Glasses, Optical Materials, and Devices: Current Issues in Science & Technology: Modeling and Simulations of Glass Materials
Program Organizers: Jincheng Du, University of North Texas; S. Sundaram, Alfred University

Monday 2:00 PM
September 30, 2019
Room: A106
Location: Oregon Convention Center

Session Chair: Jincheng Du, University of North Texas; Carlo Massobrio, l’Institut de Physique et Chimie des Matériaux de Strasbourg (IPCMS)

2:00 PM  Invited
Melting Mechanisms in Alkali Metasilicates: Alastair Cormack1; 1Alfred University
     From a glass science point of view, devitrification, that is, nucleation and growth of crystals from within the glass, is obviously a key phenomenon to be understood. However, from a computational materials science perspective, atomic scale modelling of devitrification is not so attractive, largely because of incompatible timescales. Notwithstanding, one way to approach this positively is to consider the reverse process, that is, mechanisms of melting. Silicates provide an additional level complexity because of their tetrahedral network character. Nevertheless, since both processes involve restructuring of tetrahedral network components, probing melting behavior is an attractive approach.In this presentation, we will discuss the use of molecular dynamics to follow the melting behavior of sodium and lithium metasilicates. We show that experimental differences between the two, such as homogeneous v. inhomogenous nucleation, are also reflected in their different melting processes. Some suggestions as to why this is will be advanced.

2:30 PM  Invited
Glassy Materials via First-principles Molecular Dynamics: Recent Results: Carlo Massobrio1; 1IPCMS-CNRS
    This talk will feature a set of recent results obtained in the area of glassy materials by using first-principles molecular dynamics (FPMD). In particular, we shall focus on the behavior of ternary chalcogenide systems, of interest in the fields of enhanced infrared properties and phase change materials. We shall also demonstrate that FPMD is a mature technique to have access to thermal properties of glasses, by providing explicit examples of calculations of thermal conductivity. Additional results related to the behavior of glass surfaces and to specific methodological issues will also be presented.

3:00 PM  
Reactive Potential based Simulations of Sodium Silicate Glasses: Lu Deng1; Shingo Urata2; Yasuyuki Takimoto2; Tatsuya Miyajima2; Seung Ho Hahn3; Adri C. T. van Duin3; Jincheng Du1; 1University of North Texas; 2AGC Inc.; 3The Pennsylvania State University
    Atomistic computer simulations can provide insights on the glass structures and their interactions with the environments by using reactive potential. In this talk, the structures of sodium silicate glass were studied using molecular dynamics simulations with recently proposed and refined ReaxFF parameters. The capability of the ReaxFF potential to generate the short and medium range structure features of sodium silicate glasses is investigated by comparing widely used partial charge pair-wise potential and two versions of the reactive potentials. Structural information as a function of glass forming procedures were studied. Cation coordination number, pair distribution function, neutron broadened structure factor and X-ray broadened structure factor of the simulated glasses are obtained and compared with those from simulations and available experimental data. Challenges of the potentials for glass forming procedures and potential for glass-water interactions simulations will be discussed.

3:20 PM Break

3:40 PM  Invited
Machine Learning for Glass Science and Engineering: Mathieu Bauchy1; 1University of California, Los Angeles
    Unlike crystalline materials, glasses can virtually feature any composition and stoichiometry, which creates limitless opportunities to develop new glass formulations with unusual properties. However, this large compositional space renders traditional Edisonian trial-and-error discovery approaches poorly efficient. In addition, the complex, disordered atomic structure of glasses makes it challenging to develop some mechanistic models relating composition to macroscopic properties. In this presentation, I will present some of our recent effort in applying machine learning to glass science and engineering, including (i) optimization methods for the parametrization of novel empirical forcefields, (ii) clustering and classification algorithms to identify previously hidden patterns in glass atomic networks, and (iii) regression methods for the predicting of glass engineering properties. I will focus on highlighting how machine learning, molecular dynamics simulations, and topological modeling can mutually inform and advance each other.

4:10 PM  Invited
Quantitative Structural Property Relationship Analysis of Multicomponent Silicate Glasses: Jincheng Du1; 1University of North Texas
    Quantitative Structural Property Relationship (QSPR) analysis is an effective approach to find correlations of the material properties with its structural characteristics and have found applications from pharmaceutical industry to organic materials design. The applications of QSPR in glass materials are relatively recent but with significant potential. In this talk, we will present our latest studies, by using structural features obtained from molecular dynamics simulations, a number of physical properties such as density, glass transition temperature, and thermal expansion coefficient can be well correlated with carefully chosen structure descriptors. Furthermore, this approach has been shown promising in find correlations in more complex behaviors such as chemical durability and dissolution rate. The challenges of the approach and potential applications in wide range of glass compositions and properties by coupling regression and machine learning approaches will be discussed.

4:40 PM  
Role of Exchange-correlation Functionals and Dispersion Forces in Determining the Structural Properties of Glasses: Carlo Massobrio1; 1IPCMS-CNRS
    In this talk we shall focus on the impact of exchange-correlation functional and dispersion forces on the determination of the structural properties of glassy chalcogenides. Over the years, we have collected pieces of evidence on the different electronic localization properties of several exchange-correlation functionals, some of them being more suited to account for the delicate interplay between ionic and covalent bonding found in binary chalcogenides. More recently, the same kind of investigation has been carried out for the case of dispersion forces. We have compared the performances of two kinds of schemes, differing in the way changes in the bonding properties are included in the expression of the van der Waals coefficients. We have come to the conclusion that, in the case of glassy chalcogenides, great care should be exercised to avoid artifacts resulting from the application of dispersions forces recipes not depending explicitly on the electronic structure.

5:00 PM  
Cooling Rate Effects on the Structure of 45S5 Bioglass: Yashasvi Maurya1; Pratik Bhaskar1; Rajesh Kumar1; N. M. Anoop Krishnan1; 1Indian Institute of Technology Delhi
    Despite being extensively, molecular simulations (MD) suffer from an inherent deficiency of being limited to short time scales. This leads to astronomically high cooling rates while preparing glasses by MD simulations, making it extremely challenging to compare with that of experimental structures obtained from laboratory scale cooling rates.Herein, using MD simulations with cooling rates varying over five orders of magnitude, we study structure of 45S5 bioglass. We show that the thermal history primarily affects the medium-range order structure, while the short-range order is largely unaffected over the range of cooling rates simulated. Interestingly, we observe that the Si and P tetrahedra exhibits a preferential separation with decreasing cooling rates, with the P tetrahedra preferring to form isolated Qº clusters. This suggests the existence of a Loewenstein-like rule for P atoms which prefer isolated clusters over Si atoms as neighbors while completely avoiding P–O–P bridging bonds.