Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

# Incorporation mechanism of Fe and Al into bridgmanite in a subducting mid-ocean ridge basalt and its crystal chemistry

## Abstract

The compositional difference between subducting slabs and their surrounding lower-mantle can yield the difference in incorporation mechanism of Fe and Al into bridgmanite between both regions, which should cause heterogeneity in physical properties and rheology of the lower mantle. However, the precise cation-distribution has not been examined in bridgmanites with Fe- and Al-contents expected in a mid-ocean ridge basalt component of subducting slabs. Here we report on Mg0.662Fe0.338Si0.662Al0.338O3 bridgmanite single-crystal characterized by a combination of single-crystal X-ray diffraction, synchrotron 57Fe-Mössbauer spectroscopy and electron probe microanalysis. We find that the charge-coupled substitution AMg2+  + BSi4+  ↔ AFe3+(high-spin) + BAl3+ is predominant in the incorporation of Fe and Al into the practically eightfold-coordinated A-site and the sixfold-coordinated B-site in bridgmanite structure. The incorporation of both cations via this substitution enhances the structural distortion due to the tilting of BO6 octahedra, yielding the unusual expansion of mean <A–O> bond-length due to flexibility of A–O bonds for the structural distortion, in contrast to mean <B–O> bond-length depending reasonably on the ionic radius effect. Moreover, we imply the phase-transition behavior and the elasticity of bridgmanite in slabs subducting into deeper parts of the lower mantle, in terms of the relative compressibility of AO12 (practically AO8) and BO6 polyhedra.

## Introduction

Bridgmanite, with an approximate composition of MgSiO3 and the orthorhombic perovskite-type structure (space group Pbnm), is believed to be the most dominant constituent of the Earth’s lower mantle. Physical and crystal-structural properties of bridgmanite and their pressure-, temperature- and chemical-dependence provide essential information for detailed understanding of the lower mantle viewed from mineralogical aspects. The crystal structure has the two cation sites, the larger eightfold (nominally 12-fold) coordinated A-site and the smaller sixfold coordinated B-site, consisting of a network of corner-linked BO6 octahedra with the A-site atoms located at the centers of cavities in the network and being distorted largely from the ideal cubic structure with Pm$$\overline{3 }$$m symmetry owing to the tilting of BO6 octahedra (Fig. 1). In the end-member MgSiO3 bridgmanite, the A and B sites are occupied only by Mg and Si, respectively.

The incorporation of Fe and Al, important major elements in the mantle composition after Mg and Si, into the two cation sites can affect the physical properties such as the electric conductivity, thermal conductivity, elasticity and rheology of the lower mantle, together with the structural stability of bridgmanite itself. Because of such importance, the effect of Fe and/or Al incorporation on the physical and structural properties of bridgmanite has extensively studied1,2,3,4,5,6,7,8,9,10. The valence- and spin-states of Fe, its site-distribution, and the creation of cation- and/or oxygen-vacancies depending on these are especially relevant issues associated with the incorporation mechanism of Fe into bridgmanite, because these can strongly influence the electrical conductivity of the lower mantle1, 9, 11,12,13. The creation of oxygen vacancies14,15,16,17,18,19,20 and the incorporation manner of Fe can also be controlled by the incorporation of Al. In particular, the effect of trivalent Fe3+ incorporation is of interest in view of cation partitioning among lower-mantle minerals21. The major conclusions on the role of Fe3+ and/or Al reported by the previous studies2, 6, 7, 21,22,23,24,25,26 using Fe- and/or Al-bearing bridgmanites are as follows: (i) Fe3+ can be distributed preferentially to bridgmanite at lower-mantle pressures; (ii) the presence of Al increases Fe3+ contents in bridgmanite; (iii) oxygen vacancies (VO) may be created through the substitutions 1/2O2−  + BSi4+  ↔ 1/2VO + BFe3+ and/or 1/2O2−  + BSi4+  ↔ 1/2VO + BAl3+, where the left superscripts in the chemical formulae represent the occupied sites; (iv) Fe3+ can occupy both A and B sites. In particular, Catalli et al.6 reported from in-situ synchrotron Mössbauer spectroscopy that both Fe3+ and Al are distributed evenly between A and B sites at high pressures, which is accompanied by the high-spin (HS) to low-spin (LS) transition of Fe3+. Hummer and Fei7 also reported, from Mössbauer spectroscopy, the even site-mixing of both cations in the quenched sample synthesized at 1973–2073 K and 25 GPa. On the other hand, Vanpeteghem et al.3 reported from the single-crystal X-ray diffraction study of several (Fe, Al)-bearing bridgmanites with different Fe and Al contents that Fe3+ occupies exclusively A-site via the charge-coupled substitution AMg2+  + BSi4+  ↔ AFe3+  + BAl3+. They furthermore reported that when the Fe content exceeds the Al content, the substitution AMg2+  ↔ AFe2+ occurs for the extra Fe content. Thus, there has been the large discrepancy in incorporation mechanisms of Fe3+ and Al between studies. Acquisition of detailed total knowledge of cation distribution, iron valence, and vacancy is necessary for the reliable determination of the incorporation mechanism, but each of the previous studies has not examined them from all of single-crystal X-ray diffraction, Mössbauer spectroscopy and chemical analysis.

Here we characterize the bridgmanite single-crystal with the Fe- and Al-contents expected in a mid-ocean ridge basalt (MORB) component of subducting slabs, by a combination of these three techniques. Bridgmanite formed from the MORB composition contains larger amounts of Fe and Al (~ 0.35 per formula unit for both)27 than their amounts (~ 0.05 per formula unit for both)27 of bridgmanite expected in a pyrolytic28 lower-mantle. This difference in bridgmanite compositions can yield the difference in incorporation mechanism of Fe and Al into the crystal structure between subducting slabs and their surrounding lower-mantle. This should cause heterogeneity in physical properties and rheology of the lower mantle. Elucidating crystal chemistry of bridgmanite formed from the MORB composition is thus a key to solve controversial issues in the lower mantle such as anti-correlated seismic velocity anomalies observed in large low shear velocity provinces (LLSVPs)10. From this viewpoint, the precise crystal-chemical examination employing a single crystal is quite significant for detailed understanding of lower-mantle dynamics. In particular, the present study includes the first report on single-crystal structure of bridgmanite with the Fe- and Al-contents expected in MORB. In this paper, we discuss the incorporation mechanism of Fe and Al into bridgmanite in MORB and its structural variation with the incorporation of both cations, and provide earth-scientific implications.

