geophysical monograph series gale (eds.) 163 remote sensing of northern hydrology: measuring...

Geophysical Monograph Series

IncludingIUGG Volumes

Maurice Ewing VolumesMineral Physics Volumes

GM01018_CH01.qxd 22/10/07 1:55 PM Page i

Geophysical Monograph Series

139 Volcanism and the Earth’s Atmosphere Alan Robock and Clive Oppenheimer (Eds.)

140 Explosive Subaqueous Volcanism James D. L. White, John L. Smellie, and David A. Clague (Eds.)

141 Solar Variability and Its Effects on Climate Judit M. Pap and Peter Fox (Eds.)

142 Disturbances in Geospace: The Storm-SubstormRelationship A. Surjalal Sharma, Yohsuke Kamide, and Gurbax S. Lakhima (Eds.)

143 Mt. Etna: Volcano Laboratory Alessandro Bonaccorso,Sonia Calvari, Mauro Coltelli, Ciro Del Negro, and Susanna Falsaperla (Eds.)

144 The Subseafloor Biosphere at Mid-Ocean RidgesWilliam S. D. Wilcock, Edward F. DeLong, Deborah S.Kelley, John A. Baross, and S. Craig Cary (Eds.)

145 Timescales of the Paleomagnetic Field James E. T.Channell, Dennis V. Kent, William Lowrie, and JosephG. Meert (Eds.)

146 The Extreme Proterozoic: Geology, Geochemistry, and Climate Gregory S. Jenkins, Mark A. S. McMenamin,Christopher P. McKay, and Linda Sohl (Eds.)

147 Earth’s Climate: The Ocean–Atmosphere InteractionChunzai Wang, Shang-Ping Xie, and James A. Carton (Eds.)

148 Mid-Ocean Ridges: Hydrothermal InteractionsBetween the Lithosphere and Oceans Christopher R.German, Jian Lin, and Lindsay M. Parson (Eds.)

149 Continent-Ocean Interactions Within East AsianMarginal Seas Peter Clift, Wolfgang Kuhnt, PinxianWang, and Dennis Hayes (Eds.)

150 The State of the Planet: Frontiers and Challengesin Geophysics Robert Stephen John Sparks and Christopher John Hawkesworth (Eds.)

151 The Cenozoic Southern Ocean: Tectonics,Sedimentation, and Climate Change Between Australiaand Antarctica Neville Exon, James P. Kennett, and Mitchell Malone (Eds.)

152 Sea Salt Aerosol Production: Mechanisms, Methods,Measurements, and Models Ernie R. Lewis andStephen E. Schwartz

153 Ecosystems and Land Use Change Ruth S. DeFries,Gregory P. Anser, and Richard A. Houghton (Eds.)

154 The Rocky Mountain Region—An EvolvingLithosphere: Tectonics, Geochemistry, and GeophysicsKarl E. Karlstrom and G. Randy Keller (Eds.)

155 The Inner Magnetosphere: Physics and ModelingTuija I. Pulkkinen, Nikolai A. Tsyganenko, and ReinerH. W. Friedel (Eds.)

156 Particle Acceleration in Astrophysical Plasmas:Geospace and Beyond Dennis Gallagher, JamesHorwitz, Joseph Perez, Robert Preece, and John Quenby (Eds.)

157 Seismic Earth: Array Analysis of BroadbandSeismograms Alan Levander and Guust Nolet (Eds.)

158 The Nordic Seas: An Integrated Perspective HelgeDrange, Trond Dokken, Tore Furevik, Rüdiger Gerdes,and Wolfgang Berger (Eds.)

159 Inner Magnetosphere Interactions: New PerspectivesFrom Imaging James Burch, Michael Schulz, andHarlan Spence (Eds.)

160 Earth’s Deep Mantle: Structure, Composition, andEvolution Robert D. van der Hilst, Jay D. Bass,Jan Matas, and Jeannot Trampert (Eds.)

161 Circulation in the Gulf of Mexico: Observations andModels Wilton Sturges and Alexis Lugo-Fernandez(Eds.)

162 Dynamics of Fluids and Transport Through FracturedRock Boris Faybishenko, Paul A. Witherspoon, andJohn Gale (Eds.)

163 Remote Sensing of Northern Hydrology: MeasuringEnvironmental Change Claude R. Duguay and AlainPietroniro (Eds.)

164 Archean Geodynamics and Environments Keith Benn, Jean-Claude Mareschal, and Kent C. Condie(Eds.)

165 Solar Eruptions and Energetic ParticlesNatchimuthukonar Gopalswamy, Richard Mewaldt, and Jarmo Torsti (Eds.)

166 Back-Arc Spreading Systems: Geological, Biological,Chemical, and Physical Interactions David M. Christie,Charles Fisher, Sang-Mook Lee, and Sharon Givens (Eds.)

167 Recurrent Magnetic Storms: Corotating Solar Wind Streams Bruce Tsurutani, Robert McPherron,Walter Gonzalez, Gang Lu, José H. A. Sobral, andNatchimuthukonar Gopalswamy (Eds.)

168 Earth’s Deep Water Cycle Steven D. Jacobsen andSuzan van der Lee (Eds.)

169 Magnetospheric ULF Waves: Synthesis and New Directions Kazue Takahashi, Peter J. Chi, Richard E. Denton, and Robert L. Lysal (Eds.)

170 Earthquakes: Radiated Energy and the Physics of Faulting Rachel Abercrombie, Art McGarr, Hiroo Kanamori, and Giulio Di Toro (Eds.)

171 Subsurface Hydrology: Data Integration for Propertiesand Processes David W. Hyndman, Frederick D. Day-Lewis, and Kamini Singha (Eds.)

172 Volcanism and Subduction: The Kamchatka RegionJohn Eichelberger, Evgenii Gordeev, Minoru Kasahara,Pavel Izbekov, and Johnathan Lees (Eds.)

173 Ocean Circulation: Mechanisms and Impacts—Pastand Future Changes of Meridional OverturningAndreas Schmittner, John C. H. Chiang, and Sidney R. Hemming (Eds.)

GM01018_CH01.qxd 22/10/07 1:55 PM Page ii

Geophysical Monograph 174

Post-Perovskite: The Last Mantle Phase Transition

Kei HiroseJohn Brodholt

Thorne LayDavid Yuen

Editors

American Geophysical UnionWashington, DC

GM01018_CH01.qxd 22/10/07 1:55 PM Page iii

Published under the aegis of the AGU Books Board

Darrell Strobel, Chair; Gray E. Bebout, Cassandra G. Fesen, Carl T. Friedrichs, Ralf R. Haese, W. Berry Lyons, Kenneth R. Minschwaner, Andrew Nyblade, and Chunzai Wang, members.

Library of Congress Cataloging-in-Publication Data

Post-perovskite : the last mantle phase transition / Kei Hirose ... [et al.], editors.p. cm. – (Geophysical monograph ; 174)

ISBN: 978-0-87590-439-91. Perovskite. I. Hirose, Kei.QE391.P47P67 2007551.1'16–dc22

2007042659

ISBN: 978-0-87590-439-9ISSN 0065-8448

TKCopyright 2007 by the American Geophysical Union2000 Florida Avenue, N.W.Washington, DC 20009

Figures, tables and short excerpts may be reprinted in scientific books and journals if the source is properly cited.

Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, isgranted by the American Geophysical Union for libraries and other users registered with the Copyright ClearanceCenter (CCC) Transactional Reporting Service, provided that the base fee of $1.50 per copy plus $0.35 per page is paiddirectly to CCC, 222 Rosewood Dr., Danvers, MA 01923. 0065-8448/07/$01.50+0.35.

This consent does not extend to other kinds of copying, such as copying for creating new collective works or for resale.The reproduction of multiple copies and the use of full articles or the use of extracts, including figures and tables, forcommercial purposes requires permission from the American Geophysical Union.

Printed in the United States of America.

GM01018_CH01.qxd 22/10/07 1:55 PM Page iv

CONTENTS

PrefaceCharles T. Prewitt ................................................................................................................................ vii

An Introduction to Post-Perovskite: The Last Mantle Phase TransitionKei Hirose, John Brodholt, Thorne Lay, and David A. Yuen................................................................. 1

Section I: Mineral Physics (Experimental)

Review of Experimental Studies on Mantle Phase TransitionsTakehiko Yagi ...................................................................................................................................... 9

Discovery of Post-Perovskite Phase Transition and the Nature of D” LayerKei Hirose ........................................................................................................................................... 19

Effect of Iron on the Properties of Post-Perovskite SilicateWendy L. Mao, Andrew J. Campbell, Vitali B. Prakapenka, Russell J. Hemley, and Ho-kwang Mao ............................................................................................................................ 37

Electronic Transitions and Spin States in the Lower MantleJie Li.................................................................................................................................................... 47

Lattice-Preferred Orientation of Lower Mantle Materials and Seismic Anisotropy in the D” LayerDaisuke Yamazaki and Shun-ichiro Karato.......................................................................................... 69

Section II: Mineral Physics (Theoretical)

Thermodynamic Properties and Stability Field of MgSiO3 Post-PerovskiteRenata M. Wentzcovitch, Koichiro Umemoto, Taku Tsuchiya, and Jun Tsuchiya ................................ 79

The High-Temperature Elasticity of MgSiO3 Post-PerovskiteStephen Stackhouse and John P. Brodholt ........................................................................................... 99

Effect of Chemistry on the Physical Properties of Perovskite and Post-PerovskiteRazvan Caracas and Ronald E. Cohen................................................................................................. 115

Section III: Seismology

Reconciling the Post-Perovskite Phase With Seismological Observations of Lowermost Mantle StructureThorne Lay and Edward J. Garnero ..................................................................................................... 129

Predicting a Global Perovskite and Post-Perovskite Phase BoundaryDaoyuan Sun, Don Helmberger, Xiaodong Song, and Stephen P. Grand............................................. 155

Seismic Anisotropy of Post-Perovskite and the Lowermost MantleJames Wookey and John-Michael Kendall ........................................................................................... 171

Constraints on the Presence or Absence of Post-Perovskite in the Lowermost Mantle From Long-Period SeismologyChristine Houser ................................................................................................................................. 191

GM01018_CH01.qxd 22/10/07 1:55 PM Page v

Section IV: Dynamics

Mantle Dynamics and the D″″ Layer: Impacts of the Post Perovskite PhaseW.R. Peltier ......................................................................................................................................... 217

Influence of the Post-Perovskite Transition on Thermal and Thermo-Chemical Mantle ConvectionPaul J. Tackley, Takashi Nakagawa, and John W. Hernlund................................................................. 229

The Dynamical Influences From Physical Properties in the Lower Mantle and Post-Perovskite Phase TransitionDavid A. Yuen, Ctirad Matyska, Ondrej Cadek, and Masanori Kameyama.......................................... 249

Deformation-Induced Mechanical Instabilities at the Core-Mantle BoundaryNick Petford, Tracy Rushmer, and David A. Yuen ............................................................................... 271

GM01018_CH01.qxd 22/10/07 1:55 PM Page vi

vii

Post-Perovskite: The Last Mantle Phase TransitionGeophysical Monograph Series 174Copyright 2007 by the American Geophysical Union10.1029/174GM01

I was astounded and excited when my copy of Sciencearrived containing the paper by Murakami, Hirose,Kawamura, Sata, and Ohishi (2004) describing still anotherphase transition in MgSiO3 at high pressure. My first reac-tion was, “Why did it take so long to discover this phase andwasn’t it obvious that there was a silicate phase transitionthat would explain the D″ discontinuity?” One must con-gratulate these authors for being smart enough to pursuesuch an objective and to follow it through to show that itdoes explain much or all of the evidence for a deep-mantlediscontinuity that had been discovered previously byseismologists.

