Anomalous compressibility in (Fe,Al)-bearing bridgmanite: implications for the spin state of iron

Abstract

The valence and spin states of Fe in (Fe,Al)-bearing bridgmanite (bdg) affect its physical properties, which is important for the interpretation of geophysical observations. Currently, tens of studies on the compressibility and spin states of Fe-bearing bdg have been reported. A consensus is that Fe-bearing bdg shows spin transition, which affects its elastic parameters. However, there is a conflict between reports on the compressibility and spin states of (Fe,Al)-bearing bdg in experiments using samples pre-synthesized in a multi-anvil apparatus (MA), and samples directly synthesized in a diamond anvil cell (DAC). There are no reports showing evidence of spin transition of Fe in compression experiments using (Fe,Al)-bearing bdg samples pre-synthesized in a MA, while those synthesized at relatively high pressure (at least above 45 GPa) in a DAC all exhibited the spin transition. Here, we performed synchrotron X-ray diffraction measurements on Mg_0.85Fe_0.09Al_0.21Si_0.86O₃ and Mg_0.85Fe_0.14Al_0.05Si_0.96O₃ bdg synthesized at relatively high pressure in a laser-heated DAC from amorphous starting material up to 47 and 56 GPa, respectively, at room temperature. The obtained pressure (P)–lattice volume (V) relations show noticeable softening at 22–30 GPa and 35–45 GPa, respectively, which is probably due to the spin transition of Fe. Combining our results and previous reports, we suggest that the lower mantle bdg is capable of containing low-spin Fe³⁺, which questions the general view. Such a transition changes density and may affect the physical properties of bridgmanite such as thermal conductivity and iron partitioning coefficient, thus having profound implications for mantle dynamics, and the chemical composition of the Earth.

Appendix

In order to estimate the uniaxial stress in our bdg sample, we calculated the St value of Au sputtered on both sides of the sample from the diffraction of Au. The measured lattice constant a_m can be written as the following (Singh and Takemura 2001):

$$a_{\text{m}} \left( {hkl} \right) = M_{0} + M_{1} \left\{ {3\left( {1 - 3\sin^{2} \theta } \right){{\varGamma }}\left( {hkl} \right)} \right\},$$

(3)

$$M_{0} = a_{p} \left\{ {1 + \left( {\frac{\alpha t}{3}} \right)\left( {1 - 3\sin^{2} \theta } \right)\left[ {S_{11} - S_{12} - \left( {2G_{V} } \right)^{ - 1} \left( {1 - \alpha^{ - 1} } \right)} \right]} \right\},$$

(4)

$$M_{1} = - a_{p} \alpha tS/3,$$

(5)

$${{\varGamma }}\left( {hkl} \right) = \left( {h^{2} k^{2} + k^{2} l^{2} + l^{2} h^{2} } \right)/\left( {h^{2} + k^{2} + l^{2} } \right)^{2} ,$$

(6)

$$S = S_{11} - S_{12} - S_{44} /2,$$

(7)

where a_p is the lattice parameter under hydrostatic pressure, S_ij is the single-crystal elastic compliance, G_V is the shear modulus of the polycrystalline aggregate under the assumption of strain continuity across the grain boundaries, α is the ratio of uniform stress model and uniform strain model (0.5 < α<1). By assuming M₀ ~ a_p, we can obtain the St value from Eq. (5):

$$St\sim - 3M_{1} /\left( {\alpha M_{0} } \right),$$

(8)

where M₀ and M₁ were obtained from fitting the slope of a Γ plot, a plot a_m(hkl) versus 3(1 − 3sin²θ)Γ(hkl) (Fig. S2).

We used (111), (200), (220) and (311) reflections to obtain each of the four plots. Note that we excluded a certain reflection which overlapped with peaks of samples or the pressure medium. Also, using α = 1 (uniform stress model), we estimated the St values (Fig. 5).