## Methods

### Single-crystal synthesis under high pressure and chemical analyses

Single crystals of bridgmanite were synthesized at 28 GPa and 1873 K using a 5000-ton Kawai-type high-pressure apparatus installed at the Institute for Planetary Materials, Okayama University. The procedure and technique of the experiment are essentially the same as those described in our previous study29 as follows. A 10 mm regular octahedron of a sintered MgO containing 5% of Cr2O3 was employed as a pressure-transmitting medium. The starting materials were the special grade reagents of powdered MgO, SiO2, Al2O3 and Fe2O3, and mixed in a cation ratio of Mg:Fe:Si:Al = 0.65:0.35:0.65:0.35, which is very close to that27 reported for bridgmanite formed from the MORB composition. LaCrO3 surrounded by ZrO2 thermal insulator was used as a furnace material. The powder mixture was placed in a Pt capsule, which was electrically insulated from the furnace by a MgO spacer. This cell assembly was set in the anvil assembly of tungsten carbide cubes with truncated edge lengths of 3 mm, and then was compressed up to a target pressure of 28 GPa at room temperature. The temperature was then raised to a target temperature of 1873 K at a rate of 100 K/min. The temperature was controlled with a W97%Re3%-W75%Re25% thermocouple, whose junction was put at the midpoint of the outer surface of the Pt capsule. No correction was made for the pressure effect on emf. After being kept under a desired condition (28 GPa, 1873 K) for 2 h, the product was quenched by shutting off the electric power supply. The pressure was released slowly and the product was recovered at ambient condition. Numerical single-crystals of bridgmanite with a size of about 100–200 μm were found in the recovered sample. The color of the crystals is reddish-brown, suggesting the incorporation of Fe ions into the crystals. Compositions of the single crystals (Table 1) were determined by means of a JEOL JXA-8800M electron probe microanalyzer (EPMA). No contamination from the cell assembly materials into the single crystals was detected from qualitative analyses by the EPMA.

### Synchrotron 57Fe-Mössbauer spectroscopy

Energy-domain synchrotron 57Fe-Mössbauer spectroscopy measurements at room temperature using a nuclear Bragg monochromator were conducted to evaluate Fe3+/ΣFe ratio of the present bridgmanite at the BL10XU beamline of SPring-8 (Ref.30). The sample was irradiated by the X-ray beam tuned at 14.4 keV from a high heat-load Si(111) double-crystal monochromator. The transmitted X-ray through the sample was monochromatized to around the nuclear resonance energy of 57Fe by a high resolution monochromator with a bandpass of about 4 meV, which consists of asymmetric Si(511) and symmetric Si(975) channel-cut crystals. The nuclear monochromator employs a single-line pure nuclear Bragg reflection 333 from an oscillating 57FeBO3 single-crystal near Néel temperature in the external magnetic field. The bandwidth of the electronically forbidden pure nuclear Bragg reflection was about 15 neV. The source Doppler shift was produced by oscillating the crystal in a sinusoidal velocity mode, which was mounted on a velocity transducer. The absorption spectrum was obtained by counting the intensity of the single-line nuclear Bragg reflection as a function of velocity. The velocity scale was calibrated with respect to a 57Fe-enriched standard metallic iron foil with 3 µm thickness under ambient conditions, and the isomer shift was also referenced to the same standard. The spectrum data were collected with a measurement time of 8.3 h. The MossA software package31 was used for the computational analysis and the spectrum was fitted using a Lorentzian model. The results are shown in Table 2.

### Single-crystal X-ray diffraction intensity measurements and structure refinements

The single-crystal X-ray diffraction intensity measurements, data processing and structure refinements were conducted following essentially the same procedures and techniques as those described in our previous studies32,33,34 as follows. The single crystal with a size of 0.10 × 0.08 × 0.04 mm3 was selected and then mounted on the tip of a glass fiber for the intensity measurements. The measurements were conducted at room temperature (296 K) using a Rigaku AFC-7R four-circle diffractometer with a graphite-monochromatized MoKα radiation (λ = 0.71069 Å) at an operating condition of 60 kV and 250 mA. The unit-cell parameters were determined by the least-squares method from a set of 25 reflections within the range of 44° ≤ 2θ ≤ 46°. The intensity data of a total of 1774 reflections within 2° ≤ 2θ ≤ 100° were collected using the continuous ω–2θ scan mode and corrected for Lorentz-polarization factors and absorption effects (ψ-scan method). After that, the intensity data were averaged in Laue symmetry mmm to give 930 unique reflections. Of these, unique reflections with $$\left|{F}_{\mathrm{o}}\right|\le 3{\upsigma } \left(\left|{F}_{\mathrm{o}}\right|\right)$$ were eliminated. Even if unique reflections had intensities of $$\left|{F}_{\mathrm{o}}\right|>3{\upsigma } \left(\left|{F}_{\mathrm{o}}\right|\right)$$ after averaging, those averaged from data set of equivalent reflections including reflection(s) with $$\left|{F}_{\mathrm{o}}\right|\le 3{\upsigma } \left(\left|{F}_{\mathrm{o}}\right|\right)$$ were also discarded since these reflections were potentially affected by multiple scattering. Moreover, unique reflections with sinθ/λ < 0.26 Å−1 were eliminated to reduce secondary extinction effects and to avoid dependence on atomic charge as far as possible in the choice of atomic scattering factors. Finally, 640 unique reflections were used in the present refinements. Internal residuals of the equivalent reflections (Rint) was 0.0131.

