DFT and machine-learning datasets for main figures in “Hydrogen in Brownmillerite Perovskites: First-Principles Insights into Energetics and Induced Electronic–Magnetic Changes”

This dataset contains the numerical data used to generate the main figures of the article“Hydrogen in Brownmillerite Perovskites: First-Principles Insights into Energetics and Induced Electronic–Magnetic Changes.”The data were obtained from first-principles density functional theory (DFT) calculations (SCAN + U and PBE + U) and machine-learning interatomic potential (uMLIP) predictions (UMA and CHGNet). We explored hydrogen absorption energetics, electron localization, magnetic exchange interactions, canting-angle energetics, and benchmarking of machine-learning models against DFT reference calculations in a series of brownmillerite oxides.Specifically, the dataset includes:Hydrogen absorption energies at all symmetry-distinct intercalation sites for multiple A₂B₂O₅ brownmillerites (Figures 2, 3, 7).Polaron localization energies as a function of proton–transition-metal separation (Figure 2).Total energies of multiple magnetic configurations used to fit Heisenberg Hamiltonians and extract magnetic exchange couplings (Figure 5).Exchange coupling constants before and after hydrogen intercalation (Figure 5).Canting-angle–dependent total energies for pristine and hydrogenated compounds, used to analyze hydrogen-induced Néel-vector canting (Figure 6).Parity datasets comparing hydrogen absorption energies predicted by UMA and CHGNet against DFT/SCAN reference values across 95 intercalation sites in 14 brownmillerite compounds (Figure 8).All files are provided in CSV format, are fully site-resolved, and correspond directly to the plotted quantities in the main text. The dataset enables full reproducibility of the figures, independent reanalysis, and reuse for benchmarking electronic-structure methods and machine-learning models applied to hydrogen intercalation in correlated oxides.Representative VASP input files (INCAR) and structural files (POSCAR) for hydrogenated Sr₂Fe₂O₅ calculations are also provided to illustrate typical computational settings and structural models used in this work.