2026-10-07T07:03:44Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33332610 2026-09-25T18:02:49Z category_30166 category_25258 item_type_3 month_year_09_2026

FP-DeErr DFT-NEB reference: 154 Li- and Na-ion migration pathways Kiyan Amirian (24649185) Ramanuja Srinivasan Saravanan (23858004) Yifei Mo (3717493) Condensed matter modelling and density functional theory Computational chemistry foundation potentials machine-learning interatomic potentials error decomposition nudged elastic band ion migration solid electrolytes DFT FP-DeErr benchmark <pre>Density-functional-theory nudged-elastic-band (DFT-NEB) reference calculations for 154 Li- and Na-ion migration pathways, used as the ion-migration benchmark in FP-DeErr. The pathways span 106 structurally distinct materials and a range of chemistries including oxides (32.1%), sulfides (20.8%), halides and multi-anion compounds. Each pathway carries seven finalized DFT-NEB images (1,078 images in total), with the total energy and per-atom forces for every image, plus the unrelaxed source endpoint structures used as the starting geometry for foundation-potential NEB calculations. Structurally distinct materials were identified by matching the DFT-NEB endpoint structures with pymatgen's StructureMatcher at its default tolerances (ltol = 0.2, stol = 0.3, angle_tol = 5), with primitive-cell reduction and volume scaling enabled. Two source entries are treated as the same material when any of their endpoint structures match, which requires no choice of representative structure; the count is unchanged whether the DFT-relaxed initial, DFT-relaxed final, or unrelaxed source endpoints are used. Note on sum_formula: this field preserves the nominal composition reported for the source entry and may contain partially occupied or minor dopant species that are absent from the modeled DFT structure. Nine pathways are affected. The deposited structure fields define the actual compositions used in the calculations and should be used for any composition-based analysis. The migration pathways were selected from the high-throughput ion-migration dataset of Saravanan et al., available at https://doi.org/10.6084/m9.figshare.32221506 and licensed CC BY 4.0. The DFT-NEB reference calculations were performed in this work and are released in this record. DFT was performed with VASP using the PAW approach and the PBE functional, largely following the Materials Project input set: an energy cutoff of 520 eV, Gaussian smearing with a width of 0.05 eV, and spin polarization, with Hubbard U corrections applied where applicable using the Materials Project U values and the Dudarev scheme. Endpoints were relaxed at fixed cell shape and volume, five images were interpolated between them, and the NEB used a force convergence criterion of 0.05 eV/Å. These settings are also recorded in the file's dft_calculation_metadata block. SHA-256 of the gzip file: 7f73b8d3b1f8167c67393f35bf1eebeb5dce87f831a2377cfc1af74f1ce3012e SHA-256 after decompression: 7d48d1b75033dedc013ec7b31bd9cc94604f7bbf3eb5528a8a87b6ebadac7a2c The accompanying README documents the full JSON schema. Analysis and generation code: https://github.com/mogroupumd/FP-DeErr . Leaderboard: https://mogroupumd.github.io/FP-DeErr/ </pre> 2026-09-25T18:02:49Z Dataset Dataset 10.6084/m9.figshare.33332610.v2 https://figshare.com/articles/dataset/FPBench_DFT-NEB_reference_154_Li-_and_Na-ion_migration_pathways/33332610 CC BY 4.0