Research
Collaborative Research: DMREF: Discovery of novel magnetic materials through pseudospin control
Materials discovery requires new tools that enable design principles. Notable examples are the recent advances in topological materials due to insights from the interplay of topology and symmetry. Here we propose to develop pseudospin control as a materials design tool for the discovery new magnetic states and superconductivity. The concept of pseudospin describes the two-fold Kramers degeneracy of Bloch electrons that arises at each momentum point k when the product of time-reversal T and inversion I symmetries is present. Normally, this pseudospin behaves as spin-1/2 under rotations, which drives much of our understanding of quantum materials, including Cooper pairing in superconductors, Stoner ferromagnetism, and the control of spin by Zeeman fields. Our recent work shows, however, for crystals with non-symmorphic space symmetry, the pseudospin can behave very differently than the usually spin-1/2, which can drive novel magnetic states, including altermagnets and odd-parity multipole magnets, and qualitatively alters the superconducting response to magnetic fields.This project will utilize pseudospin control as a new paradigm for materials discovery, following the collaborative and iterative closed-loop approach outlined in the MGI strategic plan, by combining analytic and predictive computational theory with experimental molecular beam epitaxy (MBE) growth and characterization with in situ scanning tunneling microscopy/spectroscopy (STM/S) and angle-resolved photoemission spectroscopy (ARPES), as well as ex-situ magneto-
optical and electrical transport measurements.
Visualizing Intertwined Quantum Phases in Epitaxial Fe-chalcogenide Films
Quantum materials exhibit emergent phenomena resulting from strong correlation effects, i.e. cooperative behavior that cannot be inferred from the behavior of individual electrons. These phenomena are often characterized by “intertwined” or competing/cooperating quantum phases. A good example is high-temperature superconductivity, which typically occurs in close proximity to symmetry-breaking electronic nematic and stripe order. This project aims to explore pathways to high-temperature superconductors by better understanding these intertwined phases in epitaxial iron chalcogenide FeX (X=Se, Te, S) thin films.
Specifically, we plan to tune 1) electron correlations by the isovalent substitution of Se and S, 2) spin-orbit coupling by the substitution of Se and Te, and 3) magnetic coupling by the substitution of S and Te. These separately tunable interactions provide an ideal platform to determine the origin of the nematic/stripe order and to reveal the interplay of superconductivity, nematicity, and magnetism in iron chalcogenides. We will first focus on FeX single-layers grown on perovskite oxide substrates by molecular beam epitaxy, where the interface enables the tuning of superconductivity through lattice strain and charge doping. FeX multi-layers will then be investigated to probe the origin of the nematic/stripe orders and their impact on superconductivity upon surface charge doping. Finally, emergent properties in magnetic/topological FeX heterostructures will be explored where the interplay of magnetism and superconductivity may lead to topologically protected Majorana states.
Comprehensive investigations will be carried out to determine these epitaxial thin films' atomic and electronic structures using scanning tunneling microscopy/spectroscopy, angle-resolved photoemission spectroscopy, and density functional theory calculations. The outcome of the project will advance the comprehensive understanding of the complexity of the electronic and topological phases in Fe-based superconductors, enabled by our unique capabilities of integrated epitaxial growth with multimodal in situ atomic-scale imaging and spectroscopy characterization, aided by theoretical modeling/calculations.
Data-Driven Autonomous Experiments for Energy Sciences: from First Principles to Machine Learning
Science paradigms have changed rapidly over the past several centuries. From empirical to theoretical (first principles) and computational (e.g., density functional theory (DFT)), we have witnessed the emerging data-driven science in which artificial intelligence (AI) and machine learning (ML) are becoming indispensable tools in physical sciences. Various experiments, such as molecular dynamics and chemical reactions, can be made “autonomous” by developing data-driven surrogate models that are more time- and cost-effective than their physical implementations. However, rapid advances in AI/ML have offered niche opportunities for next-generation data-driven models with even better performance. Due to the culture barrier, there is still a significant gap between domain knowledge dictated by first principles and surrogate models developed by ML experts. To fill this gap, we propose to expand convergence research at the intersection of AI/ML and energy sciences by developing a class of novel physics-informed surrogate models, including contrastive learning, self-supervised learning, and graph learning.