arXiv Paper Presents ML Model for High-Precision Prediction of Metallic Glass Critical Cooling Rates Using Elemental and Molecular Simulation Features

June 28, 2026

arXiv USA

Overview

This study presents a machine learning model for predicting critical cooling rates of metallic glasses using computationally derived properties, specifically elemental and molecular dynamics simulation-based features. Demonstrating high accuracy in cross-validation, the model is applicable to high-throughput studies of material properties with diverse compositions. This innovation will streamline the exploration and development of new metallic glass materials.

In Depth

Key Findings

This research introduces a high-precision machine learning model for predicting the critical cooling rates of metallic glasses, leveraging computationally derived properties. Specifically, the model incorporates features based on elemental properties and those extracted from molecular dynamics simulations. Demonstrating high predictive accuracy in cross-validation, this model significantly expands the applicability to high-throughput studies of material properties for metallic glasses with diverse compositions.

Technical / Clinical Details

The developed machine learning model predicts the critical cooling rate (the minimum cooling rate required to solidify from a liquid state into a glassy state) of metallic glasses, a crucial property. The features used in this model fall into two main categories: firstly, fundamental elemental properties such as atomic radius, electronegativity, and atomic mass of the constituent elements. Secondly, dynamic features calculated by molecular dynamics (MD) simulations, such as atomic packing efficiency, liquid viscosity, and the tendency for short-range order formation. By combining and training the model with these features, it learns a much more complex and accurate relationship between critical cooling rate and composition/structure than traditional empirical rules or simple theoretical models. Cross-validation results showed that the model maintains high predictive accuracy even for unknown metallic glass compositions, confirming its generality and robustness. This allows researchers to efficiently screen promising metallic glass compositions before embarking on expensive and time-consuming experimental trial-and-error.

Background & Context

Metallic glasses (amorphous metals) possess unique properties (e.g., high strength, high hardness, excellent corrosion resistance, elasticity) not found in conventional crystalline metals, making them promising for applications across a wide range of fields, including sports equipment, medical devices, precision machine parts, and aerospace materials. However, forming metallic glasses requires very rapid cooling rates, and accurately predicting their critical cooling rate has been one of the most challenging tasks in exploring new metallic glass compositions. Previous predictive models often suffered from insufficient accuracy or limited applicability. The introduction of machine learning holds the potential to break through this predictive barrier and dramatically improve the efficiency of materials design.

Strategic Significance & Outlook

This machine learning model will significantly accelerate the exploration and development of new metallic glass compositions. It will particularly contribute to the discovery of metallic glasses that can be manufactured through low-cost, high-efficiency processes, paving the way for industrial applications. Its application to high-throughput screening will enable rapid identification of compositions with optimal critical cooling rates from thousands or tens of thousands of candidates. In the future, this model is expected to be integrated with predictions of other material properties (e.g., mechanical strength and thermal stability) to aid in designing next-generation metallic glasses with multifunctional capabilities. Furthermore, this approach is extensible to the design of other amorphous materials and complex alloy systems beyond metallic glasses, holding the potential to contribute to the overall advancement of the materials informatics field.

Source: https://arxiv.org/html/2606.21467v1

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.