2026/07/22 by Xianyu Xue, Masami Kanzaki
Earth and Planetary Sciences · #High-pressure geophysics and materials #Geological and Geochemical Analysis #earthquake and tectonic studies
paper · doi:10.2138/am-2026-10282
Abstract Nominally anhydrous minerals represent important water reservoirs of the Earth’s mantle, storing water in the form of OH defects that affect their physical and chemical properties in ways that depend on the nature of the OH defects. However, even for major upper mantle minerals like low-Ca orthopyroxene, the water incorporation mechanisms are still not fully understood, despite extensive studies using techniques such as infrared spectroscopy. We performed first-principles calculations on the enthalpy, polarized infrared absorption spectra, and 1H and 29Si NMR spectra for MgSiO3 enstatite, a major component of low-Ca orthopyroxene, focusing on the low- (LCEn) and high- pressure clinoenstatite (HCEn) phases, to help clarify their water incorporation mechanisms and the effect of water on the phase transition. Our calculations confirmed that the experimentally observed infrared bands near 3378 and 3042 cm−1 for LCEn can be attributed to (2H)M2 defects (proton pairs charge-compensating Mg (M2) vacancies), which also produce a pair of peaks near 7.8 and 5.8 ppm in the 1H MAS NMR spectra. The experimentally observed infrared bands near 3687, 3675 and 3602 cm−1 for LCEn may be attributed to two of the four groups of OH stretching bands, corresponding to vibrations of OH groups on the two nonbridging oxygens (O1, O2) of (4H)SiB and (4H)SiA defects (four protons charge-compensating SiB and SiA vacancies). Two groups of low-frequency OH stretching bands, corresponding to vibrations of OH groups on the two bridging oxygens (O3), are also predicted for these defects, but have not been identified in previous studies. They likely correspond to weak broad infrared bands in the 2500–3000 cm−1 region observed in our parallel experimental study. The 1H MAS NMR peaks near 0, 3, 8 and 12 ppm observed in the latter study match with those predicted for the (4H)SiB and (4H)SiA defects. The configurations and spectroscopic features of these OH defects overall resemble those of MgSiO3 orthoenstatite (OEn), except that for LCEn reverted during room-temperature decompression from the unquenchable HCEn phase, both (4H)SiB and (4H)SiA defects contribute equally, despite higher enthalpy for the latter defects, because the HCEn has a single type of Si sites. Overlooking the strongly-hydrogen bonded OH groups for (4H)Si defects in clinoenstatite (and OEn) would render the water contents determined accordingly unreliable. The incorporation of water has been previously suggested to lower the transition pressure between OEn or LCEn and HCEn, which led to the speculation of the OEn/HCEn transition as a possible explanation for the observed variable depths of the “seismic X-discontinuity”. However, our study suggested that the presence of (2H)M2 defects causes an increase in the LCEn/HCEn transition pressure, whereas the (4H)Si defects does not cause appreciable change at 0 K. The previously reported lower LCEn/HCEn transition pressure for hydrous than anhydrous clinoenstatite at room temperature may be explained by the high-enthalpy state of the LCEn starting materials used that contain both (4H)SiB and (4H)SiA defects. It is likely that different types of OH defects may also affect the OEn/HCEn transition pressure differently. The most favorable (4H)Si defects in HCEn were found to change from a configuration similar to those in the LCEn and OEn phases to one exhibiting a unique configuration of 2H on a single oxygen (O2) with increasing pressure. The change in the configuration of the most strongly hydrogen-bonded OH group of the (4H)Si defects between LCEn or OEn or HCEn (lower pressure) and HCEn (higher pressure) may be regarded as a shift of role between a hydrogen-bond donor and acceptor in a hydrogen bonding linkage (from O3-H···O2 to O3···H-O2) with increasing pressure. Such a pressure-induced change in the configuration of hydrogen bonding could be common for minerals with distinct bond valence sums for the two oxygens participating in hydrogen bonding, as is the case between a bridging and nonbridging oxygen.