Pure and mixed gas n-C4H10 and CH4 permeability coefficients in poly(1-trimethylsilyl-1-propyne) (PTMSP) are reported at temperatures from -20 to 35 degrees C. CH4 partial pressures range from 1.1 to 14.6 atm, and n-C4H10 partial pressures range from 0.02 to 1.8 atm. CH4 permeability decreases with increasing n-C4H10 upstream activity (f/f(sat)) in the feed. For example, at -20 degrees C, CH4 permeability decreases by more than an order of magnitude, from 52,000 to 1700 Barter, as n-C4H10 activity increases from 0 to 0.73. In contrast, n-C4H10 mixed gas permeability is essentially unaffected by the presence of CH4. The depression of CH4 permeability in mixtures is a result of competitive sorption and blocking effects, which reduce both CH4 mixture solubility and diffusivity, respectively. Diffusion coefficients of n-C4H10 and CH4 in mixtures were calculated from mixture permeability and mixture solubility data. The CH4 concentration-averaged diffusion coefficient generally decreases as n-C4H10 activity increases. On the other hand, the n-C4H10 diffusion coefficient is essentially unaffected by the presence of CH4. Pure and mixed gas activation energies of permeation and diffusion of CH4 and n-C4H10 are reported. The mixed gas n-CAO/CH4 permeability selectivity increases with increasing n-C4H10 activity and decreasing temperature, and it is higher than pure gas estimates would suggest. Mixture diffusivity selectivity also increases with increasing n-C4H10 activity. The difference between pure and mixed gas permeability selectivity arises from both solubility and diffusivity effects. The dual mode mixed gas permeability model describes the mixture permeability data reasonably well for n-C4H10. However, the model must be modified to accurately describe the methane data by accounting for the decrease in methane diffusivity due to the presence of n-C4H10 (i.e., blocking). Even though the penetrant concentrations are rather significant at some of the conditions considered, no evidence is observed for phenomena such as multicomponent coupling that would require a model more complex than the binary form of Fick's law. That is, Fick's law in its simplest form adequately describes the experimental data. (c) 2007 Elsevier Ltd. All rights reserved.