The structure refinements were carried out by minimizing the function Σw(FoFc)2 using a full matrix least-squares program RADY35. Scattering factors of Mg2+, Al3+, Si4+, Fe3+, Fe2+ (Table 6.1.1.3 in International Tables for Crystallography36), and O2– (Tokonami37) were used. Anomalous dispersion coefficients for each scattering factor were taken from Table 4.2.6.8 in International Tables for Crystallography36. Several correction models for the secondary extinction effects were attempted during the refinements, and the isotropic correction of Type II38, 39 with a Gaussian particle size distribution model yielded the best fit. The final structure refinement converged smoothly to R = 0.0189 and wR = 0.0146. The summary of crystallographic data, data-collection and refinement parameters is given in Table 3. The refined structural parameters are given in Table 4. The selected interatomic distances are listed in Table 5. Crystallographic Information File (CIF) is deposited in the Cambridge Structural Database (CSD) (Deposition No. 2089819).

## Results and discussion

### Chemical composition, and valence- and spin-states of Fe

Compositions at five points in a crystal measured by the EPMA and the average of them are shown in Table 1. No significant compositional fluctuation is observed among these five measurement points; this shows that the crystal is almost homogeneous in composition. The averaged composition from the EPMA analyses is calculated as the cation ratio Mg:Fe:Si:Al = 0.642(7):0.341(6):0.656(10):0.356(4) assuming O = 3. This composition shows no significant deviation from the mixing composition of the starting materials in the synthetic experiment, indicating that the crystal includes no significant cation- and/or oxygen-vacancies and is well charge-balanced by trivalent Fe3+ ions within the error. Although the preferential occupation of larger Fe3+ for A site and of smaller Al3+ for B site is inferred in terms of ionic radii40 [e.g., HS Fe3+  = 0.645 Å and Al3+  = 0.535 Å in CN (coordination number) = 6], the degree of their distribution between both sites cannot be inferred from the EPMA data alone.

We here show in Fig. 2a the Mössbauer spectrum of the present bridgmanite single-crystal to gain the more detailed knowledge of valence states, spin states and coordination environments of Fe. The Mössbauer spectrum seems to consist of two absorption peaks with different intensities. The coordination environments around A and B sites (the possible occupied sites of Fe) are largely distorted; this should yield quadrupole splitting, as observed by many researchers5,6,7. The doublets measured using a single crystal can be asymmetric because a certain angle is kept between the principal axis of the electric field gradient tensor in the Fe sites and the incident X-ray beam direction. The Mössbauer spectrum of the present bridgmanite single-crystal should thus be interpreted not as a superposition of singlets but as one asymmetric doublet or a superposition of several asymmetric doublets, depending on the differences in electronic states and coordination environments of Fe. The spectrum is well represented by a Lorentzian model assuming one asymmetric doublet, and the residual peak-components are undetectable from the fitting residuals (Fig. 2b). Models with additional doublets were also attempted, but were not able to significantly improve the fitting quality. The final fit, assuming one asymmetric doublet, gives an isomer shift (IS) of 0.40(3) mm/s and a quadrupole splitting (QS) of 0.86(4) mm/s. These values match well with the reference values41, 42 for HS Fe3+ on A site and are also close to those for LS Fe2+ on A site (Table 2). The latter case is however implausible in terms of the charge balance indicated by the EPMA result. Thus, Fe ions in the present sample exclusively occupy A site in trivalent high-spin state, which leads to that Al3+ ions exclusively occupy B site in consideration of the cation ratio indicated by the EPMA result.

To further examine the site distribution of Fe and Al and the presence of vacancies, a preliminary structure-refinement was performed by varying P(AFe3+), P(O1O2−) and P(O2O2−) as valuable occupancy parameters under the following constraints: P(AMg2+) = P(BSi4+) ≡ 0.65 (fix), P(BFe3+) = P(AAl3+) ≡ 0.35 − P(AFe3+), P(BAl3+) ≡ P(AFe3+). The resulting P(AFe3+) was 0.344(2), corresponding to P(BFe3+) = P(AAl3+) = 0.006. The resulting P(O1O2−) and P(O2O2−) were 1.001(5) and 0.999(4), respectively. Another preliminary structure-refinement was also performed by varying P(AFe3+) and P(BAl3+) as valuable occupancy parameters under the following constraints: P(AMg2+) = P(BSi4+) ≡ 0.65 (fix), P(O1O2−) = P(O2O2−) ≡ 1.0 (fix). The resulting P(AFe3+) and P(BAl3+) were 0.347(2) and 0.356(4), respectively. These show that the mixing of Fe and Al between the two cation-sites and the cation- and oxygen-vacancies are undetectable, agreeing well with the EPMA and Mössbauer results. The final refinement was thus conducted by varying P(AFe3+) as the only variable occupancy parameter in a model without any cation- or oxygen-vacancies and without any distribution of Fe into B site or Al into A site, under the following constraints: P(AMg2+) = P(BSi4+) ≡ 1 − P(AFe3+), P(BAl3+) ≡ P(AFe3+). The data provided in Tables 3, 4, 5 are from this final refinement. The final P(AFe3+) is 0.338(3) (Table 4), leading to the cation ratio Mg:Fe:Si:Al = 0.662:0.338:0.662:0.338. This is consistent excellently with the cation ratio from the EPMA.

