Effect of Temperature on Ultrasonic Degradation of Sodium Poly(Styrene Sulfonate): Analysis of Online Viscometric Data with Theoretical Models and Machine Learning Approaches

Abstract

The effect of temperature (15, 20, 25, 30, and 35 °C) on the ultrasonic degradation of sodium poly(styrene sulfonate) polyelectrolyte (5.10−3 g/ml) in aqueous NaCl solutions (2.75 m) was investigated. Using the thousands of data obtained by an online viscometer monitoring technique, the degradation kinetics, and the effect of temperature on this process were analyzed in detail with 12 theoretical models reported in the literature. The Giz, OHM, and Tang-Goto models were found to give good results, and these models were selected to evaluate the effect of temperature on the ultrasonic polymer degradation. The degradation constant and the limiting molecular weight at which no polymer chain scission occurs were found to decrease with increasing temperature for all models. Furthermore, modeling was performed using popular machine learning algorithms, such as Multilayer Perceptron, Adaptive Neuro-Fuzzy Inference System, Long Short-Term Memory, Gradient Boosting Regression, Light Gradient Boosting Machine, eXtreme Gradient Boosting Regressor, CatBoost, and Random Forest. The best results were obtained with the Long Short-Term Memory, a deep learning approach. In addition, optimization techniques, such as Particle Swarm Optimization, Genetic Algorithm, Artificial Bee Colony, Clustering, and Global Optimization Method based on a parabolic approach, were used to determine the optimum molecular weight evolution and degradation temperature.