Molybdenum recovery from acidic industrial wastewater using Bayesian optimization algorithm-based ANN model: A case study by applying chemical precipitation followed by solvent extraction

This study investigates the separation and recovery of molybdenum from acidic industrial wastewater utilizing sequential chemical precipitation and solvent extraction approaches integrated with optimized machine learning (ML) algorithms. Acidic industrial wastewater, characterized by concentrated mixed mineral acids and heavy metals, was treated with NH4Cl to facilitate precipitation of molybdenum. The raffinate residue was then processed by solvent extraction to separate the residual molybdenum, which contained a mixture of LIX 63 and 1-decanol diluted in kerosene. To predict the behavior and structure of the molybdenum separation process, a Bayesian optimization algorithm (BOA) coupled artificial neural network (ANN) model, which demonstrated the lowest mean-square error among the evaluated models, was employed. The optimum precipitation process achieved a peak molybdenum recovery of 75.1% at NH4Cl dosage, reaction time, and temperature of 4%, 120 min, and 65 °C, respectively. Subsequently, selective extraction using LIX63 and parameter optimization using the ANN model predicted a molybdenum recovery efficiency of 86.1%. A McCabe–Thiele diagram, constructed using the predicted isotherm data, indicated that four stages are necessary for nearly complete molybdenum extraction. The stripping data from the ANN results indicated that molybdenum was effectively transferred into the stripping solution under specific conditions involving 3 M NH4OH and an equal phase ratio at 25 °C for 20 min. A sensitivity analysis revealed that the NH4Cl dosage had the most significant impact on the precipitation efficiency, whereas the phase ratio and NH4OH were crucial in the extraction and stripping experiments, respectively. Scanning electron microscopy (SEM) showed a porous structure with plate-like MoO3 particles, and X-ray diffraction (XRD) analysis confirmed the presence of and as the primary crystalline and thermal treatment phases, respectively. •Selective separation of Mo from highly concentrated waste acids solution•The sequential chemical precipitation using NH4Cl and solvent extraction with LIX 63•Novel optimization for Mo separation process using artificial intelligence models•Mitigation of environmental pollution through the proper recycling management