Author's personal copy Earth and Planetary Science Letters 293 (2010) 90–96

Contents lists available at ScienceDirect

Earth and Planetary Science Letters j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l

Experimental evidence for perovskite and post-perovskite coexistence throughout the whole D″ region Denis Andrault a,⁎, Manuel Muñoz b, Nathalie Bolfan-Casanova a, Nicolas Guignot c, Jean-Philippe Perrillat d, Giuliana Aquilanti d, Sakura Pascarelli d a

Laboratoire Magmas et Volcans, Université Blaise Pascal, Clermont-Ferrand, France Laboratoire de Géodynamique des Chaînes Alpines, Université Joseph Fourier, Grenoble, France Synchrotron SOLEIL, Gif-sur-Yvette, France d European Synchrotron Radiation Facility, Grenoble, France b c

a r t i c l e

i n f o

Article history: Received 25 September 2009 Received in revised form 3 January 2010 Accepted 16 February 2010 Available online 17 March 2010 Editor: L. Stixrude Keywords: lower mantle D″-layer post-perovskite transition Fe partitioning

a b s t r a c t Since the phase diagram for (Fe,Al)-bearing MgSiO3 compositions at the P–T conditions of the core–mantle boundary remains ambiguous, we investigated the Fe distribution among the silicate perovskite (Pv) and post-perovskite (PPv) polymorphs using tandem synchrotron analyses of X-ray diffraction and X-ray absorption spectroscopy. We performed measurements at the Fe K-edge of the partitioning of iron between Pv and PPv up to more than 150 GPa after annealing at about 3300 K. We obtain a unique solution for KPv/PPv Fe of 4.2 (+/− 0.5). Our results evidence that the two silicates should coexist over the whole D″ region, with the main post-perovskite phase being largely depleted in Fe compared to the perovskite. As Fe and Al have a dominant effect on the phase diagram, these new results challenge recent determinations of the temperature profile in the lowermost mantle based on the Clapeyron slope of the Pv to PPv transition for pure MgSiO3 composition. Also, it appears clear that variations in the molar fractions of perovskite and post-perovskite phases should be expected radially or laterally in the D″ region in relation with thermal or compositional heterogeneities. This can help explaining the seismological anomalies observed for this mantle region. Finally, we predict a significant increase of the FeO activity in the D″ region, which should greatly affect the chemical exchange between mantle and core. © 2010 Elsevier B.V. All rights reserved.

1. Introduction The sharp phase transition in MgSiO3 from Pv to PPv has been often invoked in order to explain seismic observations of the D″ layer (Murakami et al., 2004; Oganov and Ono, 2004). This phase transition can be associated with deep seismic reflectors (Sidorin et al., 1999) and the strong elastic anisotropy of the PPv CaIrO3 analogue structure could explain the mantle heterogeneity in sound velocities (Guignot et al., 2007; Wysession et al., 1999; Hirose et al., 1999). On the other hand, actual seismic evidences correspond to local phenomena in specific mantle regions, rather than to geographically extended mantle features, in sharp contrast with what is observed at 410 and 660 km depth discontinuities. The regional character of the D″ features has been tentatively explained by the relatively flat Clapeyron slope of the Pv to PPv transition, which, associated with large temperature heterogeneities, can modify significantly the transition depth and in an extreme case even prevent the formation of PPv at very high mantle temperatures (Wookey et al., 2005;

⁎ Corresponding author. E-mail address: [email protected] (D. Andrault). 0012-821X/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2010.02.026

Hernlund et al., 2005). Even so, it remains questionable as to whether this phase transition can explain all seismic observed features, including the low velocity zones, or if the hypotheses advanced previously, such as the presence of a primitive mantle reservoir (Kellogg et al., 1999), the case for a subducted plate cemetery (Wysession et al., 1999; Hirose et al., 1999), or partial melting at the core–mantle boundary (CMB) (Williams and Garnero, 1996) remain valid for the D″ region. Still, previous approaches tended to neglect the effects of Fe2+ or 3+ Fe , and Al3+ major elements on the Pv to PPv phase transition due to a lack of data. Their presence and distribution can modify severely the transition depth as well as the depth over which the two phases coexist in the D″ layer. This issue remains very controversial, based on the three sources of information available to date (i) in-situ X-ray diffraction in the laser-heated diamond anvil cell (LH-DAC), (ii) postmortem examination of LH-DAC samples using transmission electron microscope and (iii) ab initio structure simulations. Large uncertainties concern both the Pv and PPv phase fractions as a function of P, T and composition and the Fe and Al partitioning coefficients between the two phases. Firstly, it has been shown that incorporation of 25 mol% Al2O3 stabilizes the perovskite phase to higher pressures by more than

