(this is a sample cover image for this issue. the actual cover is...

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Olivine–wadsleyite–pyroxene topotaxy: Evidence for coherent nucleation anddiffusion-controlled growth at the 410-km discontinuity

Joseph R. Smyth a,⇑, Nobuyoshi Miyajima b, Gary R. Huss c, Eric Hellebrand c, David C. Rubie b,Daniel J. Frost b

a Department of Geological Sciences, University of Colorado, Boulder, CO 80309, USAb Bayerisches Geoinstitut, Universitaet Bayreuth, D-95440 Bayreuth, Germanyc School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, HI 96822, USA

a r t i c l e i n f o

Article history:Received 7 November 2011Received in revised form 26 March 2012Accepted 9 April 2012Available online 18 April 2012Edited by Prof. G. Helffrich

Keywords:410-km DiscontinuityOlivineWadsleyiteEpitaxyWaterH-distribution

a b s t r a c t

We have synthesized a hydrous peridotite-composition sample at 13 GPa and 1400 �C with co-existingcoarse grains (�100 lm) of olivine, wadsleyite, clinoenstatite, plus melt in a multi-anvil press. Some ofthe olivine grains contain fine-scale (0.5–2 lm-wide) lamellae of wadsleyite and clinoenstatite that likelyresulted from transformation caused by small temperature fluctuations during the four-hour experiment.Phase compositions were determined by electron probe microanalysis (EPMA) and secondary ion massspectroscopy (SIMS). The olivine ranges from Fo94 to Fo90 in composition and contains about4000 ppm wt. H2O. The wadsleyite is Fo87±1 in composition and contains about 10,000 ppm wt. H2O.The clinoenstatite is En93±1 in composition and about 1400 ppm wt. H2O. Transmission electron micros-copy of the wadsleyite lamellae and host olivine shows that the two phases share their close-packed oxy-gen planes so that the wadsleyite lamellae are nearly planar and perpendicular to the [100] of olivine.The wadsleyite lamellae thus have their {101} and {021} planes parallel to the (100) plane of olivine.Additionally, larger incoherent grains of wadsleyite in olivine are found. Dislocation microtexures inthe olivine and iron concentration profiles across the lamella interface suggest heterogeneous nucleationand diffusion-controlled growth of coherent wadsleyite lamellae on defects in the olivine followed by thenucleation of faster-growing incoherent grains on the lamellae. The results show that, under hydrousconditions, the olivine–wadsleyite transformation occurs close to equilibrium at conditions of the 410-km discontinuity. Furthermore, inheritance of crystallographic preferred orientations (and therefore seis-mic anisotropy) across the 410-km discontinuity is unlikely to be significant. In addition, hydrogen dis-tributions among the various phases indicate that dehydration by melt extraction at 410 km will beinefficient and that H contents greater than about 4000 ppm wt. H2O are needed to initiate melting at410 km.

� 2012 Published by Elsevier B.V.

1. Introduction

Transformations between the principal polymorphs of (Mg,Fe)2-

SiO4, olivine (a), wadsleyite (b), and ringwoodite (c), are thought tobe the cause of seismic discontinuities at depths of 410 and 525 kmin the Earth’s mantle. Improved knowledge of transformationmechanisms between the polymorphs may thus shed light on crys-tal preferred orientation at seismic discontinuities (Mainpriceet al., 1990), and seismic anisotropy in the deep upper mantleand transition zone. In addition, metastable persistence of olivinebelow 410 km depth and its subsequent rapid transformation towadsleyite or ringwoodite in subducting lithospheric slabs has

been proposed as a possible cause of deep-focus earthquakes (Kir-by 1987; Kirby et al., 1991, 1996; Green et al., 2010).

There have been numerous studies of the mechanisms bywhich olivine transforms to wadsleyite and/or ringwoodite (e.g.,Poirier, 1981; Rubie and Brearley, 1990; Burnley et al., 1991;Brearley et al., 1992; Dupas-Bruzek et al., 1998; Mosenfelderet al., 2001; Kubo et al., 2004). Most experimental studies havebeen performed on nominally-dry samples at P–T conditions thatgreatly overstep the equilibrium phase boundaries (e.g., by 1–4 GPa). There are two reasons for this: (1) Nucleation energy bar-riers are high (Rubie et al., 1990) so that transformation does notoccur at conditions close to equilibrium, at least on an experi-mental timescale and in the absence of catalytic water or OH.(2) The aim of most previous studies has been to understandtransformation mechanisms and kinetics at P–T conditions of sub-ducting slabs, where a high degree of overstepping is applicable.

