Optimizing the preparation of palm kernel shell activated carbon for lithium polysulfide adsorption using response surface methodology and artificial neural network for high performance lithium-sulfur battery

This study used response surface methodology (RSM) and artificial neural network (ANN) to predict and optimize lithium polysulfide (LiP) adsorption on nitrogen-doped activated carbon (NDAC). Firstly, the NDAC production from palm kernel shell was optimized using RSM, where statistical analysis indicated the best conditions to be an impregnation ratio (IR) of 2.0, an activation temperature of 880 degrees C, and an activation time of 80 min-with IR having the most significant impact on LiP adsorption. Experimental results from the RSM were then used to train the predictive capabilities of the ANN for LiP adsorption. Although both approaches effectively predicted the adsorption process, ANN exhibited a superior prediction accuracy, characterized by a higher coefficient of determination (R2) 2 ) and a lower mean square error (MSE). The NDAC synthesized under optimized conditions was subsequently made into a cathode composite with sulfur (NDAC/S) and evaluated for its performance in a lithium-sulfur (Li-S) battery. Experimental data indicated that the Li-S coin cell battery consisting of NDAC/S had a remarkable initial specific capacity of 1054.96 mAh/g and maintained a favorable capacity retention of 66 % after 100 cycles at 0.1C. This outstanding electrochemical performance is attributed to the synergistic effect of a hierarchical pore structure, large surface area, substantial pore volume, and the presence of doped nitrogen that provides strong chemical bonding with LiP.