Machine-Learning-Optimized Palm-Biomass-Derived Activated Carbon Adsorbent for Gold Recovery from Mobile Leachate

Abstract

Given the significant increase in smartphone consumption, an incremental increase in global electronic waste (e-waste) production has become an environmental concern. Herein, we report on the use of palm biomass as a precursor of activated carbon (AC) to recover gold from e-waste. The current study utilizes random forest algorithms for the optimization of AC’s production from four types of palm biomass. Based on the optimization result, the highest conversion was achieved by palm fiber biomass, with >80% conversion at 600 °C. AC showed a moderate adsorption capacity toward Au at 64 mg/g. Optimization of the adsorption process was carried out by response surface methodology, resulting in an acceptable error value that indicates the suitability of the model. Under optimal conditions, 89.9% adsorption efficiency was achieved with an adsorbent dosage of 10 g/mL, an activation temperature of 600 °C, a Au concentration of 152 mg/L, and an adsorption temperature of 30 °C. The mechanistic modelling of the process shows a suitable fit with the pseudo-first-order kinetic model and the Langmuir isotherm model. Furthermore, the thermodynamic analysis indicated that gold adsorption was an exothermic and spontaneous process. In addition, the AC resin was extensively characterized by Fourier transform infrared, scanning electron microscopy, X-ray diffraction, and gas adsorption analysis.