One curious aspect of the discovery of the post-perovskitesilicate is that a phase with the same structure, CaIrO3, hadbeen known for almost 40 years (Rodi and Babel, 1965), butnone of us had picked it up as a possible structure for a phasein the mantle. For example, the excellent compilation anddescriptions of perovskite-like structures having many differ-ent compositions by Mitchell (2002) does not list this partic-ular composition, and I am not aware of other publicationsthat proposed it as the basis for a possible silicate structure.Ir4+ has a slightly larger radius than Ti4+, but CaIrO3 has adifferent orthorhombic space group (Cmcm) from that of per-ovskite CaTiO3 (Pbnm).

Comprehensive examination of the composition andphysical properties of Earth’s mantle and core has takenplace during a period of just over fifty years. It began withBirch’s classic paper, “Elasticity and constitution of theEarth’s interior (Birch, 1952).” Here, he conclusivelydemonstrated that (1) the mantle is composed mostly of sil-icate minerals; (2) the upper mantle and lower mantle areessentially homogeneous but of somewhat differing compo-sitions and are separated by a thin transition zone associatedwith silicate phase transitions; and (3) the inner and outercore are alloys of crystalline and molten iron, respectively.It has since been the goal of experimentalists and theoreti-cians to examine the mantle and core in more detail and todevelop models explaining as many of the geophysical andgeochemical observations as possible. This effort has beenextraordinarily successful and has provided a forum for

investigators from all over the world and from a wide rangeof scientific disciplines to propose, collaborate, argue,compete, and generate many publications describing theirconclusions.

My interest in high-pressure mineral physics goes back tothe early 1960s. At that time the most active research groupin the world of high-pressure synthesis and characterizationof possible mantle phases was that of Ted Ringwood inCanberra, Australia. Synthesis experiments were conductedwith a variety of apparatus consisting of opposed tungstencarbide pistons of various designs with resistive heating andrelying on the fact that many phases could be quenched toroom conditions without undergoing reversible structuraltransitions. Phase transitions and structures of many compo-sitions were reported by this group, and we now know theprincipal silicate phases in the transition zone as wadsleyite,majorite, and ringwoodite, all named for Australianresearchers active at that time.

A great breakthrough occurred with the introduction ofthe diamond-anvil cell, which enabled us to achieve muchhigher pressures and, eventually, extremely high tempera-tures and to obtain x-ray diffraction patterns on polycrys-talline samples and to some extent on single crystals. Usingthe diamond cell Liu (1974) synthesized silicate perovskitefrom pyrope-garnet, and then showed (Liu, 1975) that astarting glass composition of (Mg.75Fe.25)SiO3 plus 5 wt. %Al2O3 transformed to garnet, spinel, and orthorhombicperovskite, thus initiating an enormous amount of researchon the nature of the lower mantle. Mao and Bell (1976)were the first to achieve pressures above one megabar andsubsequent improvements in diamond-cell techniques per-mitted higher and higher pressures to be obtained with theeventual result being relatively routine experiments in themegabar range with laser heating to thousands of degreestemperature.

Along with improvements in diamond-cell techniques, itwas necessary to detect what was going on in the cell as pres-sure was increased. At first, investigators depended onobtaining diffraction patterns using laboratory-based x-rayapparatus. Then, in the late 1970s, synchrotron sources beganto be used for producing just the right kind of intense, high-energy x-ray beam for obtaining diffraction patterns at highpressures and later with simultaneous laser heating, which isessential for searching for new phases that might not be pre-sent if the sample is quenched to room temperature. Theseadvances and many others in pressure media, x-ray detectors,

PREFACE

GM01018_CH01.qxd 22/10/07 1:55 PM Page vii

computers, theoretical calculations, seismology, and geody-namics have all permitted enormous progress to be made inour understanding of the mantle and core.

Thus, the past ~50 years have been very productive in oursearch for how the Earth is put together. Undoubtedly, therewill be other discoveries that will provide even more thor-ough understanding, but it may be that the crystal structure ofthe post-perovskite magnesium-iron silicate discovered byMurakami et al. will be the last new structure that canprovide a major explanation of the seismic structure of thelower mantle. I believe that this monograph is an excellentpresentation of the wide range of investigations that havetaken place since the post-perovskite discovery and congrat-ulate all the authors for their willingness to contributechapters based on their own research efforts.

REFERENCES

Birch, F. (1952) Elasticity and constitution of the Earth’s interior. J. Geophys.Res., 57, 227-286.

Liu, L.G. (1974) Silicate perovskite from phase transformations of pyrope-garnet at high pressure and temperature. Geophys. Res. Lttrs., 1, 277-280.

Liu, L.G. (1975) Post-oxide phases of olivine and pyroxene and mineralogyof the mantle. Nature, 258, 510-512.

Mao, H.K., and P.M. Bell. (1976) High pressure physics: the 1 megabar markon the ruby R1 static pressure scale. Science, 191, 851-852.

Mitchell, R.H. (2002) Perovskites: Modern and Ancient. 318 p. Almaz Press,Thunder Bay, Ontario.

Murakami, M., K. Hirose, K. Kawamura, N. Sata, and Y. Ohishi. (2004)Post-perovskite phase transition in MgSiO3. Science, 304, 855-858.

Rodi, F., and D. Babel. (1965) Erdalkaliiridium (IV)-oxide: Kristallstrukturvon CaIrO3. Zeit. anorg. allg. Chemie, 336, 17-23.

Charles T. Prewitt

VIII

GM01018_CH01.qxd 22/10/07 1:55 PM Page viii

INTRODUCTION

The importance of mineralogical phase transitions in thedeep Earth was anticipated in the 1950s by Francis Birch,and over the past half century interpretation of seismologi-cal structures by mineralogical phase equilibria has guidedcompositional, thermal and dynamical models of the planet’sinterior. Confident seismological detection of the occur-rence of a specific predicted phase change at depth enablesa cascade of constraints to be placed on deep properties thatcan otherwise only be loosely bounded. Earth’s transitionzone, the depth region from 410- to ∼800-km with complexseismological discontinuities and multiple associated phasechanges, has now been characterized in substantial detail.However, important attributes of the underlying lower man-tle, such as absolute temperature, have been inferred only bylarge depth extrapolations from conditions established byphase changes in the transition zone or at the inner core

boundary. Demonstration of the presence of a well-charac-terized lower mantle phase change raises the prospect of rev-olutionary improvements in our understanding of lowermantle properties and dynamics.

Following the discovery of silicate perovskite by Liu[1974], MgSiO3 perovskite has become recognized as theprincipal mineral occurring in Earth’s lower mantle. For sev-eral decades, MgSiO3 perovskite has been extensively stud-ied to clarify its physical properties, crystal chemistry, androle in mantle dynamics. After a workshop in Bisbee,Arizona, held in 1987, a monograph exploring this impor-tant mineral entitled “Perovskite: A Structure of GreatInterest to Geophysics and Materials Science” edited by A. Navrotsky and D. J. Weidner, was published by theAmerican Geophysical Union in 1989. Observation of seis-mological discontinuities in the lowermost mantle (Wrightand Lyons [1975], Lay and Helmberger [1983]) motivatedinvestigation of very high-pressure and high-temperatureproperties and stability of perovskite. This has been a subjectof some controversy in the high-pressure mineral physicscommunity. It was suggested at one time that orthorhombicperovskite transforms to cubic structure with increasingtemperature. Dissociation of perovskite into mixed simpleoxides was also suggested to occur in the mid-lower mantle.However, these possibilities were not verified by subsequentstudies. The likelihood of perovskite transforming into adenser MgSiO3 polymorph was not generally anticipated, pri-marily because perovskite is an ideal close-packed structure

An Introduction to Post-Perovskite: The Last Mantle Phase Transition

Kei Hirose1, John Brodholt2, Thorne Lay3, and David A. Yuen4

Discovery of the perovskite to post-perovskite phase transition in MgSiO3,expected to occur for deep mantle conditions, was first announced in April 2004.This immediately stimulated numerous studies in experimental and theoreticalmineral physics, seismology, and geodynamics evaluating the implications of amajor lower mantle phase change. A resulting revolution in our understanding ofthe D″ region in the lowermost mantle is well underway. This monograph presentsthe multidisciplinary advances to date ensuing from interpreting deep mantle seis-mological structures and dynamical processes in the context of the experimentallyand theoretically determined properties of the post-perovskite phase change; thelast silicate phase change likely to occur with increasing pressure in lowermostmantle rocks.

1

1Tokyo Institute of Technology, Tokyo, Japan.2University College London, London, United Kingdom.3University of California, Santa Cruz, CA.4University of Minnesota, Minneapolis, MN.


GM01018_CH02.qxd 20/10/07 1:45 PM Page 1

that is favorable for high-pressure conditions. The notion thatperovskite is the ultimate stable form of silicates in theEarth’s mantle thus began to take hold, although awaitingconfirmation by advances in experimental and theoreticalmineral physics.

This notion of perovskite stability clearly could notaccount for the observed seismic discontinuities in the lower-most few hundred kilometers of the mantle (the D″ region),and the possibility of some phase transition occurring therewas proposed on the basis of seismological and geodynami-cal considerations [e.g., Sidorin et al., 1999]. The D″ regionhas long been enigmatic because its seismological propertiesare distinct from relatively homogeneous overlying lowermantle. With improving characterization of seismologicalproperties of D″, it became clear that some combination ofmineralogical, compositional, and thermal heterogeneity isrequired to account for the observed structures [Lay andGarnero, 2004].

As experimental techniques advanced to span the fullrange of pressure and temperature (P-T) conditions of thelower mantle, a phase transition from MgSiO3 perovskite topost-perovskite was discovered and confirmed by severallaboratories. This was first reported in 2004, 30 years aftersilicate perovskite was first synthesized. This new mantlemineral has profound implications for the nature of anddynamics in the D″ region. The occurrence of the phasechange and the distinct properties of post-perovskite nowprovide a viable explanation for several major seismologicalcharacteristics of the D″ region, along with having importantimplications for the dynamics of the lower mantle boundarylayer. The specific properties of the phase change, togetherwith seismological observations, also provide the first directconstraints on absolute temperature and temperature gradi-ents in the lowermost mantle, eliminating the need for vastextrapolations of temperature estimates over large depthranges.

Several long-term enigmas may be reconciled by occur-rence of post-perovskite in the lower mantle. However, the D″region, by virtue of its location above the boundary betweenthe liquid iron core and the rocky silicate mantle, is stillexpected to have complex thermal and chemical structures.Strong radial and lateral temperature gradients should existin the boundary layer caused by heat flowing out of the coreand by mantle convection. Chemical heterogeneity is likelyto exist in this boundary layer due to the huge density con-trast at the core-mantle boundary (CMB) and the long historyof chemical differentiation in the interior, with ancientresidues of mantle differentiation and/or subsequent contri-butions from deep subduction of oceanic lithosphere, partialmelting in the ultra-low velocity zone (ULVZ) just above theCMB, and core-mantle chemical reactions. The complexityof a thermo-chemical boundary layer in D″ was extensively

documented in the 1998 American Geophysical Unionmonograph “The Core-Mantle Boundary Region” edited byM. Gurnis, E. Knittle, M. E. Wysession, and B. A. Buffett.This context indicates that extensive characterization of thethermal and chemical influences on the post-perovskite phasetransition is critical to our ability to relate seismologicalobservations to the behavior of the new mineral.

The current monograph presents a full span of post-perovskiteattributes, including characterization by experimental andtheoretical mineral physics, seismological interpretations anddynamical considerations. The papers are grouped by disci-plinary emphasis, but all of the geophysical attributes areinterconnected. It should quickly become evident why thislast silicate phase transition in the mantle is eliciting suchexcitement and concentrated effort.