Author's personal copy D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

30 GPa at the expense of post-perovskite (Tateno et al., 2005). However, ab initio studies present contradictory conclusions including large increase (Akber-Knutson and Bukowinsky, 2004), moderate increase (Caracas and Cohen, 2005; Zhang and Oganov, 2006), or moderate decrease (Tsuchiya and Tsuchiya, 2008) of the transition pressure with increasing Al2O3-content. The effect of Fe on the Pv to PPv transition also remains controversial. Two experimental studies based on (Mg,Fe)SiO3 composition concluded that Fe stabilizes PPv to slightly lower pressures (Mao et al., 2004; Shieh et al., 2006), in agreement with theoretical calculations (Caracas and Cohen, 2005; Ono and Oganov, 2005; Stackhouse et al., 2006), while other experimental works observed the opposite trend with Fe entering preferentially the Pv phase and a Pv and PPv coexistence extended into the stability field of PPv (Tateno et al., 2007; Hirose et al., 2008). For an Fe3+AlO3-bearing composition, coexistence of Pv and PPv was reported up to 157 GPa at 1600 K, thus a stabilisation of more than 40 GPa compared to the pure MgSiO3 phase (Nishio-Hamane et al., 2007a,b). For sample compositions typical of pyrolite and mid-ocean ridge basalts, previous studies suggest minor Al and Fe effects (Murakami et al., 2005; Ohta et al., 2008). On the other hand, chemical analyses based on post-mortem examination of Pv and PPv phases could provide independent information on the role of Fe and Al, since the partition coefficient PPv Pv PPv PPv (KPv/PPv = XPv and KPv/PPv = XPv Fe Fe · XMg / XMg · XFe Al Al / XAl , but the Al partition coefficient is not yet available in the literature) reflects the energy difference for the Fe insertion in Pv and PPv. KPv/PPv can be Fe calculated by dividing KPv/Fp by KPPv/Fp from previous studies reporting Fe Fe the Fe partition coefficients between ferropericlase (Fp) and Pv and between Fp and PPv in similar experimental conditions. However, measuring partition coefficients using transmission electron microscopy remains a difficult task due to the very small samples investigated. As a matter of fact, KPv/PPv was reported to vary between 0.13 Fe to 1 in Al-free systems (Hirose et al., 2008; Auzende et al., 2008; Kobayashi et al., 2005), while in an Al-bearing system, KPv/PPv was Fe reported to be 4 for pyrolite (Murakami et al., 2005). These results show that Fe is energetically favored in Pv compared to PPv and that it should thus stabilize Pv to higher pressures at the expense of PPv. In order to better constrain the mechanism of the Pv to PPv phase transformation in natural conditions, we coupled the performance of the LH-DAC with tandem in-situ analyses of the structure and composition of the samples using synchrotron X-ray diffraction (XRD) and X-ray absorption near edge structure (XANES) spectroscopy, respectively at ID27 and ID24 micro-beamlines of European Synchrotron Radiation Facility (ESRF, Grenoble, France). 2. Experimental methods As starting materials, we used two different natural enstatite samples extracted from spinel lherzolite (KLB, Kilbourne Hole, USA) and Orogenic Massif (Codera Valley, Italy), and a synthetic glass with typical MORB-perovskite composition (Table 1). A maximum care was devoted to prevent the presence of any moisture. Starting materials were ground just before loading the DAC in a glove bag with argon flow. We used membrane type DAC mounted with beveled diamonds with culet diameters of 75/300 µm. We selected diamond anvils of

Table 1 Composition of the starting materials (in atoms per formula units).

Mg Ca Fe Si Al

91

1.6 mm height in order to reduce their absorption of X-rays at the Fe K-edge energy (7112 eV). Temperatures were provided by two YAG lasers and monitored by thermal radiometry. Error in temperature is estimated to be of 50 K. The two lasers were slowly scanned on the sample surface for more than 30 min at increasing temperatures up to more than 3000 K. At these extreme conditions, the sample thickness is less than 10 µm. For such sample thickness, the double side laser heating with a wavelength of 1.064 µm reduces considerably the axial temperature gradient. If there were some parts of the sample that did not react correctly, the Bragg peaks should be anomalously broad or even doubled due to local chemical variations. However, this is not observed in our X-ray patterns (Figs. 1, 2), thus proving a good sample axial homogeneity. On the other hand, we performed systematic XRD and XANES (Fig. 3) mapping measurements in the pressure chamber. The similarity of the spectral features evidences homogeneity, not only in the Fe-content, but also in Fe-local structure over the whole sample. Pressures at 300 K were derived from the equation of state of a thin film of gold placed in the sample chamber or from Re-gasket. Above 90 GPa, both measurements agree within 2 GPa (Jamieson et al., 1982; Sha et al., 2004). It has been established that samples encounter higher pressures when submitted to laser heating due to the effect of thermal pressure (Andrault et al., 1998). Unfortunately, we could not mix the samples with any pressure calibrant (Pt or Au) to prevent potential problems associated with chemical reactions with Fe, which would (i) disable the pressure measurement and (ii) pollute the XANES spectra of an extra metallic iron phase. Therefore, we applied a pressure correction to our data set of ∼50% of the theoretical thermal pressure (αKΔT) calculated from the available thermoelastic parameters of Pv or PPv phase (Guignot et al., 2007; Andrault et al., 1998; Fiquet et al., 2000) (Table 2), in agreement with our previous measurements using internal pressure calibrant (Guignot et al., 2007). Error in pressure is estimated to be 3%. For the X-ray diffraction measurements, we used the classical setup available at the ID27 beamline for LH-DAC (Mezouar et al., 2005; Schultz et al., 2005). Wavelength was fixed at 0.3738 Å or 0.2647 Å. Two bent mirrors achieve an X-ray spot of 2 × 3 µm2 onto the sample. Typical acquisition time is 30 s using an imaging plate or a CCD detector. The X-ray beam position in the pressure chamber is determined from optical observations of the fluorescence of a material such as Re or diamond. Image of X-ray beam, as well as those of the two YAG lasers, are positioned on the pinhole of the spectrometer entrance. For Rietveld refinements, we used the same isotropic thermal parameters for all atoms in Pv and PPv phases. Atomic positions were fixed to values reported elsewhere (Guignot et al., 2007; Fiquet et al., 2000). These parameters affect ratio of diffraction peak intensities, but they have minor effect on determination of Pv and PPv phase fractions. For the XANES measurements, we used a bent Si (311) dispersive polychromator at the ID24 beamline (Pascarelli et al., 2006). The polychromatic X-ray beam was focused to a ∼5 × 5 µm spot onto the sample. The absorption spectrum μ(E) is equal to ln(I0 / I1) where I0 and I1 are the X-ray intensities recorded by the spatially resolved CCD detector with and without the DAC respectively. Using the dispersive mode, acquisition time is reduced to a few minutes. Micro-XANES mapping of samples was done using a grid of 5 × 5 to 8 × 8 pixel frames with 2.5 µm steps (Muñoz et al., 2006; 2008). 3. Results and discussion

