Revealing the hidden structure of disordered materials by parameterizing their local structural manifold

Area of Science:

• Materials Science
• Condensed Matter Physics
• Computational Chemistry

Background:

• Developing structural descriptors for glassy materials is challenging due to the trade-off between applicability and interpretability.
• Existing frameworks often lack the simplicity and predictive power of those used for crystalline materials.

Purpose of the Study:

• To develop a novel, interpretable framework for characterizing the local atomic environments in glassy materials.
• To parameterize the low-dimensional manifold of local atomic structures in glasses.

Main Methods:

• Developed a generalized distance function: the Gaussian Integral Inner Product (GIIP) distance.
• Utilized agglomerative clustering and diffusion maps to parameterize the atomic environment manifold.
• Applied the method to model systems, including a 2D crystal and a 3D metallic glass.

Main Results:

• Identified parameters interpretable as coordination number, composition, volumetric strain, and local symmetry.
• Demonstrated that slower quenching rates in glasses correlate with increased local tetrahedral symmetry and decreased cyclic symmetry.
• Showcased the ability to reveal subtle structural trends often missed by traditional methods.

Conclusions:

• The GIIP distance framework provides a powerful, unbiased approach to understanding glass structure.
• This method offers a promising alternative to crystalline-based descriptors, enhancing interpretability and predictive power for amorphous materials.