EXPERIMENTAL MINERAL PHYSICS PAPERS

The deep lower mantle is a challenging region to quantifymineralogically, in part because experimental investigationsat relevant P-T conditions are very difficult. Recent develop-ments at synchrotron radiation facilities and advances inlaser-heated diamond-anvil cell (LH-DAC) techniques nowenable experimentation at relevant high P-T conditions within-situ X-ray measurements of deep mantle minerals. Thephase transition from MgSiO3 perovskite to post-perovskitewas discovered through a change in X-ray diffraction spectraabove 125 GPa and 2500 K [Murakami et al., 2004], corre-sponding to conditions near the top of the D″ region. In addi-tion, electronic spin-pairing transition of iron in perovskiteand magnesiowüstite was found to occur in the lower mantle,based on X-ray emission spectroscopy measurements [e.g., Badro et al., 2003]. These new findings in experimentalmineral physics have significant implications for structure,seismic heterogeneity, dynamics and chemistry of the middleto deep lower mantle.

Interpreting the cause of seismic discontinuities in themantle has long been a central subject of high-pressureexperimental studies. Yagi reviews the progress of high-pres-sure experimental techniques and studies of mantle phasetransitions since the 1950s. The olivine to spinel phase trans-formation was first observed in Fe2SiO4 at 5 to 6 GPa[Ringwood, 1958]. With advances in generating higher pres-sures by using newly designed apparatus, phase relations onthe join Mg2SiO4-Fe2SiO4 were systematically investigated.Modified-spinel (so called β-phase) was found on the Mg-rich part on this join, but the crystal structure was not knownat that time. In the early 1970s, the pressure of the 660-kmseismic discontinuity was beyond the capability of anyexisting high-pressure apparatus. Liu [1974] first synthesizedsilicate perovskite using LH-DAC, inferring that the 660-kmseismic discontinuity is caused by formation of the dense

2 AN INTRODUCTION TO POST-PEROVSKITE: THE LAST MANTLE PHASE TRANSITION

GM01018_CH02.qxd 20/10/07 1:45 PM Page 2

perovskite-structured phase. MgSiO3 perovskite was subse-quently found to have great stability, leading to speculationsthat it may persist in this form all the way to the CMB.

Hirose reviews the discovery of MgSiO3 post-perovskiteand subsequent experimental studies on perovskite to post-perovskite phase transition in both simple and natural multi-component systems. While XRD patterns indicated the phasetransition in MgSiO3, the crystal structure of post-perovskitewas obtained with the aid of theoretical calculations. TheMgSiO3 post-perovskite phase transition boundary has beenexperimentally determined using several different pressurestandards. Most results show that the transition occurs withinthe mantle but the experimentally estimated transition pres-sure varies by as much as 15 GPa, due primarily to uncer-tainty in P-V-T equations of state of pressure standards. TheMgO pressure scale currently appears to be the most reliable,and the results based on the MgO scale indicate that the tran-sition pressure matches the general depth of the seismic dis-continuity observed near the top of D″ [e.g., Wysession et al.,1998] for a plausible mantle temperature of 2500 K. Thepost-perovskite phase transition occurs in a natural pyroliticmantle composition at pressures very similar to that in pureMgSiO3, when the MgO scale is used. Al-bearing(Mg,Fe)SiO3 perovskite is the most abundant mineral in sub-ducted MORB crust. The post-perovskite phase transitionoccurs in MORB materials at shallower depths, by about 70-km, than in pyrolite at the same temperature. Hirose alsodiscusses how several long-term seismological enigmas maybe reconciled by the properties of post-perovskite without theneed for invoking chemical heterogeneities; these include theD″ discontinuity, strong seismic anisotropy in the D″ region,and anti-correlation between anomalies in S-wave and bulk-sound velocities in the deep mantle. Some remainingunsolved problems in the lowermost mantle are summarized.

Iron is the most important impurity in MgSiO3 post-per-ovskite, and may significantly affect post-perovskite stability,density, and elastic properties. Mao, Campbell, Prakapenka,Hemley, and Mao report new experimental data on the vol-umes of (Mg0.8Fe0.2)SiO3 post-perovskite at high P-T and anestimate of the thermal expansivity at lowermost mantle con-ditions. The incorporation of iron significantly increases itsmass but only moderately expands the volume, resulting in alarge increase in density. They also review the previousresearch on Fe-bearing post-perovskite, inferring that ironlowers the perovskite to post-perovskite transition pressure,increases bulk modulus, and lowers the sound velocities.Nuclear resonant inelastic X-ray scattering (NRIXS) mea-surements on Fe-enriched post-perovskite (40% FeSiO3) at130 GPa and 300 K demonstrate that compressional andshear wave velocity estimates are consistent with seismolog-ical observations for the ULVZ. However, the extent of anyiron enrichment at the base of the mantle is controversial. The

effect of iron on the stability of post-perovskite is also still anopen issue.

The electronic spin state of iron in magnesiowüstite, per-ovskite, and post-perovskite may affect the physical proper-ties of the lower mantle. Li reviews the recent studies on spincrossover (high-spin to low-spin transition) of iron in thesemajor lower mantle minerals. Both experiment and theoryshow that the spin crossover occurs in magnesiowüstitearound 60 GPa at 300 K, resulting in remarkable changes involume and sound velocities. Note that the iron spin transi-tion takes place over a much broader pressure range at hightemperatures and therefore these changes should be gradualin the mantle. In contrast, the spin crossover in perovskite iscurrently controversial. The nature of spin transition is com-plicated in perovskite because iron has multiple valencestates and crystallographic sites to be incorporated. The pres-sure and sharpness of spin crossover in perovskite and itstemperature and compositional dependence are still poorlyknown.

Seismic anisotropy is observed in the D″ region, with anincrease in strength across the D″ seismic discontinuity.Since post-perovskite is possibly a predominant mineral inthis region, the seismic anisotropy may be caused by lattice-preferred orientation (LPO) of post-perovskite. Yamazakiand Karato review experimental and theoretical studies onthe deformation mechanism of post-perovskite. They discussthe experimental results on an analogue material, CaIrO3

with post-perovskite structure, obtained at high temperatureunder modest stress conditions, which may be applicable tothe deformation occurring in D″. These experiments showthat the layering plane (010) is a dominant slip plane.Yamazaki and Karato calculate the S-wave splitting for post-perovskite aggregates under horizontal flow and concludethat the sense of splitting is consistent with the seismologicalobservations (VSH is faster than VSV), but the magnitude is lessthan observed. A contribution from LPO of magnesiowüstitemay be important as an additional source of S-wave splittingobserved in the D″ layer.

THEORETICAL MINERAL PHYSICS PAPERS

The role that computational mineral physics had in the dis-covery, acceptance, and recognition of the importance of thepost-perovskite phase in the Earth cannot be overstated.Although it was experimental evidence of the perovskite topost-perovskite transformation that first stirred the computa-tional mineral physicists into action, the fact that the experi-mental evidence was quickly confirmed by ab initio resultsgave tremendous credence to it. Moreover, computationalmineral physics was immediately used to provide estimates ofthe Clapeyron slope - with which the variation in the depth ofthe D″ seismic velocity discontinuity could be compared - and

HIROSE ET AL. 3

GM01018_CH02.qxd 20/10/07 1:45 PM Page 3

estimates of the elasticity to compare with the observed seis-mic anisotropy. A flurry of further papers quickly providedhigh temperature elastic constants of post-perovskite, togetherwith estimates of the effect of chemistry on its stability andelastic properties.

The paper by Wentzcovitch, Tsuchiya, and Umemotoaddresses the stability of post-perovskite relative to per-ovskite using ab initio lattice dynamics. As long as the anhar-monicity is not too large, ab initio lattice dynamics is able toprovide accurate high temperature properties. Wentcovitch et al. present, therefore, a full range of thermodynamic prop-erties such as the heat capacity and entropy, as well as ther-moelastic properties such as the Gruneisen parameter. One ofthe major benefits of lattice dynamics over other moleculardynamics is that the free energy at a finite temperature ismore readily available. This allowed the authors to calculatethe Clapeyron slope, which they find to be about 7.5 MPa/Kat a temperature of 2500 K and a pressure of about 100 GPa.They also find the Clapeyron slope to be relatively insensi-tive to the choice of exchange-correlation energy (GGA orLDA). They also examine the possibility of post-perovskitedecomposing into the component oxides and find that it doesindeed dissociate at around a pressure of 1 TPa. They con-clude from this that post-perovskite would not exist in thecentre of the giant planets such as Saturn and Jupiter.

The paper of Stackhouse and Brodholt concentrates on thehigh P-T elasticity. They use new ab initio molecular dynam-ics calculations to generate equation of state parameters anda full set of elastic constants at 136 GPa and at temperaturesranging from 0 to 4000 K in 500 K intervals. They then com-pare their results with those previously obtained using latticedynamics and find that the two sets of results start to divergeat high temperatures (greater than about 2500 K). Althoughthe cause of this is not certain, it may be due either to a breakdown in the quasi-harmonic approximation used in latticedynamics or, alternatively, due to the choice of pseudopoten-tial or exchange-correlation functional. The divergence inhigh temperature elasticity particularly affects the seismicanisotropy and, unfortunately, at high temperatures the twomethods predict completely different polarizations for thesame crystal orientations. For instance, for a crystal aggre-gate developed with slip in the (010) plane (the slip planeintuitively expected from the layered octahedra), the MDresults predict about a 2% shear-wave anisotropy, with thehorizontal S-wave propagating faster than the vertical one;the lattice dynamics results on the other hand, show the exactopposite polarity. This makes the interpretation of theobserved anisotropy in D″ somewhat uncertain. Stackhouseand Brodholt also use their elastic constants, together withestimates for other phases and chemical components, to show that a post-perovskite bearing D″ matches PREM towithin 1%.

The last paper in the Theoretical Mineral Physics section isby Caracas and Cohen who use density functional methodsto look at the effect of Fe and Al on the physical propertiesand stability of perovskite and post-perovskite. They find thatthe effect of iron is to decrease considerably the transitionpressure between the two phases. The FeSiO3 end-memberperovskite is in fact stable at all pressures relative to per-ovskite. Aluminium, on the other hand, increases the transi-tion pressure, but to a lesser extent that iron. Due to its highatomic mass, Fe strongly increases the density of perovskiteand post-perovskite, and, therefore, strongly decreases seis-mic velocities. Fe is also found to slightly decrease the seis-mic anisotropy. Al2O3 also affects the seismic velocities, butgenerally to a lesser extent than Fe. It does, however, stronglyincrease the seismic anisotropy. In order to characterize theeffect that iron has on the width of the phase transition,Caracas and Cohen use a non-ideal solution model to con-struct a pressure-temperature-composition phase diagram forFe2+ bearing MgSiO3. They find that the effect of Fe can bequite considerable, especially in a colder mantle. Forinstance, in a very cool mantle of 1500 K (i.e., in a subduct-ing slab), the phase transition could begin at a depth 150 kmshallower than in a 3000 K mantle. In addition, they find avery wide two-phase loop in the cool case, and for about 10%Fe, the transition takes place over about 170 km depth. Incontrast, the width of the transition at 3000 K is only 50 km.The exact width depends on the concentration of iron.

SEISMOLOGICAL PAPERS

Analyses of seismic waves that traverse Earth’s interiorprovide direct constraints on material properties of the man-tle such as its elastic wave velocities and density, with geo-chemical and petrological models being guided by and testedagainst the seismic observations. The transition zone velocitydiscontinuities discovered in the 1960s have played a majorrole in developing mineralogical and petrological models forthe upper mantle, and it is not surprising that lower mantlevelocity discontinuities discovered in the 1970s and 1980sare now playing a similar role in advancing models of thedeep mantle as mineral physics experiments progressivelyreveal high pressure properties of major mineral types.

