Validation of a Machine Learning–Assisted LIBS Model for Quantitative Steel Analysis

Aline Gonçalves Capella; Marta  Martín López; Ignacio  Garcia Diego; Juan  José de Damborenea González; María  Ángeles Arenas

doi:10.3989/revmetalm.e298.1779

Authors

Aline Gonçalves Capella Institute of Science and Technology, Federal University of Sao Paulo, ICT-UNIFESP - Department of Surface Engineering Corrosion and Durability, National Center for Metallurgical Research (CENIM-CSIC) https://orcid.org/0000-0002-9149-213X
Marta Martín López Department of Surface Engineering Corrosion and Durability, National Center for Metallurgical Research (CENIM-CSIC), https://orcid.org/0009-0004-9463-5054
Ignacio Garcia Diego Department of Surface Engineering Corrosion and Durability, National Center for Metallurgical Research (CENIM-CSIC), https://orcid.org/0000-0001-6169-602X
Juan José de Damborenea González Department of Surface Engineering Corrosion and Durability, National Center for Metallurgical Research (CENIM-CSIC), https://orcid.org/0000-0002-8053-594X
María Ángeles Arenas epartment of Surface Engineering Corrosion and Durability, National Center for Metallurgical Research (CENIM-CSIC) https://orcid.org/0000-0002-7807-8209

DOI:

https://doi.org/10.3989/revmetalm.e298.1779

Keywords:

Laser-Induced Breakdown Spectroscopy, Machine Learning, Regression Models, Steel

Abstract

Laser-induced breakdown spectroscopy (LIBS) shows great promise for the rapid chemical characterization of materials. However, quantitative analysis of elements remains challenging due to strong matrix effects and the predominance of emission lines. In the present study, a machine learning–assisted LIBS model (LIBS-ML pipeline) was employed to analyze carbon, medium-alloy, and high-alloy steels. A total of 900 spectra from 18 reference specimens were used to train Random Forest (RF), Gradient Boosting (GB), and Extremely Randomized Trees (ET) ensemble models, while five independent steel specimens were reserved for prediction evaluation. The ET model demonstrated the best training performance (MSE = 0.1551; R² = 0.9435), while the RF model exhibited greater stability during independent validation. Low prediction errors were obtained for carbon steels, with mean absolute error (MAE) values as low as 0.0142 wt% for C and 0.0178 wt% for Mn. In medium-alloy steel, the predicted values of Cr and Ni remained close to the nominal compositions. Higher deviations were observed in high-alloy steel, with MAE reaching 1.1191 wt% for Ni and 1.0919 wt% for Mo, reflecting the increased complexity of the matrix. The obtained results confirmed the LIBS–ML pipeline applicability for quantitative steel analysis and highlight its potential as a rapid tool for metallurgical monitoring and alloy characterization.

Downloads

Download data is not yet available.

References

Breiman, L. (2001). Random Forests. Machine Learning 45, 5–32.

Capella, A.G., López, M.M., de Damborenea González, J.J., Arenas, M.A., Garcia Diego, I. (2026). Machine learning for quantitative LIBS analysis of aluminum alloys: a comparison of random forest, gradient boosting, and extremely randomized trees. Sci. Rep. 16, 14758.

Cremers, D.A., Radziemski, L.J. (2006). Handbook of laser-induced breakdown spectroscopy. John Wiley & Sons, Hoboken (NJ).

Ferreira, E.C., Milori, D.M.B.P., Ferreira, E.J., Da Silva, R.M., Martin-Neto, L. (2008). Artificial neural network for Cu quantitative determination in soil using a portable laser induced breakdown spectroscopy system. Spectrochim. Acta B: At. Spectrosc. 63 (10), 1216–1220.

Friedman. J.H. (2001). Greedy function approximation: A gradient boosting machine. Ann. Stat. 29 (5), 1189–1232.

Geurts, P., Ernst, D., Wehenkel, L. (2006). Extremely randomized trees. Mach. Learn. 63, 3–42.

Gonçalves Capella, A., de Damborenea, J., Arenas, M.A., Garcia, I. (2025). LIBS Quantification Pipeline – Version 1.0.0 (v1.0.0). Zenodo.

Hahn, D.W., Omenetto, N. (2010). Laser-induced breakdown spectroscopy (LIBS), Part I: Review of basic diagnostics and plasma–particle interactions. Appl. Spectrosc. 64 (12), 335–66.

Hahn, D.W., Omenetto, N. (2012). Laser-induced breakdown spectroscopy (LIBS), Part II: Review of instrumental and methodological approaches to Material Analysis and Applications to Different Fields. Appl. Spectrosc. 66 (4), 347–419.

Hu, Z., Zhang, D., Wang, W., Chen, F., Xu, Y., Nie, J., Chu, Y., Guo, L. (2022). A review of calibration-free laser-induced breakdown spectroscopy. TrAC Trends. Anal. Chem. 152, 116618.

Liang, L., Zhang, T., Wang, K., Tang, H., Yang, X., Zhu, X., Duan, Y., Li, H. (2014). Classification of steel materials by Laser-Induced Breakdown Spectroscopy (LIBS) coupled with support vector machines. Appl Opt. 53, 544–552.

Liu, K., Tian, D., Xu, H., Wang, H., Yang, G. (2019). Quantitative analysis of toxic elements in polypropylene (PP) via laser-induced breakdown spectroscopy (LIBS) coupled with random forest regression based on variable importance (VI-RFR). Anal. Methods. 11 (37), 4769–4774.

Grünberger, S., Ehrentraut, V., Eschlböck-Fuchs, S., Hofstadler. J., Pissenberger, A., Pedarnig, J.D. (2023). Overcoming the matrix effect in the element analysis of steel: Laser ablation-spark discharge-optical emission spectroscopy (LA-SD-OES) and Laser-induced breakdown spectroscopy (LIBS). Anal. Chim. Acta 1251, 341005.

Tognoni, E., Cristoforetti, G. (2016). Signal and noise in Laser Induced Breakdown Spectroscopy: An introductory review. Opt. Laser Technol. 79, 164-172.

Van den Eynde. S., Díaz-Romero, D.J., Zaplana, I., Peeters, J. (2023). Deep learning regression for quantitative LIBS analysis. Spectrochim. Acta B: At. Spectrosc. 202, 106634.