In ABO3 perovskites with the Pbnm structure, such as CaTiO3 (Ref.43), MgSiO3 (Ref.44,45,46) and CaGeO3 (Ref.33), the great structural distortion due to the tilting of BO6 octahedra yields much longer separations between an A-site atom and four of twelve O atoms surrounding its A-site atom. As shown in Table 5, in the present bridgmanite, the four longer AO separations, which are not potentially involved in chemical bonding, range between 2.9159(9) Å and 3.2292(7) Å. The remaining eight shorter separations range between 2.0057(9) Å and 2.4884(7) Å; the average of these (2.214 Å) agrees better with the expected A–O bond length (2.25 Å) from HS Fe3+ (CN = 8) than the one (2.30 Å) from HS Fe2+ (CN = 8), being concordant with the Mössbauer result. The separations between a B-site atom and six O atoms surrounding its B-site atom are close with each other, in contrast to the case of A-site atom, and range between 1.8213(7) Å and 1.8518(3) Å. The average of these (1.832 Å) agrees well with the expected B–O bond length (1.85 Å).

From the consequences of our observations and analyses described above, we conclude that in the present case, in which relatively large amounts of Fe and Al are equally contained, the following charge-coupled substitution is predominant in the incorporation of both cations into bridgmanite:

$$^{{\text{A}}} {\text{Mg}}^{{{2} + }} + \, ^{{\text{B}}} {\text{Si}}^{{{4} + }} \leftrightarrow \, ^{{\text{A}}} {\text{Fe}}^{{{3} + }} \left( {{\text{HS}}} \right) \, + \,^{{\text{B}}} {\text{Al}}^{{{3} + }}.$$
(1)

Even if there are cation- and/or oxygen-vacancies, divalent Fe2+ ions and mixing of Fe and Al between A and B sites, their amounts/degrees are negligibly small.

### Structural variation with the incorporation of Fe and Al into bridgmanite

In Fig. 3, the unit-cell edge lengths (a, b, c) and volume (V) increase with increasing the ratio (Fe + Al)/(Mg + Fe + Si + Al). This is due to the increase in the mean cation size in the whole of bridgmanite crystal accompanying the incorporation of HS Fe3+ and Al3+ via the charge-coupled substitution (1) although the mean cation size on A site (<rA>) decreases by the substitution AMg2+  → AFe3+(HS).

As observed in Fig. 4a, the increase in <rA> results in (i) a lengthening of the four shortest A–O bond lengths (A–O1v, A–O1, A–O2vi, A–O2v), (ii) the two almost-unchanged intermediate A–O bond lengths (A–O2vii, A–O2viii), and (iii) a shortening of the two longest A–O bond lengths (A–O2ix, A–O2x) and of the four longer potentially non-bonding AO separations (AO1i, AO1ii, AO2iii, AO2iv). On the other hand, as observed in Fig. 4b, the increase in the mean cation size on B site (<rB>) expands all of B–O bond lengths following the ionic radius effect. The expansivities of B–O bond lengths with increasing <rB> are the largest in the longest B–O1 and B–O1xi bond lengths, running in the direction close to the c-axis. This can account for the observation that the expansivities of the unit-cell edge lengths with increasing the ratio (Fe + Al)/(Mg + Fe + Si + Al) are the largest in the c-axis length (Fig. 3).

The symmetrical constraints always request 180° for O1–B–O1xi, O2vii–B–O2xiii, and O2ix–B–O2xii angles, and the remaining twelve O–B–O angles in a BO6 octahedron vary between 87.74(3)° and 92.26(3)°. This shows that the deviation from a regular BO6 octahedron is only slightly larger in the present (Fe3+, Al)-bearing bridgmanite than in the end-member MgSiO3 bridgmanite with the twelve O–B–O angles ranging between 88.49(4)° and 91.51(4)° (Ref.44). The incorporation of Fe and Al via the charge-coupled substitution (1), thus, does not largely change the degree of distortion of BO6 octahedron and only expands the B–O bond lengths. This shows that the response of the structural distortion to the charge-coupled substitution (1) is dominated mainly by the tilting between corner-linked BO6 octahedra as will be described latter. The shortening of the two very weak A–O2 bonds and the four potentially non-bonding AO separations with increasing <rA> observed in Fig. 4a can be a consequence of a flexible response of these bonds/separations to the structural distortion due to the tilting of octahedra. This brings about the decrease in the averaged A–O bond lengths (<A–O>) with increasing <rA> , in contradiction to the ionic radius effect.

We here describe the octahedral tilting using the three tilt angles $${\upphi }_{i}^{+}$$, $${\upphi }_{i}^{-}$$, and $${\upphi }_{i}^{0}$$ (i = x, y, or z) after Yashima and Ali43. The $${\upphi }_{i}^{+}$$, $${\upphi }_{i}^{-}$$, and $${\upphi }_{i}^{0}$$ indicate the in-phase octahedral tilt angle, the out-of-phase octahedral tilt angle, and no octahedral tilting about i-axis (i = x, y, or z), respectively. The x-, y-, and z-axes represent [100]0, [010]0, and [001]0, respectively, where the subscript “0” represents the pseudo-cubic lattice. The tilting system of Pbnm orthorhombic perovskites, represented by bridgmanite, is described by two identical out-of-phase tilting about the [100]0 and [010]0 axes ($${\upphi }_{x}^{-}={\upphi }_{y}^{-}$$) and an in-phase tilting about the [001]0 axis ($${\upphi }_{z}^{+}$$). This tilting system is expressed as $${\upphi }_{x}^{-}{\upphi }_{y}^{-}{\upphi }_{z}^{+}$$ ($${\upphi }_{x}^{-}={\upphi }_{y}^{-}$$), corresponding to aac+ in well-known Glazer’s notation47, 48. The tilt angles have often been defined only from the fractional coordinates of O atoms49, but those will more or less be influenced by the distortions of octahedra themselves. We here calculated the tilt angles via the symmetry-adapted mode approach50, which can completely separate the tilts and distortions of octahedra. In terms of this approach, the tilt angle $$\upphi$$ is given by $$\upphi ={\mathrm{tan}}^{-1}\left(2d\mathrm{^{\prime}}\right)$$, where $$d\mathrm{^{\prime}}$$ is the amplitude of octahedral tilt mode. The $$d\mathrm{^{\prime}}$$ values are converted from the standard supercell-normalized amplitude “As” and “normfactor” calculated using the program ISODISTORT (or the earlier ISODISPLACE)51.