0031-9201/$ - see front matter � 2012 Published by Elsevier B.V.http://dx.doi.org/10.1016/j.pepi.2012.04.003

⇑ Corresponding author.E-mail address: [email protected] (J.R. Smyth).

Physics of the Earth and Planetary Interiors 200-201 (2012) 85–91

Contents lists available at SciVerse ScienceDirect

Physics of the Earth and Planetary Interiors

journal homepage: www.elsevier .com/locate /pepi


We are not aware of any studies that have been performed atnear-equilibrium conditions, within the olivine–wadsleyite two-phase loop for example.

As mentioned above, most experimental studies have been per-formed on nominally-dry samples. The application of transmissionelectron microscopy (TEM) has shown that transformation pro-ceeds by incoherent grain boundary nucleation and interface con-trolled growth (i.e., without any compositional change). Whenolivine is transformed within the ringwoodite stability field (e.g.,at 18 GPa), a shear-like mechanism also operates on a spatially-limited scale and results in the development of coherent nm-thickringwoodite lamellae which subsequently act as sites for incoher-ent nucleation and growth (Kerschhofer et al., 1996, 2000). Severalexperimental studies have shown that the presence of H2O fluidand/or OH initially dissolved in olivine has a large effect in increas-ing transformation rates (Kubo et al., 1998; Diedrich et al., 2009)but, again, such experiments have been performed at conditionsthat significantly overstepped the equilibrium phase boundary. Inthis study, we describe an experiment that was performed withinthe olivine–wadsleyite two-phase loop that shows that water/OHhas a dramatic effect in reducing nucleation barriers during theolivine–wadsleyite transformation and also promotes ‘‘coherent’’(topotaxial) growth that could have implications for seismicanisotropy at the 410-km discontinuity. Hereafter, we use the‘‘coherent’’-term for an topotaxial interface between wadsleyitelamellae and the matrix olivine, which however may be strictly‘‘semi-coherent’’ due to the existence of minor lattice misfit be-tween the two phases.

In general, the various (Mg,Fe)2SiO4 crystal structures have cor-responding close-packed oxygen layers. The olivine structure isbased on an approximately hexagonal-close-packed oxygen sub-lattice, with the a-axis perpendicular to the quasi close-packedplanes (100). Wadsleyite is a spinelloid and is based on a nearlycubic-close-packed oxygen sub-lattice. In a cubic-close-packed ar-ray there are four close-packed planes, and in wadsleyite they are{101}, and {021}. Anhydrous wadsleyite is orthorhombic (spacegroup Imma), whereas hydrous varieties can be slightly monoclinic(I2/m) with a beta angle of up to 90.4� (Smyth et al., 1997; Holl et al2008; Ye et al, 2010), but the unit cells and structures are approx-imately the same.

Ringwoodite is a true cubic spinel (space group Fd-3m) having acubic-close-packed oxygen sub-lattice with the close-packedplanes being (111), (�111), (1�11), and (11�1) (perpendicularto the four three-fold axes of the cube). Olivine oxidation producesmagnetite spinel lamellae with the (111) parallel to (100) of oliv-ine (Kohlstedt and Vander Sande, 1975; Puga et al., 1999).Although olivine–ringwoodite topotaxy has been described previ-ously (Kerschhofer et al., 1996), olivine–wadsleyite topotaxy hasnot been reported. We have identified ‘‘coherent’’ wadsleyitelamellae precipitating from an olivine host in one of our experi-mental samples synthesized in an Fe-bearing hydrous peridotitecomposition and here describe the compositional and crystallo-graphic relations as they may pertain to the olivine–wadsleyitetransformation at 410 km depth in the Earth. Additionally, Couvyet al (2004) and Demouchy et al (2011) have discussed shear defor-mation mechanisms in wadsleyite and effects of shear on the oliv-ine–wadsleyite transformation. An understanding of observedtopotaxial relations between the two phases is potentially impor-tant in studies of crystal-preferred orientation and elastic anisot-ropy in the transition zone (Mainprice et al., 1990).

