Tribological enhancement of Al 2 O 3 -TiO 2 D-gun coated Ti-6Al-4V alloy: A machine learning approach

Abstract

Ti-6Al-4V alloy is used a lot in aerospace, biomedical and marine applications. This study investigates the tribological behaviour of Al₂O₃-TiO₂ coated and uncoated Ti-6Al-4V alloy by experimental and machine learning approaches. An Al₂O₃-TiO₂ ceramic coating, approximately 200 µm thick, was applied to Ti-6Al-4V alloy via the detonation gun (D-gun) process under regulated conditions (oxygen flow: 4800 SLPH, acetylene: 2150 SLPH, stand-off distance: 180 mm, particle velocity ∼1200 m/s). The wear behaviour of samples was tested using a pin-on-disc apparatus in accordance with ASTM G99 standards. Tests were conducted under normal loads of 40–100 N, sliding speeds of 100–200 rpm and durations of 20–40 min against an EN31 steel counter face. Scanning electron microscopy demonstrated that coated surfaces exhibited reduced wear and formed a protective tribo-layer. Machine learning models such as artificial neural networks, K-nearest neighbours and random forest were developed to predict wear based on sliding parameters. Random forest proved to be the best choice with less mean percentage error and mean squared error. The results showed that D-gun deposited Al₂O₃-TiO₂ coatings improved the wear performance of Ti-6Al-4V.