Advancing electrochemical water desalination: Machine learning-driven prediction and RSM optimization of activated carbon electrodes

Abstract

This study presents an innovative and sustainable approach for synthesizing porous carbon electrodes from palm kernel shells (PKS) using a single-step physicochemical activation process. The study investigates the effect of carbon dioxide and zinc chloride on electrochemical features such as surface morphology and functional chemistry, intrinsic resistance and electrical conductivity. The best electrode developed from palm kernel shell derived activated carbon (PKSAC) using ZnCl2 and N2/CO2 (PKSAC_N2/CO2_ZnCl2) exhibited a superior specific capacitance, electrosorption capacity, and average salt adsorption rate of 365.4 F/g, 22.5 mg/g and 0.372 mg/g/min under optimized CDI conditions, such as applied voltage, feed flow rate, and initial NaCl concentration of 1.2 V, 7.5 ml/min, and 750 mg/L, respectively. Comparatively, electrodes developed with either N2 activation (PKSAC_N2) or N2/ZnCl2 activation (PKSAC_N2_ZnCl2) demonstrated electrosorption capacities of 12.55 and 19.74 mg/g, respectively. Response surface methodology (RSM) was applied to optimize the CDI parameters for electrochemical water desalination, ensuring process efficiency and scalability. Further, the machine learning extreme gradient boosting (XGB) regressor model predicted the performance of the developed electrodes and aligned well with the experimental data. The article provides key insights into activated carbon synthesis and its role in electrochemical water desalination, offering a sustainable solution to water scarcity that aligns with the UN's sustainability goals.