Large language model-assisted design and thermodynamic screening of electrophilic thiocyanating reagents

Electrophilic thiocyanation directly installs SCN groups, yet the design of thiocyanating reagents remains largely empirical. We report a large-language-model (LLM)-assisted workflow for candidate ideation, manual structure curation, relative thermodynamic screening, and electronic-structure interpretation of Y-SCN-type electrophilic thiocyanating reagents. LLM-proposed structures were inspected and curated, and 38 candidates were evaluated using a common DFT protocol. Relative SCN-transfer tendency was quantified by ΔΔG_SCN^NTS, an isodesmic exchange free energy referenced to N-thiocyanatosuccinimide, while homolytic-cleavage tendency was assessed by the Y-SCN bond dissociation enthalpy, ΔH(BDE). The ΔΔG_SCN^NTS values range from −55.51 to +17.47 kcal mol⁻¹, revealing wide variation across the library. Multiwfn analysis relates SCN-transfer free energy to SCN charge distribution and polarization, and ΔH(BDE) to local Y-S-C bonding and geometry. A nitro-saccharin-derived candidate, 6-nitro-N-thiocyanatosaccharin, was synthesized and characterized. Overall, the workflow provides a transparent thermodynamic decision map for prioritizing LLM-assisted reagent proposals.