A new high-pressure phase of Ca2Al2SiO7 and implications for the earth's interior

Gehlenite (Ca₂Al₂SiO₇) has been found to transform to a new phase at pressures greater than 100 kbar and at about 1000°C, using a diamond-anvil pressure cell coupled with laser heating. The atoms of the new phase appear to be arranged in a perovskite-related structure similar to that described for Na₂Ti₃O₇. The structure probably consists of layers of (Al₂SiO₇)⁴⁻, which are built up from blocks of edge-sharing (Al, Si)O₆ octahedra and these blocks are joined by common octahedra corners. A small cubic unit cell with a = 3.719 ± 0.004 Å indexes completely the strong lines of the powder diffraction pattern, and a superlattice with a = 14.88 ± 0.02 Å satisfies all the observed weak lines in addition to the strong ones. However, the cell may be pseudocubic. The small cell contains a half of the gehlenite formula while the large cell contains 32 gehlenite formulae. Hence the molar volume for the new phase of Ca₂Al₂SiO₇ is calculated to be 61.96 ± 0.20 cm³ at atmospheric pressure and room temperature. The new sodium titanate-type structure is probably more closely packed than an ordinary perovskite-type structure in which all octahedral corners are shared. This view is strongly supported by the very great density of this new phase, which is about 8% denser than the equivalent mixture of CaAl₂O₄ (calcium ferrite type) plus CaSiO₃ (cubic perovskite type). The new phase is probably the most closely packed silicate known. Mg₂SiO₄ (spinel) was found to transform to an assemblage containing MgSiO₃ (perovskite) plus MgO (periclase) at P-T conditions equivalent to the upper part of the lower mantle. By reacting with MgO, the perovskite modification of both MgSiO₃ and MgSiO₃ · xAl₂O₃ may adopt the sodium titanate structure at the still greater depths of the lower mantle. If the sodium titanate structures of Mg₂(Al₂Si)O₇ and Mg₂(MgSi₂)O₇ are present in the deep part of the lower mantle, MgO does not exist as a separate phase at the mantle-core boundary. This might be an obstacle to the possibility of dissolving these oxides (specifically the FeO component) in the molten Fe in the outer core as suggested by geophysical and geochemical studies of the earth's interior. The mechanism for developing the chemical plumes in the deep mantle proposed by Anderson does not appear to be consistent with studies of phase transformations in Ca-Al-rich compounds as outlined in this paper.