The presence of a mineralogical phase change in the man-tle can produce observable changes in material properties.Lay and Garnero review the complex suite of seismologicalobservations of deep mantle velocity discontinuities, explor-ing the viability of attributing some features to the perovskiteto post-perovskite phase transition in a chemically and ther-mally heterogeneous environment. Interpretation of anobserved seismic velocity discontinuity as the result of a par-ticular phase transition is not unique, so the variability of the


GM01018_CH02.qxd 20/10/07 1:46 PM Page 4

seismic observations is considered in order to evaluate thelikelihood that a phase transition to post-perovskite structureoccurs in the deep mantle. Attributes of the seismic disconti-nuity observations such as the depth, size, and sharpness ofP- and S-wave discontinuities, and their relationship to sur-rounding volumetric velocity heterogeneity, are considered.Inconsistencies with the phase change predictions for end-member compositions motivate consideration of the possibleeffects of variable chemistry and temperature compatiblewith the seismic heterogeneities. The observation of multipleseismic velocity discontinuities in some regions is discussed,with attendant implications for multiple intersections of thegeotherm with the post-perovskite phase boundary beingconsidered. The importance of expanding seismic wave datasets, new waveform stacking and migration algorithms, and2D and 3D waveform modeling methods is discussed. Theoverall emphasis of this contribution is on the variable natureof the seismic velocity structures in the lowermost mantle,and the need to avoid simplified generalizations about theoccurrence of post-perovskite. While the existence of post-perovskite in the deep mantle cannot yet be conclusivelydemonstrated, it is shown that its presence is plausible andmany attributes of the deep structure revealed by seismologycan be reconciled with the phase change occurring in a ther-mally and compositionally heterogeneous environment.

Sun, Helmberger, Song, and Grand assume that post-per-ovskite is present and that large-scale tomographic seismicvelocity variations are primarily thermally controlled to makepredictions of the lateral position of the phase boundary.Higher velocity regions are assumed to be cooler, giving riseto shallower occurrence of the phase-transition than in lowervelocity, presumably warmer regions. Detailed seismic wave-form analysis is then used to constrain the properties of thethermally modulated phase boundary. The use of tomo-graphic models provides constraints on the lower mantlevelocity models that were not available when the first D″ dis-continuity models were developed, and this allows improvedresolution of the overall velocity models and discontinuitydepths. Strong lateral variations in the phase boundary arepredicted by the strong gradients in tomography models andin independent travel time observations, and these variationscan be reconciled with seismic reflections from the phaseboundary. Large low velocity provinces are recognized toinvolve chemical heterogeneity in addition to having warmertemperatures. Internal convection of these provinces resultsin localized regions where lower temperatures can supportpost-perovskite even if most of the chemically distinctprovince is too warm for the phase to be stable. The conclu-sion is again that many attributes of the lowermost mantleseismic velocity structure can be accounted for by pre-dictable variations in the phase boundary when thermal andchemical heterogeneity are allowed for.

Wookey and Kendall consider the seismic velocityanisotropy expected for post-perovskite, demonstrating howthis provides another observational approach to detecting andpotentially exploiting the presence of post-perovskite to con-strain deformational processes in the lower mantle boundarylayer. Observations of seismic shear wave splitting are sum-marized and considered in the context of predicted anisotropiceffects for end-member mineralogies involving perovskite,MgO, and post-perovksite, including variations with Fe and Alcomponents. Uncertainties in the slip planes that will actuallybe activated in lower mantle boundary layer flows preclude adefinitive conclusion at this time, and alternatives such asshape-preferred anisotropy associated with liquid inclusionscannot be ruled out. However, the possibility of lattice pre-ferred orientation in post-perovskite accounting for seismicobservations is demonstrated for some viable slip systems.

Observation of the weak reflections from a phase boundaryare difficult to seek on a global basis, so to constrain large-scale structures Reif explores long-period travel time con-straints on models with and without phase boundaries.Normal mode observations are shown to provide some weakconstraints, precluding the existence of a global, thick layerof high density post-perovskite, but not the possibility of astrongly laterally varying layer as suggested by the velocitydiscontinuity studies. Carefully measured long-period arrivaltime patterns are used to characterize the variable slope of thefirst-arrival time curve, finding only limited first-arrival timesupport for a rapid velocity increase like that expected for aphase boundary. Further comparison of P- and S-wavetomography models in terms of thermal, compositional andphase change effects demonstrates that the additional degreesof freedom upon including a phase change further compli-cates the inversion for separate compositional and thermaleffects. As in the chapters on seismic velocity discontinuities,it is recognized that progress in mapping the presence ofpost-perovskite hinges upon more thorough calibration of theeffects of composition on the phase change itself, if the inver-sion trade-offs are to be overcome.

GEODYNAMICAL PAPERS

From a geodynamical vantage point, the D″ layer has longbeen recognized as the site where instabilities are likely todevelop because of the presence of the thermal boundarylayer right above the CMB [Jones, 1977; Yuen and Peltier,1980; Loper and Stacey, 1983]. The dynamical implicationsof the post-perovskite interpretation of the D″ layer having asteep positive Clapeyron slope are quite profound for thedynamics of the deep mantle, since the rheology of post-per-ovskite may also be non-Newtonian and lower than the adja-cent perovskite because of the large-stresses present at theboundary layers of mantle convection.

HIROSE ET AL. 5

GM01018_CH02.qxd 20/10/07 1:46 PM Page 5

Peltier gives an historical account of the dynamics of theD″ layer from a purely thermal perspective. He reminds thereader that the interpretation of the D″ layer in terms ofchemical heterogeneity [e.g., Trampert et al., 2004] was orig-inally invoked to explain properties of this layer that may beadequately explained now by the post-perovskite phase tran-sition. This motivates a review of the plausibility of the endmember of chemically homogeneous models of mantle con-vection. Peltier stresses that the post-perovskite transitionmay force geophysicists to reconsider the idea that the chem-ical heterogeneity that is associated with D″ may be entirelyderived from the core rather than from slabs piling up at theCMB. Such a scenario may fit very nicely with the idea ofinfiltration of iron into the post-perovskite phase advocatedin Petford et al. The relative importance between thermal andchemical buoyant forces in the deep mantle relies criticallyon the knowledge of equation of state of subducted oceanicbasalt and pyrolite for different compositions of iron. The dif-ferences in the bulk moduli with depth [Tan and Gurnis,2005] being traded off with the reduction of the coefficient ofthermal expansion with depth must be examined quantita-tively by detailed equation of state calculations and not byextrapolations of thermodynamic data.

Tackley, Nakagawa, and Hernlund examine the influenceof post-perovskite transition on both thermal and thermal-chemical mantle convection. They study the dynamicaleffects arising from the complex interplay of variations intemperature, composition and the post-perovskite transition.Figure 1 shows the situation that arises for two Clapeyronslopes that depend on composition and how they would becrossed by the temperature curve labeled T associated with athermal boundary layer above the CMB. Because of uncer-tainties of the physical parameters, such as depth variationsof thermal expansivity and the differences in the bulk moduliwith depth, it is extremely difficult to ascertain the relativedynamical importance of each of these factors. Tackley et al.’scalculations reveal that the lateral variations in the occur-rence of post-perovskite contribute the most to the longwavelength lateral shear-wave anomalies in the deepest por-tion of the mantle. With a post-perovskite transition depen-dent on composition, a great variety of complex behaviormay ensue, producing structures such as multiple crossingsof the temperature curve by the two different types of post-perovskite transition (see Figure 1, where t1, t2, b1 andb2 are the four crossings due to compositionally dependentphase boundaries).


Figure 1. Schematic diagram showing the temperature curve and the phase boundaries in the D″ layer. The temperature T curverepresents the temperature profile as it approaches the CMB. Two phase boundaries depending on composition have been includedand the Clapeyron slopes are given by γ, and the temperature-intercept at the CMB by T-int. The temperature of the CMB is desig-nated by T-CMB. The levels t1, t2, b1, and b2 are the depths where the temperature curve is intersected by the two phase boundaries.The thermal expansivity α is dependent on P, the pressure. The thermal conductivity k depends on temperature, pressure and iron con-tent Fe. The non-Newtonian viscosity of post-perovskite depends on the stress τ, temperature, pressure and Fe content.

GM01018_CH02.qxd 20/10/07 1:46 PM Page 6

Yuen, Matyska, Cadek, and Kameyama focus on thedynamical effects from the physical properties in the lowermantle on the post-perovskite transition within the frameworkof thermal convection in Cartesian geometry. They investigatethe influences on lower mantle plume dynamics of stronglydepth-dependent coefficient of thermal expansion and radia-tive thermal conductivity (see Figure 1) acting in concert withthe post-perovskite transition. Double-crossing of the post-perovskite boundary only takes place when the CMB temper-ature is higher than the temperature intercept of the phasechange T-int (see Figure 1). Both radiative thermal conduc-tivity and strongly decreasing thermal expansivity conspire toinduce partially layered convection with slabs stagnating inthe transition zone and to develop multiple scale mantleplumes, with superplumes in the lower mantle and smallerscale secondary plumes emerging from 670 km depth. Fromthe same thermal expansivity, they deduce the 3-D density anomalies from the seismic velocity anomaliesinferred from seismic tomographic inversion. They thendeduce the lateral viscosity variations above the CMB bysolving the inverse problem dealing with the long-wavelengthgeoid anomalies computed for viscous responses to the man-tle flows excited by the 3-D density heterogeneities. They findthat the region underneath hot spots has significantly higherviscosity in the lower mantle than the region below subduc-tion zones. They suggest that the bottom portions of lowermantle perovskite superplumes are stiffer than the adjacentpost-perovskite deep mantle and the fixity of these plumes isdue to the constraints imposed by the surrounding horizontalflow of post-perovskite with cold downwelling origins.

Petford, Rushmer, and Yuen consider the material transferof iron into the D″ layer from the core, as proposed in Peltier.They view this phenomenon as a multiscale problem, bothspatially and temporally. On the microscale they used pureand simple shear deformation mechanisms which producetransient pressure gradients to drive local fluid flow. Theyemphasize that the mesoscale non-Newtonian flow associ-ated with the D″ boundary layer is a possible trigger of small-scale convection within this sub-layer, while the macroscaleflow comes from lower-mantle circulation. They also discussin detail the microscale physical processes and the couplingto geochemistry. Ideas on core-mantle interaction and meltmigration under large stresses are drawn from experimentaldeformation studies under moderate temperature and high

strain-rate conditions. The rheology of post-perovskite is animportant ingredient for this filtration process to work effi-ciently.

CONCLUSION

Any compilation of results such as this book provides onlya snap-shot of knowledge at a given time, and there will besteady advances in our understanding of post-perovskiteproperties and occurrence in the Earth. But all scientific rev-olutions have an initial phase of dramatic changes, followedby long-term adjustments. This book documents the remark-able discoveries and advances of the first three years of thepost-perovskite revolution.

REFERENCES

Badro, J., G. Fiquet, F. Guyot, J.P. Rueff, V.V. Struzhkin, G. Vanko, and G.Monaco, Iron partitioning in Earth’s mantle: toward a deep lower mantlediscontinuity, Science, 300, 789-791, 2003.

Jones, G.M., Thermal interactions of the core and mantle and long-termbehavior of the geomagnetic field, J. Geophys. Res., 82, 1703-1708, 1977.

Lay, T., and E.J. Garnero, Core-mantle boundary structures and processes.In: Sparks, R.S.J., and C.J. Hawkesworth (eds.), The State of the Planet:Frontiers and Challenges in Geophysics. Geophysical Monograph Series,150, IUGG Volume 19, pp. 25-41, 2004.

Lay, T., and D.V. Helmberger, A lower mantle S-wave triplication and thevelocity structure of D″, Geophys. J. R. Astron. Soc., 75, 799-837, 1983.

Liu, L., Silicate perovskite from phase transformation of pyrope-garnet athigh pressure and temperature, Geophys. Res. Lett., 1, 277-280, 1974.

Loper, D.E., and F.D. Stacey, The dynamical and thermal structure of deepmantle plumes, Phys. Earth Planet. Inter., 33, 304-317, 1983.

