Predicting particle size of Iron magnetic nanoparticles synthesized from celery extract using artificial neural networks and regression learner models

Abstract

An important factor that defines the behavior of Ferric magnetic nanoparticles (FeMNPs) in environmental applications is the particle size. In this work, the researcher sought to utilize Artificial Neural Networks (ANN) and a regression learner, developed on Matlab software, to predict the particle size of FeMNPs. More precisely, the experiments involved the prediction of particle size of FeMNPs during their synthesis using celery extract. The characterization studies using field emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), and X-ray diffraction (XRD) analyses showed that the optimal experimental conditions gave the particle size of 18.01 nm. The two models that have been created for the prediction of particle size include Artificial Neural Network and Regression Learner using pH, reaction time, temperature, and celery extract concentration as parameters. From the analysis of the obtained graphs and the value of the mean square error of the ANN model equal to 0.0142, it was evident that the predictive capability of the model was satisfactory and in good agreement with the experimental data. The reliability of the predictions was further verified by the regression learner model, as from a data point, the predicted and actual particle sizes were 18.41 nm and 18 nm at pH 9. Thus, these models are believed to be efficient and precise in particle size prediction; moreover, they appear to be a useful tool for the optimization of the process factors.