Machine-learning-guided discovery of functional polymers

Modern polymer science suffers from the curse of multidimensionality. The large chemical space imposed by including combinations of monomers into a statistical copolymer overwhelms polymer synthesis and characterization technology and limits the ability to systematically study structure−property relationships. This is especially relevant for applications that demand numerous properties to be optimized simultaneously. To tackle these challenges in high dimensional structure space inherent to statistical copolymers, we have employed a combined experimental and data science approach to materials discovery to expedite the multiobjective optimization required to identify performance-advantaged material compositions. The combination of machine learning of processes that span multiple elementary steps combined with computer-guided exploratory synthesis led to a data-driven approach that enabled the experimentally efficient identification of complex and non-intuitive structure–property relationships for a number of applied areas. Ultimately, exploiting the machine learning algorithms enabled accurate prediction of high performing materials for both MRI imaging agents and 3D printed polymer networks.