Journal of Membrane Science 462 (2014) 131138

Contents lists available at ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

Gas/gas membrane contactors An emerging membrane

unit operation
Pingjiao Hao n, J.G. Wijmans, Jay Kniep, Richard W. Baker
Membrane Technology and Research, Inc., 39630 Eureka Drive, Newark, CA 94560, USA

art ic l e i nf o

a b s t r a c t

Article history:
Received 17 January 2014
Received in revised form
13 March 2014
Accepted 17 March 2014
Available online 21 March 2014

Gas/gas membrane contactors are devices in which a feed gas and a sweep gas are circulated on either
side of a membrane. The pressures of the two gas streams are often approximately equivalent, so the
principal driving force for permeation is the concentration difference between the feed and sweep gas
components. This type of contactor has been commercialized for energy recovery devices in air
conditioning applications. More recently, these contactors have been suggested for use in carbon
dioxide capture and sequestration. In this paper, the performance of an ideal contactor using perfectly
selective membranes is examined. For this type of ideal contactor, analytical equations can be derived
that allow the partial pressure proles within the contactor to be calculated. The effect of the sweep ratio
(sweep ow rate/feed ow rate) and feed pressure on separation performance of the contactors has been
calculated. In the nal section of the paper, the performance of contactors tted with membranes
permeable to other components of the feed gas is described.
& 2014 Elsevier B.V. All rights reserved.

Keywords:
Membrane contactors
CO2 capture
Gas separation

1. Introduction
This paper describes the operation of gas/gas membrane
contactors. These are typically low-pressure devices in which a
feed gas is circulated across one surface of a permeable membrane,
and an approximately equal volume of another gas, at a similar
pressure, is circulated countercurrently on the other side of the
membrane. Permeation occurs because of the partial pressure
difference between components in the feed and sweep gases. Most
of this partial pressure difference occurs because of concentration
rather than pressure differences across the membrane. This type of
contactor has been commercialized for energy recovery devices in
air conditioning and fuel cell humidity control applications [13].
More recently, these contactors have been suggested for use in
carbon dioxide capture [4,5].
Most membrane gas separation processes do not use a permeate side sweep gas. In conventional processes, a feed gas mixture
ows across the surface of a permselective membrane; a portion of
the mixture permeates the membrane and is removed as lower
pressure permeate gas. The remaining gas, depleted in the
permeating components, is removed from the feed side of the
membrane as a residue gas. One stream enters the membrane
module (the feed), two streams leave (the residue and permeate).

Corresponding author. Tel.: 1 650 543 3344.

E-mail address: pingjiao.hao@mtrinc.com (P. Hao).

http://dx.doi.org/10.1016/j.memsci.2014.03.039
0376-7388/& 2014 Elsevier B.V. All rights reserved.

A few membrane processes have been developed in which a

small ow of sweep gas is introduced on the permeate side of the
membrane. This gas usually ows countercurrently to the incoming feed. Two streams enter the module (the feed and sweep) and
two streams leave (the residue and permeate sweep). This type
of sweep process has been used when the membrane selectivity is
much higher than the pressure ratio across the membrane, and a
small amount of an extremely permeable component must be
removed. A well-known example is the removal of water vapor
from compressed air, rst commercialized in the 1990s [6,7]. In
this process, the membrane performance is pressure ratio limited,
and the separation can be much improved by using a small
fraction of dry gas as a permeate side sweep to reduce the partial
pressure of the permeating component (water) on the permeate
side of the membrane. In these applications, the volume of sweep
gas used is only a few percent of the volume of the feed gas to be
separated.
A few years ago, Membrane Technology and Research, Inc.
(MTR) proposed using a gas/gas membrane contactor to separate
carbon dioxide (CO2) from fossil fuel power plant ue gas [4,5].
If successfully developed, this could be an important membrane
application. In this process, CO2-rich ue gas passes across the feed
side of a membrane, while an approximately equal volume of air
passes as a sweep gas on the membrane permeate side. In this
way, the partial pressure of CO2 on the permeate side is maintained at a lower level than on the feed side. Because of this partial
pressure difference, CO2 passes from the ue gas into the sweep
air stream. If membranes are used that are very permeable to CO2,

