Epscor research fellows: NSF: closed-loop autonomous AI platform for rapid discovery and optimization of compositionally complex ceramics

The National Science Foundation Office of Integrative Activities awarded Iowa State University of Science and Technology $300,000 on January 1, 2027, under the Integrative Activities program (CFDA 47.083) to support an EPSCOR Research Fellows project developing a closed-loop autonomous artificial intelligence platform for rapid discovery and optimization of compositionally complex ceramics.

The project will develop an AI framework coupling Bayesian optimization and deep reinforcement learning with machine-learning force-field atomistic simulations to systematically explore the composition space of high-entropy carbides. Using a quinary carbide (TI-ZR-HF-NB-TA)C as proof-of-concept, the work will deliver the first systematic head-to-head benchmarking of these two learning paradigms on an identical materials-optimization problem and establish quantitative composition–property relationships governing strength and ductility. The resulting open-source platform will be generalizable to other high-dimensional materials systems, including high-entropy alloys, battery materials, and semiconductors. These advanced ceramics are targeted for aerospace and nuclear energy applications.

The fellowship supports an associate professor at Iowa State University and provides training for undergraduate and graduate students through coursework and hands-on activities. Work is conducted in collaboration with Brookhaven National Laboratory. Performance occurs in Ames, Iowa, with a period of performance from January 1, 2027, through December 31, 2028. The project will strengthen research capacity in Iowa and release software openly for the research community.

Abstract

This research infrastructure improvement (RII) epscor research fellows project provides a fellowship to an associate professor and training for undergraduate and graduate students at iowa state university. This work is conducted in collaboration with collaborators at brookhaven national laboratory. Through the fellowship, the PI will develop artificial intelligence tools that can automatically design new ceramic materials able to withstand extreme heat, wear, and harsh conditions. Finding such materials by traditional trial-and-error is slow and costly because the number of possible chemical recipes is enormous. This project combines materials science, computer science, and physics-based simulation so that an AI agent can learn from computational results and search for the best recipes. These advanced ceramics are needed for aerospace and nuclear energy applications. The project will also strengthen research capacity in iowa, train undergraduate and graduate students through coursework and hands-on activities, and release software openly for the research community. This project will develop a closed-loop, autonomous artificial-intelligence framework that couples bayesian optimization and deep reinforcement learning with machine-learning force-field atomistic simulations to explore the high-dimensional composition space of high-entropy carbides. Using the quinary carbide (TI-ZR-HF-NB-TA)C as a proof-of-concept, the work will deliver the first systematic head-to-head benchmarking of these two learning paradigms on an identical materials-optimization problem and will establish quantitative composition?property relationships governing strength and ductility. The resulting open-source platform will be generalizable to other high-dimensional materials systems, including high-entropy alloys, battery materials, and semiconductors. The fellowship will advance the PI's expertise in artificial-intelligence-driven materials discovery, provide sustained access to national-laboratory high-performance computing, and build lasting research capability at iowa state university. These outcomes will be integrated with new undergraduate and graduate curricula, student research and mentoring, K-12 outreach, and a durable iowa state?brookhaven national laboratory partnership. This project is supported by the epscor research infrastructure improvement program: epscor research fellows (ERF). The ERF program supports early- and mid-career investigators in eligible jurisdictions to develop collaborations at the nation?S private, government, or academic research institutions. 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.- subawards are not planned for this award.