CHEMICAL-POTENTIALS OF POLYMER BLENDS FROM MONTE-CARLO SIMULATIONS - CONSEQUENCES ON SANS-DETERMINED CHI-PARAMETERS

Monte Carlo simulations on symmetric polymer blends were conducted to examine critically the applicability of Flory theory to polymer blends and the random phase approximation that is used to analyze scattering data from these blends. The novelty of the approach presented here is that the chemical potentials of chains are evaluated using the chain increment method. Surprisingly, we find that the chemical potentials at constant pressure can be modeled by the Flory-Huggins free energy form with a composition-independent (chi) parameter. The numerical values of (chi), however, cannot be anticipated a priori, and we therefore surmise that the good fit obtained to Flory theory is simply due to the cancellation of errors. The corresponding results for blends whose molar volumes are independent of composition, however, can only be modeled with a composition-dependent (chi), in agreement with previous lattice simulations of Binder. These results suggest that, contrary to accepted concepts, the incorporation of equation-of-state effects has important consequences in determining the thermodynamic properties of polymer blends. We have also attempted to model these simulation data by incorporating the effects of fluctuations but show that currently available theories predict improper trends. As a consequence of these results we show that the unusual composition dependence for (chi) parameters determined from scattering experiments ((chi SANS)) can be attributed, at least in part, to small excess volume changes on mixing which are not incorporated in the incompressible random phase approximation (i-RPA). The simple treatment presented here, which incorporates volume changes on mixing, then allows for the prediction of the ''upturn'' or ''downturn'' of (chi SANS) as a function of composition that is observed experimentally. Subsequently, we show that even extremely small volume changes on mixing, such as 0.05%, are sufficient to explain experimental trends for (chi SANS). Finally, we make contact with past theories and show that the results presented here are qualitatively similar to previous work of Freed and co-workers and Schweizer and Curro, both of which emphasize the importance of packing effects in this context, although differences still exist.