KLB

Codera

MORB

0.835 0.015 0.094 0.955 0.101

0.701 0.005 0.194 0.925 0.175

0.502 – 0.368 0.760 0.370

Growth of Pv and PPv grains was carefully monitored by following the shape of diffraction peaks as a function of time at increasing temperature. Sharpness of the Bragg lines and occurrence of a multitude of small spots on the 2-dimensional X-ray pattern evidence efficient atomic diffusion at the scale of the grain size and quasiequilibrium crystallization of the Pv and PPv grains (Figs. 1, 2).

Author's personal copy 92

D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

Fig. 1. X-ray diffraction images recorded for Al-bearing (Mg,Fe0.19)SiO3 Codera sample after laser heating at increasing temperature for a nominal pressure of 172 GPa. Above 3150 K, the diffraction lines progressively change from continuous lines to spotty rings. This is associated to grain growth of the Pv and PPv phases. Further heating (longer time or higher temperature above 3150 K) only favours grain growth (larger dots on the rings) but no significant change in the integrated pattern, and thus in the mineralogical content.

Maximum temperature was kept between 3150 and 3600 K, in agreement with recent works suggesting temperatures between 3000 and 4000 K in the D″ region (Hernlund et al., 2005; Lay et al., 2008; van der Hilst et al., 2007). Our temperatures are ∼ 1000 K higher than in previous works, so kinetic hindering should not be an issue. Moreover, a new sample was systematically used for each pressure investigated in this study (Table 2).

Fig. 2. Integrated diffraction patterns recorded for Al-bearing (Mg,Fe0.19)SiO3 Codera samples after laser heating at pressures of 106, 142 and 184 GPa. Crosses indicate the experimental measurement while the red line is a Rietveld fit using a mixture of Pv and PPv phases. Background due to Compton diffusion in diamond anvils has been subtracted. The PPv content increases from 7% to 36% and 95% with increase pressure (Table 2).

For the Codera composition, the diffraction patterns recorded between 70 and 184 GPa reveal the presence of two crystalline phases (Fig. 2). Relative proportions of Pv and PPv were monitored as a function of pressure and temperature using the intensity of the integrated X-ray diffraction patterns. For the Rietveld refinements, fractions of Fe in Mg-sites of Pv and PPv were inserted a posteriori after the Fe partitioning between both phases was determined (see below). For the three (Fe,Al)-bearing MgSiO3 compositions studied, we observed an extended range of coexistence of the Pv and PPv phases extending significantly above the 135 GPa pressure corresponding to the CMB (Table 2, Fig. 4). Error in phase fraction is estimated to be 3%. Our results agree with some previous experimental reports, in particular (Nishio-Hamane et al., 2007a,b), but disagree with others (see Introduction section for a detailed analysis of previous reports). Different reasons can explain such disagreement: (i) we used Albearing samples which are not the case for all previous reports; (ii) we performed temperatures of ∼1000 K higher than in previous studies, in order to match the recent estimates of the temperature in the D″ region. According to the positive Clapeyron slope of the Pv to PPv transition (Oganov and Ono 2004), higher temperatures favor higher transition pressures; and (iii) in contrast with most previous studies, we corrected sample pressures for the high temperatures encountered by taking into account the effect of thermal pressure. It yields pressure differences of more than 10 GPa (Table 2). We then performed Fe K-edge XANES, which is sensitive to local structure and degree of oxidation. We discuss here results obtained for the Codera composition. The XANES vary significantly with pressure from the room P and T orthopyroxene phase to the Pv and PPv mixture, and finally to the amorphous phase recovered after decompression (Fig. 5). As pressure increases from 100 to 180 GPa, the intensity of features denoted B, C and D increases, B and C shift towards higher energies, whereas A and D do not. In general, the effect of compression is observed through a gradual shift of features towards higher energies, with more rapid shifts further from the absorption edge. Our data therefore suggests that the local structure does not evolve following a simple compression, but that more complex structural changes are occurring as the Fe environment is modified

