| Abstract: | Complete fractionations into 5, 10, and 20 fractions were calculated by a numerical method based on the Flory-Huggins theory in order to evaluate various procedures for determining the molecular weight distribution from fractionation data. If the initial distributions are wide, the differential distribution cannot be accurately reconstructed, not even if each fraction is characterized by two average molecular weights (instead of one, as is customary). In addition to this inadequacy in the evaluation procedure there are the experimental errors which further detract from the accuracy of the result. The integral distribution can, in some cases, be approximated fairly well by means of the Schulz method, provided that the polymer is separated into many fractions with narrow distributions. However, the integral distribution thus obtained does not reflect details in the differential distribution. Polymer fractionation does not appear to be a suitable procedure for accurate determination of the differential distribution. From the assembled material, a thermodynamic method has been derived which seems to hold out better prospects. It should enable the differential distribution to be directly determined from a detailed analysis of the liquid–liquid phase relationships, provided the free energy of mixing function of the system is known. |
