Artificial Intelligence-Guided Design of Eco-Friendly Hybrid Corrosion Inhibitors for Mild Steel in Acidic Media: Integrating Molecular Descriptors with Electrochemical Performance

Abstract

Corrosion of mild steel in acidic environments presents significant operational and economic challenges in industries such as oil and gas, chemical processing, and metal finishing. This study developed an artificial intelligence (AI)-guided framework for designing eco-friendly hybrid corrosion inhibitors by integrating molecular descriptors with electrochemical performance. Fifteen hybrid inhibitors comprising plant-derived phytochemicals reinforced with zinc oxide (ZnO) nanoparticles were investigated. Seventeen molecular descriptors were calculated and used to develop four machine learning models: Support Vector Regression (SVR), Artificial Neural Network (ANN), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). Among the models, XGBoost achieved the best predictive performance with a testing coefficient of determination (R²) of 0.984, root mean square error (RMSE) of 0.91%, and mean absolute error (MAE) of 0.72%. Experimental validation was performed using weight-loss measurements, potentiodynamic polarization, electrochemical impedance spectroscopy (EIS), and adsorption studies in 1.0 M HCl. The optimized hybrid inhibitor (HI-15) reduced the corrosion rate from 6.81 to 0.21 mm y⁻¹ and decreased the corrosion current density from 742.8 to 24.7 µA cm⁻², corresponding to inhibition efficiencies of 96.9%, 96.7%, and 96.8% from weight-loss, polarization, and EIS measurements, respectively. Charge-transfer resistance increased from 21.4 to 663.8 Ω cm², while double-layer capacitance decreased from 248.6 to 39.4 µF cm⁻², confirming the formation of a compact protective adsorption film. Langmuir adsorption analysis produced an adsorption equilibrium constant of 1.32 × 10⁴ L mol⁻¹ with a regression coefficient of 0.9985 and a standard free energy of adsorption of −34.7 kJ mol⁻¹, indicating spontaneous mixed physisorption–chemisorption. The agreement between AI-predicted and experimental inhibition efficiencies (R² = 0.986; mean absolute prediction error = 0.56%) demonstrates that integrating machine learning with electrochemical evaluation provides a rapid, accurate, and sustainable strategy for developing high-performance green corrosion inhibitors.