2. Experimental

In order to measure H and Fe partitioning among nominallyanhydrous phases at the 410 km discontinuity, we have conductedexperimental syntheses of phases coexisting with hydrous melt in a

MgO–FeO–SiO2–H2O system with a mantle-like Fe:Mg molar ratioof about 1:9 and sufficient silica to have a pyroxene phase present.The experiment reported here was conducted at 13 GPa and1400 �C in a 1200-tonne multi-anvil press at Bayerisches Geoinsti-tut using in an 11 M sample assembly (11 mm sintered MgO octa-hedron and WC cubes with 8 mm corner truncations). Startingmaterials were San Carlos olivine (76 wt.%) plus brucite (10 wt.%),silica (12 wt.%), and FeO (2 wt.%) for a bulk water content (as hydro-xyl in brucite) of 2.9 wt.% H2O. The starting materials were finelyground and loaded into a 2 mm diameter welded Pt capsule. Theduration of heating was 4.5 h and temperature was measured usinga W–Re thermocouple and controlled manually. Temperature fluc-tuations during the experiment up to about ±20 �C were observed.

After opening the capsule several crystals were removed for X-ray identification and characterization. The remaining sample wassectioned for scanning electron microscopy (SEM), and chemicaland H-content analysis by EPMA and SIMS. The capsule containedcrystals of olivine (Fo90–94), wadsleyite (Fo87), and clinoenstatite(En93), plus an estimated five to ten volume percent of a quenchedmelt phase consisting of feathery wadsleyite crystals and voidspace that is presumed to have been occupied by a fluid phase.An SEM electron backscatter image of the sectioned capsule is gi-ven in Fig. 1a. There were clearly temperature gradients acrossthe radius and length of the capsule as indicated by the melt beingconcentrated in a girdle around the ‘‘waist’’ of the capsule. Clinoen-statite is primarily in contact with the melt. Olivine is present inthe outer center of the capsule near the pyroxene with wadsleyitepredominating in the cooler center and ends of the capsule. Phasecompositions, as analyzed by electron microprobe, are given in Ta-ble 1. Major-element compositions were determined by wave-length-dispersive electron microprobe analysis, using the JEOLJXA-8500F at the University of Hawaii. Analytical conditions werean acceleration voltage of 15 kV, a 10 nA probe current beam diam-eter between <1 (focused) and 5 microns. San Carlos olivine stan-dard USNM 111312/444 was used for calibration of Fe-ka, Si-ka,and Mg-ka and Kakanui augite standard USNM 122142 was usedfor Ca-ka and Al-ka. On- and off-peak counting times were 20 sfor all elements. The mineral compositions listed in Table 1 arethe averages of six individual analyses. Of these, olivine showsthe greatest range from Fo95 to Fo90, whereas the wadsleyite andclinopyroxene are relatively constant in composition.

H contents (Table 1) were determined by secondary ion massspectroscopy (SIMS) at the University of Hawaii. We used a Cs+ pri-mary beam to generate negative secondary ions: H–, D–, and 30Si–.

Fig. 1A. Backscatter electron image of the capsule showing the bright Pt capsulematerial, and the distribution of the quenched melt phase (Mlt), clinoenstatite (P),olivine (Ol), and wadsleyite.

86 J.R. Smyth et al. / Physics of the Earth and Planetary Interiors 200-201 (2012) 85–91


The electron flood gun was used for charge compensation. Themass resolving power was �1500, sufficient to resolve all interfer-ences from the hydrogen isotopes. The small signal from 29SiH atmass 30 is partially resolved and does not significantly affect theresults. Each spot on the sample was first sputter-cleaned byrastering a 1 nA beam of a 25 � 25 lm area for 600 seconds. Thebeam current and raster size were then reduced to 250 pA and5 lm, respectively for the measurement. A field aperture maskedthe ions originating from outside of a �50 � 50 lm area on thesample. A liquid-nitrogen trap was used to reduce the pressurein the sample chamber to �1 � 10�9 torr. A measurement con-sisted of 40 cycles through the masses with count times of 1 sec-ond (H), 40 seconds (D), and 1 second (30Si). For the first 10cycles and last 10 cycles, the primary ion beam was off, giving usa measure of the hydrogen background in the sample chamber.The measurements were calibrated using an experimental ring-woodite sample that had been measured by FTIR. Uncertaintiesin our water contents, which include the statistical uncertaintiesin measurement and background and the uncertainty in the sensi-tivity factor derived from our standard, are �8% relative for wads-leyite and olivine and �8.5% relative for pyroxene.