Author's personal copy D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

93

Fig. 4. Fraction of post-perovskite phase as a function of pressure for different compositions. The vertical black line corresponds to the Pv to PPv transition for MgSiO3 (Murakami et al., 2004; Oganov and Ono, 2004). Pink, light blue and orange squares correspond to MgSiO3-based compositions with additions of 15% FeAlO3 (NishioHamane et al., 2007a,b), pyrope (Tateno et al., 2005), or 50% FeSiO3 (Mao et al., 2004; Tateno et al., 2007), respectively. Note that the persistence of Pv cannot be explained by the presence of a thermal gradient. On the contrary, the positive Clapeyron slope for the Pv to PPv transformation should induce higher PPv content in the colder zone close to the diamonds.

the Mg dodecahedral site. This question remains open for PPv. For such dense silicate phase, it is generally accepted that (i) larger Fe2+ can only enter the Mg-site in agreement with a recent structural determinations (Yamanaka et al., 2008) and (ii) smaller Al size makes this element more suitable to the Si octahedral site in case of Fe3+Al3+O3 substitution. Therefore, all Fe-cations should show similar local structure in the PPv structure. To confirm this, we performed theoretical Fig. 3. (A) XANES mapping performed on the sample recovered after the run at the highest P–T reached in this study. The pixel size is 2.5 × 2.5 µm. Blue and orange areas correspond to Re-gasket and sample, respectively. The color-scale indicates the height of Fe K-edge absorption jump (directly proportional to the Fe-content). The small variation in Fe K-edge shows that the Fe-content over the sample area remains spatially constant +/−5%. (B) Pre-edge normalized XANES spectra from the orange area in (A). The similarity of the spectral features evidences homogeneity, not only in the Fe-content, but also in the Fe-local structure over the whole sample. The few spectra that look slightly different in shape (mainly the 4th and 9th from the top) correspond to sample positions where the X-ray spot was partially overlapping with the Re-gasket.

from Pv to PPv. Concerning the Fe-local structure in Pv, previous works based on XAFS (Farges et al., 1995) or electron microprobe (Bolfan-Casanova et al., 2003) indicate that both Fe2+ and Fe3+ enter

Table 2 Experimental details for each run. The heating duration was more than 30 min for all runs. Sample

P300Ka (GPa)

Tmax (K)

PTmaxb (GPa)

PPv content (%)

KLB KLB KLB Coderac Coderac Coderac Coderac MORB MORB MORB MORB

60 122 147 94 130 158 172 94 145 150 150

2000 2000 2200 3250 3600 3500 3150 2050 2200 3100 2350

67 129 156 106 142 170 184 102 154 162 159

0 24 50 7 36 75 95 0 58 63 66

a b c

Nominal pressures measured at 300 K after laser heating. Pressures at high temperatures (see experimental methods section). When drawing Fig. 6, we consider a mean temperature of 3300 K.

Fig. 5. X-ray absorption spectra recorded as a function of the synthesis pressure (black lines) or calculated (color lines) at the Fe K-edge for the Codera sample. The XANES spectra of the enstatite starting material and amorphous quenched material are shown for comparison. This is associated to changes in the Fe local structure when the Fe environment changes progressively from Pv to PPv with increasing pressure from 106 to 184 GPa. Explanations for calculation procedures are given in the text.

Author's personal copy 94

D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

ab initio calculations for Fe K-edge XANES in Pv and PPv compounds, using the FEFF8 code (Rehr et al., 1992), and compared them with our experimental data (Fig. 5). The great similarity between both results confirms that Fe is likely to be localized in substitution to Mg in both structures. Fe is located in 12-fold (Fiquet et al., 2000), or 8-fold (Murakami et al., 2004) coordinated sites in Pv, or PPv structures, respectively, which can explain the differences in XANES features. The examination of the XANES patterns reveals that the energy position of the absorption jump, feature A, remains unchanged with increasing pressure from 106 to 184 GPa (Fig. 5), although the Fe fraction in a PPv environment increases from virtually zero to 86% in this pressure range. This means that the Fe oxidation state is very similar in both structures. Indeed, an experiment performed at 40 GPa for Al-bearing and Al-free (Mg,Fe)SiO3 starting materials shows that the absorption jump recorded for the former sample displays a shift of about +1 eV compared to the latter sample. This shift is associated to higher Fe3+ contents in the Al-bearing Pv (McCammon, 1997). Our observation is compatible with recent results showing that the coupled substitution of Fe3+ + Al3+ = Mg2+ + Si4+ is efficient in PPv as well as in very high-pressure Pv (Sinmyo et al., 2006). Thus, the effect of Fe on the Pv to PPv phase transition is likely to be significantly different in Al-free and Al-bearing systems. In order to retrieve the iron distribution coefficient between Pv and PPv polymorphs, KPv/PPv , we need the information about the XANES Fe spectra of the isolated Pv and PPv phases. At 106 GPa, the XRD indicates that only 7% of PPv is present. We assume that a negligible amount of Fe enters the PPv at 106 GPa, an assumption that is well verified by our results a posteriori. Thus, the XANES spectrum recorded at 106 GPa is used as a reference for Fe located in the Pv phase (hereafter noted Xanes-Pv). The XANES spectrum recorded at the highest pressure of 184 GPa is likely to be close to that of pure PPv, but since some Pv remains visible on the diffraction pattern at this pressure (Figs. 1 and 2), we leave as unknown the shape of the XANES for Fe located in the PPv phase (noted Xanes-PPv). The second unknown is the Fe-partitioning coefficient between the two phases, KPv/PPv . Both unknowns are solved by combining experimental Fe XANES spectra (Xanes-Mix) recorded at 142 GPa and 184 GPa, with the Pv (XPv) and PPv (XPPv) phase fractions determined from X-ray diffraction. We first simulated Xanes-PPv using the experimental XANES spectra and following the expression of mass conservation:

