Comparative study of response surface methodology and artificial neural network for optimization of process parameters for synthesis of gold nanoparticles by Desmostachya bipinnata extract

Abstract

Green synthesis of nanoparticles has gained attention due to its eco-friendly and sustainable approach to synthesize nanoparticles at a reduced cost. Artificial neural network (ANN) and response surface model (RSM) are important to reduce experimental efforts in nanoparticle synthesis. In this work, optimization of gold nanoparticle synthesis by Desmostachya bipinnata extract was performed using the volume of plant extract, concentration of auric chloride, reaction time, pH, and temperature as process parameters, and the output was absorbance. The experimental design was obtained from RSM and the model was optimized further using ANN. Thirty-two experimental runs generated by RSM were performed and the results obtained experimentally were compared with those generated by RSM and ANN. Different algorithms of ANN were tested to obtain the best one. The optimization studies resulted in a maximum response for 20th run with 15 ml, 2.5 mM, 45 min, 7, and 40 °C as parameters. Optimized input parameters obtained by RSM were 10 ml, 2 mM, 30 min, 6, and 30 °C. The formation of gold nanoparticles was confirmed by UV spectroscopy, XRD, and SEM. Different algorithms of ANN, such as leven marquardt, scaled conjugate gradient, and bayesian network were used. Leven marquardt algorithm was found to be the most suitable algorithm for the current study.