Building Physics-Based and Data-Driven Methods for Efficient Molecular Design and Reaction Optimization

Our research group focuses on building tools that enable inverse materials design and give new insights into the fundamental chemical physics of liquids, interfaces, and chemical reactions. For this talk, we will discuss our progress in two of our primary research thrusts.

The first part of the talk will focus on our work in developing methods that are used to accelerate the design of functional materials. We focus on two types of materials: electronic/redox-active polymers and intrinsically disordered polymers. Although radical-based polymers are promising energy storage materials, successful materials design requires careful molecular engineering of the polymer and electrolyte. To solve the molecular-scale part of the problem, we develop physically motivated machine learning models that predict molecular properties (e.g., hole reorganization energies) from low-cost representations, and pair these with multiscale simulations of the polymers. Next, we will discuss our efforts to use reinforcement learning methods to accelerate materials design. We are able to couple these methods directly with high-throughput computational simulation tools to accelerate the design process. Our initial demonstrations of this method are on optoelectronic organic materials design.

If time permits, we will discuss our work to understand the fundamental design principles for optimizing chemical reactions under external forces (mechanochemistry). Here, we use a combination of high-throughput screening, optimization methods, and graph-based neural network potentials to conduct a broad search for reactions that can be significantly accelerated by external forces that are achievable in modern mechanochemical reactors. Our methods use machine learning potentials in combination with reaction path searching protocols (e.g., nudged elastic band and the growing string methods) to find potential transition states. We then explore candidate “activatable” coordinates—specific deformation modes that lead to enhanced reaction rates—by analyzing a mix of localized and normal coordinates. The most promising reactions and degrees of motion are then verified by higher-level calculations.