[Paper Review] First-principles simulations of glass-formers

Walter Kob, Simona Ispas|arXiv (Cornell University)|Apr 27, 2016

Glass properties and applications4 references3 citations

TL;DR

This paper reviews first-principles ab initio simulations of glass-forming materials using density functional theory (DFT), demonstrating their ability to provide accurate, quantitative insights into the microscopic structure and dynamics of glasses—particularly where classical potentials fail due to environment-dependent bonding. The key contribution is establishing ab initio methods as essential for understanding complex glass properties, including defect structures and water interactions, despite high computational costs.

ABSTRACT

In this article we review results of computer simulation of glasses carried out using first principles approaches, notably density functional theory. We start with a brief introduction to this method and compare the pros and cons of this approach with the ones of simulations with classical potentials. This is followed by a discussion of simulation results of various glass-forming systems that have been obtained via ab initio simulations and that demonstrate the usefulness of this approach to understand the properties of glasses on the microscopic level.

Motivation & Objective

• To evaluate the advantages and limitations of first-principles simulations using density functional theory (DFT) compared to classical potentials in modeling glass-forming systems.
• To demonstrate the capability of ab initio methods to capture environment-dependent bonding and local structural variations critical in glasses, which classical potentials cannot reliably describe.
• To illustrate the utility of ab initio simulations in studying complex phenomena such as water incorporation, defect formation, and aging in oxide glasses.
• To highlight the role of ab initio simulations in developing accurate classical force fields by providing high-fidelity structural and energetic data.
• To outline future directions, including order-N algorithms and machine learning, to overcome computational scaling challenges for larger systems.

Proposed method

• Ab initio simulations solve the Kohn-Sham equations of density functional theory (DFT) to compute electronic structure and forces on atoms from first principles.
• The Born-Oppenheimer approximation is applied, treating electrons as instantaneously adjusting to nuclear positions, enabling the calculation of forces via energy gradients.
• Newton's equations of motion are integrated using the computed forces to generate particle trajectories and access dynamic and structural properties.
• The method requires only atomic species as input, avoiding empirical parameterization, and enables accurate treatment of bonding variations in disordered glassy systems.
• Systematic comparisons are made between ab initio results and classical simulations to assess accuracy in predicting radial distribution functions, diffusion coefficients, and elastic constants.
• Advanced techniques such as order-N algorithms and machine learning potentials are discussed as pathways to reduce computational cost and scale simulations to larger system sizes.

Experimental results

Research questions

• RQ1 How do ab initio simulations with DFT compare to classical molecular dynamics in describing the structural and dynamic properties of glass-forming systems?
• RQ2 In what ways do environment-dependent bonding effects in glasses challenge the accuracy of classical force fields, and how can ab initio methods overcome these limitations?
• RQ3 What specific insights into local atomic structure, defects, and water interactions in oxide glasses can be obtained only through first-principles simulations?
• RQ4 How can ab initio simulations contribute to the development of more accurate classical force fields for large-scale simulations of glasses?
• RQ5 What emerging computational techniques (e.g., order-N algorithms, machine learning) are most promising for extending the reach of ab initio simulations to larger glass systems?

Key findings

• Ab initio simulations provide a quantitatively accurate description of glass-forming systems by explicitly computing forces from electronic structure, avoiding the limitations of empirical potentials.
• The method successfully captures complex local environments in glasses, such as varying Si–O bond strengths and coordination changes, which classical potentials fail to represent accurately.
• First-principles simulations have revealed detailed mechanisms of water incorporation and its impact on structural and electronic properties in silica-based glasses.
• Ab initio studies have provided critical insights into the nature and effects of charged and neutral defects in amorphous SiO2, which influence optical and electronic device performance.
• Despite high computational cost scaling as O(N³), the method remains indispensable for studying intrinsic properties of glasses that are inaccessible to classical simulations.
• Emerging techniques such as order-N DFT and machine learning potentials are expected to significantly extend the accessible system sizes and time scales in future ab initio simulations of glasses.