Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Vivek B. Shenoy


The creation, control, and communication of information is the foundation of our digital economy. But as our demands for information processing continually increase, fundamental limitations in the current generation of electronics threaten to bottleneck progress towards increased computational power. A new paradigm of information processing based on coherent quantum states will sidestep these limitations entirely and have far-reaching impacts in fields ranging from energy consumption to computational modeling of new functional materials. The primary goal of this thesis is to use physics-informed rational design to develop platforms for the creation and control of quantum states in layered materials. This challenge is addressed by 1) accelerating the synthesis of new functional materials with machine learning; 2) designing quantum materials with multiscale modeling and simple phenomenological models; and 3) engineering layered materials for ultra-compact solid-state devices that will enable efficient information processing and storage. A new type of semi-supervised machine learning was developed to predict newly synthesizable layered transition metal carbides and nitrides. A crystal field model was applied and an electro-mechanical model parameterized by first-principles calculations was developed to investigate the origins and control of magnetism in layered transition metal carbides and nitrides. The insights from these investigations, combined with high-throughput simulations and machine learning, were deployed at scale on a database of over 100,000 inorganic crystal structures to identify multi-order quantum materials with coexisting magnetic and topological orders. An analytic model, tight-binding, and continuum approaches were used to predict intrinsic confinement of Dirac fermions in lateral heterostructures of transition metal dichalcogenides and engineer device architectures for optimal confinement. Deep transfer learning and machine learning models were used to map monolayer materials and their point defect structures directly to key properties relevant for controllable two-level systems. We explained trends in defect formation energy within a minimal physical picture of defect formation and identified point defects that are of interest for quantum and neuromorphic information processing. We anticipate that the models, methods, and findings presented in this thesis will contribute to a greater understanding of engineerable quantum states in layered materials.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."