems increase continuously with silicate con-centration (Figure 1.5, above); however, most ofthe increase occurs on the very silicate-rich side ofthe system, so that fluids with moderately highsilicate content still retain very low viscosities(Audetat & Keppler, 2005).

The composition of aqueous fluids coexistingwith MgSiO3 enstatite and Mg2SiO4 forsterite inthe system MgO–SiO2 changes profoundly withpressure; while at 1 GPa, the solute appears tobe rich in SiO2, the MgO/SiO2 ratio of the fluidas well as bulk solute concentration strongly in-crease with pressure (Ryabchikov et al., 1982;Stalder et al., 2001; Mibe et al., 2002).

High field strength elements (HFSE), i.e. ele-ments that form cations with a high charge andrelatively small ionic radius, such as Ti4+, Zr4+,Hf4+, Nb5+, and Ta5+, are generally poorly sol-uble in water, even at upper mantle pressures(Manning, 2004; Audetat & Keppler, 2005). Asan example, Figure 1.8 shows experimental dataon the solubility of TiO2 in water. In the pres-ence of silicates and aluminosilicates solutes, thesolubility of these elements increases, but it still