Ultrasound–assisted adsorption of toxic dyes by cottonseed cake: Artificial neural networks, regression models and response surface optimization

Abstract

In this study, dehydrated cottonseed cake as a low-cost and abundant byproduct in Turkey was utilized as an adsorbent for the decolorization of Reactive Blue 19 (RB19) and Reactive Yellow 145 (RY145) from aqueous solutions based on adsorption and ultrasound-assisted adsorption (UAA). Decolorization efficiency was optimized as a function of changes in process type, initial pH value, adsorbent concentration, temperature, reaction time, and initial dye concentrations of RY145 and RB19 based on response surface methodology (RSM) using Box-Behnken Design. The maximum decolorization efficiency of 99.9% for both RY145 and RB19 was obtained with ultrasound-assisted adsorption under the RSM-optimized conditions (with unity desirability) of 76.98 and 79.40 min reaction times, 233.20 and 254.29 mg L-1 initial dye concentrations, 1.37 and 1.44 g L-1 adsorbent concentrations, and 35.42 and 49.37 °C, respectively. The best-fit multiple non-linear regression models of decolorization efficiency with the highest adjusted coefficients of determination (R2adj) explained 99.52% and 99.48% of variations through adsorption of RY145 and RB19 and 98.14% and 98.01% of variations through UAA of RY145 and RB19, respectively, while artificial neural networks accounted for 99.82%.