Automated Reaction Transition State Search for Bimolecular Liquid-Phase Reactions Using Internal Coordinates: A Test Case for Neutral Hydrolysis

Abstract

Transition-state (TS) identification for bimolecular liquid-phase reactions is notoriously sensitive to the initial spatial arrangement of reactants, making automated searches difficult, especially in solvation where conformational effects dominate barrier heights. We address this gap with a fully automated, heuristic framework integrated into ARC that generates TS guesses for neutral hydrolysis, as a model reaction type, by positioning water using atom-centered internal-coordinate rules derived from representative DFT-optimized cases. The approach is parameterized for three hydrolysis families, carbonyl-based (esters, amides, acyl halides), ethers, and nitriles, and operates from reactant/product SMILES alone. Validation across 91 diverse reactions shows that chemically guided internal-coordinate placement yields relatively high success rates under SMD-water conditions: 96.9% for carbonyl-based substrates, 86.2% for nitriles, and 72.4% for ethers, consistent with the greater conformational variability and weaker intrinsic directionality of ether substrates. An ablation study highlights that small, targeted reactant-dihedral adjustments and ±φ sign-sampling are essential to robustly align the water nucleophile, while the electronegativity-based neighbor ranking primarily fine-tunes local orientation. By automating the classically manual step of water placement and orientation, and producing chemically faithful geometry initializations, this framework enables scalable, high-throughput TS searches for neutral hydrolysis reactions. It provides a practical foundation for mechanistic studies and kinetic modeling in condensed-phase organic chemistry. The methodology is readily extensible to additional solvents and catalytic regimes, and to other bimolecular liquid-phase reactions where directed fragment placement is the key bottleneck.