ChE PhD Dissertation Defense: Yujia Wang

Name:
Yujia Wang

Title:
Machine Learning and Spatially Resolved Spectroscopy for Controlled Synthesis and Novel Phenomena Discovery in Monolayer Molybdenum Disulfide

Date:
06/20/2024

Time:
10:30:00 AM

Committee Members:
Prof. Swastik Kar (Advisor)
Prof. Joshua Gallaway
Prof. Steve Lustig
Prof. Francisco Hung

Location:
Burlington Campus: Building 5, Room C

Abstract:

Two-dimensional (2D) materials, particularly transition metal dichalcogenides like molybdenum disulfide (MoS₂), represent a promising platform for next-generation semiconductor technologies due to their atomic-thin structure, tunable electronic properties, and unique optical characteristics. However, realizing their full potential is hindered by two important challenges: the time-consuming, trial-and-error nature of chemical vapor deposition (CVD) synthesis optimization, and the difficulty in accurately characterizing spatially inhomogeneous material properties. In this dissertation defense, I will present detailed investigations into the development of new methods for addressing these challenges through the integration of machine learning-guided synthesis and high-resolution spatially resolved optical characterization techniques.

First, I developed an experimental “hyperspace” design system that serves as an interface that adapts synthesis parameters to machine learning frameworks. The machine learning algorithms were developed through collaborations with Prof. Xiaoning (Sarah) Jin’s group (Department of Mechanical and Industrial Engineering). The algorithms efficiently identify pathways towards synthesis optimization in complex multidimensional experimental hyperspaces, accelerating the synthesis towards the highest sample quality. Our results showed up to 85% reduction (two months instead of 1 year) in trial-and-error efforts, and rapid optimization of the ideal synthesis conditions.

Second, I established a spatially resolved optical spectroscopy platform for characterizing the as-grown 2D materials by combining Raman and photoluminescence (PL) mapping at micrometer resolution. Going much beyond traditional mapping techniques, this technique enables simultaneous extraction of strain, doping, and excitonic properties across individual flakes, revealing growth-intrinsic mechanisms for controlled engineering of electronic and optical properties. Through systematically varied CVD conditions, I demonstrated deterministic control over both orbital bandgaps and spin-orbit coupling, achieving the first purely optical experimental observation of strain-tunable spin-orbit splitting in as-grown monolayer MoS₂.

Finally, I applied this integrated platform to explore iron-doped MoS₂ – to further elucidate the role of doping in 2D semiconductors. This work established valuable synthesis protocols and identified distinctive optical signatures that serve as indicators of dopant incorporation in 2D materials.

My research demonstrates how coupling experimental design and scanning-based characterization approaches with advanced computational methods can substantially accelerate materials discovery and establish new pathways for deterministic property engineering in 2D semiconductors, with implications extending to quantum technologies, optoelectronics, and spintronic applications.