The tilt angles $${\upphi }_{x}^{-}$$ (= $${\upphi }_{y}^{-}$$) and $${\upphi }_{z}^{+}$$ of BO6 octahedra calculated in this way increase with increasing the ratio (Fe + Al)/(Mg + Fe + Si + Al) (Fig. 5a). The $${\upphi }_{x}^{-}$$ and $${\upphi }_{z}^{+}$$ are almost equal in the end-member MgSiO3 bridgmanite, but the increasing rate of $${\upphi }_{x}^{-}$$ with increasing the ratio (Fe + Al)/(Mg + Fe + Si + Al) is larger than that of $${\upphi }_{z}^{+}$$. The A-site atom also goes away from its ideal position (0, 0, 0.25), corresponding to the A-site position in the Pm$$\overline{3 }$$m cubic structure, with increasing the ratio (Fe + Al)/(Mg + Fe + Si + Al) (Fig. 5b). The structural distortion, i.e. the deviation from the Pm$$\overline{3 }$$m cubic structure, is thus getting larger with the incorporation of Fe and Al dominated by the charge-coupled substitution (1). The expansion of the two very weak A–O2 bonds and the four potentially non-bonding AO separations with decreasing <rA> (Fig. 4a), i.e. with increasing Fe3+ content on A site, is a consequence of such increase in structural distortion and a sign of the increased deviation from the Pm$$\overline{3 }$$m cubic structure, with twelve equivalent A–O bond lengths and without any octahedral tilting ($${\upphi }_{x}^{0}{\upphi }_{y}^{0}{\upphi }_{z}^{0}$$).

### Implications for the Earth’s lower mantle

Bridgmanite is now believed to undergo the phase transition to post-perovskite phase with CaIrO3 structure, associated with the D′′ seismic discontinuity, at 125 GPa and 2500 K (Ref.52, 53). However, some high-pressure studies54,55,56 suggested that another perovskite phase with a different symmetry can intervene between the Pbnm perovskite phase and the post-perovskite phase. To examine the possibility of such phase transitions in ABO3 perovskites at high pressures and high temperatures, it is effective to discuss the relative compressibility of AO12 and BO6 polyhedra57, 58. The compressibility ratio (βB/ βA) of the two polyhedra is given by βBA = MA/MB (Ref.57), where the subscripts “A” and “B” represent the AO12 and BO6 polyhedra, respectively. In Pbnm orthorhombic perovskites with largely distorted AO12 (practically AO8) polyhedra, such as bridgmanite, the parameters MA and MB are defined as follows:

$${M}_{\mathrm{A}}=\left(8{R}_{\mathrm{A}8}/B\right) {\mathrm {exp}} \left[\left({R}_{0}-{R}_{\mathrm{A}8}\right)/B\right]+\left(4{R}_{\mathrm{A}4}/B\right) {\mathrm {exp}} \left[\left({R}_{0}-{R}_{\mathrm{A}4}\right)/B\right]$$
(2)
$${M}_{\mathrm{B}}=\left({6R}_{\mathrm{B}}/B\right)\mathrm{exp}\left[\left({R}_{0}-{R}_{\mathrm{B}}\right)/B\right]$$
(3)

where RA8, RA4 and RB are the average distances of eight shorter A–O bonds, of four longer AO separations and of six B–O bonds, respectively; R0 and B the bond valence parameters. According to Angel et al.58, in perovskites exhibiting zone-boundary type phase transitions, when the BO6 octahedra are more rigid than the AO12 polyhedra (i.e., MA/MB < 1), the phase transition temperature Tc rises with increasing pressure as a consequence of the increase in the octahedral tilting; thus, the phase boundary has a positive Clapeyron slope (dP/dTc > 0). Conversely, when the BO6 octahedra are less rigid than the AO12 polyhedra (i.e., MA/MB > 1), Tc reduces with increasing pressure as a consequence of the decrease in the tilting; thus, the phase boundary has a negative Clapeyron slope (dP/dTc < 0).

The MA/MB ratios at ambient condition, from Eqs. (2) and (3), are calculated to be 0.67 for the end-member MgSiO3 bridgmanite44,45,46, 0.72 for the present (Fe3+, Al)-bearing bridgmanite, and 0.63 for CaTiO3 perovskite43 as a good analog of bridgmanite, using the R0 and B values determined by Brown and Altermatt59 and the average interatomic distances reported for each compound. It follows therefore that if these A2+B4+O3-type Pbnm perovskites undergo phase transitions to perovskite phases with different symmetries at high pressures and high temperatures, their phase boundaries have positive Clapeyron slopes.

The tilt angles of the end-member MgSiO3 bridgmanite at ambient condition are calculated to be $${\upphi }_{x}^{-}$$ = 11.7° and $${\upphi }_{z}^{+}$$ = 11.6° from the reported structural parameters45. These values are much larger than those of CaTiO3 perovskite ($${\upphi }_{x}^{-}$$ = 8.3°, $${\upphi }_{z}^{+}$$ = 8.8°), which were reported to undergo the phase transitions of Pbnm → I4/mcm at 1512 K and of I4/mcm → Pm$$\overline{3 }$$m at 1635 K under ambient pressure43. As mentioned above, when MA/MB < 1, the larger tilting yields the higher Tc, and the rise in pressure further promotes the rise in Tc because of a positive Clapeyron slope. If the same sequence of the high-temperature phase transitions to higher symmetric LP (low pressure)–HT (high temperature) phases as CaTiO3 perovskite also appears in the end-member MgSiO3 bridgmanite at high pressures, thus, Tc should become much higher than those observed in CaTiO3 perovskite at ambient pressure. The incorporation of Fe and Al into bridgmanite would further raise Tc because it increases the tilt angles $${\upphi }_{x}^{-}$$ and $${\upphi }_{z}^{+}$$ as shown in Fig. 5a.

