Modeling and optimization of imidacloprid degradation by catalytic percarbonate oxidation using artificial neural network and Box-Behnken experimental design.

Due to its toxicity and persistence, pesticide pollution poses a serious threat to human health and the environment. Imidacloprid or IMD is an archetypal neonicotinoid insecticide commonly used to protect a variety of crops worldwide. The present study examines the applicability of two numerical tools -- artificial neural network (ANN) and response surface methodology - Box Behnken design (RSM-BBD) -- to model and optimize oxidative IMD degradation by sodium percarbonate (SPC). The influences of SPC dose, Fe catalyst dosage, and solution pH on IMD removal were evaluated. An ANN composed of an input layer with three neurons, a hidden layer with eight optimum neurons, and an output layer with one neuron was developed to map the complex non-linear process at different levels. Seventeen designed runs of different experimental conditions were derived from RSM-BBD. These experimental conditions and their response values showed to be best fitted in a reduced cubic model equation. Sensitivity analyses revealed the relative importance of the various components: Fe (40.4%) > pH (31.1%) > SPC dose (28.5%). The two model were highly predictive with overall coefficients of determination and root-mean-square errors of 0.9983 and 0.31 for ANN, while 0.9996 and 0.20 for RSM-BBD. Overall, the present study established ANN and RSM-BBD as valuable and effective tools for catalytic SPC oxidation of IMD contaminants. SPC is a cleaner alternative to other oxidants for pesticide degradation as it is non-toxic, safe to handle, and produces by-products that inherently exist in the natural water matrix.