Mechanism-Guided Discovery of Superior Corrosion Resistance in High-Entropy Alloys

NON-TECHNICAL SUMMARY This research project is addressing a major challenge in designing advanced metal alloys that resist corrosion in harsh environments. High entropy alloys, which contain several elements in nearly equal amounts, show strong potential for corrosion resistance, but identifying the best compositions is difficult because the design space is extremely large. This project is developing a machine learning framework that integrates experimental measurements and physics-based descriptors to predict corrosion rates and identify why alloys fail. By learning from both laboratory data and computer simulations, this data-driven approach accelerates materials discovery far beyond traditional, slow, and expensive trial-and-error testing. The framework is also revealing the mechanisms that govern the formation and breakdown of protective surface films on alloy surfaces. This research is promoting the progress of science by generating new understanding of corrosion processes in complex alloys and by enabling the design of next-generation materials with improved durability. This project is advancing the national health, prosperity, and welfare by supporting the development of longer-lasting materials for critical sectors such as energy, transportation, marine, aerospace, and biomedical engineering, where corrosion causes costly failures. The project is also strengthening workforce development by training students in materials science, electrochemistry, machine learning and data science. Through outreach events for K–12 students, the project is inspiring future scientists and engineers and expanding participation in STEM fields. All data, models, and software created through this award are being shared openly, ensuring broad public benefit. TECHNICAL SUMMARY This research project is developing an integrated experimental, computational, and machine learning framework to predict and understand corrosion behavior in high entropy alloys. The work focuses on quaternary alloy systems and is advancing a physics-informed machine learning model that combines electrochemical measurements, density functional theory calculations, and mechanistic descriptors. The project is identifying the material features that govern corrosion rates, passivation behavior, selective dissolution, and surface film stability in high entropy alloys. The technical objectives include: (1) developing a predictive model that provides corrosion rate estimates and uncertainty bounds for a broad set of alloy compositions; (2) refining model performance through active learning, where new experiments are guided by regions of high prediction uncertainty; (3) identifying controlling mechanisms by analyzing the descriptors selected by the machine learning model, then validating them through surface characterization and atomistic simulations; and (4) generalizing percolation-based passivation theory, originally developed for binary alloys, to multi-principal element systems. This generalized framework is revealing how the connectivity of elements that promote passivation influences the formation and repair of protective surface films across a wide compositional space. This work is generating new knowledge on the multiscale relationships among composition, electronic structure, surface reactivity, and corrosion resistance in concentrated alloys. The approach is producing a mechanism-guided pathway for alloy design that reduces reliance on empirical screening and lowers the cost and time required to discover corrosion-resistant materials. The research aligns with the mission of NSF by advancing fundamental scientific understanding and providing tools for rapid materials discovery. Broader impacts include an open-source machine learning toolkit, publicly available data resources, and training for students in interdisciplinary materials design. Project outcomes support national interests in infrastructure durability, energy efficiency, and materials reliability. NSF Award ID: 2530512 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Wenjun Cai | Institution: Virginia Polytechnic Institute and State University, BLACKSBURG, VA | Award Amount: $486,349 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2530512 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2530512.html