In the end-member MgSiO3 bridgmanite, Wang et al.54 observed the discontinuous changes in the unit-cell parameters and volume at ~ 600 K and 7.3 GPa, suggesting the phase transition to another perovskite phase. This PT condition corresponds to a much lower temperature despite a higher pressure than the phase-transition points observed in CaTiO3 perovskite. Even if there exists such phase transition in the end-member MgSiO3 bridgmanite, therefore, it would be the phase transition not to a higher symmetric LP-HT phase, such as in CaTiO3 perovskite, but to a lower symmetric HP (high pressure)–LT (low temperature) phase.

At extreme conditions corresponding to deeper parts of the lower mantle, a few in-situ energy-dispersive X-ray diffraction studies using a laser-heated diamond anvil cell have suggested the phase transition of Pbnm bridgmanite to another perovskite phase. For example, Meade et al.55 reported the phase transition from Pbnm (orthorhombic) to Pm$$\overline{3 }$$m (cubic) at 64 GPa and 1850 K for (Mg, Fe)SiO3 bridgmanite. Shim et al.56 reported the phase transition of the end-member MgSiO3 bridgmanite from Pbnm to one of the three possible symmetries [P21/m (monoclinic), Pmmn (orthorhombic), or P42/nmc (tetragonal)] above 83 GPa and 1700 K. These suggested space groups after the phase transitions or the presence itself of the phase transitions were not entirely definitive because of low resolution and unreliable peak intensities in the powder X-ray diffraction patterns measured at these extreme conditions. These suggested phase-transition points are however at higher temperatures and higher pressures than those of CaTiO3 perovskite, being consistent with the prediction that the larger octahedral tilting and higher pressure raise Tc. The possibility that bridgmanite changes into another perovskite phase with a different symmetry from Pbnm before the phase transition to the post-perovskite phase cannot thus be ruled out, whether it is the higher symmetric LP-HT phase or the lower symmetric HP-LT phase. Although bridgmanite including high Al and/or Fe contents, such as the present sample (Mg0.662Fe0.338Si0.662Al0.338O3) with their contents very close to those27 reported for bridgmanite formed from the MORB composition, is predicted to have further higher Tc, the phase transition of such bridgmanite to another perovskite phase might be found in slabs that fell/subducted into the lowermost parts of the lower mantle.

The compressibility ratio βBA of bridgmanite can provide important knowledge of its elastic velocity as well. Comparison of the end-member MgSiO3 bridgmanite45 and the present (Fe3+, Al)-bearing bridgmanite shows that the incorporation of Fe3+ and Al through the charge-coupled substitution (1) makes MA unchanged, decreases MB, increases the density ρ, and consequently increases βBA (= MA/MB), where MA = 12.87, MB = 19.23 and ρ = 4.103 g/cm3 for the former; MA = 12.92, MB = 17.85 and ρ = 4.357 g/cm3 for the latter. We can consider that the increase in βBA, depending only on MB, corresponds to the decrease in the bulk modulus K. The bulk sound velocity $${V}_{\mathrm{B}}=\sqrt{K/\uprho }$$ is thus expected to decrease with increasing Fe3+ and Al contents, which is consistent with the theoretical calculation60 for (Fe3+, Al)-bearing bridgmanite. This approach from crystallography can thus be a helpful method to gain important insights into the seismic properties within the lower mantle. For this purpose, systematic crystal-chemical studies of bridgmanites with a variety of valence- and spin-states of Fe and compositions are necessary.

## References

1. 1.

Katsura, T., Sato, K. & Ito, E. Electrical conductivity of silicate perovskite at lower-mantle conditions. Nature 395, 493–495 (1998).

2. 2.

Jackson, J. M. et al. A synchrotron Mössbauer spectroscopy study of (Mg, Fe)SiO3 perovskite up to 120 GPa. Am. Mineral. 90, 199–205 (2005).

3. 3.

Vanpeteghem, C. B. et al. Al, Fe substitution in the MgSiO3 perovskite structure: A single crystal X-ray diffraction study. Phys. Earth Planet. Inter. 155, 96–103 (2006).

4. 4.

Saikia, A., Ballaran, T. B. & Frost, D. J. The effect of Fe and Al substitution on the compressibility of MgSiO3-perovskite determined through single-crystal X-ray diffraction. Phys. Earth Planet. Inter. 173, 153–161 (2009).

5. 5.

Catalli, K. et al. Spin state of ferric iron in MgSiO3 perovskite and its effect on elastic properties. Earth Planet. Sci. Lett. 289, 68–75 (2010).

6. 6.

Catalli, K. et al. Effects of the Fe3+ spin transition on the properties of aluminous perovskite—New insights for lower mantle seismic heterogeneities. Earth Planet. Sci. Lett. 310, 293–302 (2011).

7. 7.

Hummer, D. R. & Fei, Y. Synthesis and crystal chemistry of Fe3+-bearing (Mg, Fe3+)(Si, Fe3+)O3 perovskite. Am. Mineral. 97, 1915–1921 (2012).

8. 8.

Ismailova, L. et al. Stability of Fe, Al-bearing bridgmanite in the lower mantle and synthesis of pure Fe-bridgmanite. Sci. Adv. 2, e1600427 (2016).

9. 9.

Yoshino, T., Kamada, S., Zhao, C., Ohtani, E. & Hirao, N. Electrical conductivity model of Al-bearing bridgmanite with implications for the electrical structure of the Earth’s lower mantle. Earth Planet. Sci. Lett. 434, 208–219 (2016).

10. 10.

Fukui, H. et al. Effect of cation substitution on bridgmanite elasticity: A key to interpret seismic anomalies in the lower mantle. Sci. Rep. 6, 33337. https://doi.org/10.1038/srep33337 (2016).

11. 11.

Li, X. & Jeanloz, R. Laboratory studies of the electrical conductivity of silicate perovskite at high pressures and temperatures. J. Geophys. Res. 95, 5067–5078 (1990).

12. 12.

