Adsorptive removal of two basic dyes (Methylene Blue and Malachite Green) from binary system using low cost adsorbents

Abstract

Abstract The removal of dyes from the industrial effluent is become a critical hazards in present society. The challenge is more formidable when more than one dye is to be removed from the effluent in a cost effective manner. The objective of the present study is the adsorptive removal of mixture of two basic dyes malachite green (MG) and methylene blue (MB) from the aqueous solution by using four low cost adsorbents neem leaf ash, jack fruit leaf ash, bagasse fly ash and rice husk ash. The influence of different operating parameters such as adsorbent bed height (4 to 8 cm), initial dye concentrations (25 to 100 mg//L), influent flow rate (5 to 10 mL/min) and the pH of the dye mixture in aqueous solution have been investigated in column study. The influences of adsorbent dose (2 to 5gm), initial dye concentration (25 to 150 mg/L), shaker speed (30 to 165 rpm), contact time (30 to 190 min) and pH (4.1 to 9.2) of the dye mixture have considered in batch study. Langmuir, Freundlich and Temkin isotherm model along with pseudo-first and second order kinetic equations have been used in batch mode. Thomas, Yoon-Nelson, Adams-Bohart and BDST models have been applied to experimental data to study the time-concentration profile for the determination of dye uptake of the adsorbent in dynamic column study. The maximum dye uptake of the two dyes in mixture was 19.90 mg/g at 50mg/L dye concentration and at the 7.5 mL/min flow rate. The Thomas and BDST model have defined the experimental run better than the other two models. Artificial neural network (ANN) modeling is used on the data to predict the performance of the laboratory experiment for both the batch and column study. It was proved that the developed ANN model efficiently optimize the adsorption experiment of two dyes onto all four low cost adsorbent. Statistical t-test has been applied to check the closeness of the experimental data with the ANN simulation. Low t-score and high ‘p’ value(>0.05) established the similarities between experimental results and ANN outcomes.