Physics-informed learning-based synthesis of functional chemical products for renewable energy applications

This project aims to revolutionize how chemical synthesis and composite material discoveries are made. Traditionally, scientists have relied on one-variable-at-a-time experimentation, which is time-consuming and often overlooks the complex interactions among various components and processing conditions. In contrast, this project introduces an innovative approach called adaptive design of experiments, which allows for the simultaneous evaluation and optimization of multiple variables in a dynamic process.

This method can significantly enhance the quality and efficiency of chemical synthesis, leading to better product properties while saving time and resources in the laboratory. The project focuses on developing advanced computational modeling tools that integrate experimental data with first-principles modeling, machine learning, and process optimization.

This integrated approach will be primarily applied to synthesizing perovskite oxides, materials critical for energy-related applications such as fuel cells and catalysis. The outcomes of this project will not only accelerate material discovery and manufacturing optimization but also serve broader societal needs by advancing energy technologies and providing a systematic framework for various manufacturing systems.

Moreover, the project will foster interdisciplinary student training and engagement, aiming to produce graduates with high-level expertise in mathematics, computation, and data science, thereby contributing to a skilled workforce in modern chemical manufacturing.

The primary objective of this project is to address the complex multivariable problem inherent in chemical synthesis by developing and applying a physics-informed machine learning-based adaptive design of experiments framework. The project will integrate first-principles modeling, machine learning, process optimization, and experimental data to create a hybrid model that predicts and optimizes synthesized materials’ microstructure, stoichiometric variation, and composition.

This hybrid model will serve as the foundation for an autonomous chemical synthesis framework, optimizing the decision variables to enhance the overall characteristics of the product. Specifically, the project will focus on synthesizing perovskite oxides to manipulate product microstructure and compositional variations to tune the electronic structure and oxygen vacancies.

The proposed approach will identify optimal operating temperature dynamics, annealing times, and other synthesis variables to improve product properties. By automating the experimental exploration process, the research will significantly reduce the cost, time, and effort typically associated with material discovery and optimization. The outcomes will enhance the synthesis process and provide a versatile platform applicable to various manufacturing systems, including chemicals, advanced materials, pharmaceuticals, and biological probes.

The project will also contribute to educational and outreach efforts, promoting interdisciplinary training and broadening participation in STEM fields.

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.