132

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

but relatively impermeable to oxygen and nitrogen, very little

nitrogen passes from the ue gas into the sweep air and very little
oxygen passes from the air sweep into the ue gas. Because the
feed and permeate pressures are low, blowers are the only
compression equipment needed to circulate the gases across the
membrane. A useful separation is performed at a minimal
energy cost.
A block diagram illustrating the application of this device to
CO2 capture in a coal-red power plant is shown in Fig. 1. Air
passing countercurrently to the ue gas strips CO2 out of the gas.
The CO2-laden air is then used to burn coal in the power plant
boiler. This combustion process is used to produce steam to drive a
turbine and make electricity; it also generates a CO2-enriched ue
gas. A portion of this ue gas is separated as a CO2-enriched
stream by a conventional membrane unit. The remaining gas
passes across the feed surface of the membrane contactor and
becomes the CO2-depleted gas (2% CO2) that is discarded through
the chimney. By using the membrane contactor, the CO2 concentration in the ue gas from the boiler can be enriched from the
normal concentration of 1013% CO2 to 20% CO2 or more. Enhancing the CO2 concentration of the ue gas from the coal combustor
increases the efciency of the selective CO2 purge step. A more
concentrated CO2 purge is produced because only a portion of the
CO2 must be removed in this step. The remaining CO2 is recycled
by the membrane contactor. The use of a membrane contactor
signicantly reduces the cost of separating a CO2 concentrate
stream from the ue gas. This process is under development by
MTR as a potential CO2 capture technology.
In this paper, a gas/gas contactor of the type shown in Fig. 1 is
treated as a unit operation. The impacts of different process
parameters on unit performance are examined. The paper is
divided into two sections. In the rst section, we will show
calculations for an ideal contactor; that is, a contactor tted with
a membrane permeable to one of the components of the feed gas
(CO2), but impermeable to all other components. We will also
assume the component to be removed is present at a low
concentration (  1%) so that the volume change in the feed and
sweep streams caused by permeation can be ignored. The properties of a gas/gas contactor will then be illustrated using this ideal
device.
In the second section of the paper, we will illustrate the
complications that result when real membranes with limited
selectivity are used and permeation of other components is
possible. The changes in performance that occur when permeation
through the membrane causes volume changes to the feed and
sweep ows will also be examined. This type of contactor is closer
to the type that would be used for the application shown in Fig. 1.

2. Ideal contactor performance

The base case operating conditions for an ideal contactor that
forms the starting point of this analysis are shown in Fig. 2. The
base case contactor is assumed to have an area of 5000 m2, and
contains a membrane having a CO2 permeance of 1000 gpu. The
feed stream contains 1% CO2, and the sweep stream is pure
nitrogen. The feed and sweep streams are both at atmospheric
pressure. The device achieves 80% CO2 removal from the feed to
the sweep gas when the feed and sweep ows are set at
1 m3(STP)/s.
One concern with high permeance membranes of the type
shown in Fig. 2 is that concentration polarization effects may occur
in stagnant boundary layers on either side of the membrane. The
problem is expected to be most signicant on the side in contact
with the microporous support layer of the composite membrane.
This support creates a stagnant layer which is signicantly thicker

Fig. 1. Block diagram illustrating the use of a selective membrane contactor to

recycle CO2 to the boiler of a coal power plant. In this way, the concentration of CO2
in the ue gas exiting the boiler increases from 1013% to 20% at very little energy
cost [4].

Fig. 2. Base case conditions used in this paper for an ideal gas/gas membrane
contactor.

than the gas boundary layers in the gas channels. The likelihood of
concentration gradients forming in the stagnant layer can be
estimated by calculating the Peclet number, Jv/D, where Jv is
the actual gas velocity or volume ux in the layer, is the stagnant
layer thickness and D is the gas diffusion coefcient in the
stagnant layer gas at the stagnant layer pressure [8]. This dimensionless number represents the ratio of the convective transport Jv
and the diffusive transport D/. When the Peclet number is large
(Jv Z D/), the convective ux through the membrane cannot easily
be balanced by diffusion in the boundary layer, and concentration
gradients form in the boundary layers. When the Peclet number is
small (Jv rD/), convection is easily balanced by diffusion in the
boundary layer and signicant concentration gradients do not
form in the boundary layer. The CO2 permeance of the membranes
shown in Fig. 2 is 1000 gpu (1000  10  6 cm3(STP)/cm2 s cmHg).
Under typical operating conditions of the process in a power plant
type of environment (12 bar feed, 1 bar permeate and  10% CO2
in the feed gas), the volume ux through the membrane is about
1.5  10  2 cm3(STP)/cm2 s. Assuming the stagnant layer is at
atmospheric pressure, the supercial velocity through the microporous support in the layer is 1.5  10  2 cm/s. The actual velocity,
Jv, will be higher because of the effects of porosity and tortuosity;
we will assume here that the actual velocity is about six times
higher and equal to 1.0  10  1 cm/s. Assuming the microporous
support layer that separates the selective membrane layer from
the well-mixed counter-owing gas is 200 m thick (), and
taking the gas diffusion coefcient at atmospheric pressure to be
 0.2 cm2/s, it follows that the permeate-side Peclet number Jv/D
is 1  10  2. A Peclet number this small implies that diffusion is

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

133

Fig. 3. CO2 partial pressures on the feed and sweep sides of the ideal base case membrane contactor shown in Fig. 2. The driving force for permeation is constant because the
two ow rates are equal.

dominant over convection and that concentration gradients in the

stagnant layer are small and can be ignored.

