Regeneration of Co-based bead type catalyst in ammonia borane hydrolysis for hydrogen generation: Artificial neural networks and response surface methodology

Abstract

The usage of ammonia borane (NH3BH3, AB) as a hydrogen storage medium and the release of hydrogen from its structure via hydrolysis has gained attention as for the hydrogen energy systems. Cobalt (Co)-based catalysts are the one of the activist material that can effectively catalyze the hydrogen production at current applications. However, Co-based catalysts have gradually lost their initial activities in long-term reactions due to by-product deposition on their surface. To address this issue, a catalyst regeneration strategy based on solvent washing to reactivation of Co-based catalysts with help of new perspective is presented in this study. Regeneration parameters as solution properties (pH:3-7), temperature (15-75 degrees C), and duration (10-50 min) have been modeled by both Response Surface Methodology (RSM) and Artificial Neural Network (ANN) techniques with different levels to optimize the process conditions for the regeneration process and to regain the high initial catalytic activity. The results have showed that solvent washing at optimization conditions was the effective method for restoring the activity of the catalyst. The findings indicate that time and temperature have a significant impact on the regeneration of the Co-based catalyst. The characterization techniques such as XRD, FTIR, HR-TEM, SEM/EDS, and XPS were performed to illuminate and correlate the effect of regeneration on the internal and external properties of fresh, used and regenerated catalysts. The optimum regeneration conditions, the pH of 6.5, the temperature of 44 degrees C, and the duration of 42 min under the optimal conditions identified by RSM, provided the highest hydrogen production (3.23 l min  1.g  1), showed much closer results to ANN technique predicted value (3.45 l min  1.g  1). It is shown that ANN has a relatively higher prediction accuracy and optimization ability compared to RSM for hydrogen production rate.