Murakami, M., K. Hirose, K. Kawamura, N. Sata, and Y. Ohishi, Post-per-ovskite phase transition in MgSiO3, Science, 304, 855-858, 2004.

Ringwood, A.E., Olivine-spinel transition in fayalite, Bull. Geol. Soc. Am.,69, 129-130, 1958.

Sidorin, I., M. Gurnis, and D.V. Helmberger, Dynamics of a phase change atthe base of the mantle consistent with seismological observations, J.Geophys. Res., 104, 15,005-15,023, 1999.

Tan, E., and M. Gurnis, Metastable superplumes and mantle compressibility,Geophys. Res. Lett., 32, L20307, doi:10.1029/2005GL024190, 2005.

Trampert, J., F. Deschamps, J. Resovsky, and D.A. Yuen, Probabilistictomography maps chemical heterogeneities throughout the lower mantle,Science, 306, 853-856, 2004.

Wright, C., and J.A. Lyons, Seismology, dT/d∆ and deep mantle convection,Geophys. J. R. Astron. Soc., 40, 115-138, 1975.

Wysession, M.E., T. Lay, J. Revenaugh, Q. Williams, E. Garnero, R. Jeanloz,and L. Kellogg, The D″ discontinuity and its implications, in The Core-Mantle Boundary Region, edited by M. Gurnis, M.E. Wysession, E.Knittle, and B.A. Buffet, pp. 237-297, AGU, Washington, D.C., 1998.

Yuen, D.A., and W.R. Peltier, Mantle plumes and the thermal stability of theD″ layer, Geophys. Res. Lett., 7, 625-628, 1980.

HIROSE ET AL. 7

GM01018_CH02.qxd 20/10/07 1:46 PM Page 7

INTRODUCTION

The post-perovskite transition was a completely unex-pected finding for Earth scientists. Although numerousexperimental studies under high pressure have been madesince the 1960’s to understand the Earth’s deep interior, sci-entists had never thought of the possibility of a post-per-ovskite phase because the structure of perovskite is soefficiently packed that perovskite was believed to be the ulti-mate dense form of silicate in the Earth’s mantle. For betterunderstanding of the Earth’s interior, scientists made variousefforts after seismologists clarified the layered structure ofthe Earth. If we look back at the history of this kind of study,it is clear that the development of high-pressure experimen-tal techniques played an essential role in the advancement ofour knowledge. Sometimes studies on analog material pro-vided quite useful insights but they never led to any unex-pected discoveries. And once a new discovery was made,numerous studies followed to elucidate the properties of the

newly found high-pressure mineral. This happened when sil-icate perovskite was found in the 1970’s and a similar trendis now being observed in the post-perovskite phase.

The purpose of this paper is to briefly review the progressof these experimental studies to clarify various phase trans-formations in the mantle from the 1960’s up until the recentdiscovery of “The last phase transition”, together with theprogress of these experimental techniques and theoreticalsimulations. Because of the limitation of the space, thisreview covers only the limited part of the history.

SEISMIC DISCONTINUITIES IN THE MANTLE

Seismologists clarified the distribution of seismic wavevelocity within the Earth as early as the 1930’s. Most ofthese analyses were made assuming the spherical symmetryof the Earth, until seismic tomography became popular inthe 1990’s. In other words, various properties such as Vp, Vs,density, and so on, were determined as a function of depthalone. In the early days, resolution of these analyses was notso high and many discontinuities we now know were simplyrecognized as a region of rapid change. Discontinuity at 660 km depth was recognized clearly only in the late 1960’s[Johnson, 1967, Kanamori, 1967] when the high-resolution

Review of Experimental Studies on Mantle Phase Transitions

Takehiko Yagi

Institute for Solid State Physics, University of Tokyo, Tokyo, Japan

Progress in experimental studies to clarify the nature of the discontinuities in theEarth’s mantle is reviewed. Developments of high-pressure and high-temperatureexperimental techniques played an essential role to extend studies to deeper partsof the mantle. Studies on analog materials also played an important role, particu-larly to clarify the nature of 410 km and 660 km discontinuities, but the experi-ments on the real material lead to the most unexpected discoveries and conclusiveresults. Findings of the modified-spinel phase and the post-perovskite phase aregood examples. Developments of high-pressure and high-temperature in situ X-ray diffraction using a laser-heated diamond anvil combined with synchrotronradiation dramatically extended the pressure range of the experiments and led tothe discovery of the post-perovskite phase at a condition corresponding to the bot-tom of the mantle. For findings of the post-perovskite phase, various computersimulations also played a very important role once experimental evidence of thephase transition had been obtained.

9


GM01018_CH03.qxd 20/10/07 1:46 PM Page 9

analysis became possible. Birch [1952] made a detailed dis-cussion of the constitution of the Earth based on seismicstructure observed by Jeffreys [1939] and Gutenberg [1951],and Bullen’s Earth model [Bullen, 1947]. In this model(Figure 1), the Earth’s interior was divided into seven layers,and the layers from B, C, and D corresponded to the mantle.Birch [1952] made an argument that both the B and D layers,which correspond to the upper and lower mantle, respec-tively, can be understood as layers for simple compression ofhomogeneous material. On the other hand, he pointed out thatchanges in the C layer, which is now known as the transitionzone including both 410 km and 660 km discontinuities, aretoo rapid to be explained by the simple compaction of theconstitutive materials. He also pointed out that the elasticproperty of the D layer is difficult to explain by a normal sil-icate such as olivine and pyroxene and is better explained bysimple oxides such as corundum and rutile. Among variouspossibilities such as the change in chemistry and phase trans-formations of the constitutive material, he introduced thepossibility of an olivine-spinel transition, as will be described

in detail in the next section. Although Birch concluded thatthe D layer is a homogeneous layer, it is interesting to notethat we can clearly see, in the figure of his paper (Figure 1),a small region of anomalous change of velocity at the bottom,which is now known as the D″ layer. Compared to other lay-ers, the anomalous property of this layer has strong regionalvariation and detailed studies became possible only in recentyears. This D″ layer has many strange properties such asstrong elastic anisotropy, anti correlation of Vs and Vø, andthe existence of an ultra low-velocity zone. These propertieswere very difficult to understand until the findings of thepost-perovskite transition in 2004.

NATURE OF THE 410 KM DISCONTINUITY

A possibility to explain the rapid change of seismic veloc-ity observed in the transition zone by the phase transition ofcomponent olivine into denser spinel polymorph was pointedout as early as the 1930’s by Goldschmidt [1931], Bernal[1936], and Jeffreys [1936]. Direct experimental studies,

10 REVIEW OF EXPERIMENTAL STUDIES ON MANTLE PHASE TRANSITIONS

Figure 1. Seismic velocity within the Earth plotted by Birch [1952] using data by Jeffreys [1939] and Gutenberg [1951], and the lay-ered structure of the Earth proposed by Bullen [1947][after Birch, 1952].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 10

called “modified-spinel structure” in the Mg-rich part of thephase diagram. Ringwood also noticed the existence of acomplicated X-ray pattern from samples enriched inMg2SiO4 but the exact nature of this material remainedunclear. Through a detailed study in the join of Mg2SiO4-Co2SiO4, Akimoto and Sato [1968] found a new phase whichgave identical X-ray pattern to that recognized by Ringwood.Its crystal structure was successfully analyzed using aquenched and recovered single crystal of Mn2GeO4, whichwas also found to transform into the same structure aboveabout 4 GPa [Morimoto et al., 1969]. This structure is quiteunique in the sense that no material with this structure waspreviously known at all. The structure could be understood asa slight modification of spinel structure and was subse-quently named as the “modified spinel” structure. Later, amineral having this structure in the composition of(Mg,Fe)2SiO4 was found in shocked meteorite and wasnamed as wadsleyite [Price et al., 1983], while the spinelpolymorph which was also found in shocked meteorite wasnamed ringwoodite [Binns et al., 1969].

Extension of the high-pressure experiments beyond 10 GPawas not so easy but by adopting a newly developed internallyheated Bridgman-anvil apparatus [Nishikawa and Akimoto,1971], Akimoto successfully extended the phase diagram ofMg2SiO4-Fe2SiO4 join to the Mg2SiO4 endmember [Akomoto1972], as shown in Figure 3b. Because of the existence of themodified spinel structure, the phase relation in the system ofMg2SiO4-Fe2SiO4 became much more complicated and the410 km discontinuity was explained by the transition of(Mg0.9Fe0.1)2SiO4 from olivine to modified spinel structure.The transition from modified spinel to spinel was expected tooccur at a deeper part of the mantle and because of the diffi-culty of pressure calibration in this range, Akimoto originallybelieved that this transition would occur close to the 650 kmdiscontinuity [Akomoto et al., 1976]. By that time some othergroups had also started studying the phase relation of olivinesolid solution (Mg2SiO4-Fe2SiO4 system) using various dif-ferent high-pressure apparatus. Kawai and his colleaguesstarted research using a split-sphere apparatus [Kawai andEndo, 1970] which was further developed to a 6-8 type dou-ble stage multi-anvil apparatus (now called as “Kawai-typeapparatus”). Kumazawa and his colleagues developed a multi-anvil sliding system (“MASS” apparatus) and made variousstudies on silicates [Kumazawa et al., 1976]. Through thesestudies they clarified that the pressure values in Akimoto’spaper were overestimated above about 15 GPa. Moreover, theelastic property of modified-spinel phase was clarified to besimilar to that of the spinel phase. This led to the conclusionthat the transition between these two phases is unlikely to givea sharp discontinuity and would occur in the region around adepth of 500 km, where the slope of the velocity and densityare steeper compared to other regions.

YAGI 11

Figure 2. Tetrahedral press at the Institute for Solid State Physics,University of Tokyo, installed in 1963. A tetrahedral shaped pres-sure-transmitting medium was compressed by four independentrams and pressures up to about 10 GPa was achieved. This appara-tus was originally developed by T. Hall [1958] and played essentialrole in Akimoto’s group to clarify the olivine-spinel transitions.

however, only started in the 1960’s, when high pressure andtemperature experimental techniques in the laboratorybecame high enough to cover the P-T conditions correspond-ing to this depth. Pioneering works were made by two groups,A. E. Ringwood of the Australian National University and S.Akimoto of the Institute for Solid State Physics, Universityof Tokyo. Ringwood [1958] succeeded in transformingFe2SiO4 olivine into the spinel structure at 5 to 6 GPa and at400°C using an externally heated simple squeezer apparatus.He also proved that other silicate olivines like Ni2SiO4 andCo2SiO4 also transform into spinel polymorphs [Ringwood,1962, 1963] and tried to clarify the nature of the transition ofMg2SiO4 by studying various solid solutions containing it.He also used various germanates to estimate the behavior ofMg2SiO4. Precise and systematic studies of the join ofFe2SiO4-Mg2SiO4 were made by Akimoto and his colleagues.The first paper of his group on the experimental work of thissystem, which was obtained by a newly constructed tetrahe-dral press (Figure 2), appeared in 1966 [Akimoto andFujisawa, 1966]. At that time the highest pressure was lim-ited to below 9 GPa and the olivine-spinel transition wasobserved only in the compositional range from pure Fe2SiO4

to (Mg0.6Fe0.4)2SiO4, as shown in Figure 3a. In this composi-tional range, the obtained result was that of a typical binarysystem and he concluded that the 410 km discontinuity couldbe well explained by the olivine-spinel transition of(Mg0.9Fe0.1)2SiO4 by extrapolating experimental results usingthermodynamic considerations. This conclusion was laterchanged, however, by the findings of a new phase, the so

GM01018_CH03.qxd 20/10/07 1:46 PM Page 11

By that time the depth resolution of the seismic modelwithin the mantle became high enough to clearly show theexistence of two distinct discontinuities, 410 km and 660 km,in the transition zone. Since the nature of the 410 km discon-tinuity was successfully explained by the olivine to modifiedspinel transition, scientists started to work on the next target,the 660 km discontinuity, and tried to find out the “post-spinel transition”.