One of the most important parameters affecting the operation

of a membrane contactor is the sweep ratio, dened as the
volumetric ow of the inlet sweep gas divided by the volumetric
ow of the inlet feed gas (both as standard m3/s, not actual m3/s).

2.1. Factors affecting ideal contactor performance

One of the best ways to understand the inuence of process
operating conditions on contactor performance is to calculate the
partial pressure difference between the gases on either side of the
membrane. The equations required to make this possible are
derived in an appendix to this paper. These equations have been
formulated into an Excel program that allows the partial pressure
on the feed and permeate side of the membrane to be calculated at
any point. This program is given in the supplemental material to
the paper.
The results obtained using these equations, for the base case
process shown in Fig. 2, are given in Fig. 3 below. The feed gas at a
concentration of 1% CO2 enters on the feed side of the membrane
at the left end of the module. An equal volume of nitrogen sweep
gas enters on the permeate side of the membrane at the right end
of the module. Because of the difference in partial pressure created
across the membrane, CO2 permeates from the feed to the
permeate side. In the base case process illustrated in Fig. 3, the
ows on either side of the membrane are equal. This means an
incremental ow of CO2 from the feed to the permeate side of the
membrane causes a decrease in the feed side CO2, while simultaneously producing an equal increase in the CO2 concentration of
the sweep gas on the other side of the membrane. The result
(derived mathematically in Appendix 1) is that the partial pressure
of CO2 decreases linearly on the feed side of the membrane and
increases linearly on the counter-owing sweep gas side. The
difference between the feed and sweep side partial pressure is the
driving force for CO2 permeation, and is equal at all points in the
membrane module.

Sweep ratio

Inlet sweep flow rate m3 STP=s

Inlet feed flow rate m3 STP=s

The denition of sweep ratio given in Eq. (1) is a reliable way of

characterizing the ideal membrane contactor shown in Fig. 2,
where only a small fraction of the feed and sweep gas ow
permeate the membrane (the process stage-cut is small). In the
second half of this paper, we will show a modied denition might
be used when a signicant volume ow takes place across the
membrane.
Fig. 4 shows the calculated partial pressure proles within
modules when the sweep ow is increased or decreased to
produce a sweep ratio of 0.5 or 2.0. Consider rst the case when
the sweep ratio is 0.5, shown in Fig. 4(a). Because the volume of
sweep gas is half that of the feed, only half of the CO2 in the feed
can permeate, even when an innitely permeable membrane is
used. If more than half of the feed gas CO2 permeates the
membrane, the sweep gas leaving the module would have a
higher concentration than the feed gas entering. CO2 would then
ow from the sweep to the feed. As Fig. 4(a) shows, a sweep ratio
of 0.5 removes almost half of the CO2 from the feed, reducing the
CO2 concentration of the feed to just a little over 0.5% CO2.
Concurrently, the CO2 concentration in the sweep gas increases,
reaching a little under 1.0% CO2 in the sweep gas leaving the
module. The driving force for CO2 permeation is the difference in
CO2 partial pressure (concentration) across the membrane, and as
the gure shows, the driving force is highest at the incoming
sweep end of the module. The bulk of CO2 permeation occurs at
the sweep input end.

134

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

Fig. 4. CO2 concentration proles in the base case membrane module shown in Fig. 2, except that the sweep ow rate, and thus the sweep ratio, is changed. (a) Sweep ow
rate 0.5 m3/s, sweep ratio 0.5: most CO2 permeation occurs at the incoming sweep end of the module. (b) Sweep ow rate 2.0 m3/s, sweep ratio 2.0: most CO2 permeation
occurs at the incoming feed end of the module.