Li, X., Ming, L.-C., Manghnani, M. H., Wang, Y. & Jeanloz, R. Pressure dependence of the electrical conductivity of (Mg0.9Fe0.1)SiO3 perovskite. J. Geophys. Res. 98, 501–508 (1993).

13. 13.

Xu, Y., McCammon, C. & Poe, B. T. The effect of alumina of the electrical conductivity of silicate perovskite. Science 282, 922–924 (1998).

14. 14.

Kesson, S. E., Fitz Gerald, J. D., Shelley, J. M. & Whithers, R. L. Phase relations, structure and crystal chemistry of some aluminous silicate perovskites. Earth Planet. Sci. Lett. 134, 187–201 (1995).

15. 15.

Andrault, D., Neuville, D. R., Flanck, A.-M. & Wang, Y. Cation sites in Al-rich MgSiO3 perovskites. Am. Mineral. 83, 1045–1053 (1998).

16. 16.

Stebbins, J. F., Krocker, S. & Andrault, D. The mechanism of solution of Al oxide in MgSiO3 perovskite. Geophys. Res. Lett. 28, 615–618 (2001).

17. 17.

Stebbins, J. F., Kojitani, H., Akaogi, M. & Navrotsky, A. Aluminum substitution in MgSiO3 perovskite: Investigation of multiple mechanisms by 27Al NMR. Am. Mineral. 88, 1161–1164 (2003).

18. 18.

Yamamoto, T., Yuen, D. A. & Ebisuzuki, T. Substitution mechanisms of Al ions in MgSiO3 perovskite under high pressure conditions from first-principles calculations. Earth Planet. Sci. Lett. 206, 617–625 (2003).

19. 19.

Navrotsky, A. et al. Aluminum in magnesium silicate perovskite: Formation, structure, and energetics of magnesium-rich defect solid solutions. J. Geophys. Res. 108, 2330. https://doi.org/10.1029/2002JB002055 (2003).

20. 20.

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

21. 21.

McCammon, C. A. Perovskite as a possible sink for ferric iron in the lower mantle. Nature 387, 694–696 (1997).

22. 22.

Lauterbach, S., McCammon, C. A., van Aken, P., Langenhorst, F. & Seifert, F. Mössbauer and ELNES spectroscopy of (Mg, Fe)(Si, Al)O3 perovskite: A highly oxidized component of the lower mantle. Contrib. Mineral. Petrol. 138, 17–26 (2000).

23. 23.

Frost, D. J. & Langenhorst, F. The effect of Al2O3 on Fe-Mg partitioning between magnesiowüstite and magnesium silicate perovskite. Earth Planet. Sci. Lett. 199, 227–241 (2002).

24. 24.

Frost, D. J. et al. Experimental evidence for the existence of iron-rich metal in the Earth’s lower mantle. Nature 428, 409–412 (2004).

25. 25.

McCammon, C. A., Lauterbach, S., Seifert, F., Langenhorst, F. & van Aken, P. A. Iron oxidation state in lower mantle mineral assemblages I. Empirical relations derived from high-pressure experiments. Earth Planet. Sci. Lett. 222, 435–449 (2004).

26. 26.

Grocholski, B. et al. Spin and valence states of iron in (Mg0.8Fe0.2)SiO3 perovskite. Geophys. Res. Lett. 36, L24303 (2009).

27. 27.

Hirose, K. & Fei, Y. Subsolidus and melting phase relations of basaltic composition in the uppermost lower mantle. Geochim. Cosmochim. Acta 66, 2099–2108 (2002).

28. 28.

Ringwood, A.E. In Advances in Earth Sciences (ed Hurley, P.M.) 287–356 (M.I.T. Press, 1966).

29. 29.

Nakatsuka, A. et al. Symmetry change of majorite solid-solution in the system Mg3Al2Si3O12-MgSiO3. Am. Mineral. 84, 1135–1143 (1999).

30. 30.

Hirao, N. et al. New developments in high-pressure X-ray diffraction beamline for diamond anvil cell at SPring-8. Matter Radiat. Extremes 5, 018403 (2020).

31. 31.

Prescher, C., McCammon, C. & Dubrovinsky, L. MossA: A program for analyzing energy-domain Mössbauer spectra from conventional and synchrotron sources. J. Appl. Cryst. 45, 329–331 (2012).

32. 32.

Nakatsuka, A. et al. Static disorders of atoms and experimental determination of Debye-temperature in pyrope: Low- and high-temperature single crystal X-ray diffraction study. Am. Mineral. 96, 1593–1605 (2011).

33. 33.

Nakatsuka, A. et al. Temperature dependence of crystal structure of CaGeO3 high-pressure perovskite phase and experimental determination of its Debye temperatures studied by low- and high-temperature single-crystal X-ray diffraction. Am. Mineral. 100, 1190–1202 (2015).

34. 34.

Nakatsuka, A., Yoshiasa, A., Fujiwara, K. & Ohtaka, O. Variable-temperature single-crystal X-ray diffraction study of SrGeO3 high-pressure perovskite phase. J. Mineral. Petrol. Sci. 113, 280–285 (2018).

35. 35.

Sasaki, S. RADY: A Fortran Program for the Least-Squares Refinement of Crystal Structures (National Laboratory for High Energy Physics, 1987).

36. 36.

International Tables for Crystallography Vol. C (ed Wilson, A. J. C.) (Kluwer Dordrecht, 1992).

37. 37.

Tokonami, M. Atomic scattering factor for O2−. Acta Cryst. 19, 486 (1965).

38. 38.

Becker, P. J. & Coppens, P. Extinction within the limit of validity of the Darwin transfer equations. I. General formalisms for primary and secondary extinction and their application to spherical crystals. Acta Cryst. A30, 129–147 (1974).

39. 39.

Becker, P. J. & Coppens, P. Extinction within the limit of validity of the Darwin transfer equations. II. Refinement of extinction in spherical crystals of SrF2 and LiF. Acta Cryst. A30, 148–153 (1974).