NATURE OF THE 660 KM DISCONTINUITY

The pressure that corresponds to the 660 km discontinuityis about 24 GPa and it was beyond the capability of the high-pressure apparatus available in early 1970’s. Scientists stud-ied various analog materials and proposed many possibilitiesfor the post-spinel transition of Mg2SiO4. These can be clas-sified into three groups as follows [e.g. Ringwood 1975];

1. Transition into a single dense phase with A2BO4 compo-sition such as Sr2PbO4 and K2NiF4 structures, e.g.Mn2GeO4, Ca2GeO4;

2. Decomposition into a dense ABO3-type structure such asilmenite, perovskite, or corundum structure plus rock-salttype AO compounds, e.g. Mg2TiO4, Fe2TiO4, FeAl2O4;

3. Complete decomposition into simple oxides, 2AO + BO2

with rock salt and rutile structures, respectively, e.g.Fe2SiO4, Co2SiO4, Ni2SiO4

There were some reports that Mg2SiO4 spinel decomposedinto MgO plus SiO2 under high pressure but this was notproved. Meanwhile, Ming and Bassett [1974] developed alaser-heating technique for diamond-anvil experiments. At thattime the diamond-anvil experiments were performed by com-pressing powdered samples directly without using a gasket andthe pressure was limited to below 30 GPa. A considerable pres-sure and temperature gradient existed within the sample andquantitative studies were difficult. Nevertheless, it became apowerful tool to make exploratory study of the phase transi-tions and L.G. Liu of the Australian National University madenumerous experiments to clarify the pressure induced phasetransitions in silicates. In 1974, he first reported that silicateperovskite formed from natural garnet [Liu, 1974; Figure 4].Following this work, he clarified that silicate perovskitecould form from all the major upper mantle minerals, olivine,pyroxene, and garnet [Liu, 1975], which led to the definiteconclusion that the nature of the 660 km discontinuity is thephase transition of the component silicates into the assem-blage mainly composed of perovskite-structured phases.


Figure 3. Two phase diagrams of the Mg2SiO4-Fe2SiO2 solid solution system published in two different years. Progress caused by theextension of the pressure range can be seen. (a) Akimoto and Fujisawa [1966]; (b) Akimoto [1972].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 12

Since the volume of the lower mantle exceeds 50% of theentire Earth, silicate perovskite became the most abundantmineral in our planet, although we cannot find it at all at thesurface of the Earth.

PROPERTY OF SILICATE PEROVSKITE

Following the discovery of silicate perovskite, numerousexperimental studies began to elucidate its property, becauseof its importance to understand the lower mantle. Liu con-

cluded that silicate perovskite was formed based on the sim-ilarity of the powder X-ray diffraction pattern (Figure 4) withknown perovskites, although he made no structural analysis.The first structural analysis of MgSiO3 perovskite wasreported four years after its discovery using two differenttechniques [Yagi et al., 1978; Ito and Matsui, 1978]. Yagi et al. synthesized the perovskite sample using a gasketed dia-mond-anvil apparatus combined with laser heating and col-lected the intensity data of X-ray diffraction using a Debyecamera. Ito and Matsui, on the other hand, synthesized a

YAGI 13

Figure 4. Observed intensity vs. inter-planer distance from the first silicate perovskite formed from natural garnet and MgSiO3, andcomparisons with typical compounds with the perovskite structure. MgSiO3 perovskite in this figure was formed from synthetic end-member pyrope. At that time Liu believed that the pyrope was decomposed into perovskite plus corundum and the diffractionscorrespond to the corundum, together with the unreacted pyrope, are eliminated from the top figure[after Liu, 1974].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 13

much larger amount of perovskite sample using a Kawai-typemulti-anvil apparatus and obtained X-ray data using a diffrac-tometor. Despite the difference of both synthesis methods andanalytical techniques, the obtained structure was identical,within experimental error, and it was confirmed that the sili-cate transforms above about 24 GPa into a GdFeO3-typeorthorhombic perovskite structure with the Pbnm space group.

Intensive studies were made to make clear many otherproperties such as the transition pressure, compression curve,solubility of iron, and so on. The first measurement of thebulk modulus was made by high pressure X-ray diffractionusing a system composed of a conventional X-ray tube andMao-Bell type diamond anvil cell (Figure 5) below 10 GPaadopting a methanol-ethanol mixture as a hydrostatic pres-sure-transmitting medium [Yagi et al., 1978, 1982]. They

clarified that the bulk modulus was about 260 GPa. Sincethen numerous studies were made repeatedly to elucidate thedetails of the elastic property by adopting various and moreadvanced experimental techniques including synchrotron X-ray radiation and Brillouin scattering, but the obtained bulkmodulus basically remained unchanged. When the spinel-type (Mg,Fe)2SiO4 transforms into a post-spinel phase, itdecomposes into (Mg,Fe)SiO3 with a perovskite structureplus (Mg,Fe)O with a rock salt structure. One of the moststriking features of this decomposition was that iron prefer-entially distributes into magnesiowustite, which was firstreported by Bell et al. [1979]. This affected considerably thevarious arguments about the lower mantle, and numerousstudies were later made to clarify the partition coefficient bychanging various factors. Through these studies it becameclear that many factors such as oxygen fugacity and the coex-istence of trivalent cation considerably affected this parti-tioning and that the situation in the Earth’s interior seemed tobe complicated [e.g. Mao et al., 1997].

The stable structure of silicate perovskite in the lower man-tle was an important issue from the beginning. It was knownthat many compounds with a perovskite structure changetheir symmetry with temperature. In many materials, thecubic phase is stable at high temperature and with decreasingtemperature, it distorts into a tetragonal and then to anorthorhombic symmetry. In MgSiO3 perovskite, which wasfound to have orthorhombic symmetry in the recovered sam-ple, the possibility of a similar change in the mantle waspointed out but at ambient pressure the quenched samplebecome amorphous on heating and it was impossible to knowthe stable structure at above several hundred degrees. Usinga newly developed Drickamer-type apparatus for high-tem-perature experiments combined with synchrotron radiation(Figure 6), Funamori and Yagi [1993] succeeded in makingan in situ observation of the crystal structure under high-pressure and high-temperature conditions corresponding tothat of the lower mantle and confirmed that the orthorhombicform is stable in the Earth, at least at the uppermost part ofthe lower mantle. They also clarified an important elasticparameter, thermal expansion. This work was followed byvarious precise equations of state studies using a multi-anvilapparatus [Wang et al., 1994; Utsumi et al., 1995; Funamoriet al., 1996].

High pressure stability limit of silicate perovskite was asubject of great interest. Many of the studies to elucidate theprecise property of silicate perovskite were made using amulti-anvil apparatus because they have the advantage of amuch larger sample volume and a much better temperaturecontrol compared to the diamond-anvil apparatus. For exam-ple, Ohtani [1983] studied the melting temperature ofMgSiO3 in the lower mantle, Ito and Takahashi[1989] com-pleted reliable phase relations of the (Mg,Fe)2SiO4 system,


Figure 5. High pressure powder X-ray diffraction system atGeophysical Laboratory, Carnegie Institution of Washington and A.Van Valkenburg, one of the inventors of diamond anvil cell.Monochromatic X-ray from the generator was irradiated to the celland very thin X-ray beam passed through the pinhole collimator inthe cell reached the sample. Diffracted X-ray was recorded on a filmin a cassette which was mounted on the cell. Geiger counter wasused to align the system. More than 300 hours exposure time wasrequired for each measurement.

GM01018_CH03.qxd 20/10/07 1:46 PM Page 14

and Irifune [1994] clearly showed that most of the aluminumdissolves into perovskite structure in the lower mantle. Allthese studies were completed using above mentioned advan-tage of Kawai-type multi-anvil apparatus.

However, experimental studies to clarify the stability limitof perovskite were made possible only by using a diamond-anvil apparatus. Because, by the improvements of variousexperimental techniques, Mao and Bell [1978] has first suc-ceeded in generating pressures above 100 GPa and pressurerange of many experiments were extended beyond 100 GPain the 1980’s. Knittle and Jeanloz [1987] made high-pres-sure in situ X-ray studies for olivine samples heated at 127GPa and concluded that “silicate perovskite is stablethroughout the lower mantle”. Kesson et al. [1998] performed

a transmission electron microscopy (TEM) study of the sam-ple recovered from 135 GPa and concluded that “Mg per-ovskite was found to be present and no additional phases ordisproportionations were encountered”. As is clear fromthese statements, it was widely believed until 2004 that per-ovskite is stable in the lower mantle all the way down to thecore-mantle boundary.

CLOSE PACKED STRUCTURES OF OXIDES IN THE DEEP MANTLE

Minerals take a structure which has the lowest free energyat a given condition and the free energy can be expressed bythe following equation:

G = U + PV − TS

where U is the internal energy, and V and S are the volumeand entropy, respectively. In the deep mantle, pressureincreases rapidly and becomes more than 105 times largerthan that of the Earth’s surface while the temperatureincreases only about 10 times. As a result, among variousterms in the right side of the above equation, the contributionfrom the term PV becomes dominant in the deep mantle anda phase with a smaller volume becomes favorable. Most ofthe dense silicates in the upper mantle, such as spinel andgarnet, are made of the close packing of oxygen ions. Cationsare distributed among the spaces of the close-packed networkof oxygen. This is because, as clearly shown in Table 1, thesize of the oxygen ion is much larger than that of cations andthe close packing of oxygen alone is sufficient to achievesmaller volume. When the pressure is increased further, largeoxygen ions are compressed more than the small cations andthe difference in size is reduced. Under these conditions,close packing of oxygen ions alone becomes insufficient anda new structure is required.

Figure 7 shows one of the cross-sections of perovskitestructure with a composition of ABO3. It can be seen fromthis figure that the structure of perovskite is formed by a fccclose-packing of both oxygen ions and large “A” cations.Small “B” cations are distributed among them. Because ofthis structure, perovskite has an unusually high efficiency of

YAGI 15

Figure 6. Drickamer-type high pressure and high temperature appa-ratus used to study the stable structure of MgSiO3 perovskite in thelower mantle [Funamori and Yagi, 1993]. (a) is a whole view of theapparatus and (b) is an enlarged view of the sample assemblysqueezed between two anvils. Sintered diamond was adopted foranvil material and temperatures above 1000°C was achieved at 30GPa. Combining with a synchrotron radiation, in situ X-ray obser-vations under the condition corresponding to that of the lower man-tle were made possible.

Table 1. Ionic radii and volumes of typical ions forming silicates.

Ion#1 Ionic radius(pm) #2 Volume ratio(O2− =1.00)

O2− 140 1.00Si4+ 40 0.02Mg2+ 72 0.14Fe2+ 78 0.17

#1 coordination number = 6.#2 after Shannon and Prewitt [1969].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 15

packing among various ABO3-type structures, particularlywhen the size of the A cation and the oxygen anion are closeto each other. As a result, many ABO3-type compoundsfinally transform into the perovskite structure at high pres-sures. However, no further pressure-induced phase transfor-mation from perovskite was known to exist and no one couldthink of a structure which had a higher efficiency of packingthan perovskite in ABO3-type compounds.

DISCOVERY OF THE POST PEROVSKITE PHASE

As described in previous sections, experimental results onMgSiO3 perovskite together with the consideration of theclose-packing nature of perovskite structure led to the con-clusion that perovskite is an ultimate stable form of silicatesin the Earth’s mantle, until the findings of the “post per-ovskite phase” in 2004. The silicate perovskites, except forCaSiO3, are quenchable to ambient condition, but it wasknown that the necessity of high pressure (and high temper-ature) in situ observations became more and more importantfor the study of various phase transformations in the lowermantle.