When the sweep ratio is 2.0, shown in Fig. 4(b), the situation is
reversed. There is now more than enough sweep gas to remove all
of the CO2 from the feed. With a membrane having the base case
properties, the feed gas leaving the module contains less than 0.1%
CO2. The permeated CO2 leaves with the sweep air at a concentration of just under 0.5% CO2. The driving force is highest at the
incoming feed end of the module and the bulk of CO2 permeation
occurs at the feed input end.
An alternative way to show the effect of sweep ratio on CO2
removal from the feed is shown in Fig. 5. The calculations shown
in Fig. 5 and those reported elsewhere in this paper were
performed using a computer process simulator (ChemCad 6.3,
Chemstations, Inc., Austin, TX), enhanced with differential element
code for the membrane separation step, written at MTR. In the
Fig. 5 calculation, the base case module is operated with various
sweep gas ow rates to change the sweep ratio. The fractional
removal of CO2 from the feed gas is calculated at each sweep ratio.
Plots determined this way are shown for membranes with different CO2 permeances. The base case membrane has a CO2 permeance of 1000 gpu and the performance of this module (already
illustrated in Figs. 3 and 4) is shown on the line marked CO2
permeance 1000 gpu. At a sweep ratio of 0.5, the fractional
removal of this module is 49.6%; at a ratio of 1.0 (the base case),
the removal is 80%; and at a sweep ratio of 2.0, the removal
increases to 93.4%.
Two limiting regions are shown in Fig. 5. The rst limiting
region is to the left of the bold line that shows the fractional CO2
removal achieved by an innitely permeable membrane. This
boundary is dened by
Fractional removal Sweep ratio

In pressure driven processes, two similar limiting regions are

also known, governed by the pressure ratio of the processes and
called the pressure ratio limited region and the selectivity limited

Fig. 5. Effect of sweep ratio on CO2 removal from the feed, calculated for
membranes of different CO2 permeance. The feed ow rate is maintained constant
at 1 m3/s, while the sweep ow rate is changed to adjust the sweep ratio. The base
case membrane module (Fig. 2) is shown as a dot on the 1000 gpu membrane line.
Two limiting regions are shown: one is in the region below a sweep ratio of 1.0,
where the CO2 removal is limited at least in part by the sweep ratio, and the other
is a region at higher sweep ratios where the CO2 removal is limited at least in part
by the inability of the membranes to permeate sufcient CO2.

region [9]. The sweep ratio, like the pressure ratio, provides a link
between the driving force, selectivity and separation.
The boundary of the sweep ratio limited region can be derived
from simple mass balance considerations. As the curves in Fig. 5
show, at a sweep ratio of 0.5, all membranes with a permeance

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

Fig. 6. Feed and sweep concentration proles within the ideal contactor (Fig. 2),
operated with membranes having a permeance of 200 gpu. The sweep ratio is xed
at 100. Under these conditions, the exiting sweep gas contains very little CO2
( o 0.01%) and the driving force for CO2 permeation reaches the maximum
possible value.

135

Fig. 7. Contactor membrane permeances required to achieve 50% and 80% CO2
removal as a function of sweep ratio for the ideal 5000 m2 contactor illustrated in
Fig. 2.

above  1000 gpu produce essentially the same fractional CO2

removal; that is, very close to 0.5. Increasing the membrane
permeance above 1000 gpu does not produce a higher CO2
removal. The contactor performance is limited by the sweep ratio.
This limit occurs at all sweep ratios below 1.0. In this region, the
maximum fractional removal is set by the sweep ratio, no matter
how high the membrane permeance.
A second limiting region is also shown in Fig. 5. This region, on
the right-hand side of the gure, occurs when low membrane
permeance limits the removal of CO2. Performance in this region is
best explained by calculating the partial pressure on the feed and
sweep side of the membrane. This calculation is shown in Fig. 6,
for a membrane with a permeance of 200 gpu. The sweep ratio is
set at 100. Because of the very high sweep ratio, the partial
pressure of CO2 on the sweep side of the module is almost zero
at all points along the module. The driving force for permeation is
therefore at its maximum value and cannot be further improved
by increasing the sweep ow rate. The fractional removal of CO2
from the feed is then also at its maximum value, set by the
permeation rate of CO2 through the membrane. In the example
illustrated, a membrane with a permeance of 200 gpu achieves a
limiting maximum removal (xmax) of  55%.
The limiting (maximum possible) removal of CO2 at high sweep
ratios is linked to the contactor membrane permeance, area and
the operating conditions of the device. In Appendix 2 at the end of
this paper, it is shown that the term xmax is given by the
expression:

