Detector response as a function of molecular weight and its effect on size exclusion chromatography molecular weight determination of polymer distributions

In size exclusion chromatography (SEC), if the response of the detector is not constant with molecular weight, the calculated molecular weights are not a true representation for the sample. Without response factors, or some other method of correcting the response for the specific polymer calibration standards used, the resulting distribution is biased. From literature references, it is known that polystyrene has a non-constant, refractive index (RI) detector response with molecular weight. However, this is generally ignored since the effect is relatively small and does not preclude their use as calibration standards due to their narrow dispersities. Analysis within Dow Coming has confirmed the literature result. Polydimethylsiloxane (PDMS) has a relatively constant refractive index response (confirmed by Analysis of Variance statistical method). Therefore, analysis of PDMS polymers using (RI) detection gives a true molecular weight distribution with little bias in the distribution. For siloxane resins, the refractive index response is not constant with molecular weight. So far, only infrared (IR) and viscometry detection have the necessary consistancy for a true molecular weight distribution determination. The IR detection is readily explained from IR principles, but the viscosity can Only be explained by hypothesising the siloxane resin to adopt essentially a spherical configuration in solvent. Polyethylene glycol, used for analysis of silicone polyethers, shows a marked non-constant response with molecular weight using both RI and IR detection. Unless the molecular weight data are transformed using the equation of the fit line or response factors are used to correct the detector responses, the molecular weight distributions calculated will have a bias associated with them. The variability in response seems to correlate with molecular weight, suggesting either that polymer endgroups may be having a significant effect upon detection, or the polymers may be preferentially adsorbing on to the columns. Polyisobutylene seems to have two areas of constant RI detector response with a step change between them. The standards used in this study were purchased from two separate vendors and may have been manufactured by different processes, which could account for this difference in response. Otherwise, the calibration is virtually constant, and would give effectively unbiased distributions. (C) 1997 Elsevier Science Ltd. All rights reserved.