Xanes2PPv =

evidencing a clear Fe-depletion in the PPv phase compared to the coexisting Pv phase. Our new measurements do not provide direct information on the behaviour of Al. However, the observations (i) that Al stabilizes the Pv phase at the expense of the PPv in both Fe-free (Tateno et al., 2005) and Fe-bearing (Nishio-Hamane et al., 2007a,b) systems, and (ii) given that the coupled substitution of Al3+ and Fe3+ occurs in PPv as it does in Pv indicate that the Al partitioning coefficient KPv/PPv is signifAl icantly larger than unity. As a first approximation, we can consider a 1 to 1 coupling of Fe and Al atoms in both Pv and PPv structures, and thus KPv/PPv = KPv/PPv = 4.2. Al Fe

4. Implications We can now draw the binary loop to represent the coexistence of Pv and PPv as a function of the bulk (Fe,Al) content and as a function of pressure (Fig. 6). In this diagram we report the composition of coexisting Pv and PPv phases retrieved from our measurements at about 3300 K as well as those reported for PPv in pyrolitic composition (Murakami et al., 2005). We also report the transition pressure recorded for pure MgSiO3 at a somewhat lower temperature (Murakami et al., 2004; Oganov and Ono, 2004). Based on this diagram, we calculate that a mixture of 64% of PPv containing 6.0% Fe with 36% of Pv containing 21.1% Fe could be stable at a pressure of 135 GPa corresponding to the CMB for a (Fe,Al)0.114–MgSiO3 composition typical of pyrolitic material (Ringwood, 1975). However, we need to consider the presence of (Mg,Fe)O ferropericlase (Fp), as it can incorporate significant amounts of Fe in its structure, and also of CaSiO3 perovskite (Ca-Pv) in lower mantle material. In pyrolite, the bulk Fe / (Mg + Fe) ratio and the Ca-Pv mole fraction amount to 0.114 and 0.05, respectively. Concerning the Fe partitioning between Pv and Fp, we used KFp/Pv = 2 in agreement with Fe the available reports for Al-bearing pyrolitic-type compositions (Murakami et al., 2005; Kesson et al., 1998). The mole fraction of Fp in pyrolite is about 25%. Due to its high affinity for iron, the effect of Fp

ðXanes2Mix−CPv *XPv *Xanes2PvÞ ðCPPv *XPPv Þ

in which the Fe-concentration in Pv and PPv, i.e., CPv and CPPv, can be derived from the expression of KPv/PPv , with the knowledge of the bulk Fe iron content and the Pv and PPv phase fractions. The KPv/PPv partition Fe coefficient is varied until the simulated Xanes-PPv is identical for the two experimental pressures. There is a unique solution for KPv/PPv and Fe Xanes-PPv. In a second step, we can improve our experimental values of XPv and XPPv by adjusting the Fe-contents in Pv and PPv phases during the Rietveld refinement, in agreement with the refined KPv/PPv Fe value. A few iterations were performed before convergence. In the final step, we recalculated linear combination of Xanes-Pv and XanesPPv, at 142 GPa for example (Fig. 5), and compared it to the experimental spectra, in order to confirm the results. We obtain a unique solution for KPv/PPv of 4.2 (+/−0.5). This value is Fe in very good agreement with that of an early report obtained in peridotitic composition (Murakami et al., 2005), although this partition coefficient was obtained indirectly by dividing KPv/Fp by KPPv/Fp , where Fe Fe Fp stands for ferropericlase. For the Codera-enstatite sample with a total Fe-content of 0.194 pfu, the results of our calculation points out to PPv fractions of 36% and 95%, at 142 and 184 GPa, associated with 14% and 88% of the total Fe-content inserted in the PPv phase at these pressures, PPv respectively. It corresponds to (XPv Fe = 0.077 and XFe = 0.259) and Pv PPv (XFe = 0.179 and XFe = 0.478) for 142 and 184 GPa, respectively,

Fig. 6. Binary loop calculated for the MgSiO3–FeAlO3 system on the basis of our tandem X-ray diffraction and X-ray absorption measurements. The dark ellipse represents the transition pressure for MgSiO3 composition (Murakami et al., 2004; Oganov and Ono, 2004). Information reported in green and red correspond to the Pv and PPv compositions retrieved at 142 GPa and 184 GPa, respectively. The diagram shows that a (Fe,Al)-depleted post-perovskite phase coexists with perovskite (blue dots) for a (FeAl)0.11(MgSi)0.89O3 composition (grey zone), typical of pyrolitic mantle at a pressure of 135 GPa.

