Neural network optimization during the purification of industrial effluents using steel slag: kinetics and mechanism

The current paper investigated the potential of alkali-activated granulated blast furnace slag (AAGBFS) to remove metal ions from non-synthetic industrial effluent waste. Sorption experiments were carried out batchwise, using parametric optimization where the AAGBFS solid loading, sorption time and temperatures were varied to determine the sorption isotherms, kinetics, and thermodynamics. Various analytical techniques were used to characterize the raw and AAGBFS. Fourier transform infrared spectroscopy, FTIR and X-ray diffraction spectrometry analyses confirmed that AAGBFS mainly consists of calcium silicate hydrates with aluminum substitution. The optimal dosage condition with the highest metal ion removal was 8% m/v solid loading. Metal ion removal efficiency percentages remained relatively constant after 7 h, and were above 98.9% for Cu²⁺, Pb²⁺, Al^3＋, Cr^3＋, Zn²⁺, Fe^2＋ and Ni²⁺. The maximum adsorption capacity values were 0.682 mg/g; 2.134 mg/g; 2.409 mg/g, 0.008 mg/g, 1.216 mg/g, 135.318 mg/g and 0.005 mg/g, respectively, for Cu²⁺, Pb²⁺, Al^3＋, Cr^3＋, Zn²⁺, Fe^2＋ and Ni²⁺ respectively. Sorption was discovered to be an endothermic process, with more favorable sorption occurring at higher temperatures. The sorption process could be modeled well using the Langmuir isotherm and the Sips model. In addition, the process was optimized using a neuro-evolutive approach combining Differential Evolution and Artificial Neural Networks. AAGBFS loaded with heavy metals could be desorbed and reused for two cycles of adsorption before the removal efficiency of Ni2+ dropped to around 60%, allowing for the efficient and responsible use of resources. AAGBFS is an emerging and versatile sorbent for removal of heavy metal ions effectively.