Fig. 5: LLM-based agents facilitated reaction kinetics study.
From: An automatic end-to-end chemical synthesis development platform powered by large language models
a Workflow for reaction kinetic study copiloted by Experiment Designer, Hardware Executor, Spectrum Analyzer, and Result Interpreter agents. b Comparison of the yields for products and overoxidation byproducts of diol 12 s in different solvents (MeCN from the literature48 and DMSO in this work) and copper catalysts. c The interaction between human chemists with large language model (LLM)-based agents for reaction kinetics study (see details in Supplementary Table 19-24). The dagger symbol indicates that the numerical results were generated by the agents’ code interpreter, while the asterisk denotes that the numerical results were directly provided by the LLM. d Characteristic proton nuclear magnetic resonance (1H NMR) peaks identified by Spectrum Analyzer for calculations of reaction samples’ compositions. e The time-course concentration profile in DMSO solvent, and the fitted reaction kinetic curves given by Result Interpreter, with rate constants \({k}_{1}=22.34\,{{{\rm{M}}}}^{-1}\), \({k}_{2}=2.84\times {10}^{-3}\,{{\rm{M}}}\bullet {\min }^{-1}\), \({k}_{3}=2.51\times {10}^{-4}\,{\min }^{-1}\), and the coefficient of determination \({R}^{2}=0.996\).
