Machine Learning Approach to Develop Novel Metallic Glasses and Predict Glass-Forming Ability

Abstract:

Metallic Glasses (MGs) have unique mechanical and physical properties making them highly desirable for applications across various fields. However, some MGs have poor Glass-Forming Ability (GFA) and conventional methods to improve it are time-consuming, resource-demanding, and costly. In this study, advanced machine learning (ML) techniques are leveraged to develop robust and data-driven models capable of predicting the critical diameter (Dmax) of MGs from the concentrations of their constituent elements. Dmax is an essential indicator for GFA, whereby higher Dmax values indicate better GFA. A comprehensive dataset encompassing 8,734 MG alloys and their associated Dmax was compiled, cleaned, and analyzed from various sources. The Gradient Boosting model was the best-performing predictive model achieving a R² of 0.86 and a RMSE of 1.66 mm for estimating Dmax, outperforming other models such as Random Forest and XGBoost. Furthermore, the SHAP (SHapley Additive exPlanations) analysis was utilized to rank the importance of individual elements of the alloys, identifying Zirconium (Zr) as the most influential feature in predicting Dmax. Additionally, pseudo-ternary diagrams were generated based on the Gradient Boosting model to identify potential novel BMGs with enhanced GFA. The model's robustness and utility were validated by comparing the Dmax values predicted by the ML model to experimentally obtained values for the Ni_76-xFe_xP₁₄B₆Ta₄ alloy across varying Fe concentrations (x). The results of the study enhance the accuracy of GFA predictions and establish a robust data-driven framework for expediting and automating the discovery of novel BMGs.

• Periodical: Key Engineering Materials (Volume 1027)
• Pages: 45-52
• DOI: https://doi.org/10.4028/p-YPDzS9
• Online since: October 2025
• Authors: Fahid Abu-Salah*, Elsa Maalouf
• Keywords: Bulk Metallic Glasses, Critical Diameter, Glass Forming Ability, Machine Learning, Pseudo-Ternary Diagram

References

[1] Sharma, A., & Zadorozhnyy, V. (2021). Review of the recent development in metallic glass and its composites. Metals, 11(12), 1933.

DOI: 10.3390/met11121933

[2] Jabed, A., Bhuiyan, M. N., Haider, W., & Shabib, I. (2023). Distinctive features and fabrication routes of metallic-glass systems designed for different engineering applications: A review. Coatings, 13(10), 1689

DOI: 10.3390/coatings13101689

[3] X. Li, G. Shan, C.H. Shek, Machine learning prediction of magnetic properties of Fe-based metallic glasses considering glass-forming ability, J. Mater. Sci. Technol. 91 (2021) 113–122.

DOI: 10.1016/j.jmst.2021.05.076

[4] Suryanarayana, C., & Inoue, A. (2013). Iron-based bulk metallic glasses. International Materials Reviews, 58(3), 131-166.

DOI: 10.1179/1743280412Y.0000000007

[5] Liu, C., Wang, X., Cai, W., He, Y., & Su, H. (2023). Machine learning aided prediction of glass-forming ability of metallic glass. Processes, 11(9), 2806

DOI: 10.3390/pr11092806

[6] Mastropietro, D. G., & Moya, J. A. (2021). Design of Fe-based bulk metallic glasses for maximum amorphous diameter (Dmax) using machine learning models. Computational Materials Science, 188, 110230.

DOI: 10.1016/j.commatsci.2020.110230

[7] Li, K. Y., Li, M. Z., & Wang, W. H. (2024). Inverse design machine learning model for metallic glasses with good glass-forming ability and properties. Journal of Applied Physics, 135(2), 025102.

DOI: 10.1063/5.0179854

[8] Bobzin, K., Heinemann, H., Burbaum, E., Johann, L. M., Seßler, J., & Gärtner, J. (2023). Data driven development of iron-based metallic glasses using artificial neural networks [Data set]. RWTH Aachen University.

DOI: 10.1016/j.jallcom.2023.172895

[9] F. Ghorbani, A. Kumar, Y. Lee, et al., Thermodynamically-guided machine learning modelling for predicting the glass-forming ability of bulk metallic glasses, Sci. Rep. 12 (2022) 10890.

