AI-driven Exploration of Topological Quantum Materials

Abstract

Topological quantum materials have emerged as promising platforms for next-generation technologies in energy, computing, and information processing. However, their exploration faces two major challenges: reliably detecting exotic topological states from experimental signals and identifying materials that balance quantum functionality with economic and environmental sustainability. In this thesis, I develop artificial intelligence (AI)-driven approaches to address both fronts. First, I introduce a machine learning framework that distinguishes Majorana zero modes (MZMs) from spurious zero bias peaks in scanning tunneling spectroscopy, integrating quantum transport simulations, topological data analysis, and deep learning to achieve robust classification even in noisy experimental regimes. Second, I establish a data-driven methodology to screen over 16,000 topological materials by combining thermodynamic stability, supply-chain resilience, toxicity, and environmental footprint with the quantum weight, an AI-predicted metric quantifying electronic quantumness. The analysis uncovers a striking correlation between enhanced quantum behavior and increased sustainability costs, and identifies a small set of viable candidates that optimize both. Together, these efforts demonstrate how AI can bridge the gap between theoretical prediction, experimental verification, and sustainable design of topological quantum materials, charting a pathway toward their scalable deployment in future quantum and energy technologies.