Green-fabricated Cu–Ti Bimetallic Nanoparticles Derived from Psidium guajava Leaf Extract for Catalytic Hydrogen Generation
Abstract
In this investigation, Cu–Ti bimetallic nanoparticles were synthesized through a sustainable green route employing Psidium guajava leaf extract as the reducing and stabilizing agent. The XRD characterization indicated an average crystallite size of approximately 40 nm, whereas SEM observations revealed predominantly irregular structures together with a limited number of spherical particles. The elemental composition of Cu, Ti, and C was confirmed with EDX measurements, while Raman analysis verified the structural stability and formation of metal–oxygen bonds. The catalytic efficiency of the prepared nanoparticles was evaluated by varying reaction temperature, catalyst dosage, and the solution composition during NaBH4 hydrolysis. Under 0.50 wt% NaOH, the catalyst exhibited a maximum hydrogen generation rate of 446.5 mL min-1 g-1. Kinetic evaluation determined activation energy of 53.98 kJ mol-1 for Cu–Ti catalyst. Subsequently, machine learning techniques including Gaussian Process Regression, Light Gradient Boosting Machine (LightGBM), HistGradientBoosting, CatBoost, Support Vector Regression, Random Forest, and Extra Trees Regressor were implemented to estimate hydrogen generation performance. The experimental dataset was enhanced using polynomial, logarithmic, and interaction-based descriptors, while 5-fold cross-validation minimized overfitting during model training. Among the developed models, Gaussian Process Regression (R2 = 0.9937) achieved the highest predictive capability, whereas LightGBM (R2 = 0.9589) and CatBoost (R2= 0.9584) also demonstrated reliable prediction performance. These findings indicate that, green-fabricated Cu–Ti nanoparticles provide efficient catalytic activity for hydrogen evolution, and that, advanced machine learning methods offer reliable prediction and optimization of hydrogen production. This combined experimental and computational strategy provides a valuable pathway for accelerating catalyst design toward sustainable clean-energy technologies.