High-pressure and high-temperature in situ X-ray observa-tions under deep lower-mantle conditions became possible bythe use of a laser-heated diamond anvil apparatus combinedwith synchrotron radiation. Various systems were developedand constructed at the synchrotron radiation facilities aroundthe world. A pioneering system was constructed by Boehler etal. [1990] at DESY in Germany using CO2 laser heating, fol-lowed by the systems at ESRF (France), APS (USA), andPhoton Factory (Japan). The system developed at the PhotonFactory was also constructed at SPring-8 [Watanuki, et al.,2001; Yagi et al., 2001], the third generation synchrotron sourcein Japan. Using these systems high quality X-ray diffraction

patterns were obtained in less than 10 minutes, compared with 300 hours required by the system in 1970’s, even above100 GPa. However, no one had seriously pursued the possi-bility to clarify further transitions in MgSiO3 perovskite.

In 2002, K. Hirose of Tokyo Institute of Technology and hiscolleagues started a systematic study of the lower mantlemineralogy using a laser-heated diamond anvil system atSPring-8, which was further improved by that time. Throughthe study of pyrolite composition, they found new unidenti-fied diffraction lines above about 120 GPa. They had no ideawhat component caused these new lines and studied simplecomponents separately. Through these experiments theyfound that the diffraction pattern of MgSiO3 perovskitechanges completely between 110 and 125 GPa. Still, theywere not sure if this change was caused by the equilibriumphase transformation of perovskite into a new phase. Therewas no idea about the new structure which had higher densitythan perovskite. From the powder diffraction pattern alone, itwas almost impossible to solve the structure and theoreticalcalculations played a very important role to solve this problem.

In spite of the extreme conditions of the experiment, thequality of the X-ray pattern they obtained was very high andthey succeeded in determining the unite cell parameter whichsatisfies the obtained X-ray data unambiguously. Then theyperformed molecular dynamics calculations using this unitcell and tried to find the structure which most fitted theobserved X-ray diffraction pattern. Through these analyses,they succeeded in clarifying that this new phase had a struc-ture with a Cmcm space group and was isostructural withUFeS3 [Murakami et al. 2004]. To our surprise, the newstructure had a layered structure made up of a stacking of twokinds of layers; one layer formed by a two dimensional net-work of SiO6 octahedrons and the other formed by Mgcations (Figure 8). This structure is quite different from thestructure of perovskite which is formed by a three dimen-sional network of SiO6 octahedrons connected by cornershearing. As a result, a strong anisotropic property wasexpected in this new phase.

The same structure was also obtained by other theoreticalsimulations and experiments. Oganov and Ono [2004]reported that MgSiO3 transforms into the same structure atabout 118 GPa, which was predicted by their first principlescalculations based on the observation of the transition inFe2O3. Tsuchiya et al. [2004] made a first principles calcula-tion of the structure using the unit cell parameter observed byMurakami et al. and concluded that the same structure is sta-bilized. Iitaka et al. [2004] also performed a first principlescalculation and confirmed the stability of this new phase rel-ative to the perovskite phase above about 100 GPa. They alsoclarified the elastic property of this new phase and found thatthis phase has a strong elastic anisotropy, which has greatimportance to explain various unsolved problems about the


Figure 7. Structure of the ABO3-type silicate perovskite [modifiedfrom Yagi et al., 1978].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 16

D″ layer. Although the methods of these theoretical calcula-tions are quite different from each other, they all resulted inthe same structure.

Since then, an explosion of new studies to clarify variousproperties of the post-perovskite phase started, just as hap-pened after the findings of silicate perovskite in the 1970’s, butat a much faster rate. It is interesting to note that through thesestudies, it became clear that MgGeO3, a classical germanateanalog material of MgSiO3, also transforms from perovskiteinto the same structure at a much lower pressure of around 60GPa [Hirose et al., 2005]. Experiments in this pressure rangeare much easier and scientists could have recognized the pos-sibility of the post-perovskite transition much earlier hadsomeone studied it seriously. The reality, however, was that noone had studied it seriously and the post-perovskite phase wasfound directly in the real material, rather than in the modelmaterial. It is clear that this direct finding was made possibleby the large progress of the high-pressure and high-tempera-ture experimental techniques in the last decade.

CONCLUSIONS

The progress of high-pressure and high-temperature exper-imental studies to elucidate the nature of the discontinuitiesin the Earth’s mantle has been reviewed. It is clear that thedevelopment of these experimental techniques played anessential role in the progress of these findings. Theoreticalconsiderations and studies of the analog materials also playedan important role in this progress but experiments on the realmaterial provided the most unexpected discoveries. Findingsof the modified spinel phase and the post-perovskite phaseare good examples. For the study of the post-perovskitephase, various computer simulations also played a veryimportant role, once experimental evidence of the transitionhad been given. Theoretical calculation is a field of rapidprogress these days and it will become a more and more pow-erful tool to study the Earth’s deep interior.

Acknowledgments. The author is grateful for constructive com-ments of the two anonymous reviews and an editor that contributeda lot to improve the original manuscript.

REFERENCES

Akimoto, S., and H. Fujisawa, Olivine-spinel transition in the systemMg2SiO4-Fe2SiO4, at 800°C, Earth Planet. Sci. Lett., 1, 237-240, 1966.

Akimoto, S., The system MgO-FeO-SiO2 at high pressures and temperatures-phase equilibria and elastic properties, in A.R. Ritsema (Editor), TheUpper Mantle, tectonophysics, 13(1-4), pp. 161-187, 1972.

Akimoto, S., Y. Matsui, and Y. Syono, High-pressure crystal chemistry oforthosilicates and the formation of the mantle transition zone, in R.G.J.Strens ed. “The Physics and Chemistry of Minerals and Rocks”, J. Wiley,pp. 327-363, 1976.

Akimoto, S., and Y. Sato, High-pressure transformation in Co2SiO4 olivineand some geophysical implications, Phys. Earth Planet. Interiors, 1, 498-504, 1968.

Bell, P.M., T. Yagi, and H.K. Mao, Iron-magnetism distribution coefficientsbetween spinel [(Mg,Fe)2SiO4], magnesiowüstite [(Mg,Fe)O], and per-ovskite [(Mg,Fe)SiO3],Carnegie Inst. Washington Yearb. 78, 618, 1979.

Bernal, J.D., Discussion, Observatory 59, 268, 1936.Binns, R.A., R.J. Davis, and S.J.B. Reed, Ringwoodite, natural (Mg,Fe) 2SiO4

spinel in the Tenham meteorite, Nature, 221, 943-944, 1969.Birch, F., Elasticity and constitution of the Earth’s interior, J. Geophys. Res.,

57, 227-286, 1952.Boehler, R., V. von Bargen, and A. Chopelas, Melting, thermal expansion,

and phase transitions of iron at high pressures, J. Geophys. Res., 95,21731-21736, 1990.

Bullen, K.E., An introduction to the theory of seismology, CambridgeUniversity Press, p. 276, 1947.

Funamori, N., and T. Yagi, High pressure and high temperature in situ X-rayobservation of MgSiO3 perovskite under lower mantle condition,Geophys. Res. Lett., 20, 387-390, 1993.

Funamori, N., T. Yagi, W. Utsumi, T. Kondo, T. Uchida, and M. Funamori,Thermoelastic properties of MgSiO3 perovskite determined by in situ Xray observations up to 30 GPa and 2000 K, J. Geophys. Res., 101, 8257-8269, 1996.

Goldschmidt, V.M., Zur Kristallchemie des Germaniums. Nachr. Akad. Wiss.Göttingen Math. Physik. Kl. 1, No. 2, 184-190, 1931.

Gutenberg, B., PKKP, P′P′, and the Earth’s core, Trans. Am. Geophys. Union,32, 373-390, 1951.

Hall, H.T., Some high-pressure, high-temperature apparatus design consid-erations: Equipment for use at 100 000 atmospheres and 3000°C, Rev. Sci.Instrum., 29, 267-275, 1958.

Hirose, K., K. Kawamura, Y. Ohishi, S. Tateno, and N. Sata, Stability andequation of state of MgGeO3 post-perovskite phase, Am. Mineral., 90,262-265, 2005.

YAGI 17

Figure 8. Structure of the post-perovskite phase [after Murakami et al., 2004].

GM01018_CH03.qxd 20/10/07 1:46 PM Page 17

Iitaka, T., K. Hirose, K. Kawamura, and M. Murakami, The elasticity of theMgSiO3 post-perovskite phase in the Earth’s lowermost mantle, Nature430, 442-445, 2004.

Irifune, T. Absence of an aluminous phase in the upper part of the Earth’slower mantle, Nature, 370, 131-133, 1994.

Ito, E., and Y. Matsui, Synthesis and crystal-chemical characterization ofMgSiO3 perovskite, Earth Planet. Sci. Lett., 38, 443-450, 1978.

Ito, E., and E. Takahashi, Post spinel transformations in the systemMg2SiO4-Fe2SiO4 and some geophysical implications, J. Geophys. Res.,94, 10,637-10,646, 1989.

Jeffreys, H., The structure of the Earth down to the 20° discontinuity, Mon.Not. R. Astr. Soc., Geophys. Sup., 3, 401-422, 1936.

Jeffreys, H., The times of P, S, and SKS, and the velocities of P and S, Mon.Not. R. Astr. Soc., Geophys. Sup., 4, 537-547, 1939.

Johnson, L.R., Array measurement of P velocity in the upper mantle, J.Geophys. Res., 72, 6309-6325, 1967.

Kanamori, H., Upper mantle structure from apparent velocity of P waverecorded at Wakayama micro-earthquake observatory, Bull. Earthq. Res.Inst., 45, 657-678, 1967.

Kawai, N., and S. Endo, The generation of ultrahigh hydrostatic pressures bya split sphere apparatus, Rev. Sci. Instrum., 41, 1178-1181, 1970.

Kesson, S.E., J.D. Fitz Gerald, and J.M. Shelley, Mineralogy and dynamicsof a pyrolite lower mantle, Nature, 393, 252-255, 1998.

Knittle, E., and R. Jeanloz, Synthesis and equation of state of (Mg,Fe)SiO3perovskite to over 100 gigapascals, Science, 235, 668-670, 1987.

Kumazawa, M., H. Sawamoto, E. Ohtani, and K. Masaki, Post-spinel phaseof forsterite and evolution of the mantle, Nature, 247, 356, 1976.

Liu, L., Silicate perovskite from phase transformations of pyrope-garnet athigh pressure and temperature, Geophys. Res. Lett., 1, 277-280, 1974.

Liu, L., Post-oxide phases of forsterite and enstatite, Geophys. Res. Lett., 2,417, 1975.

Mao, H.K., and P.M. Bell, High-pressure physics: Sustained static genera-tion of 1.36 to 1.72 megabars, Science, 200, 1145-1147, 1978.

Mao, H.K., G. Shen, R.J. Hemley, Multivariable dependence of Fe-Mg par-titioning in the lower mantle, Science, 278, 2098-2100, 1997.

Ming, L., and W.A. Bassett, Laser heating in the diamond anvil press up to2000°C sustained and 3000°C pulsed at pressures up to 260 kilobars, Rev.Sci. Instrum., 45, 1115-1118, 1974.

Morimoto, N., S. Akimoto, K. Koto, and M. Tokonami, Modified spinel,beta-manganous orthogermanate: stability and crystal structure, Science,165, 586-588.

Murakami, M., K. Hirose, K. Kawamura, N. Sata, and Y. Ohishi, Post-perovskite phase transition in MgSiO3, Science, 304, 855-858, 2004.

Nishikawa, M., and S. Akimoto, Bridgman anvil with an internal heatingsystem for phase transformation studies, High temperatures- High pres-sures, 3, 161-176, 1971.

