Experimental and machine learning assisted predictive modelling of tribological performance in TiO2/Al/TiO2 multilayer thin film coated stainless steel

Tri-layer TiO 2 /Al/TiO 2 thin films deposited using radio frequency (RF) and direct current (DC) magnetron sputtering on to stainless steel 304 rod. Effects of RF Power and annealing on physical and tribological properties of coated SS 304 have been investigated. The surface morphology of films was examined by a scanning electron microscope. Cross-sectional SEM analysis performed to obtained the thickness of films, which increased from 176 to 562 nm with RF power. The elemental EDS mapping confirms the coated layer elements. Pin-on-disc tribometer used to evaluate the wear and coefficient of friction (COF) of the coated SS. The results reveal that the film deposited at higher power shows better wear and COF. Furthermore, machine learning models, including Gradient Boosting, LightGBM, XGBoost, and CatBoost, were employed to predict wear and frictional responses based on experimental data. The strong correlation between predicted and observed results validates the robustness of the ensemble models and demonstrates the potential of ML-assisted approaches for optimized the coating parameters and forecasting tribological performance, which demonstrates the potential application in the tribological application.