In order to determine the crystallographic orientations betweenphases, a portion of an optical thin section was selected, removedfrom the glass slide and thinned using Ar-ion milling. TEM obser-vations were performed with a Philips CM20FEG TEM (BayerischesGeoinstitut), operating at 200 kV. Topotaxial relations were exam-ined using bright field (BF) and weak-beam dark-field (WBDF)images and selected area electron diffraction (SAED). High resolu-tion TEM images were recorded under defocused conditions nearthe Scherzer defocus (�67 nm). For chemical analysis across wads-leyite lamellae, energy dispersive X-ray spectroscopy (EDXS) usingscanning beam mode was performed. The chemical data were eval-uated with calibrated k-factors for oxygen, magnesium and ironagainst silicon by using a pyrope-almandine mineral standard(van Cappellen, 1990) and a thickness-absorption correction basedon electron neutrality (van Cappellen and Doukhan, 1994).

3. Results and discussion

3.1. Petrographic relations

The SEM electron backscatter image of Fig. 1b shows olivinecrystals near the center of the capsule that contain coherent lamel-lae of wadsleyite. The backscatter image shows variable contrast in

the olivine grain that indicates significant gradients in Fe content.The regions with higher Fe contents contain exsolution lamellae ofwadsleyite typically �200 nm wide, and may represent areas thatwere initially wadsleyite but that had reverted to olivine with asmall increase in temperature during the experiment. The higherFe (and H) contents then partitioned into coherent lamellae ofwadsleyite that subsequently nucleated within the olivine(Figs. 1b and 2). Coherent wadsleyite lamellae within olivine havenot been reported or described previously and here we documentthe crystallographic topotaxy of the two phases by optical andtransmission electron microscopy. In the vicinity of the wadsleyitelamellae, topotaxial clinopyroxene lamellae were also alwaysfound in the TEM observation, which indicates that the precursorof the olivine with these two-phase lamellae is a wadsleyite thathad a lower (Mg + Fe): Si ratio than 2.0 of a stoichiometric olivine.The existence of the clinopyroxene lamellae is proof of the trans-formation mechanisms from wadsleyite to olivine with exsolutionlamellae. Its presence is consistent with hydration of wadsleyite by

Table 1Chemical compositions (weight percent oxide) of co-existing phases as determined byelectron microprobe chemical analysis and H by ion microprobe (SIMS).

Oxide Olivine Wadsleyite Clinoenstatite

SiO2 40.16 40.01 57.70TiO2 0.00 0.00 0.00Al2O3 0.00 0.07 0.07MgO 51.79 45.92 37.70FeO 7.48 11.95 3.16NiO 0.43 0.48 0.38CaO 0.00 0.00 0.07H2O 0.40 1.00 0.14Total 100.26 99.43 99.38

Atom ratiosOxygens 4 4 6Si 0.971 0.986 1.970Ti 0.000 0.000 0.000Al 0.000 0.002 0.003Mg 1.866 1.687 1.924Fe 0.151 0.246 0.091Ni 0.008 0.009 0.011Ca 0.000 0.000 0.003H 0.064 0.164 0.032Total 3.061 3.095 4.039

Fig. 1B. Backscattered electron image of exsolution texture of wadsleyite (bright)from olivine (darker). The wadsleyite (bright rounded grains and lamellae) arericher in Fe and H relative to the darker olivine. The image is from near the center ofthe capsule (Fig 1A) after the section was re-polished.

Fig. 2. Crossed-polars optical image of wadsleyite lamellae and coarse grains (blue-gray) in an olivine host showing orange-red birefringence colors. The image is about100 micrometers across. The optical thin section was cut from near the center of thecapsule (Fig 1A).

J.R. Smyth et al. / Physics of the Earth and Planetary Interiors 200-201 (2012) 85–91 87


octahedral cation vacancies which gives the wadsleyite a lower(Mg + Fe): Si ratio than olivine with a lower H content.

A 30 lm thick petrographic thin section was prepared of thesectioned capsule. A crossed-polars optical image is given inFig. 2. This shows a single olivine grain with a strong interferencecolor gradient probably due largely to thickness, but also to com-

positional gradients. There are several large grains of wadsleyite.Several orientations of wadsleyite inclusions are visible withinthe olivine grain. The inclusions are rounded blebs, several ofwhich appear to be near extinction and are possibly in optical con-tinuity. There are also the prominent coherent lamellae of wads-leyite that appear only in the Fe-rich regions of the olivine grain.

