| dc.description.abstract | Materials properties depend strongly on chemical composition, i.e., the relative amounts of each chemical element. In metallic alloys, small changes in composition can alter mechanical strength, thermal conductivity, corrosion resistance, and phase stability. This sensitivity arises from changes in the local chemical environments that atoms experience—features that vary significantly across chemically ordered and disordered phases. In ordered phases, specific atomic motifs repeat periodically, whereas in chemically disordered phases, local chemical environments vary widely with composition and temperature. Both types of phases are essential to high-performance structural alloys, high-entropy materials, and chemically complex catalytic surfaces. Yet, the configurational complexity of disordered phases poses a particular challenge for predictive modeling.
This thesis addresses the fundamental question: "What does it mean to be chemically complex?"
We will see that the understanding of chemical complexity is key to enabling predictive modeling of high-entropy materials across composition, particularly for the modeling of metallic alloys, where disordered solid solutions are the prevalent phases across their phase diagrams.
Traditional approaches to materials design rely on simplified models that fail to capture the intricate interplay between chemical disorder, local atomic environments, and emergent properties. This thesis bridges this gap by combining geometric deep learning with large-scale molecular dynamics simulations to characterize, predict, and control chemical complexity at the atomic scale, ultimately opening up the rational design of next-generation disordered materials with tailored properties. | |
