Application of Response Surface Methodology and Genetic Algorithm for Optimization and Determination of Iron in Food Samples by Dispersive Liquid–Liquid Microextraction Coupled UV–Visible Spectrophotometry

A simple and facile method was developed for the determination of trace amount of iron. The method is based on the complex formation between Fe (III) and picrate anion in the presence of piroxicam, as a complexing agent. Dispersive liquid–liquid microextraction (DLLME) was applied to extract the formed ion associate, Fe (III)-piroxicam. The absorbance of the extracted iron in the sedimented phase was measured by UV–Vis spectrophotometry. Two statistical methods of response surface methodology and genetic algorithm (GA) based on artificial neural network (ANN) were employed for prediction and optimization of a four-constituent DLLME. Plackett–Burman design was used for screening the influential parameters including pH, the volume of picrate anion, disperser, and extraction solvents. Central composite design (CCD) was used to obtain the optimum levels in the proposed method. The experimentally obtained data were used to train the GA model. CCD and GA models were compared for their predictive abilities. The result showed that both models have the ability to predict the proposed process, but ANN model is more reliable than CCD. The absorbance of the extracted iron obeys Beer’s law in the range of 0.03–0.96 \(\upmu \hbox {g}\,\hbox {mL}^{-1 }({R}^{2} = 0.998)\), and the limit of detection of 0.008 \(\upmu \hbox {g}\ \hbox {mL}^{-1}\) and enhancement factor of 88.84 were achieved for the process. The developed procedure was successfully applied to the determination of iron in water samples and two types of common vegetation sample, i.e., tea and mint.