The electrochemical activation of thiocarbonylthio chain-transfer agents has been used for initiating reversible addition-fragmentation chain transfer (RAFT) polymerizations due to processing under mild conditions without generating initiator-derived byproducts. To design more effective electrochemical RAFT agents, it is important to understand their fundamental structure-activity relationships. In this study, we developed two separate data-driven machine learning models, multiple linear regression (MLR) and Bayesian regularized artificial neural networks (ANNs), to computationally extract structure features in correlation with half-peak reduction potentials of RAFT agents. The models demonstrate the robustness with highly accurate predictive abilities and enable the design of new electrochemical thiocarbonylthio RAFT agents with desired electrochemical behaviors.

Machine learning-enabled prediction and design of thiocarbonylthio chain-transfer agents for RAFT polymerizations

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.