Hydrogen production via sodium borohydride hydrolysis catalyzed by cobalt ferrite anchored nitrogen-and sulfur co-doped graphene hybrid nanocatalyst: Artificial neural network modeling approach

The sluggish kinetics of the Sodium borohydride (NaBH₄) hydrolysis process particularly in alkaline conditions requires the design of high-performance low-cost catalysts. Herein, it was aimed to tailor cobalt ferrite anchored nitrogen-and sulfur-doped graphene architecture (CoFe₂O₄ @N,S-G) via a facile production pathway, to explore its potential application as a catalyst in alkaline NaBH₄ hydrolysis reaction for hydrogen production, and to develop an optimal artificial neural network (ANN) architecture to predict hydrogen production rate. In this regard, the influence of several variables such as reaction temperature, NaBH₄ concentration, and catalyst loading was explored to determine the optimal operational conditions for effective hydrogen generation. Furthermore, the performance metrics of ANN topologies were investigated to establish the best ANN model for predicting hydrogen generation rate under different operational conditions. The experimental results offered the outstanding catalytic activity of CoFe₂O₄ @N,S-G towards NaBH₄ hydrolysis with the volumetric hydrogen production rate of 8.5 L.min⁻¹.g_cat⁻¹ at 25 ℃, and catalyst loading of 0.02 g, and 1.0 M NaBH₄ concentration. The CoFe₂O₄ @N,S-G nanocatalyst was found to retain 94.9% of its initial catalytic activity after 5 consecutive uses, according to the reusability tests. The optimum performance metrics that were determined by the mean squared error (MSE) of 0.00052 and the coefficient of determination (R²) of 0.9989 were achieved for the ANN model with the configuration of 3–10–5–1 trained by Levenberg-Marquardt backpropagation algorithm. The activation function of tansig and purelin functions at hidden and output layers, respectively. The findings revealed that the experimental data were in harmony with the ANN-predicted one, thereby inferring the optimized ANN model could be employed in the forecasting of hydrogen production rate at various operational conditions.