Statistical optimization and artificial neural network modeling for acridine orange dye degradation using in-situ synthesized polymer capped ZnO nanoparticles

详细信息    查看全文

• 作者：Nitesh Dhiman^a ; ^b ; Markandeya^c ; ^d ; Amrita Singh^a ; Neeraj K. Verma^a ; ^e ; Nidhi Ajaria^f ; Satyakam Patnaik^a ; ^b ; ^{satyakampatnaik@iitr.res.in} ; ^{satyakampatnaik@yahoo.com}
• 关键词：Guar gum ; ZnO nanoparticles ; Acridine orange ; Response surface methodology ; Box-Behnken Design ; Artificial Neural Network
• 刊名：Journal of Colloid and Interface Science
• 出版年：2017
• 出版时间：1 May 2017
• 年：2017
• 卷：493
• 期：Complete
• 页码：295-306
• 全文大小：2296 K
• 卷排序：493

文摘

ZnO NPs were synthesized by a prudent green chemistry approach in presence of polyacrylamide grafted guar gum polymer (pAAm-g-GG) to ensure uniform morphology, and functionality and appraised for their ability to degrade photocatalytically Acridine Orange (AO) dye. These ZnO@pAAm-g-GG NPs were thoroughly characterized by various spectroscopic, XRD and electron microscopic techniques. The relative quantity of ZnO NPs in polymeric matrix has been estimated by spectro-analytical procedure; AAS and TGA analysis. The impact of process parameters viz. NP's dose, contact time and AO dye concentration on percentage photocatalytic degradation of AO dyes were evaluated using multivariate optimizing tools, Response Surface Methodology (RSM) involving Box-Behnken Design (BBD) and Artificial Neural Network (ANN). Congruity of the BBD statistical model was implied by R2 value 0.9786 and F-value 35.48. At RSM predicted optimal condition viz. ZnO@pAAm-g-GG NP's dose of 0.2 g/L, contact time of 210 min and AO dye concentration 10 mg/L, a maximum of 98% dye degradation was obtained. ANOVA indicated appropriateness of the model for dye degradation owing to “Prob. > F” less than 0.05 for variable parameters. We further, employed three layers feed forward ANN model for validating the BBD process parameters and suitability of our chosen model. The evaluation of Levenberg-Marquardt algorithm (ANN1) and Gradient Descent with adaptive learning rate (ANN2) model employed to scrutinize the best method and found experimental values of AO dye degradation were in close to those with predicated value of ANN 2 modeling with minimum error.