Supporting Dataset for Data-Driven Discovery of Structure–Reactivity Relationships in Oxidative Addition at Zerovalent Nickel Centers

This record contains the computational chemistry datasets supporting the study “Data-Driven Discovery of Structure–Reactivity Relationships in Oxidative Addition at Zerovalent Nickel Centers.” The datasets describe CO2 insertion into zerovalent bipyridine–Ni–alkyne complexes to form Type I and Type II nickelacycle products. They include MACE-MP-0a-large optimized reactant and product geometries, vibrational and thermochemical quantities generated through the Interactive Quantum Chemistry (IQC) workflow, paired reaction-level data, uncorrected and benchmark-corrected reaction Gibbs free energies, and the reference subset used to derive the reaction-energy correction. Files iqc_thermo_results_20260404_095146_reduced.parquetIQC output containing optimized reactant and product geometries and associated thermochemical properties. The file contains 58,157 structure records and includes optimized XYZ geometries, Gibbs free energies, electronic energies, vibrational frequencies, convergence information, and calculation provenance. reaction_data_06_12_raw.parquetReaction-level dataset containing 15,003 paired reactant–product pathways. Each record is identified by ligand pair, stereochemical orientation, and CO2-insertion type and includes the corresponding reactant and product geometries and uncorrected reaction Gibbs free energy. reaction_data_06_12_corrected.parquetThe same 15,003 reaction pathways with the benchmark-derived correction of −20.35 kcal mol−1 applied to the reaction Gibbs free energies. This is the recommended dataset for the reaction-energy analyses and machine-learning applications reported in the associated study. mace_mp.parquetMACE-MP-0a-large results for the representative reaction subset used in the reaction-energy benchmark. pbe.parquetCorresponding reference electronic-structure results for the benchmark subset used to determine the reaction-energy correction. The associated data-processing scripts, descriptor calculations, and IQC-Dashboard visualization software are maintained in the public IQC-Dashboard GitHub repository. The repository provides tools for browsing Parquet datasets, inspecting optimized molecular structures, analyzing convergence and thermochemical quantities, and visualizing structure–reactivity relationships. Associated preprint Elliott, S. N.; Reji, R.; Chen, Q.; Zheng, X.; Riaz, S.; Yan, X.; Pham, T. D.; Huerta, E. A.; Glusac, K. D.; Keçeli, M. “Data-Driven Discovery of Structure–Reactivity Relationships in Oxidative Addition at Zerovalent Nickel Centers.” ChemRxiv 2026. DOI: 10.26434/chemrxiv.15004105/v1.