High-throughput screening and mechanistic insights into solid acid proton conductors

This dataset provides a two-stage, motif-driven screening and refinement workflow for solid-acid candidate materials, organized into step1 (3,967 screened structures with ready-to-run ASE MD inputs for a foundation-model potential) and step2 (~30 top candidates with complete fine-tuning and evaluation data). For each selected step2 candidate, the deposit includes the fine-tuned MACE-MP-0 model checkpoint, fine-tuning configuration and training script, ab initio reference dataset, training logs, MD input files and simulation cell, plus an errors.txt report summarizing energy/force errors from DFT evaluation on configurations drawn from an MLIP trajectory. Overall, it enables reproducible foundation-model MD for large-scale candidate sets and reproducible MLIP fine-tuning/evaluation for a curated subset using consistent per-material folder naming ({prefix}_{MAT_ID}).