A Dataset of Plausible Proton Transfer Steps for Arrow-Pushing Mechanisms
2026-01-10
Proton transfers are fundamental steps in polar reaction mechanisms. We generated a large dataset of over 51 million kinetically plausible proton transfer steps between heteroatoms from about 8,000 acids and conjugate bases with experimental aqueous p K a s, spanning p K a values from −15 to +37. Rate factors were estimated at 25 °C using a simplified Eigen equation with p K a s but without statistical factors. Steps with estimated rate constants ≥ 10 3 M −1 s −1 were included in the final dataset. Additionally, 5,043 proton transfer steps from carbon acids to heteroatom bases were estimated using the Eigen-Bernasconi equation based on reported intrinsic rate constants and Brønsted β values. Carbon proton transfers with rate constants ≥ 10 3 M −1 s −1 were added to the final dataset. Each entry was encoded in SMIRKS format with electron-flow specification for machine learning compatibility. Diversity of structure was prioritized over diversity of conditions; calculated rate constants are expected to be accurate in aqueous environments. This approach and dataset should prove valuable for training models to predict stepwise mechanistic pathways.