Tribological and Machine Learning Investigation of TiO₂, Al₂O₃, and MoS₂ Nanolubricants for Methanol-Fueled Engine Applications

This study systematically investigated the tribological performance of TiO₂, Al₂O₃, and hybrid MoS₂–TiO₂ nanolubricants under methanol-contaminated lubrication conditions. Seventeen lubricant formulations, comprising fresh engine oil, methanol-contaminated oil, single-component oxide nanolubricants, and hybrid nanolubricants, were evaluated using a reciprocating tribometer under identical operating conditions (60 N, 6.5 Hz, 10 min). The coefficient of friction (COF) and relative contact displacement were analyzed using Savitzky–Golay signal processing, while wear mechanisms were investigated through field-emission scanning electron microscopy (FESEM), energy-dispersive spectroscopy (EDS), and wear scar diameter (WSD) measurements. Methanol contamination increased the WSD by approximately 11.4%, accompanied by severe adhesive wear, greater substrate exposure, and deterioration of the protective tribofilm. Although the nanolubricants exhibited only moderate differences in average COF, the hybrid formulations produced a more stable friction response and substantially improved surface protection. TiO₂ and Al₂O₃ nanoparticles effectively improved the anti-wear performance of the contaminated lubricant, although their effectiveness was strongly dependent on nanoparticle concentration. The hybrid lubricant containing 0.30 wt.% MoS₂ + 0.15 wt.% TiO₂ achieved the most stable COF, the lowest relative contact displacement, the smallest wear scar diameter, and the most uniform tribofilm. FESEM and EDS analyses confirmed the formation of a robust hybrid tribofilm containing lubricant-derived carbonaceous species, oxide-related compounds, and MoS₂-derived sulfur species, resulting from the synergistic effects of TiO₂-assisted load sharing and tribofilm formation with the low-shear layered lubrication mechanism of MoS₂. Furthermore, machine-learning-based COF prediction using ANN, RF, and LightGBM demonstrated excellent prediction capability, with LightGBM achieving the highest accuracy among the evaluated models. The findings highlight hybrid MoS₂–TiO₂ nanolubricants as a promising approach for improving tribological durability of methanol-contaminated engine lubrication systems.