3.2. Orientation relations of the coexisting phases

Two types of topotaxial relations between olivine and wadsley-ite lamellae were confirmed by selected area electron diffractionpatterns from the two-phase boundaries. In Type 1, the {101}planes of wadsleyite are parallel to the (200) plane of olivineand the [010] zone axis of wadsleyite is parallel to [001] of olivine,which corresponds to the orientation of the incident electron beam(Fig 3). In this orientation relation, transformation between thephases preserves hexagonal close-packed and cubic close-packedlayers of oxygen, respectively. On the other hand, in the Type 2relationship the {04�2} plane of wadsleyite is parallel to the(200) plane of olivine (Fig. 4), which implies that {021} planes

Fig. 3. Bright field TEM image and diffraction pattern of the olivine (a)-wadsleyite(b) intergrowth. (Type 1 orientation relationship: (100) olivine // (101) wadsleyiteand [001] olivine // [010] wadsleyite). Cpx: clinoenstatite.

Fig. 4. Typical bright field-TEM image and selected area electron diffractionpatterns of the topotaxy in olivine (a)-wadsleyite (b)-clinopyroxene (cpx) (Type 2orientation relationship: (200) olivine // (04�2) wadsleyite, (060) olivine // (044)wadsleyite and [001] olivine // [100] wadsleyite).

Fig. 5. High resolution lattice images of the two types of interfaces between olivine(a) and wadsleyite lamellae (b). (A) Type 1, where [001] olivine and [010]wadsleyite zone axes are parallel, (B) Type 2, [001] olivine and [100] wadsleyitezone axes The Type 1 in the image (a) displays a sharp interface, but the Type 2 inthe image (b) displays a vague one, implying a tilting interface, i.e., not parallel tothe incident electron beam. Insets are Fast Fourier Transform patterns from the twophases separated by the interfaces, which correspond to SAED patterns of theabove-mentioned zone axes.



of wadsleyite are parallel to the (100) plane of olivine (see the ste-reo plots of axes and plane directions in Appendix Fig. A1), and the

[100] zone axis of wadsleyite is parallel to the [001] of olivine. Inthe same olivine grain, clinopyroxene lamellae are also present andthe (100) plane is also parallel to the (100) of olivine (upper rightinset in Fig. 4). In all the relations, the close-packed layers in oliv-ine and wadsleyite, as well as clinopyroxene, are preserved in thesame orientation normal to the lamella walls. High resolution lat-tice images of the two-types of interfaces between olivine andwadsleyite indicate a sharp interface for Type 1 and a broaderinterface for Type 2, respectively (Fig. 5), which implies that thelatter composite planes can be tilted from the close-packed layersof both phases. Moreover, a textural transition from semi-coherentlamellae to incoherent blebs is observed in wadsleyite (Fig. 6). Suchtextures appear to indicate that the coherent wadsleyite lamellaeact as nucleation sites for incoherent wadsleyite grains that havemuch higher growth rates, as observed previously for the oliv-ine–ringwoodite transformation by Kerschhofer et al. (2000).

Our interpretation of the texture is that the sample initiallyequilibrated with olivine, wadsleyite, clinoenstatite and melt beingdistributed according to the thermal gradient in the capsule. In the

Fig. 6. Bright field TEM image of a textural transition in wadsleyite from semi-coherent lamellae to relatively large incoherent grains. Such textures suggest thatincoherent wadsleyite grains nucleate on the narrow lamellae, as observed for theolivine–ringwoodite transformation by Kerschhofer et al. (2000).

Fig. 7. Weak-beam dark-field TEM image (with g = 400 a) of [00] edge dislocations(triangle-arrowhead) in olivine (a), which are parallel to wadsleyite lamellae (b).The wadsleyite lamellae are likely to have nucleated on or in the vicinity of the[100] edge dislocations.

Fig. 8. Chemical analysis of a narrow lamella (�285 nm in width) by STEM–EDXS.(a) A dark field STEM image with beam trace spots in the EDXS analysis across alamella and the adjacent host olivine. (b) The wt.% FeO concentration across thewadsleyite (b) lamella and adjacent olivine (a); the exact location of the profile isindicated by the white arrow in (a). The data point indicated by the arrow isinfluenced by the compositions of both phases.



center of the capsule a small rise in temperature, or perhaps loss ofH through the capsule wall, during the experiment shifted thephase boundary between the olivine and wadsleyite in favor ofthe olivine phase. The olivine grew into a previously wadsleyite re-gion that had higher Fe and H contents and a lower (Mg + Fe):Si ra-tio (due to higher H content: hydrogen substitutes into thewadsleyite structure forming OH on the O1 position with octahe-dral vacancies at M3 (Smyth, 1987; Holl et al., 2008; Ye et al.,2010)). The new olivine then approached chemical equilibriumby exsolving wadsleyite and clinoenstatite lamellae (Figs. 3 and4). The higher iron content of the olivine that contains wadsleyitelamellae is seen in the electron backscatter contrast in Fig. 1b.

