Collaborative Research: MFS-SPEED: AI-Driven Design of Architecturally Varied and Deconstructable Polymers for a Comprehensive Circular Plastics Economy

About This Grant

In this project, artificial intelligence (AI) will be used to design new sustainable polymeric materials with a range of properties and that can be recycled without the need for costly and inefficient separation from mixed waste streams. Today’s plastic waste challenge exists at a scale of megatons per day across tens of thousands of applications and products. The researchers will create new types of depolymerizable plastics derived from simple feedstocks, and they will develop physics-informed AI models to aid design of these plastics such that they meet a variety of product specifications across a wide range of properties. Ultimately, this approach enables products of all different types, functions, and lifetimes to be integrated into a single recycling stream and accelerates their discovery-to-use timeline. The results and methods developed by this research will be publicly accessible for industrial benchmarking and include code and tutorials for users to perform AI-guided design on their own materials. Through this research, a new generation of scientists will be trained to work at the emerging intersection of polymer materials design and AI model development and use. With this award, the project will develop physics-informed AI and synthesize architecturally varied and deconstructable (ADD) polymers by cationic ring-opening polymerization (CROP) with controlled chain length, branching, and dynamic bond incorporation. This work will create new synthetic strategies to control chain end and side chain functionality, branch type and frequency, and dynamic bond incorporation for polymers produced by CROP. Using polyacetals and polyethers synthesized from a select few monomers, these complex molecular architectures will be linked to key properties using physics-informed AI, which both describes polymer architecture through sets of probability distributions and incorporates theoretical estimates of structure-property relationships. This physics-informed AI will be iteratively improved through active learning approaches and subsequently used to perform inverse design for the creation of new ADD polymers with targeted properties within specified tolerances that will be experimentally validated against industry benchmarks. This Molecular Foundations for Sustainability: Sustainable Polymers Enabled by Emerging Data Analytics (MFS-SPEED) award is co-funded by the NSF through the Division of Chemistry (CHE), the Directorate for Mathematical and Physical Sciences (MPS), and the Division of Innovation and Technology Ecosystems (ITE) in the Directorate for Technology, Innovation, and Partnerships (TIP). Additional MFS-SPEED funding is provided by Procter & Gamble, PepsiCo, Dow, BASF, and IBM. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.