Computational Discovery and Design of Materials : Poster Session
Sponsored by: TMS Materials Processing and Manufacturing Division, TMS: Computational Materials Science and Engineering Committee, TMS: Integrated Computational Materials Engineering Committee
Program Organizers: Houlong Zhuang, Arizona State University; Duyu Chen, University of California, Santa Barbara; Ismaila Dabo, Pennsylvania State University; Yang Jiao, Arizona State University; Sara Kadkhodaei, University Of Illinois Chicago; Mahesh Neupane, Army Research Laboratory; Xiaofeng Qian, Texas A&M University; Arunima Singh, Arizona State University; Natasha Vermaak, Lehigh University
Tuesday 5:30 PM
March 21, 2023
Room: Exhibit Hall G
Location: SDCC
Session Chair: Houlong Zhuang, Arizona State University
M-16: Building an ImageNet for Materials Grain Boundaries: Huolin Xin1; Jose Venegas2; Chengyun Zhao1; 1University of California - Irvine; 2Syracuse University
Recent development in machine learning has nearly solved all computer vision problems that challenged the field in the past thirty years, however, most of these successful approaches are based on supervised learning meaning that ground truth labeling is needed for the training set. In this respect, the ImageNet project has become the cornerstone of the computer vision AI field. It shows the power of existing AI algorithms when labeled data are provided. In this work, we will present our effort to fill in the gap for the scientific imaging field. Specifically, we build a large library of scientific images where grain boundaries are present and the ground truth label are provided. We show that with this GrainBoundaryImageNet, we can build highly accurate models for providing inference for research on materials grain boundaries. This research is supported by the National Science Foundation through the UCI CCAM (DMR-2011967).
M-17: Generative Adversarial Networks and Diffusion Models in Material Discovery: Michael Alverson1; Sterling Baird1; Taylor Sparks1; 1Department of Material Science and Engineering
The idea of material discovery has excited and perplexed research scientists for centuries. Several different methods have been employed to find new types of materials, ranging from the arbitrary replacement of atoms in a crystal structure to advanced machine learning methods for predicting entirely new crystal structures. In this work, we had three primary objectives. I) Introduce CrysGraph, a crystal encoding that can be used in a wide variety of deep learning generative models. II) Investigate and analyze the performance of Generative Adversarial Networks (GANs) and Diffusion models to find an innovative and effective way of generating theoretical crystal structures that are synthesizable and stable. III) Show that the models that have a better "understanding" of the structure of CrysGraph produce more symmetrical and realistic crystals and exhibit a better apprehension of the dataset as a whole.
M-28: Molecular Dynamics Investigation of Electrochemical Systems: Lingxiao Mu1; Ismaila Dabo1; Susan Sinnott1; 1The Pennsylvania State University
The computational investigation of voltage-dependent solid-liquid interfaces is challenging for extended time and length scales. Molecular dynamics can simulate large systems over a longer period compared to DFT. Reactive potentials such as COMB (Charged-Optimized Many-Body) Potentials are widely used for describing complex systems in molecular dynamics simulations. Au-Pt, Pd-Pt, and Ni-Pt nanoalloys systems are of our interest since they are excellent cases for testing metal migration mechanisms. In this work, molecular dynamics simulations with COMB3 and eCOMB potentials are applied for uncharged systems and systems with external voltages, respectively. The influence of size and surface curvature for Au-Pt, Ni-Pt, and Pd-Pt core-shell nanoparticles in explicit water, the role of intermetallic diffusion, and the impact of external voltage are examined. Besides, due to the incomplete parametrization of COMB3 potentials, datasets generated from DFT calculations are used for refining the Au-Pt, Pd-Pt, and Ni-Pt COMB3 parameters.