3919870

Bayesian-optimization-assisted discovery of stereoselective catalysts for ring-opening polymerization of racemic lactide

Date
August 15, 2023
Explore related products in the following collection:

High-performance homogeneous single-site catalysts for polymer synthesis are required for economically producing environmentally friendly degradable polymers. Trial-and-error-based discovery and optimization of polymerization catalysts can be both time-consuming and expensive because this method relies on polymer chemists’ experience and empirical knowledge, and on serendipity.
Recently, a complementary approach has emerged in the chemistry community that applies data-driven machine learning methods to capture multidimensional structure–activity relationships for catalysts. Machine learning approaches can accept numerous reagent features and reaction conditions as inputs without recourse to a specific mechanistic hypothesis, and can recognize hidden patterns in a multidimensional chemical space. This approach has been successfully used to develop enantioselective catalysts and to predict reaction yields in organic chemistry. However, to the best of our knowledge, Bayesian optimization has never been used to discover stereoselective polymer catalysts. Moreover, no efficient implementable strategy based on data science has been developed for use as a mechanistic tool for understanding nonintuitive trends in catalyst performance in polymer science.
Herein, we describe a workflow and analysis framework to achieve these goals (Figure 1). We focused on Al-mediated stereoselective ring-opening polymerization (ROP) of racemic lactide, which affords stereoregular poly(lactic acid). Starting from literature data points for tetradentate salen-type Al complexes, we showed that our Bayesian optimization model can guide the discovery of multiple high-performance isoselective and heteroselective Al complexes for the ROP of rac-LA. Analysis of the machine-learned results revealed important albeit nonintuitive descriptors that can be used for mechanistic studies. Ultimately, our framework serves as an important quantitative tool for both iterative catalyst discovery and mechanism rationalization in polymerization chemistry.
Figure 1. Overview of Bayesian optimization for discovery stereoselective catalysts.

Figure 1. Overview of Bayesian optimization for discovery stereoselective catalysts.

Presenter

Speakers

Speaker Image for Hongliang Xin
Virginia Polytechnic Institute and State University

Related Products

Thumbnail for Polymer-to-polymer chemical transformations to produce specialty plastics from waste polyolefins
Polymer-to-polymer chemical transformations to produce specialty plastics from waste polyolefins
Recent advances in the dehydrogenation of polyolefins are opening new opportunities for polymer-to-polymer chemical recycling of waste plastics…
Thumbnail for Bayesian optimization of aluminum complexes for ring-opening polymerization of racemic Lactide
Bayesian optimization of aluminum complexes for ring-opening polymerization of racemic Lactide
Reaction optimization is experimentally costly due to the tremendous chemical search space spanned by functional materials and condition variables that determine the reaction outcomes…
Thumbnail for POLY Business Meeting
POLY Business Meeting
DIVISION/COMMITTEE: [POLY] Division of Polymer Chemistry
Thumbnail for POLY Programming Meeting
POLY Programming Meeting
DIVISION/COMMITTEE: [POLY] Division of Polymer Chemistry