Qin Gao

Qin Gao

Ph.D. candidate in theoretical Physics
M.S. student in Machine Learning
Carnegie Mellon University

Physics Machine Learning


I enjoy constructing theoretical study of emerging phenomena in physics from first principles. I'm starting working on the interdispline area of Physics and Machine Learning. Some of my primary research areas are described below; related publications are listed here.

First Principle study of solid surface and interface

Bi at Transition Metal Grain Boundaries

Physics at surfaces and interfaces is especially attractive to me. Solids are usually not perfectly arranged single crystals but consist of many randomly oriented micro-size crystals that are called grains. The boundaries between adjacent grains are chemically active and favorable for other atoms to segregate. The grain boundaries and segregates strongly influence the material properties. Recently, our collaborators at UCSD and Lehigh published their experimental results of bilayer Bismuth (Bi) films segregation at Nickel (Ni) grain boundaries in Science. Their observation could possibly explain the long standing puzzle of the liquid embrittlement of metal, where ductile metals become brittle in the presence of specific liquid metals. Together with my advisor, I succeeded in theoretically understanding and explaining these phenomena. Moreover, we also proposed a model to predict the occurrence of segregation in general cases. Lastly, the underlying reason of different Bi stability at various transition metal grain boundaries is also resolved based on electronic structure and magnetism in the systems. Our seminal result has been submitted to Physical Review Letters recently. A copy of the paper is here.

Bi on Ni(111) surface

As a precursor to the interfacial study, we have investigated Bi growth on Ni(111) surface. We refer to a recent published experimental paper by T. Bollmann et al. on Phys. Rev. Lett. which shows that Bi forms 1, 3, 5, and 7 layer films on the Ni (111) surface. The authors suggest these selected height films are stabilized by a quantum size effect (QSE) with the hexagonal structure. In their explanation, free electrons in Bi films are confined in the vertical direction and favor certain thickness because of their quantum nature. To test this idea, we calculated the total energy and QSE of Bi on Ni using density functional theory. Surprisingly, it turned out that their simple interpretation of the experiment appears to be far from the truth. We first find the hexagonal films with observed thickness are unstable. Besides, the hexagonal films are destabilized by adding capping atoms which lead to puckering of the hexagonal layers. Furthermore, we find that rhombohedral films based on the bulk Bi structure are energetically much more favorable than their proposed hexagonal films. These structures also favor odd numbers of layers (a flat wetting layer followed by bulk-like rhombohedral bilayers), but owing to covalent chemical bonding rather than QSE. A copy of the paper is here.

Electron reflectivity on solid surface

Together with our colleague Prof. R. Feenstra, Prof. Widom and I developed a novel first-principles electron reflectivity calculation method recently. When low energy (several eV) electrons approach the solid surface, part of them are reflected. The intensities of the reflectivity at different energies contain information of the surface structures. In this method, we first compute the wavefunctions of electrons from first-principles calculations. We then match the wavefunctions with free electrons in a sophisticated way to mimic the experiment and extract the reflectivity. This method provides a direct comparison between theory and experiment. We applied the method to graphene on metals and the resulting reflectivity curves coincide with experiments very well. We have published a series of papers on this method and several more papers of its application are in preparation. Prof. Widom and I will further develop our electron reflectivity calculation methods with Prof. Feenstra. A representative paper is here.

Machine Learning in solid state Physics

Starting soon.