3.3. Heterogeneous nucleation mechanisms of coherent wadsleyitelamellae

The dislocation microtextures in the matrix olivine and chemi-cal compositions in the vicinity of wadsleyite lamellae suggest het-erogeneous nucleation and growth of wadsleyite lamellae ondefects in the olivine. The [100] edge dislocations with [010] or[001] line direction are observed in a parallel orientation to thewadsleyite lamellae (Fig. 7). The [100] dislocation is the most com-monly observed in olivine deformed at temperatures above1400 �C and low stress conditions (e.g., Carter and Ave’ Lallemant,1970; Phakey et al., 1972). At higher pressures around 7.5 GPa, the[001] dislocation is also activated depending on water contentsand stress conditions (e.g., Raterron et al. 2011; Shekhar 2011),although the geometry of the TEM foil studied here is not suitablefor visualizing [001] dislocations. It is likely that coherent lamellaeof wadsleyite nucleated on [100] edge dislocations. A chemicalprofile across a narrow wadsleyite lamella (<300 nm in width)and the adjacent olivine using EDXS analysis with a scanning elec-tron beam displays a detectable downhill gradient in iron content,i.e., fayalite component, in the olivine (Fig. 8). This shows thatgrowth of the Fe-rich wadsleyite lamella has depleted the adjacentolivine in Fe, as would be expected in the case of diffusion-con-trolled growth. However, analyses across wider lamellae (e.g.,�600 nm in width) do not display any significant chemical gradi-ents (Appendix Fig. A2). These results support the interpretationthat heterogeneous nucleation of wadsleyite takes place on crystaldefects in the matrix olivine and subsequent downhill diffusion inolivine controls the coarsening rate of the lamella at conditionsclose to thermodynamic equilibrium.

The microtextures of the lamellae indicate that the olivine–wadsleyite transformation can proceed effectively by heteroge-neous nucleation mechanisms with semi-coherent interfaces thatcould potentially preserve lattice preferred orientations. The ob-served rapid transformation from olivine to wadsleyite undernear-equilibrium conditions means that metastable preservationof olivine below 410 km is especially unlikely under hydrous con-ditions. The temperature of the experiment was, of course, muchhigher than expected in a subducting slab and close to that ofthe expected mantle geotherm.

The textures observed in this study do not support the notionthat the olivine to wadsleyite transformation will preserve crystal-lographic orientation relationships across the 410-km discontinu-ity (e.g., Mainprice et al., 1990). Although wadsleyite nucleates atintracrystalline sites in olivine with well defined orientation rela-tionships, the resulting lamellae do not volumetrically dominatethe transition. Instead the lamellae act as nucleation sites for fas-ter-growing incoherent wadsleyite grains (Fig. 6), which constitutea larger volume and are thus likely to dominate the transformation.This two-stage transformation process is identical to observationsfor the olivine–ringwoodite transformation in experiments per-formed at 18 GPa (Kerschhofer et al., 2000). A relatively fastgrowth rate for incoherent grains, compared to those of coher-

ently-nucleated grains, has also been reported for the wadsleyiteto ringwoodite transformation (Dupas-Bruzek et al., 1998, Fig. 10).

