Percolative core formation in oxidized planetesimals – I: experiments on melt migration in olivine aggregates
Varun Manilal, Kyusei Tsuno, Hideharu Kuwahara, Axel Wittmann, Anne Pommier, Christy Till, Damanveer S. Grewal
Geochimica et Cosmochimica Acta
In Press, Journal Pre-proof, Available online 15 June 2026
“Percolation is a leading mechanism for metal-silicate segregation during the earliest stages of planetesimal differentiation triggered by 26Al-heating. The efficacy of percolative core formation strongly depends on the compositions of metallic melts and co‑existing silicate mineralogies. We report results from 21 high-pressure–temperature (P = 0.5–2 GPa; T = 1350 °C) experimental charges designed to simulate percolative core formation conditions in oxidized planetesimals. These experiments equilibrated olivine aggregates having variable FeO contents (0–30 wt%) with Fe-O-S metallic melts generated from two starting compositions (eutectic-like Fe1.46S and stoichiometric FeS). Our results quantify how pressure, FeO content in olivine, and metal composition affect oxygen solubility and dihedral angles in Fe-O-S alloys. Both O solubility and dihedral angles are largely insensitive to pressure over planetesimal conditions (P < 2 GPa) but are strongly dependent on olivine FeO content and metallic composition. For eutectic systems, dihedral angles decrease from 80° to 44° for olivine containing 1 wt% and 20 wt% FeO, respectively, crossing the critical interconnection threshold near 10 wt% FeO. For FeS systems, dihedral angles decrease more gradually (67° at 5 wt% FeO to 53° at 20 wt% FeO), consistent with the lower solubility of O in FeS-rich melts and a higher FeO threshold for connectivity. We developed a Langmuir competitive adsorption framework to explain these observations: S and O compete for interfacial binding sites at the olivine-metal boundary, with their relative coverage controlled by the activities of FeO and FeS. This predictive framework explains why O solubility is systematically lower in FeS-rich melts than in eutectic melts at the same fO2 and demonstrates that both S and O act synergistically as surface-active elements that reduce dihedral angles and interfacial energies, with O exerting a stronger effect. Applying these results to differentiation of oxidized planetesimals (fO2 > IW; based on the compositions of oxidized chondrites), we find that the first-formed metallic melts would have formed interconnected networks at extremely low melt fractions and were capable of forming cores prior to widespread silicate melting in their parent bodies. Using the experimentally defined connectivity and O-solubility thresholds, we model core compositions in oxidized planetesimals as combinations of eutectic and FeS-rich Fe-O-S melts. The resulting metallic cores are predicted to be consistently O-bearing, ranging from ∼3–4 wt% O in CM-/CI-like bodies to ∼5–6 wt% O in CV-ox- and RC-like bodies. These predicted O contents are first-order estimates, as both the higher experimental temperature (1350 °C vs. ∼988–1200 °C during percolative core formation in planetesimals) and the absence of Ni in our metallic melts would likely overestimate O solubility in natural systems.”