40. 40.

Shannon, R. D. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Cryst. A32, 751–767 (1976).

41. 41.

Gütlich, P. Mössbauer spectroscopy in chemistry. In Mössbauer Spectroscopy, Topics in Applied Physics Vol. 5 (ed Gonser, U.) 53–96 (Springer, 1975).

42. 42.

Hsu, H., Blaha, P., Cococcioni, M. & Wentzcovitch, R. M. Spin-state crossover and hyperfine interactions of ferric iron in MgSiO3 perovskite. Phys. Rev. Lett. 106, 118501 (2011).

43. 43.

Yashima, M. & Ali, R. Structural phase transition and octahedral tilting in the calcium titanate. Solid State Ionics 180, 120–126 (2009).

44. 44.

Horiuchi, H., Ito, E. & Weidner, D. J. Perovskite-type MgSiO3: Single-crystal X-ray diffraction study. Am. Mineral. 72, 357–360 (1987).

45. 45.

Sugahara, M. et al. Reinvestigation of the MgSiO3 perovskite structure at high pressure. Am. Mineral. 91, 533–536 (2006).

46. 46.

Dobson, D. P. & Jacobsen, S. D. The flux growth of magnesium silicate perovskite single crystals. Am. Mineral. 89, 807–811 (2004).

47. 47.

Glazer, A. M. The classification of tilted octahedra in perovskites. Acta Cryst. B28, 3384–3392 (1972).

48. 48.

Glazer, A. M. Simple ways of determining perovskite structures. Acta Cryst. A31, 756–762 (1975).

49. 49.

Kennedy, B. J., Howard, C. J. & Chakoumakos, B. C. Phase transitions in perovskite at elevated temperatures—A powder neutron diffraction study. J. Phys. Condens. Matter 11, 1479–1488 (1999).

50. 50.

Wang, D. & Angel, R. J. Octahedral tilts, symmetry-adapted displacive modes and polyhedral volume ratios in perovskite structures. Acta Cryst. B67, 302–314 (2011).

51. 51.

Campbell, B. J., Stokes, H. T., Tanner, D. E. & Hatch, D. M. ISODISPLACE: A web-based tool for exploring structural distortions. J. Appl. Cryst. 39, 607–614 (2006).

52. 52.

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

53. 53.

Tsuchiya, T., Tsuchiya, J., Umemoto, K. & Wentzcovitch, R. M. Phase transition in MgSiO3 perovskite in the Earth’s lower mantle. Earth Planet. Sci. Lett. 224, 241–248 (2004).

54. 54.

Wang, Y. et al. Phase transition and thermal expansion of MgSiO3 perovskite. Science 251, 410–413 (1991).

55. 55.

Meade, C., Mao, H. K. & Hu, J. High-temperature phase transition and dissociation of (Mg, Fe)SiO3 perovskite at lower mantle pressures. Science 268, 1743–1745 (1995).

56. 56.

Shim, S.-H., Duffy, T. S. & Shen, G. Stability and structure of MgSiO3 perovskite to 2300-kilometer depth in Earth’s mantle. Science 293, 2437–2440 (2001).

57. 57.

Zhao, J., Ross, N. L. & Angel, R. J. New view of the high-pressure behavior of GdFeO3-type perovskites. Acta Cryst. B60, 263–271 (2004).

58. 58.

Angel, R. J., Zhao, J. & Ross, N. L. General rules for predicting phase transitions in perovskites due to octahedral tilting. Phys. Rev. Lett. 95, 025503 (2005).

59. 59.

Brown, I. D. & Altermatt, D. Bond-valence parameters obtained from a systematic analysis of the Inorganic Crystal Structure Database. Acta Cryst. B41, 244–247 (1985).

60. 60.

Li, L. et al. Elasticity of (Mg, Fe)(Si, Al)O3-perovskite at high pressure. Earth Planet. Sci. Lett. 240, 529–536 (2005).

61. 61.

Ito, E. & Matsui, Y. Synthesis and crystal-chemical characterization of MgSiO3 perovskite. Earth Planet. Sci. Lett. 38, 443–450 (1978).

## Acknowledgements

The present study was supported by JSPS KAKENHI (Grant Nos. JP19H02004, JP15K05344, JP15H05748 and JP22000002). The crystal synthesis was performed using joint-use facilities of the Institute for Planetary Materials, Okayama University. The synchrotron Mössbauer spectroscopy was conducted at the BL10XU of SPring-8, Hyogo, Japan under the approval of the Japan Synchrotron Radiation Research Institute (Proposal No. 2018B1500). The crystallinity of the sample was characterized using the vertical type four-circle diffractometer installed at the BL-10A of Photon Factory, High Energy Accelerator Research Organization, Tsukuba, Japan under the approval of the Photon Factory Program Advisory Committee (Proposal No. 2016G641). We also acknowledge support from the GIMRT Program of the Institute for Materials Research, Tohoku University (Proposal Nos. 15K0015 and 15K0054).

## Author information

Authors

### Contributions

A.N. and H.F. planned the present study. T.Y. synthesized the single-crystal sample from the high-pressure experiment. A.N. and K.S. performed the single-crystal X-ray diffraction experiment and crystal structure analysis. S.K. and N.H. performed the synchrotron Mössbauer measurements and analyzed the Mössbauer spectrum. S.K. and M.O. analyzed the chemical composition of the sample from the electron probe microanalyses. The manuscript was written by A.N. and reviewed by all authors.

### Corresponding author

Correspondence to Akihiko Nakatsuka.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

### Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Rights and permissions

Reprints and Permissions

Nakatsuka, A., Fukui, H., Kamada, S. et al. Incorporation mechanism of Fe and Al into bridgmanite in a subducting mid-ocean ridge basalt and its crystal chemistry. Sci Rep 11, 22839 (2021). https://doi.org/10.1038/s41598-021-00403-6

• Accepted:

• Published:

• DOI: https://doi.org/10.1038/s41598-021-00403-6