Application of design of experiments, response surface methodology and partial least squares regression on nanocomposites synthesis

Abstract

We present an integral chemometric treatment for characterization and synthesis of low-density poly(ethylene)/organically modified montmorillonite nanocomposites using direct melt processing method, complementing design of experiment (DOE), response surface methodology (RSM) and partial least squares (PLS) regression. A central composite circumscribed DOE was used to study the influence of four processing parameters—concentration of clay (Clay%), concentration of compatibilizer (Comp%), mixing temperature (T mix) and mixing time (t mix)—on six nanocomposites properties: interlayer distance, decomposition temperature, melting temperature, Young’s modulus, loss modulus and storage modulus. PLS-regression was used to simultaneously correlate parameters and responses. RSM was used to explore interactions among parameters and predict nanocomposite properties on the experimental region. The six responses were simultaneously PLS-modeled with R 2 = 0.768 (p ≤ 0.05) being Clay% and Comp% the most important parameters and t mix the least influential. Moreover, significant (p ≤ 0.05) and complex interactions among Clay%, Comp% and t mix were found. A complementary interpretation of score and loading plots, coefficient plots, variable importance plots and response surface plots are explained to show how to find the optimal combination of processing parameters according the desired nanocomposite properties.