Following Di Benedetto it is proposed that noncrystalline polymer regions possess an approximate semicrystalline order with chain bundles that are locally parallel along distances of several nanometers. Packing with on‐average four nearest neighbors is assumed. A spherical molecule may move through such a substrate in two distinct ways: (a) along the axis of a “tube” formed by locally parallel chains or (b) perpendicular to this axis by two polymer chains separating sufficiently to permit passage of the molecule. The first process is relatively fast, generally requires little activation energy, and determines the effective jump length in diffusion. The second is responsible for the activation energy of diffusion, which is taken as the minimum energy necessary to produce a symmetrical chain separation which allows transfer of a molecule. This is calculated as a function of the penetrant diameter d and parameters Γ and β which characterize the interchain cohesion and chain stiffness, respectively. Γ is estimated from the polymer density and cohesive energy density by suitably linearizing a relation given by Di Benedetto for the potential between two polymer chains approximated as infinite strings of Lennard‐Jones force centers. β is shown to be approximately obtainable from the polymer chain backbone geometry and bond rotation potentials. An expression for the diffusion coefficient D is developed which contains only one disposable parameter, the effective jump length.

No abstract

The theory developed in Part I of this series is applied to a number of nonvinyl “smooth” chained homopolymers. The agreement between predicted and observed activation energies of diffusion for simple penetrants is generally good, particularly for polyethylene. Discrepancies observed for the smallest penetrants, He and H2, in some polymers may be rationalized in terms of atomic scale irregularities on the polymer chain surface. It is shown that in favorable cases the theory may permit diffusion data to be used as an additional check on the accuracy of conformational energy maps for polymers.

The theory developed in Part I of this series is modified to accommodate polymers that possess closely spaced, bulky side groups on the chains. The side groups give rise to free space between the chain “cores,” which reduces the chain separation required for penetrant motion transverse to the local chain axis. The theory is then identical to that of Part I, except that penetrant diameters minus a constant factor are employed in place of the normal diameters. In most of the cases studied the reduction factor for a given polymer may be estimated with reasonable precision from chain geometry data. This diameter‐reduction effect is the likely explanation of the apparent proportionality between the activation energy of diffusion and the square of the penetrant diameter reported earlier for vinyl polymers. The data quoted here and in Part II are analyzed to give a semitheoretical correlation between the effective jump length L̄ and ΔE, the activation energy of diffusion. This correlation appears to be equally valid for glassy and rubbery noncrystalline polymers.