Evidence for a liquid silicate layer atop the Martian core

Nearly four years of InSight seismic monitoring indicated a large but low-density Martian core (mean density ~6.0–6.3 g cm⁻³), implying substantial light-element enrichment (S, C, O, H). Such high light-element contents are difficult to reconcile with cosmochemical inventories of Mars’s building blocks and suggest early core formation while nebular gas was present. Prior core composition estimates relied largely on geochemical arguments or seismic/geophysical averages and tidal response, but volatile element abundances in precursor materials are variable and poorly constrained, leaving the identity and abundance of core light elements uncertain. Recent detections of a P wave diffracted along the core-mantle boundary and core-transiting P waves enable joint constraints on P-wave velocity and density, but translating these into composition and temperature is hampered by limited property data for liquid Fe-Ni-X alloys under Martian core conditions. This study asks whether a stratified structure at the base of the mantle exists and, if so, whether a molten silicate layer overlies a smaller, denser metallic core, reconciling seismological and cosmochemical constraints.

• Previous InSight analyses inferred a large (≈25% of Mars’s mass) but low-density core (~6.0–6.3 g cm⁻³), implying substantial light elements (S, C, O, H). However, these elements are volatile and scarce in plausible Martian precursors, challenging the inferred large density deficit.
• Geochemical and cosmochemical studies provide ranges for light elements but with large uncertainties due to volatility and heterogeneity in chondritic sources.
• Seismic observations of core-diffracted and core-transiting phases now provide more direct probes of the CMB region.
• A shortage of experimental measurements for P-wave velocity and density of liquid Fe-Ni-X alloys at Martian pressures/temperatures limited prior compositional inference; first-principles (AIMD) calculations offer a way to bridge this gap.
• Geodynamic models allow for a molten silicate layer above the core; on Earth, stratified layers at the top of the core are discussed, but the densities inferred here argue against a similar metallic stratified layer for Mars.

• Ab initio molecular dynamics (AIMD): Computed thermoelastic properties (density, P-wave velocity) of liquid Fe-Ni-S-C-O-H alloys across Mars core pressure-temperature conditions. Generated random senary mixtures and compared profiles to seismically inferred profiles. Considered both spin-polarized and non-spin-polarized calculations and temperature variations (±200 K). Varied Fe–S EoS within experimental uncertainties to assess robustness.
• Seismic inversion: Inverted an updated set of differential body-wave travel times and stacked P-to-s receiver functions (Ps RF), jointly with bulk geophysical constraints (planet mass and moment of inertia), to derive radial profiles of P- and S-wave speeds and density, particularly in the CMB region. Explored models with and without a molten silicate layer (LSL).
• Phase identification: Predicted and searched for LSL-interacting seismic phases (e.g., PdP, SdS, P diffracted along mantle–LSL and LSL–core interfaces, and LSL reverberations). Conducted synthetic waveform analyses to assess phase moveout, waveforms, and detectability. Applied polarization analysis, waveform matching with template traces, and scalogram time–frequency analysis on observed data from near- and far-side events (e.g., S1094b, S1000a).
• Model refinement: Downsampled inverted model ensemble to satisfy observed diffracted P-wave reverberation timing within the LSL, sharpening estimates of LSL thickness and updating core radius and density.
• Cosmochemical filtering: Generated 10⁴ senary core compositions; retained those matching inverted density profiles and further constrained by cosmochemical plausibility. Used SDKDS (S-to-P conversion through LSL and core) travel-time residuals as an indicator for additional screening, acknowledging P-wave velocity uncertainties in AIMD/experiments.