Author's personal copy D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

95

Fig. 7. (A) Phase fractions and (B) Fe number plotted as a function of pressure for the four phases expected in a pyrolitic-type composition in the D″ region up to the core–mantle boundary pressure of 135 GPa. Fp and Ca-Pv stand for (Mg,Fe)O ferropericlase and CaSiO3 perovskite, respectively. This graph represents a lower bound of Pv to PPv loop width. Indeed, coexistence of Pv and PPv phases extends to higher pressures if (i) temperatures higher than 3300 K are found at the CMB, (ii) Fp-content in the D″-layer is lower than 25%, and (iii) Al-effect remains significant, which are both likely (see text).

is to deplete the Pv and PPv mixture in iron, and therefore to reduce the Pv content when Pv and PPv coexist (Fig. 7). For pyrolite, we calculate that the mixture of PPv and Pv phases should coexist for pressures between 118 and 131 GPa, which corresponds to heights of 300 km and 70 km above the CMB, respectively. This pressure range is slightly larger, 13 GPa at 3300 K to be compared with 6 GPa at 2500 K, in the previous study of Ohta et al., 2008. The temperature difference of 700 K between the two studies could explain the discrepancy. There are uncertainties, however, that need to be considered: (i) here, we model coexistence of Pv and PPv for a constant temperature of 3300 K (Fig. 7). However, it is well accepted that the D″ region undergoes a relatively steep temperature gradient, yielding CMB temperature potentially up to 4400 K (Wookey et al., 2005; Hernlund et al., 2005). Since higher temperatures favor the Pv phase, pure PPv is unlikely to be found at the CMB. For the olivine-wadsleyite phase transition, for example, increasing temperature by 400 K shifts the transition to higher pressure and induces a 50% increase in loop width (Frost and Dolejs, 2007). (ii) Several reports devoted to modelling the Earth's dynamics or reproducing the seismic velocity profiles argue that the pyrolitic Mg/Si ratio of 1.2 may be too high for the D″ region and propose a SiO2 enrichment in the lowermost mantle (Kellogg et al., 1999; Matas et al., 2007; Samuel et al., 2006). If the Fp fraction decreases, then the Pv fraction in the D″-layer should be higher, because Fe favors the formation of Pv compared to PPv. At 3300 K, we calculate that a Fp fraction below 19% would yield to coexistence of Pv and PPv over the whole D" region up to 135 GPa; (iii) it is important to note that while Fe can partition preferentially into Fp and facilitate the disappearance of Pv, it is not the case for Al that remains preferentially located in the Pv phase in absence of secondary Al-bearing phases (Murakami et al., 2005; Nishio-Hamane et al., 2007a,b). In other words, the presence of Fp should reduce less the Pv to PPv loop width in an Al-bearing system compared to an Alfree system. (iv) For MORB materials, the mineralogy is affected by the presence of other phases, such as the “calcium-ferrite type” or the “New Aluminous” phases, which can absorb much of the Al. On the other hand, Fp is absent due to the high SiO2-content in MORBs, which

makes the role of Fe more important for the broadness of the twophase region. The occurrence of mixed PPv and Pv phases down to the CMB has severe geophysical consequences. The fact that the main PPv phase is largely depleted in Fe is likely to enhance the radiative thermal conductivity in this mantle region. This could induce higher temperatures in the lowermost part of the mantle, which in turn would increase the fraction of Pv phase due to the positive Clapeyron slope for the Pv to PPv phase transition. The increase of FeO content in Fp, up to ∼ 0.30 mol% (Fig. 7), implies a definite increase of the FeO activity at the CMB vicinity which needs to be taken into account for the chemical equilibrium at the CMB. The FeO activity in the outer core could be severely increased by this effect. If some FeO entered the outer core with a concomitant loss of FeO in the lowermost mantle, the dynamical properties of the D″ layer could be strongly modified. On the other hand, the occurrence of such a binary loop of Pv plus PPv (Fig. 6) offers the opportunity for a large variability of their phase fractions. Any heterogeneity in mantle temperature or Fe/Al composition is expected to induce noticeable changes in the Pv and PPv relative contents. This effect could largely amplify the seismic signature usually associated to such type of heterogeneities and help explain the peculiar seismic observation in the D″ region.

Acknowledgments This paper is dedicated to the memory of Matéo Muñoz. Assistance and helpful comments from G. Garbarino, T. Hammouda, J. Jacobs, P. van der Linden were greatly appreciated. Funding of this project is in part provided by the European Commission through the Marie Curie Research Training Network "c2c" (Contract No. MRTN-CT-2006035957).

References Akber-Knutson, S., Bukowinsky, M.S., 2004. The energetics of aluminum solubility into MgSiO3 perovskite at lower mantle conditions. Earth Planet. Sci. Lett. 220, 317–330.