3.4. Hydrogen distributions

The distribution of H among the coexisting phases may alsoshed some light on melting relations that might occur if a hydroustransition zone were to convect upward across the 410 km discon-tinuity as suggested by Bercovici and Karato (2003). In the exper-iment described here, the wadsleyite contains about 10,000 ppmwt. H2O, the olivine about 4000 ppm, and the clinoenstatite about1400 ppm, as measured by SIMS. If the areas in the section ofFig. 1a are representative of the volume fractions of the variousphases, the melt phase may constitute up to about 10% of the totalcapsule volume and was probably a single-phase hydrous melt.The remainder of the sample consists of �60% wadsleyite, �20%olivine, and �10% clinoenstatite (by volume). If the total H2O con-tent was about 3% by weight and no water was lost from the cap-sule during the experiment (not likely), then the melt phase mayhave contained up to 20 wt.% H2O. On quench, the melt crystallizedto form mostly wadsleyite dendrites and an aqueous phase thatwas liberated when the capsule was opened. This implies that dis-tribution of H2O between solid and melt was at most 20:1 byweight which demonstrates that significantly hydrous wadsleyiteand olivine can co-exist with hydrous melt so that dehydrationby melt extraction is not highly efficient under these conditions.The H distribution between wadsleyite and olivine is about 2.5and wadsleyite to clinoenstatite is about 5 at temperatures abovethe hydrous solidus for 3% H2O by weight and near an expectedgeotherm. The H content of the olivine therefore implies that melt-ing at 410 km on upwelling convection would require more thanabout 4000 ppm wt. H2O.

Acknowledgements

This work was supported by US National Science FoundationGrant EAR 07-11165 and 11-13369 and by NASA AstrobiologyInstitute cooperative agreement No. NNA09DA77A with the Uni-versity of Hawaii. High pressure synthesis was supported by BGIVisitors Program and the Alexander von Humboldt Foundation.Hubert Schulze (BGI) is thanked for preparation of the optical thinsection (Fig. 2) from which the TEM sections were prepared.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.pepi.2012.04.003.

References

Bercovici, D., Karato, S.-I., 2003. Whole mantle convection and the transition-zonewater filter. Nature 425, 39–44.

Brearley, A.J., Rubie, D.C., Ito, E., 1992. Mechanisms of the transformations betweenthe a, b, and c polymorphs of Mg2SiO4 ar 15 GPa. Phys. Chem. Minerals 18, 343–358.

Burnley, P.C., Green, H.W., Prior, D.J., 1991. Faulting associated with the olivine tospinel transformation in Mg2GeO4 and its implications for deep-focusearthquakes. J. Geophys. Res. 96, 425–443.

Carter, N.L., Ave’ Lallemant, G., 1970. High temperature flow of dunite andperidotite. Geol. Soc. Am. Bull. 81, 2181–2202.

Couvy, H., Frost, D.J., Heidelbach, F., Nyilas, F., Ungar, T., Mackwell, S., Cordier, P.,2004. Shear deformation experiments of forsterite at 11 GPa–1400 �C in themultianvil apparatus. Eur. J. Mineral. 16, 877–889.

Demouchy, S., Mainprice, D., Tommasi, A., Couvy, H., Barou, F., Frost, D.J., Cordier, P.,2011. Forsterite to wadsleyite phase transformation under shear stress andconsequences for the Earth’s mantle transition zone. Phys. Earth Planet. Inter.184, 91–104.

Diedrich, T., Sharp, T.G., Leinweber, K., Holloway, J.R., 2009. The effects of smallamounts of H2O on olivine to ringwoodite transformation growth rates andimplications for subduction of metastable olivine. Chem. Geol. 262, 87–99.



Dupas-Bruzek, C., Sharp, T.G., Rubie, D.C., Durham, W.B., 1998. Mechanisms oftransformation and deformation in Mg1.8Fe0.2SiO4 olivine and wadsleyite undernon-hydrostatic stress. Phys. Earth Planet. Int. 108, 33–48.

Green, H.W., Chen, W.-P., Brudzinski, M.R., 2010. Seismic evidence of negligiblewater carried below 400-km depth in subducting lithosphere. Nature 467, 828–831.

Holl, C.M., Smyth, J.R., Jacobsen, S.D., Frost, D.J., 2008. Effects of hydration onstructure and compressibility of wadsleyite, beta-Mg2SiO4. Am. Mineral. 93,598–607.

Kerschhofer, L., Sharp, T.G., Rubie, D.C., 1996. Intracrystalline transformation ofolivine to wadsleyite and ringwoodite under subduction zone conditions.Science 274, 79–81.

Kerschhofer, L., Rubie, D.C., Sharp, T.G., McConnell, J.D.C., Dupas-Bruzek, C., 2000.Kinetics of intracrystalline olivine–ringwoodite transformation. Phys. EarthPlanet. Int. 121, 59–76.

Kirby, S.H., 1987. Localised polymorphic phase transformations in high-pressurefaults and applications to the physical mechanism of deep earthquakes. J.Geophys. Res. 92, 13789–13800.

Kirby, S.H., Durham, W.B., Stern, L.A., 1991. Mantle phase changes and deepearthquake faulting in subducting lithosphere. Science 252, 216–225.

