Both silicon and carbon have been proposed to be important light elements in the Earth’s core, in particular in the inner core. Here we performed melting experiments on the Fe-Si-C ternary system at ~50, ~136, and ~200 GPa and determined the liquidus phase relations and the solid/liquid partition coefficients (D) of C and Si. The liquidus field of Fe shrinks at higher pressures, which narrows down the possible outer core liquid composition. Our data also demonstrate that the Fe-Si binary eutectic liquid reduces its Si concentration to ~8 wt% with increasing pressure to 330 GPa. We found that the inner core is not Fe-Si-C-S alloy but likely includes hydrogen when considering the low DC and the strong enhancement of DSi with increasing liquid C abundance. The present-day core does not include as much as ~6 wt% Si, suggesting that at least a part of “missing” Si could be sequestrated elsewhere.

We examined liquidus phase relations in Fe-O±H at ~40 and ~150 GPa, and in Fe-H at 45 GPa. While it has been speculated that Fe and FeH form continuous solid solution to core pressures, our experiment on Fe-H showed that FeH0.20 forms with the hcp structure, different from fcc for stoichiometric FeH, and melts at temperature lower than that for FeH, suggesting eutectic melting between Fe and FeH. It is consistent with the liquidus phase diagram in Fe-O-H, which implies the Fe-FeH binary eutectic liquid composition of FeH0.42 at ~40 GPa. We estimated the outer core liquid composition to be Fe + 2.9–5.2% O + 0.03–0.32% H + 0–3.4% Si + 1.7% S by weight, based on the liquidus phase relations, solid-liquid partitioning, and outer/inner core densities and velocities, indicating that O and either H or Si are important core light elements.

While hydrogen is one of plausible major light elements in the core, the high-temperature equation of state (EoS) of Fe-H alloy has not been experimentally examined to the core pressure range. Here we measured the volume (V) of non-magnetic (NM) fcc FeH at high pressure and temperature (P-T) to 142 GPa and 3660 K in a laser-heated diamond-anvil cell (DAC) and obtained its P-V-T EoS. An increase in the lattice volume of Fe per H atom, ΔVH, determined as functions of P and T is found to be substantially smaller than the volume of metallic H that has been used to estimate H concentration in Fe-H alloy. The ΔVH is almost identical between fcc and dhcp phases in the NM state, suggesting that it is applicable to hcp. The extrapolation of ΔVH to inner core conditions indicates its maximum H content to be 0.8–0.9 wt%.

While hydrogen could be an important light alloying element in planetary iron cores, phase relations in the Fe-FeH system remain largely unknown at high pressures and temperatures (P-T). A speculative Fe-H2 phase diagram has been proposed assuming continuous solid solution between Fe and FeH and eutectic melting between FeH and H2. Recent studies revealed that stoichiometric FeH becomes non-magnetic above ~40 GPa, which might affect its melting behavior. Here we examined the melting curve of non-magnetic FeH between 43 and 152 GPa by a combination of laser-heated diamond-anvil cell (DAC) techniques and synchrotron X-ray diffraction (XRD) analyses. The melting temperature was determined by employing the appearance of additional hazy XRD signals upon quenching temperature as a melting criterion. We also performed thermodynamic modeling, which well reproduces the change in the curvature of FeH melting curve upon the loss of magnetism and extrapolates the experimental constraints to inner core pressures. The XRD data showed that non-magnetic FeH melts congruently at temperatures higher than the known eutectic melting curve for FeHx (x > 1). Combined with the fact that the endmembers exhibit different crystal structures, these results indicate that Fe and non-magnetic FeH form a eutectic system. The dT/dP slope of the FeH melting curve is comparable to that for Fe, suggesting that the eutectic liquid composition of FeH0.42 (Fe + 0.75 wt% H) previously estimated at ~40 GPa changes little with increasing pressure.

We conducted X-ray diffraction (XRD) measurements of a pyrolitic mantle material up to 4480 K at 122-166 GPa in a laser-heated diamond-anvil cell (DAC). Results demonstrate that the phase transition between bridgmanite and post-perovskite occurs in pyrolite within the lowermost mantle pressure range even at >4000 K. It suggests the ubiquitous occurrence of post-perovskite above the core-mantle boundary (CMB), which may be consistent with recent high-quality seismology data that non-observations of D” reflections are exceptions. Combining with earlier experiments performed at and below the normal lower-mantle geotherm, our data show that the bridgmanite + post-perovskite two-phase region is ~5 GPa thick and the Clapeyron slope of the boundary is +7(+2/-3) MPa/K in agreement with previous theoretical calculations. The global presence of rheologically weak post-perovskite at the bottom of the mantle has profound implications in seismology, geodynamics, and heat transfer from the core.

We examined liquidus phase relations in Fe-O+/-H at ~40 and ~150 GPa, and subsolidus phase equilibria in Fe-FeH. While it has been speculated that Fe and FeH form continuous solid solution to core pressures, our experiments show the coexistence of the H-poor hcp and H-rich fcc phases in the Fe-FeH system. Considering higher melting temperature of stoichiometric FeH than that of Fe-FeH, it indicates eutectic melting between Fe and FeH. It is consistent with the liquidus phase diagram in Fe-O-H, which implies the Fe-FeH binary eutectic liquid composition of FeH0.42 at ~40 GPa. We estimated the outer core liquid composition to be Fe + 2.9-5.2% O + 0.03-0.32% H + 0-3.4% Si + 1.7% S by weight, based on the liquidus phase relations, solid-liquid partitioning, and outer/inner core densities and velocities, indicating that O and either H or Si are important core light elements.