Author's personal copy 96

D. Andrault et al. / Earth and Planetary Science Letters 293 (2010) 90–96

Andrault, D., Fiquet, G., Itié, J.P., Richet, P., Gillet, P., Häusermann, D., Hanfland, M., 1998. Thermal pressure in a laser-heated diamond-anvil cell: an x-ray diffraction study. Eur. J. Mineral. 10, 931–940. Auzende, A.-L., Badro, J., Ryerson, F.J., Weber, P.K., Fallon, T.J., Addab, A., Siebert, J., Fiquet, G., 2008. Element partitioning between magnesium silicate perovskite and ferropericlase: new insights into bulk lower-mantle geochemistry. Earth Planet. Sci. Lett. 269, 164–174. Bolfan-Casanova, N., Keppler, H., Rubie, D.C., 2003. Water partitioning at 660 km depth and evidence for very low water solubility in magnesium silicate perovskite. Geophys. Res. Lett. 30. doi:10.1029/2003GRL017182. Caracas, R., Cohen, R.E., 2005. Prediction of a new phase transition in Al2O3 at high pressures. Geophys. Res. Lett. 32, L06303. Farges, F., Guyot, F., Andrault, D., Wang, Y., 1995. Local structure around Fe in Mg0.9Fe0.1SiO3: an X-ray absorption spectroscopy study at Fe–K edge. Eur. J. Mineral. 6, 303–312. Fiquet, G., Dewaele, A., Andrault, D., Kunz, M., Le Bihan, T., 2000. Thermoelastic properties and crystal structure of MgSiO3 perovskite at lower mantle pressure and temperature conditions. Geophys. Res. Lett. 27, 21–24. Frost, D.J., Dolejs, D., 2007. Experimental determination of the effect of H2O on the 410km seismic discontinuity. Earth Planet. Sci. Lett. 256 (1–2), 182–195. Guignot, N., Andrault, D., Morard, G., Bolfan-Casanova, N., Mezouar, M., 2007. Thermoelastic properties of the post-perovskite phase MgSiO3 determined experimentally at core–mantle boundary P–T conditions. Earth Planet. Sci. Lett. 256 (1–2), 162–168. Hernlund, J.W., Thomas, C., Tackley, P.J., 2005. A doubling of the post-perovskite phase boundary and structure of the Earth's lowermost mantle. Nature 434, 882–886. Hirose, K., Fei, Y., Ma, P., Mao, H.K., 1999. The fate of subducted basaltic crust in the Earth's lower mantle. Nature 397, 53–56. Hirose, K., Takafuji, N., Fujino, K., Shieh, S.R., Duffy, T.S., 2008. Iron partitioning between perovskite and post-perovskite: a transmission electron microscope study. Am. Mineral. 93, 1678–1681. Jamieson, J.C., Fritz, J.N., Manghnani, M.H., 1982. Pressure measurement at high temperature in x-ray diffraction studies: gold as a primary standard. In: Akimoto, S., Manghnani, M.H. (Eds.), High Pressure Research in Geophysics, pp. 27–48. Riedel, Boston. Kellogg, L.H., Hager, B.H., van der Hilst, R.D., 1999. Compositional stratification in the deep mantle. Science 283, 1881–1884. Kesson, S.E., Fitz Gerald, J.D., Shelley, J.M., 1998. Mineralogy and dynamics of a pyrolite lower mantle. Nature 393, 252–255. Kobayashi, Y., Kondo, T., Ohtani, E., Hirao, N., Miyajima, N., Yagi, T., Nagase, T., Kikegawa, T., 2005. Fe–Mg partitioning between (Mg, Fe)SiO3 post-perovskite, perovskite and magnesiowüstite in the Earth's lower mantle. Geophys. Res. Lett. 32, L19301. Lay, T., Hernlund, J.W., Buffett, B.A., 2008. Core–mantle boundary heat flow. Nature 1, 25–32. Mao, W.L., Shen, G., Prakapenka, V.B., Meng, Y., Campbell, A.J., Heinz, D.L., Shu, J., Hemley, R.J., Mao, H.K., 2004. Ferromagnesian post-perovskite silicates in the D″ layer of the Earth. Proc. Natl. Acad. Sci. U.S.A. 101 (45), 15867–15869. Matas, J., Bass, J.D., Ricard, Y., Mattern, E., Bukowinsky, M.S., 2007. On the bulk composition of the lower mantle: predictions and limitations from generalized inversion of radial seismic profiles. Geophys. J. Int. 170, 764–780. McCammon, C.A., 1997. Perovskite as a possible sink for ferric iron in the lower mantle. Nature 387, 694–696. Mezouar, M., Crichton, W.A., Bauchau, S., Thurel, F., Witsch, H., Torrecillas, F., Blattmann, S., Marion, P., Dabin, Y., Chavanne, J., Hignette, O., Morawe, C., Borel, C., 2005. Development of a new state-of-the-art beamline optimized for monochromatic single-crystal and powder X-ray diffraction under extreme conditions at the ESRF. J. Synchrotron Radiat. 12, 559–664. Muñoz, M., De Andrade, V., Vidal, O., Lewin, E., Pascarelli, S., Susini, J., 2006. Redox and speciation micro-mapping using dispersive X-ray absorption spectroscopy: application to iron in chlorite mineral of a metamorphic rock thin section. Geochem. Geophys. Geosyst. 7, Q11020. Muñoz, M., Pascarelli, S., Aquilanti, G., Narygina, O., Kurnosov, A., Dubrovinsky, L., 2008. Hyperspectral micro-XANES mapping in the diamond-anvil cell: analytical procedure applied to the decomposition of (Mg,Fe)-ringwoodite at the upper/ lower mantle boundary. High. Press. Res. 28, 665–673. Murakami, M., Hirose, K., Kawamura, K., Sata, N., Ohishi, Y., 2004. Post-perovskite phase transition in MgSiO3. Science 304, 855–858.

