Transport Through Polymer Membranes: G The Polymer Is in Its Glassy
Transport Through Polymer Membranes: G The Polymer Is in Its Glassy
Transport Through Polymer Membranes: G The Polymer Is in Its Glassy
than vibrations. All structural properties as the distribution of cavities in the polymer
have effectively become static quantities. Depending on the conditions during the
formation of the glassy state (for example, the temperature gradient) the polymer is
more or less trapped in a non-equilibrium state. In glassy polymers the penetrant
diffusion is low but size selectivity is very good.
Above the glass transition temperature the polymer is in its rubbery state. In this state the
polymers are generally tough and flexible, which is associated with freer chain motion.
Rubbery polymers have very short relaxation times (compared to glassy polymers) and
respond very rapidly to external stresses. Thus a change in temperature causes an
immediate adjustment to the new equilibrium state. A similar immediate adjustment
occurs when small penetrants are absorbed in a rubbery polymer. Larger segments of
the polymer are thought to participate in the penetrant diffusion process due to internal
chain motions such as chain rotations, translations and stronger vibrational motions.
The penetrant diffusion is much faster than in a glassy polymer but size selectivity is
lower.
In both the glassy and rubbery state the polymer properties can be further modified
by the presence of crystalline phases, by stress induced orientations or as a function
of cross-link density. They tend to place additional constraints on the mobility of the
amorphous phase through which diffusion takes place. (It is partly because of these
possible variations in polymer properties that there is such a wide range in experimental
values of permeability coefficients)
J = P dc
:1)
(2
P = DS
:2)
(2
J (r) = c(r)v(r)
:3)
(2
where v (r) is the average velocity of the dissolved particles. In the linear regime v (r )
can be written as
v(r)
=
=
F th
, 1 r(r)
1
:4)
(2
the chemical
:5)
(2
Phase 1
Polymer Membrane
c1
Phase 2
c2
ex
Figure 2.1: The schematic polymer membrane. The membrane separates two phases with
concentrations c1 and c2 and chemical potential 1 and 2 . We assume that is continuous
at the interface and that ex (the excess chemical potential) is constant throughout the
homogeneous membrane.
in which 0 is the standard chemical potential of the ideal gas phase based on unit molar
concentration, c is the local concentration and ex (r) is the excess chemical potential of
the dissolved species with respect to the ideal gas state. Equations 2.3, 2.4 and 2.5 give
c(r) r (r):
J (r) = ,RT
r
c
(r ) ,
ex
:6)
(2
Equating RT= with the diffusion coefficient D , eq. 2.6 can be written as
J (r) = ,D
:7)
(2
Equation 2.7 is still general. We now consider a membrane with thickness d in the
x-direction and infinite dimensions in the yz-plane. The interfaces at x = 0 and x = d
are in contact with concentrations c1 and c2 (c = c2 , c1 ) and we assume that an
ideal gas phase is in equilibrium across both interfaces. Hence is continuous at the
interfaces. Furthermore ex is assumed to be constant throughout the homogeneous
membrane. This implies that any concentration dependence of ex is neglected. Thus
0 + RT ln c1 = 0 + RT ln c(0) + ex
:8)
(2
or
c(0) = c1
exp(,ex =RT )
c(d) = c2
exp(,ex =RT ):
:9)
(2
similarly
:10)
(2
If ex is constant then rex is zero. Then, for a stationary flux J (r ) we find that
according to equation 2.6, c(x) is a linear function of x and the gradient in equation 2.7
is equal to (c(d) , c(0))=d. Equation 2.7 now reduces to
J = ,DS dc ;
(2
S = exp(,ex=RT ):
(2
:11)
with
:12)
Note that it is not necessary to explicitly simulate a membrane with actual interfaces. This would
place high constraints on the MD simulations. All the information needed from the simulations is the
diffusion constant D in the bulk polymer and the excess chemical potential ex of the particles in the bulk
polymer compared to the ideal gas phase.
10
feed
dP
Membrane
dt
extrapolation of the steady-state part of the curve, the time-lag can be obtained and the
diffusivity can be calculated with:
d2 ;
6
Pfeed
Mgas d dPd ;
Vd RT
A dt
:13)
(2
can be calculated
:14)
(2
in which Pfeed is the applied upstream pressure, Vd the downstream compartment volume, Mgas the molecular weight of the penetrant gas at density and A the membrane
area.
The solubility coefficient S is usually calculated from the diffusion and the permeability
coefficients, using the relation P = DS . But if the permeation rate is too fast it is not
possible to determine the diffusion coefficient with the desired accuracy. Then S has to
be measured separately.
11
12