Week 24 AI Lab Review: Superconductor Research Summary

Week 24 Research Review

Analysis

Week 24 produced two standout outcomes: the quaternary Li₂(Mg₁₋ₓCaₓ)(Be₁₋ᵧBᵧ)H₁₆ system reached the highest Tc (432 K) but only at ~198 GPa, while the boron-free Li₂(Mg₁₋ₓCaₓ)BeH₁₆ achieved 377 K at a dramatically lower 69 GPa — by far the best Tc-per-GPa figure of merit of the week and the most experimentally actionable lead. Rare-earth substitution (Y, La) further reduced the optimal pressure to 42–76 GPa but cost ~100 K of Tc, indicating that large trivalent cations soften the lattice and lower pressure requirements at the expense of H-derived DOS at E_F. Conversely, Al/Sc co-doping and carbon substitution both degraded performance while pushing optimal pressure above 280 GPa, suggesting these dopants introduce electron-count changes or charge localization that suppress the σ-bonding hydrogen network; these branches should be deprioritized. Gemini's feedback across all five days converged on a single dominant deficiency: the complete absence of methodological reporting (DFT functional, structure-prediction algorithm, electron-phonon coupling method, μ*, anharmonic corrections, and — critically — the actual optimal (x, y) compositions). Week 25 must therefore combine chemical exploration with a methodological hardening campaign: report compositions explicitly, perform anharmonic phonon and dynamical-stability checks, and screen for metastability at recoverable pressures. Unexplored high-value directions include Ca-rich ternary optimization near the 69 GPa sweet spot, dilute (x < 0.15) rare-earth micro-doping of the Ca-Be system to combine both pressure-lowering mechanisms, Sr/Ba as alternative alkaline-earth substituents, and partial Li→Na substitution to tune the electron donation to the H sublattice.

Next Week Preview