A Descriptor for Non-Arrhenius Ion Transport Enables Design of Sulfide Superionic Conductors

Non-Arrhenius ion transport is increasingly observed in solid electrolytes but remains difficult to compare across materials or use in screening. A curated database of 867 conductivity-temperature entries from thio-LISICON and LGPS-type sulfides is used to introduce Meyer-Neldel deviation (MND), a physics-informed, material-level descriptor of temperature-dependent changes in apparent activation energy. More than 40% of the analyzed sulfide electrolytes show notable deviations from linear Arrhenius behavior. MND separates low-conductivity/high-barrier and superionic/low-barrier regimes associated with ionic-radius mismatch, electronegativity difference, configurational entropy, and mixing thermodynamics, while providing a more informative learning target than single-value activation energies. Sensitivity analyses indicate that material-level trends remain reasonably robust to variations in conductivity, reference temperature, and sampling density. Application to a reprocessed dataset of 160 compositions and 2,410 conductivity-temperature entries across seven electrolyte families supports broader applicability. MND-guided screening of 171 Li₁₀Ge_xSi_ySn_zP₂S₁₂ compositions prioritizes Li₁₀Ge_0.3Si_0.15Sn_0.55P₂S₁₂, which exhibits a room-temperature conductivity of 7.21 mS cm^{- 1}, an apparent activation energy of 0.288 eV, and weak non-Arrhenius behavior consistent with prediction. Evaluation of a second composition further supports the multistage screening strategy. These results position MND as a chemically interpretable representation for quantifying, learning, and screening temperature-dependent ion transport.