Integrated response surface and machine learning approach: Experimental optimization and DFT analysis for NaN3 removal via NaClO oxidation

The treatment of wastewater containing sodium azide (NaN 3 ) poses a major challenge to environmental protection and public safety. In this study, a multi-scale analysis method integrating experimental optimization, machine learning modeling and theoretical calculation was proposed. The mechanism and optimal conditions of NaN 3 removal by sodium hypochlorite (NaClO) oxidation were systematically studied. Through the combination of single factor experiment design and response surface method (RSM), the optimum treatment conditions were determined as follows: dilution factor was 300, NaClO dosage was 3.5 mL, and pH value was 6.0. Under these conditions, the removal rate of NaN 3 reached 96.53 %. In addition, the developed BP-GA hybrid neural network model showed excellent prediction performance (R 2  = 0.98805). After further optimization of process parameters, the removal rate of NaN 3 increased to 98.2 %. Density functional theory (DFT) calculations revealed the significant effects of pH value and intermediate HOCl on the reaction efficiency, and clarified the microscopic mechanism of the reaction. This study optimizes the treatment process of wastewater containing NaN 3 , which provides a theoretical basis for the design of a new efficient oxidation system and is of great significance for promoting sustainable wastewater treatment.