Ceramics and Glasses Modeling by Simulations and Machine Learning: Session I
Sponsored by: ACerS Glass & Optical Materials Division
Program Organizers: Mathieu Bauchy, University of California, Los Angeles; Peter Kroll, University of Texas at Arlington; N. M. Anoop Krishnan, Indian Institute of Technology Delhi
Wednesday 10:00 AM
October 20, 2021
Room: B231
Location: Greater Columbus Convention Center
Session Chair: Mathieu Bauchy, UCLA; Peter Kroll, UT Arlington; Anoop Krishnan, IIT Delhi
10:00 AM
Now On-Demand Only: Information Extraction Pipeline for Glasses: An NLP Based Approach: Vineeth Venugopal1; Sourav Sahoo1; Mohd Zaki1; Nitya Nand Gosvami1; N. M. Anoop Krishnan1; 1IIT DELHI
A large amount of information about materials is scattered in scientific journals, handbooks, patents, textbooks and other resources. The text and images comprise most of the information which is currently unstructured. To retrieve research papers related to particular topics in specialized materials science domains or get information from figure captions are trivial tasks. Therefore, to streamline information extraction from research papers, we present latent Dirichlet allocation (LDA) assisted topic labelling to obtain glass science papers on the basis of their abstract. Further, we develop “Caption Cluster Plots” (CCP) to automate information extraction from figure captions. Using both LDA and CCP, we have also developed “Elemental Maps” which disseminate the information about which chemical elements are used in abstracts of which research papers and associated figure captions. Hence, this pipeline will enable researchers to explore different material science domains and excavate the hidden information from the vast corpora of research articles.
10:20 AM
A Machine-learning Based Hierarchical Framework to Discover Novel Scintillator Chemistries: Anjana Talapatra1; Blas Uberuaga1; Christopher Stanek1; Ghanshyam Pilania1; 1Los Alamos National Laboratory
Scintillators are fascinating materials with wide-ranging applications. However, the discovery of new scintillator materials has traditionally relied on a laborious, time-intensive, trial-and-error approach, leaving a vast space of potentially revolutionary materials unexplored. To accelerate the discovery of optimal scintillators with targeted properties and performance, we present an adaptive design framework that couples density functional theory (DFT) computations and machine learning (ML) to (1) screen a large chemical space of potential chemistries and (2) identify promising chemistries via iterative inputs from theory and experiments. This talk will focus on the details of the screening strategy applied to the class of oxide perovskites as candidate scintillator materials. Specifically, we present a novel hierarchical down-selection approach that employs structure maps, DFT-based stability analysis, ML models for bandgap predictions and physics-based classification to efficiently predict minimal favorable electronic structure for a viable scintillator. The developed framework is general and has implications beyond scintillator discovery.
11:00 AM
The Energy Landscape Governs Ductility in Disordered Materials: Mathieu Bauchy1; 1University of California, Los Angeles
Based on their structure, non-crystalline phases can fail in a brittle or ductile fashion. However, the nature of the link between structure and propensity for ductility in disordered materials has remained elusive. Here, based on MD simulations, we investigate how the degree of structural disorder affects the fracture of disordered materials. By applying the activation-relaxation technique (an open-ended saddle point search algorithm), we demonstrate that the propensity for ductility is controlled by the topography of the energy landscape. We observe a power-law relationship between the particle non-affine displacement upon fracture and the average local energy barrier. This reveals that the dynamics of the particles upon fracture is encoded in the static energy landscape, i.e., before any load is applied. This relationship is shown to apply to several classes of non-crystalline materials (oxide and metallic glasses, amorphous solid, and colloidal gels).
10:40 AM
Kinetic Monte Carlo Simulation of Glasses Aided by Machine Learning: Ajay Annamareddy1; 1University of Wisconsin - Madison
Studying the kinetics in the glassy state, especially of an aged glass, becomes computationally prohibitive with molecular dynamics simulations because of the slowdown in dynamics. We aim to integrate kinetic Monte Carlo with machine learning (ML) to study the dynamics and diffusion in the aged state. This requires that ML methods be trained to accurately predict the hop-rate and hop-vector from a local description of atom. The hop-rate is accessible from softness developed by Andrea Liu and collaborators. Initial studies on how much hop-vectors can vary given the same environment indicates that the same starting conditions lead to hops clustered closely together in direction, supporting that hop-vector is approximately controlled by local environment and can be machine-learned. To predict the hop-vector, we adopted existing schemes used in ML interatomic potentials to define a local coordinate system and a rotationally-invariant feature vector and now using a deep learning convolutional neural network.