Murakami, M., Hirose, K., Sata, N., Ohishi, Y., 2005. Post-perovskite phase transition and mineral chemistry in the pyrolitic lowermost mantle. Geophys. Res. Lett. 32, L03304. Nishio-Hamane, D., Fujino, K., Seto, Y., Nagai, T., 2007a. Effect of the incorporation of FeAlO3 into MgSiO3 perovskite on the post-perovskite transition. Geophys. Res. Lett. 34, L12307. Nishio-Hamane, D., Seto, Y., Nagai, T., Fujino, K., 2007b. Ferric iron and aluminum partitioning between MgSiO3 and CaSiO3 perovskites under oxidizing conditions. J. Mineral. Petrol. Sci. 102, 291–297. Oganov, A.R., Ono, S., 2004. Theoretical and experimental evidence for a postperovskite phase of MgSiO3 in the Earth's D″ layer. Nature 430, 445–448. Ohta, K., Hirose, K., Lay, T., Sata, N., Ohishi, Y., 2008. Phase transitions in pyrolite and MORB at lowermost mantle conditions: implications for a MORB-rich pile above the core–mantle boundary. Earth Planet. Sci. Lett. 267, 107–117. Ono, S., Oganov, A.R., 2005. In situ observations of phase transition between perovskite and CaIrO3-type phase in MgSiO3 and pyrolitic mantle composition. Earth Planet. Sci. Lett. 236, 914–932. Pascarelli, S., Mathon, O., Muñoz, M., Mairs, T., Susini, J., 2006. Energy-dispersive absorption spectroscopy for hard-X-ray micro-XAS applications. J. Synchrotron Radiat. 13, 351–358. Rehr, J.J., Zabinsky, Z.I., Albers, R.C., 1992. High-order multiple scattering calculations of x-ray-absorption fine structure. Phys. Rev. Lett. 69, 3397–3400. Ringwood, A.E., 1975. Pyrolite and the chondritic earth model. In: GrawHill, M. (Ed.), International series in the Earth's and Planetary Sciences, pp. 189–194. Samuel, H., Farnetani, C.G., Andrault, D., 2006. Chemically denser material in the lowermost mantle: origin and induced seismic velocity anomalies. In: Bass, J.D., van der Hilst, R.D., Trampert, J. (Eds.), Structure, Evolution and Composition of the Earth's Mantle. Geophys. Monogr., vol. 160. AGU, Washington DC, pp. 101–116. Schultz, E., Mezouar, M., Crichton, W., Bauchau, S., Blattmann, S., Andrault, D., Fiquet, G., Boehler, R., Rambert, N., Sitaud, B., Loubeyre, P., 2005. Double-sided laser heating system for in situ high pressure-high temperature monochromatic x-ray diffraction at the ESRF. High Press. Res. 25 (1), 71–83. Sha, C.-S., Basset, W.A., Shim, S.H., 2004. Rhenium, an in situ pressure calibrant for internally heated diamond anvil cells. Rev. Sci. Instrum. 75 (7), 2409. Shieh, S.R., Duffy, T.S., Kubo, A., Shen, G.Y., Prakapenka, V.B., Sata, N., Hirose, K., Ohishi, Y., 2006. Equation of state of the postperovskite phase synthesized from a natural (Mg, Fe)SiO3 orthopyroxene. Proc. Nat. Acad. Sci. 103 (9), 3039–3043. Sidorin, I., Gurnis, M., Helmberger, D.V., 1999. Evidence for a ubiquitous seismic discontinuity at the base of the mantle. Science 286, 1326–1331. Sinmyo, R., Hirose, K., O'Neill, H.S.C., Okunishi, E., 2006. Ferric iron in Al-bearing postperovskite. Geophys. Res. Lett. 33, L12S13. Stackhouse, S., Brodholt, J.P., Dobson, D.P., Price, G.D., 2006. Electronic spin transitions and the seismic properties of ferrous ironbearing MgSiO3 post-perovskite. Geophys. Res. Lett. 33, L12S03. Tateno, S., Hirose, K., Sata, N., Ohishi, Y., 2005. Phase relations in Mg3Al2Si3O12 to 180 GPa: effect of Al on post-perovskite phase transition. Geophys. Res. Lett. 32, L15306. Tateno, S., Hirose, K., Sata, N., Ohishi, Y., 2007. Solubility of FeO in (Mg, Fe)SiO3 perovskite and the post-perovskite phase transition. Phys. Earth Planet. In. 160, 319–325. Tsuchiya, J., Tsuchiya, T., 2008. Postperovskite phase equilibria in the MgSiO3–Al2O3 system. Proc. Natl. Acad. Sci. U.S.A. 105 (49), 19160–19164. van der Hilst, R.D., de Hoop, M.V., Wang, P., Shim, S.H., Ma, P., Tenorio, L., 2007. Seismostratigraphy and thermal structure of Earth's core–mantle boundary region. Science 315 (5820), 1813–1817. Williams, Q., Garnero, E.J., 1996. Seismic evidence for partial melt at the base of Earth's mantle. Science 273, 1528–1530. Wookey, J., Stackhouse, S., Kendall, J.M., Brodholt, J.P., Price, G.D., 2005. Efficacy of postperovskite as an explanation for lowermost mantle seismic properties. Nature 438, 1004–1007. Wysession, M.E., Langenhorst, A., Fouch, M.J., Fischer, K.M., Al-Eqabi, G.I., Shore, P.J., Clarke, T.J., 1999. Lateral variations in compressional/shear velocities at the base of the mantle. Science 284, 120–125. Yamanaka, T., Mao, H.K., Ganesh, P., Shulenburger, L., Cohen, R.E., Meng, Y., Prakapenka, V.B., Mao, H.K., Hemley, R.J., 2008. Redetermination of post-perovskite structure of (Mg,Fe)SiO3, in: AGU meeting, San-Francisco. Zhang, F., Oganov, A.R., 2006. Mechanisms of Al3+ insertion in the MgSiO3 postperovskite at high pressures. Earth Planet. Sci. Lett. 248, 69–76.