Toward a More General Solution to the Band-Broadening Problem in Size Separation of Polymers

Abstract

The molecular weight distributions (MWDs) and hydrodynamic volume distributions of polymers can reveal considerable mechanistic information on the polymerization process, and have significant effects on physical properties such as viscosity. While the broadening function for a particular SEC setup can be found using ultranarrow standards, these are extremely difficult to obtain. The present paper implements and tests a suggested technique (Aust. J. Chem. 2005, 58, 178) to enable the deconvolution of size distributions using broad standards, synthesized under conditions which are expected to produce a number MWD P(M) which is a single exponential. Broad with a wide range of (M) over bar (n) were synthesized for both styrene and methyl methacrylate (MMA), using low-conversion free-radical polymerization with appropriate choice of chain transfer agent (CTA) and initiator concentrations; with high (M) over bar (n)were synthesized at 25 degrees C without added initiator. The broadening function was obtained by assuming a flexible functional form (exponential Gaussian hybrid) and least-squares fitting its parameters so that the theoretical exponential P(M) curves for each sample, with exponents obtained experimentally, matched the experimental SEC distribution for styrene. The procedure was tested by using the same band-broadening function to deconvolute data for the original polystyrene standards and the polyMMA samples, using the Ishige deconvolution method. This method tends to amplify noise, and too tight a tolerance can lead to spurious structure in the deconvoluted distributions. Nevertheless, a tolerance range could be found which led to stable solutions, where the deconvoluted P(M) curves for both were indeed single exponential over the range of molecular weights where data with acceptable accuracy could be obtained. This suggests that this is a generally applicable method to correct for band broadening for a wide range of systems, although improved deconvolution methods are needed to obtain truly converged and stable solutions.