Understanding and Enhancing Stereoselective Polymerization Using a Data Science Approach.

Controlling the tacticity of synthetic polymers results in the transformation of simple chemical building blocks into valuable materials with emergent physical properties. Mechanistic insight into stereoselective polymerizations enables hypothesis-driven improvements to catalysts and gives access to polymers with systematic differences in tacticity for structure-property studies. Studying the mechanism of stereoselective polymerization, especially of heteroatom-containing monomers that polymerize through ionic intermediates, is hindered by the challenges of using traditional physical-organic or computational approaches. Here, we use a combination of experiments and computationally derived molecular descriptors to identify quantitative relationships between catalyst structure and stereoselectivity through a data science approach. Stereoselective polymerization of benzyl vinyl ether derivatives with a structurally diverse library of imidodiphosphorimidate (IDPi) catalysts resulted in 40 experimental data points, which were correlated to computationally derived molecular descriptors by using multivariate linear regression analysis. The regression model identified the dihedral angle of the 1,1'-binaphthyl-2,2'-diol (BINOL) subunit of the IDPi to be strongly correlated to isotacticity, which led us to reconsider the long-standing hypothesis for the conformation of the propagating polymer chain-end during cationic vinyl ether polymerization. We anticipate that the specific insights of this study will inform the next generation of catalysts for stereoselective cationic polymerization and that the data-driven approach to understand the mechanism for stereoselective polymerizations demonstrated herein will be an invaluable tool in catalyst design and discovery for polymer chemistry broadly.