Ohtani, E, Melting temperature distribution and fractionation in the lowermantle, Phys. Earth Planet. Inter, 33, 12-25, 1983.

Oganov, A.R., and S. Ono, Theoretical and experimental evidence for a post-perovskite phase of MgSiO3 in Earth’s D″ layer, Nature, 430, 445-448,2004.

Price, G.D., A. Putnis, S.O. Agrell, and D.G. W. Smith, Wadsleyite, natural(Mg,Fe) 2SiO4 from the Peace River Meteorite, Can. Mineral., 21, 29-35,1983.

Ringwood, A.E., Olivine-spinel transition in fayalite, Bull. Geol. Soc. Am., 69,129-130, 1958.

Ringwood, A.E., Prediction and confirmation of olivine-spinel transition inNi2SiO4, Geochim. Cosmochim. Acta, 26, 457-469, 1962.

Ringwood, A.E., Olivine-spinel transformation in cobalt orthosilicate,Nature, 198, 79-80, 1963.

Ringwood, A.E., in “Composition and Petrology of the Earth’s Mantle”,McGraw Hill, New York, 1975.

Shannon, R.D., and C.T. Prewitt, Effective ionic radii in oxides and fluo-rides, Acta Cryst., B25, 925-946, 1969.

Tsuchiya, T., J. Tsuchiya, K. Umemoto, R.M. Wentzcovitch, Phase transitionin MgSiO3 perovskite in the earth’s lower mantle, Earth Planet. Sci. Lettl,224, 241-248, 2004.

Utsumi, W., N. Funamori, T. Yagi, E. Ito, T. Kikegawa, and O. Shimomura,Thermal expansivity of MgSiO3 perovskite under high pressures up to 20GPa, Geophys. Res. Lett., 22, 1005-1008, 1995.

Wang Y.B., D.J. Weidner, R.C. Liebermann, Y.S. Zhao, P-V-T equation ofstate of (Mg,Fe)SiO3 perovskite - constraints on composition of the lowermantle, Phys. Earth Planet. Inter., 82, 13-40, 1994.

Watanuki, T., O. Shimomura, T. Yagi, T. Kondo, and M. Isshiki, Constructionof laser-heated diamond anvil cell system for in situ x-ray diffractionstudy at SPring-8, Rev. Sci. Instrum., 72, 1289-1292, 2001.

Yagi, T., H.K. Mao, and P.M. Bell, Structure and crystal chemistry of per-ovskite-type MgSiO3, Phys. Chem. Minerals, 3, 97-110, 1978.

Yagi, T., H.K. Mao, and P.M. Bell, Isothermal compression of perovskite-type MgSiO3, Carnegie Inst. Washington Year Book, 77, 835-837, 1978.

Yagi, T., H.K. Mao, and P.M. Bell, Hydrostatic compression of perovskite-type MgSiO3, in “Advances in Physical Geochemistry”, Vol. 2, S. K.Saxena ed., pp. 317-325, Springer-Verlag, New York, 1982.

Yagi, T., T. Kondo, T. Watanuki, O. Shimomura, and T. Kikegaw, Laser heateddiamond anvil apparatus at the Photon Factory and SPring-8: Problemsand improvements, Rev. Sci. Instrum., 72, 1293-1296.


GM01018_CH03.qxd 20/10/07 1:46 PM Page 18

Discovery of Post-Perovskite Phase Transition and the Nature of D″ Layer

Kei Hirose

Department of Earth and Planetary Sciences, Tokyo Institute of Technology, Tokyo, Japan

MgSiO3 perovskite is a principal mineral in the upper part of the lower mantle,but its stability and possible phase transition at greater depths have long beenuncertain. Recently, a phase transition to post-perovskite was discovered througha significant change in the X-ray diffraction (XRD) pattern at high-pressure andhigh-temperature conditions corresponding to the core-mantle boundary (CMB)region. Experiments on natural pyrolitic mantle and mid-oceanic ridge basalt(MORB) compositions also show that Al-bearing (Mg,Fe)SiO3 post-perovskite isa predominant mineral in the lowermost mantle called D″ layer. Many characteri-stics of the D″ layer, such as D″ seismic discontinuity, S-wave anisotropy, and anti-correlation between the anomalies in S-wave and bulk-sound velocities, may beexplained by this new phase without the need for chemical heterogeneities.However, simply by its location, the D” layer likely has very complex chemicalstructures. Dense subducted MORB crust may have accumulated into chemically-distinct piles underneath upwellings. Partial melting at ultra-low velocity zone(ULVZ) could cause significant chemical differentiation. The bottom of the mantle is likely depleted in iron by the consequence of chemical reaction with theouter core.

19


transition of any specific mantle mineral had not beenidentified until recently at high pressure and temperature (P-T) conditions corresponding to the D″ region.

The recent discovery of MgSiO3 post-perovskite phaseabove 125 GPa and 2500 K has very important implicationsfor the nature and dynamics of this mysterious layer. Sincethe first announcement of this discovery in April 2004[Murakami et al., 2004], rapid developments in experi-mental and theoretical mineral physics, seismology and geo-dynamics have taken place, resulting in a rapid progress inour understanding of the lowermost mantle. The per-ovskite/post-perovskite phase transition boundary has beendetermined in both simplified and natural compositions[Murakami et al., 2005; Ono and Oganov, 2005; Hiroseet al., 2005a; Ohta et al., 2006]. The compositional effectson the stability of post-perovskite have been also extensivelystudied [e.g., Mao et al., 2004, this volume; Akber-Knutsonet al., 2005; Caracas and Cohen, 2005; Tateno et al., 2005,2007]. Theory plays an important role in the determination

1. INTRODUCTION

Large anomalies in seismic wave velocities are observedin the deepest several hundred kilometers of the mantle (theD″ layer) [Lay and Garnero, this volume]. Since the originsof these anomalies were difficult to explain with the knownproperties of MgSiO3 perovskite, a primary mineral in thelower mantle, the D″ layer has long been the most enigmaticregion inside the Earth. The existence of a phase transitionthat could occur in this region has been a subject of debate[e.g., Lay and Helmberger, 1983; Wysession et al., 1998].Sidorin et al. [1999] suggested the presence of a solid-solidphase transition near the base of the mantle in order toexplain the topography of the D″ layer. However, a phase

GM01018_CH04.qxd 20/10/07 1:46 PM Page 19

of the elasticity of the post-perovskite phase at T = 0 K [e.g.,Tsuchiya et al., 2004a; Iitaka et al., 2004; Oganov and Ono,2004] and high temperatures [Stackhouse et al., 2005a;Wookey et al., 2005; Wentzcovitch et al., 2006; Stackhouseand Brodholt, this volume]. In addition, the post-perovskitephase transition is a significantly exothermic reaction with aClapeyron slope of +5 to +11 MPa/K [e.g., Tsuchiya et al.,2004b; Ono and Oganov, 2005; Hirose and Fujita, 2005;Hirose et al., 2006a; Kojitani et al., 2007]. Such an exother-mic reaction near the base of the mantle should stronglyaffect the stability of the bottom thermal boundary layer andthe formation of plumes [Nakagawa and Tackley, 2004;Matyska and Yuen, 2005, 2006].

Many of the long-term enigmas in the lowermost mantlemay be reconciled with this newly discovered post-perovskitephase. They include a sharp seismic discontinuity at the topof D″ (D″ discontinuity), regionally varying S-waveanisotropy, and anti-correlation between the anomalies in S-wave and bulk-sound velocities. However, simply by itsvery location, the D″ layer is a thermal, chemical, andmechanical boundary separating a liquid iron core from asilicate rocky mantle. One can therefore naturally expect verycomplex physical and chemical structures in D″. Thereshould be a steep temperature gradient, which possiblycauses back-transformation from post-perovskite to per-ovskite in the deep D″ (double-crossing model) [Hernlund etal., 2005]. Chemical heterogeneity is a natural consequence,possibly caused by deep subduction of oceanic lithosphere,partial melting corresponding to the ULVZ, and core-mantlechemical reaction. The presence of dense crust subducted inthe early Earth (>4 Ga) is also suggested from Nd isotopeanalyses as a missing enriched reservoir at the CMB [Boyetand Carlson, 2005]. Constraining the amount and nature ofchemical heterogeneity in Earth’s mantle is a key to under-standing the chemical evolution of this planet.

Here I review (1) experimental studies on the post-per-ovskite phase transition and their implications for seismicanomalies in D″, and (2) possible origins and nature ofchemical heterogeneities in D″. The current picture of D″ isdiscussed based on the recent progress of our understandingof this region.

2. EXPERIMENTS AT LOWERMOST MANTLECONDITIONS

2.1. XRD Measurements and TEM Analyses

With recent developments in XRD measurements at syn-chrotron radiation facilities combined with laser-heated dia-mond-anvil cell (LHDAC) techniques, we can determine thecrystal structure and phase relation in-situ at high P-T condi-tions corresponding to the deep Earth (Figure 1) [e.g., Shen

et al., 2001; Watanuki et al., 2001]. The Earth’s CMB islocated at 135 GPa and 2500 to 4000 K. Generating suchhigh P-T conditions are within the capabilities of currentexperimental techniques [Boehler, 1996]. The synchrotronXRD experiments are currently conducted above 300 GPaand 2000 K at BL10XU of SPring-8, although it is more dif-ficult to achieve high temperature at higher pressuresbecause the layers that provide thermal insulation from thediamonds become thinner. The pyrite-type cubic phase ofSiO2 was recently observed above 270 GPa and 1800 K[Kuwayama et al., 2005].

The nanometer-scale characterization of the recoveredsample is also of great importance. The grain size in a samplesynthesized in the LHDAC is typically 100-nm (Figure 2).The transmission electron microscope (TEM) equipped withenergy-dispersive X-ray spectroscopy (EDS) is often used toobtain the chemical compositions of coexisting phases andtheir textural relationships. The crystal chemistry found inpyrolitic mantle and MORB compositions was determined upto the conditions of the lowermost mantle [e.g., Kesson et al.,1994, 1998; Murakami et al., 2005; Hirose et al., 2005a].Core-mantle chemical reaction experiments have also beenconducted using analytical TEM up to 139 GPa [Goarant et al.,1992; Takafuji et al., 2005; Sakai et al., 2006].

2.2. Uncertainty in Pressure Scales

In these high-pressure experimental studies based on in-situ XRD measurements, the sample is usually mixed with Ptor Au powder that serves both as laser absorber and internalpressure standard. Pressure is determined from the unit-cellvolume of the internal pressure standard by applying its P-V(volume)-T equation of state (pressure scale). The apparentpressure error derived from uncertainties in unit-cell volumeand temperature measurements is typically several GPa at120 GPa [e.g., Hirose et al., 2006a]. However, the largestsource of pressure uncertainty is the accuracy of the P-V-Tequation of state of the internal pressure standard.

There has been an extensive debate on the accuracy of thepressure scale [e.g., Hirose et al., 2001; Fei et al., 2004].Irifune et al. [1998] determined the post-spinel phase transi-tion in Mg2SiO4 using a Au pressure scale [Anderson et al.,1989] (Figure 3). They demonstrated that the transition pres-sure is lower by more than 2 GPa than that corresponding tothe depth of 660-km seismic discontinuity. In contrast, morerecent experiments by Fei et al. [2004] showed that the post-spinel phase transformation boundary closely matches thedepth of 660-km boundary when an MgO pressure scale isapplied [Speziale et al., 2001]. The uncertainty derived fromthe pressure scale is more critical at higher pressures, espe-cially over 100 GPa. The simultaneous measurements of thevolumes of Au and Pt, for example, show that the Pt pressure

20 DISCOVERY OF POST-PEROVSKITE PHASE TRANSITION AND THE NATURE OF D″ LAYER

GM01018_CH04.qxd 20/10/07 1:46 PM Page 20

geophysical monograph series gale (eds.) 163 remote sensing of northern hydrology: measuring...

Documents