Prediction of copolymer composition drift using artificial neural networks: copolymerization of acrylamide with quaternary ammonium cationic monomers

Abstract

The free radical copolymerization of acrylamide with a quaternary ammonium cationic comonomer, diethylaminoethyl acrylate (DMAEA), has been investigated in inverse-emulsion. The copolymer composition was determined from residual monomer concentrations using an h.p.l.c. method. Both reactivity ratios were observed to change with conversion. Furthermore, the reactivity ratio of the cationic monomer was found to be a function of the ionic strength and monomer concentration and, to a limited extent, the polymer concentration and the organic-to-aqueous phase ratio. Therefore, the classical binary ultimate group copolymerization scheme cannot predict copolymer composition drift throughout the reaction. An artificial neural network (ANN) has been built to predict the copolymer composition. ANNs have the ability to map nonlinear relationships without a priori process information. The results show that an ANN can predict the copolymer composition very well as a function of reaction conditions and conversion. It is expected that for any system where the reactivity ratios are conversion dependent that an ANN, such as the one developed herein, will be preferable.