• A molten silicate layer (LSL) overlies the metallic core of Mars. Seismic phases unique to a stratified CMB region (diffracted and reverberating P-waves) are observed, supporting the LSL’s presence.
• Revised core size and density: Core radius = 1,675 ± 30 km (consistent range 1,640–1,740 km); mean core density = 6.65 ± 0.10–0.15 g cm⁻³, higher than prior 6.0–6.3 g cm⁻³ estimates.
• LSL properties: Thickness = 150 ± 15 km (geophysical inversion 145 ± 25 km refined by phase reverberation constraint); mean density ≈ 4.05 ± 0.05 g cm⁻³; P-wave velocity ~4.5–5.5 km s⁻¹.
• No single homogeneous core composition can match InSight-derived P-wave velocity and density simultaneously at the previously inferred CMB and deeper within the core. The mismatch at the earlier CMB depth implies that region is molten silicate, not liquid metal.
• The core composition consistent with revised density reconciles seismology with cosmochemistry: approximately 85–91 wt% Fe–Ni and 9–15 wt% light elements (chiefly S, C, O, H). AIMD-compliant models often require H >1 wt%, implying a metal–silicate partition coefficient for H of at least ~3 or the presence of additional light species (e.g., N, P).
• LSL density (~4.0 g cm⁻³) rules out a stably stratified Fe–Ni–light-element enriched layer at the top of the metallic core (as proposed for Earth). Instead, the LSL is silicate and denser than primitive liquid silicate estimates, suggesting FeO enrichment.
• Thermal state: Inversions indicate temperatures ~2,000–2,300 K in the lowermost mantle; stability of the molten layer is consistent with required density (≥80 kg m⁻³) and viscosity contrasts (~100) between liquid and solid mantle.
• Seismic phase observations: Identification of LSL-related diffracted and reverberating P phases (PCMB, PLSL, and Pair^LSL Pair) with consistent polarization and travel-time differences; SdS reflections from the top of LSL reinterpreted from earlier ScS.
• Planet-scale constraints: Models satisfy Mars’s mass and moment of inertia while accommodating a smaller, denser core and the intervening LSL.

The findings resolve the long-standing inconsistency between seismically inferred low core density and cosmochemical limits on available light elements by reinterpreting the CMB region as containing a global molten silicate layer overlying a smaller, denser metallic core. Seismic evidence—particularly the identification of diffracted and reverberating P-wave phases tied to both the mantle–LSL and LSL–core interfaces—provides independent support for the LSL beyond non-unique radial inversions. The revised core radius and density permit a core composition with total light elements of ~9–15 wt%, aligning with plausible cosmochemical inventories, thereby alleviating the need for unrealistically high volatile contents. The LSL’s density and seismic properties suggest FeO enrichment relative to the bulk mantle, consistent with scenarios of magma ocean crystallization under reducing conditions and/or chemical exchange with the core. Stability analyses indicate that the inferred density and viscosity contrasts, along with low Mg numbers compatible with 2,000–2,300 K temperatures, allow the layer to persist geologically. The presence of an LSL has implications for mantle dynamics, tidal response, and potential geochemical signatures in plume-related magmatism; entrainment of LSL material could leave detectable isotopic and elemental anomalies (e.g., Hf–W, highly siderophile elements) in erupted rocks.

This study presents seismic and first-principles evidence for a fully molten silicate layer atop a smaller, denser liquid core on Mars. By combining differential travel-time inversions, waveform and polarization analyses of LSL-interacting phases, and AIMD-calculated thermoelastic properties of Fe–Ni–S–C–O–H alloys, the authors revise Mars’s core radius to 1,675 ± 30 km and mean core density to ~6.65 g cm⁻³, and determine an LSL thickness of 150 ± 15 km with density ~4.05 g cm⁻³. These results reconcile Martian core composition with cosmochemical constraints, implying a core of ~85–91 wt% Fe–Ni and ~9–15 wt% light elements. The LSL likely reflects FeO-enriched silicate, stable under present thermal conditions, and may imprint geochemical signatures in plume-derived magmas. Future work should include laboratory measurements to validate AIMD velocity predictions, expanded seismic observations to refine LSL topography and heterogeneity, and targeted geochemical studies of Martian basalts to search for LSL-derived isotopic and elemental signatures.

• Non-uniqueness: Radial seismic models from differential travel times, mass, and moment of inertia are inherently non-unique; independent phase identifications mitigate but do not eliminate ambiguity.
• Assumption of global LSL: The layer is assumed global and interconnected; limited lower-mantle-sensitive phases prevent robust mapping of possible topography or lateral variations. Locally thinner regions cannot be fully excluded.
• Seismic detectability: Some predicted phases (e.g., PdP, PDcDP) are weak and not definitively observed due to interference/coda; diffracted S phases were not identified.
• Property uncertainties: AIMD and experimental P-wave velocities carry uncertainties; Fe–S EoS extrapolations and density measurements have errors; composition inferences using SDKDS residuals are indicative, not definitive.
• Composition degeneracy: Multiple Fe–Ni–light-element mixtures can fit density; partitioning behavior of H (and potential roles of N, P) remains uncertain.
• Data coverage: Single-station InSight geometry and limited event distribution constrain resolution of deep structure and lateral heterogeneity.