DOI: 10.1038/s41598-022-15981-2

[10] Dataset: Fe-based Bulk Metallic Glass Alloys: Database (up to February 2020) and R code, Mendeley Data, v5, 2020. https://data.mendeley.com/datasets/jy9skrx74g/5.

[11] Deng, B., & Zhang, Y. (2020). Critical feature space for predicting the glass-forming ability of metallic alloys revealed by machine learning. Chemical Physics, 538, 110898.

DOI: 10.1016/j.chemphys.2020.110898

[12] Xiong, J., Shi, S.-Q., & Zhang, T.-Y. (2020). A machine-learning approach to predicting and understanding the properties of amorphous metallic alloys. Materials & Design, 187, 108378.

DOI: 10.1016/j.matdes.2019.108378

[13] Peng, L., Long, Z., & Zhao, M. (2021). Determination of glass forming ability of bulk metallic glasses based on machine learning. Computational Materials Science, 195, 110480.

DOI: 10.1016/j.commatsci.2021.110480

[14] Lu, Z. P., & Liu, C. T. (2002). A new glass-forming ability criterion for bulk metallic glasses. Acta Materialia, 50(13), 3501-3512.

DOI: 10.1016/S1359-6454(02)00166-0

[15] Wang, X., Zeng, M., Nollmann, N., Wilde, G., Tian, Z., & Tang, C. (2017). Effect of copper addition on the glass forming ability in Pd-Si binary amorphous alloying system. AIP Advances, 7(9), 095108.

DOI: 10.1063/1.4986532

[16] Miller, M., & Liaw, P. (Eds.). (2008). Bulk Metallic Glasses: An Overview. Springer.

DOI: 10.1007/978-0-387-48921-6

[17] Afflerbach, B., Francis, C., Szlufarska, I., Morgan, D., Voyles, P. M., & Schultz, L. E. (2021). Characteristic Temperature Model for Metallic Glass Critical Casting Diameter [Data set]. Materials Data Facility.

DOI: 10.1016/j.commatsci.2021.110494

[18] Ren, B., Long, Z., & Deng, R. (2021). A new criterion for predicting the glass-forming ability of alloys based on machine learning. Computational Materials Science, 189, 110259.

DOI: 10.1016/j.commatsci.2020.110259

[19] Yuan, Z.-Z., Bao, S.-L., Lu, Y., Zhang, D.-P., & Yao, L. (2008). A new criterion for evaluating the glass-forming ability of bulk glass forming alloys. Journal of Alloys and Compounds, 459(1–2), 251-260.

DOI: 10.1016/j.jallcom.2007.05.037

[20] Wu, X. F., Suo, Z. Y., Si, Y., Meng, L. K., & Qiu, K. Q. (2008). Bulk metallic glass formation in a ternary Ti–Cu–Ni alloy system. Journal of Alloys and Compounds, 452(2), 268-272.

DOI: 10.1016/j.jallcom.2006.11.010

[21] Kuball, A., Bochtler, B., Gross, O., Pacheco, V., Stolpe, M., Hechler, S., & Busch, R. (2018). On the bulk glass formation in the ternary Pd-Ni-S system. Acta Materialia, 158, 13-22.

DOI: 10.1016/j.actamat.2018.07.039

[22] Santos, F. S., Kiminami, C. S., Bolfarini, C., de Oliveira, M. F., & Botta, W. J. (2010). Evaluation of glass forming ability in the Ni–Nb–Zr alloy system by the topological instability (λ) criterion. Journal of Alloys and Compounds, 495(2), 313-315.

DOI: 10.1016/j.jallcom.2009.10.212

[23] Zeng, W., Chen, Y., Li, Q., Li, H., Mu, B., Ye, J., & Chang, C. (2023). Ductile Ni-based bulk metallic glasses at room temperature. Journal of Materials Research and Technology, 26, 2432-2442.

DOI: 10.1016/j.jmrt.2023.08.062

[24] Ma, X., Li, Q., Chang, L., Chang, C., Li, H., & Sun, Y. (2017). Enhancement on GFA and mechanical properties of Ni-based bulk metallic glasses through Fe addition. Intermetallics, 86, 34– 39.

DOI: 10.1016/j.intermet.2017.06.012