Understanding matter under planetary interior conditions requires pushing beyond the limits of both experiment and conventional simulation. We present ab initio crystal structure prediction and molecular dynamics results addressing two aspects of deep planetary physics: volatile incorporation in silicate minerals, and the dynamic properties of iron alloys at inner-core conditions.

Systematic exploration of the H-Si-N-O compositional space shows that ammoniated silicas, particularly H₃Si₂NO₄ and H₆SiN₂O₂, are stable across the full lower-mantle pressure range, contrary to predictions based on the potassium-ammonium analogy. At high temperatures these phases become superionic, with protonic conductivity relevant to magnetic field stability in Uranus and Neptune.

At inner-core conditions, H, C, and O become highly diffusive at interstitial sites in hcp Fe, producing a superionic state that reduces seismic velocities and generates depth-dependent anisotropy. Anisotropic H-ion diffusion under an applied field provides a mechanism by which the geomagnetic field textures the inner core. Machine-learning force fields are extending these calculations toward a computational digital twin of Earth's deep interior.