Kirby, S.H., Stein, S., Okal, E.A., Rubie, D.C., 1996. Metastable mantle phasetransformations and deep earthquakes in subducting oceanic lithosphere. Rev.Geophys. 34, 261–306.

Kogure, T., 2003. A program to assist Kikuchi pattern analysis. J. Cryst. Soc. Japan 45,391–395 (in Japanese with English abstract, the program is available fromhttp://www-gbs.eps.s.u-tokyo.ac.jp/kogure/EDANA/EDA/EDANA_E.htm).

Kohlstedt, D.L., Vander Sande, J.B., 1975. An electron microscopy study of naturallyoccurring oxidation produced precipitates in iron-bearing olivines. Contrib.Mineral. Petrol. 53, 13–24.

Kubo, T., Ohtani, E., Kato, T., Shinmei, T., Fujino, K., 1998. Effects of water on thealpha–beta transformation kinetics in San Carlos Olivine. Science 281, 85–87.

Kubo, T., Ohtani, E., Funakoshi, K., 2004. Nucleation and growth kinetics of thealpha-beta transformation in Mg2SiO4 determined by in situ synchrotron X-raydiffraction. Am. Mineral. 89, 285–293.

Mainprice, D., Humbert, M., Wagner, F., 1990. Phase transformations and inheritedlattice preferred orientations: implications for seismic properties.Tectonophysics 180, 213–228.

Mosenfelder, J.L., Marton, F.C., Ross II, C.R., Kerschhofer, L., Rubie, D.C., 2001.Experimental constraints on the depth of olivine metastability in subductinglithosphere. Phys. Earth Planet. Int. 127, 165–180.

Phakey et al., 1972. Transmission electron microscopy of experimentally deformedolivine crystals. In: Flow and Fracture of Rocks, Geophys. Monograph, 16, AGU,pp. 139–156.

Poirier, J-P., 1981. Martensitic olivine–spinel transformation and plasticity of themantle transition zone. In: Stacy, F.D., Patterson, M.S., abd Nicolas, A. (Eds.),Anelasticity in the Earth, Geodynamics Series, vol. 4. AGU, Washington, DC, pp.113–117.

Puga, E., Ruiz-Cruz, M.D., Federico, A.D., 1999. Magnetite-silicate inclusions inolivine of ophiolitic metagabbros from the Mulhacen Complex, Betic Cordillera,southeastern Spain. Can. Mineral. 37, 1191–1209.

Raterron, P., Chen, J., Geenen, T., Girard, J., 2011. Pressure effect on forsteritedislocation slip systems: Implications for upper-mantle LPO and low viscosityzone. Phys. Earth Planet. Int. 188, 26–36.

Rubie, D.C., Brearley, A.J., 1990. Mechanism of the c–b phase transformation ofMg2SiO4 at high temperature and pressure. Nature 348, 628–631.

Rubie, D.C., Tsuchida, Y., Yagi, T., Utsumi, W., Kikegawa, T., Shimomura, O., Brearley,A.J., 1990. An in situ X-ray diffraction study of the kinetics of the Ni2SiO4

olivine–spinel transformation. J. Geophys. Res. 95, 15829–15844.Shekhar, S., 2011. The Effects of Stress and Water Contents on the Deformation of

Olivine at High Pressure. PhD thesis. University of Bayreuth.Smyth, J.R., 1987. Beta-Mg2SiO4: a potential host for water in the mantle? Am.

Mineral. 72, 1051–1055.Smyth, J.R., Kawamoto, T., Jacobsen, S.D., Swope, R.J., Hervig, R.L., Holloway, J.R.,

1997. Crystal structure of monoclinic hydrous wadsleyite. Am. Mineral. 82,270–275.

van Cappellen, E., 1990. The parameter-less corrections method in X-raymicroanalysis. Microsc. Microanal. Microstruct. 1, 1–22.

van Cappellen, E., Doukhan, J.C., 1994. Quantitative X-ray microanalysis of ioniccompounds. Ultramicroscopy 53, 343–349.

Ye, Y., Smyth, J.R., Hushur, A., Lonappan, D., Manghnani, M.H., Dera, P., Frost, D.J.,2010. Crystal structure of hydrous wadsleyite with 2.8% H2O andcompressibility to 60 GPa. Am. Mineral. 95, 1765–1772.


(this is a sample cover image for this issue. the actual cover is...

Documents