"Accelerated Discovery of Novel Layered Materials using Computational Quantum Mechanical Modeling and Machine Learning" 

Kat Nykiel, MSE PhD Candidate 

Advisor: Professor Alejandro Strachan

WebEx Link

ABSTRACT

The discovery of novel materials with exceptional properties is one of the central goals of materials science. Layered materials, characterized by highly anisotropic structures and bonding, offer one of the best opportunities to realize this goal, containing many of the highest performing materials for mechanical, electronic, thermal, and optical applications. Among them, graphene is the most well-known, but the broader domain of layered materials has seen a recent surge in exploration with the discovery of new subfamilies such as MXenes. MXenes, atomically thin transition metal carbides and nitrides derived from bulk layered phases, are particularly attractive due to their scalable synthesis and potential in energy storage, catalysis, and electronic applications. Despite their promise, many open questions remain about the compositional pervasiveness and properties of MXenes. In this work, we use high-throughput density functional theory and machine learning, in close collaboration with experimental efforts, to explore the stability and properties of novel MXene-related layered materials. We first assess the thermodynamic stability and atomic ordering of MXene precursors with multiple transition metals, followed by a search for magnetic MXenes derived from stable bulk precursor phases. Beyond MXenes, we then investigate bulk layered ceramics constructed from MXene-like motifs, characterizing their stability, mechanical, electronic, and optical properties. Finally, we explore high entropy rock salt carbides, which are structurally similar to MXene-derived layered ceramics, using density functional theory and graph neural networks in an active learning framework to identify compositions with desirable mechanical properties. Throughout this work, we emphasize high-throughput, high-fidelity quantum mechanical computational methods as a useful tool for materials discovery, and make our results findable, accessible, interoperable, and reusable.