<b>Mechanistic Insights and RSM-Based Optimization of Aluminium Corrosion Inhibition Using Banana Leaf Extract</b>

Abstract

The corrosion inhibition performance of banana (Musa spp.) leaf extract on aluminium in acidic medium was evaluated using Response Surface Methodology (RSM) considering immersion time, temperature, and inhibitor concentration. The regression model exhibited strong predictive accuracy, with predicted and experimental inhibition efficiencies closely aligned along the 45° line. Residual normality was confirmed, with most residuals following the reference line and only minor deviations at approximately –3.0 and +3.8. The externally studentized residuals analysis showed that only one observation (predicted efficiency ≈ 75%) exceeded the critical limit of ±3.71733, while the remaining runs were within acceptable limits. Cook’s distance values ranged from 0.02 – 0.74, all below the influence threshold of 0.8845, with the highest values recorded for Run 2 (0.74), Run 15 (0.56), and Run 7 (0.51). Leverage values varied between 0.05 and 0.32, remaining below the critical value of 0.4706 and above the average leverage of 0.2353. DFFITS diagnostics identified Run 2 (+2.3) and Run 16 (–1.8) as relatively influential compared with the threshold of ±1.45521. Similarly, DFBETAS analysis for the intercept showed influential values for Run 2 (+1.05) and Run 17 (–0.95) relative to the limit of ±0.7276. The Box–Cox transformation produced an optimal λ ≈ 1.0 (95% CI), indicating stable variance and no need for response transformation. Perturbation analysis indicated a sensitivity order of immersion time > temperature > inhibitor concentration, with immersion time and temperature exerting negative effects while concentration showed a mild positive influence. Overall, the robust statistical diagnostics and predictive performance confirm the effectiveness of banana leaf extract as a sustainable green inhibitor for aluminium corrosion in acidic environments.