membrane permeance required increases asymptotically. In this

region, the separation reaches the sweep ratio limit set by Eq. (2).
Similarly at high sweep ratios, the membrane permeance required
to achieve 80% CO2 removal asymptotically approaches a limiting
value of 480 gpu. The plot for 50% CO2 removal has the same form.
The sweep ratio limit is at a ratio of 0.5, and the limiting
permeance is at 195 gpu.
Thus far, the discussion of contactor performance has been
limited to units operating with equal pressures on either side of
the membrane. The driving force for permeation in these units is
only due to the concentration differences between components of
the feed and sweep gases. However, the concentration driving
forces can be enhanced by creating a pressure difference across the
membrane. This effect is illustrated in Fig. 8, which compares the
partial pressure driving force proles for the base case shown in
Fig. 3, and the same device operating at a feed pressure of 2 bar.
When the pressure is equal on either side of the membrane
(1 bar/1 bar) [Fig. 8(a)], the partial pressure driving force feed-tosweep is uniform across the module at 0.002 bar. However, when
the same volume of feed gas is compressed to 2 bar [Fig. 8(b)], the
partial pressure driving force is signicantly higher. At the feed end
of the module, the driving force (feed-to-sweep) increases vefold
to 0.011 bar, and then decreases steadily as feed CO2 concentration
falls. The net result is to increase CO2 removal from 80% to 95%.

!
 P CO2 =lpft A
xmax 1  exp
Ff

In the description of contactor performance given thus far, we

have made two signicant assumptions. First, the membranes are
permeable to one component and impermeable to all others.
Second, the volume of gas permeating the membrane is small
compared to the feed ow. These assumptions are realistic for
some contactor applications. For example, in many dehydration
applications, the permeability of water can be several-hundredfold higher than that of the other components in the feed. Also, the
feed often contains only 12% water, so the volume ow through
the membrane is small. However, in other applications, including
CO2 separation from ue gas, the simplifying assumptions are no
longer valid, so the separation performance will deviate from the
ideal contactor behavior described thus far. The consequences of
these effects are described below.
For this section of the paper, we will use a new base case
contactor operating under the conditions shown in Fig. 9. Comparing this case to the ideal base case (Fig. 2):

where P CO2 =l is the membrane permeance, ptt is the total feed

pressure of the contactor, A is the contactor surface area and Ff is
the feed volume ow rate (STP) to the contactor.
One nal way to illustrate this interaction of sweep ratio and
permeance is to replot the data in Fig. 5 as a plot of permeance
against sweep ratio for various levels of CO2 removal. Two
scenarios are shown in Fig. 7, one for 80% CO2 removal and the
other for 50% removal. On the line marked 80% removal, the base
case is marked as a solid point (black circle) at a permeance of
1000 gpu and a sweep ratio of 1.0. Increasing the permeance of the
membrane used in the base case device above 1000 gpu allows
lower sweep ratios to be used while still reaching the target of 80%
removal. However, as the sweep ratio approaches 0.8, the

2.2. Non-ideal contactors

136

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

Fig. 8. Effect of increasing the feed side pressure on the partial pressure proles through the base case membrane contactor. Increasing the feed pressure from 1 bar [Figs. 8
(a)] to 2 bar [Fig. 8(b)] increases the driving force (partial pressure difference, feed-to-sweep) vefold at the feed end of the module. The fractional CO2 removal achieved
with the same size module then increases from 80% to 95%.

Fig. 9. Non-ideal base case membrane contactor.

 The concentration of CO2 in the feed gas has been increased

tenfold.

 The sweep gas is air containing 21% oxygen.

 The membrane still has a CO2 permeance of 1000 gpu, but is
now also permeable to nitrogen (25 gpu) and oxygen (50 gpu).
The permeances of these membranes are comparable to membranes currently available for CO2 separations at industrial operating conditions.
Comparing the new base case contactor performance assumptions in Fig. 9 with the ideal contactor in Fig. 2 shows three main
differences. First, the denition of the sweep ratio used in Eq. (1)
for the ideal contactor no longer reects the reality of the new
contactor. Using the Eq. (1) denition, the contactor has a sweep
ratio of 1.0 [sweep ow rate 1 m3(STP/s) and feed ow 1 m3(STP/
s)], but at the feed end of the contactor, the sweep-to-feed ratio is
1.08 (m3/s)/1.0 (m3/s) or 1.08, while at the sweep end of the
contactor the ratio is 1.0 (m3/s)/0.92 (m3/s), or 1.09. Dening the
sweep ratio as the average of the inlet and exit sides of the sweep
ratio is clearly a better method to use than the ratio of the inlet
sweep and feed ows. Second, because of the volume
ow through the membrane, 80% CO2 removal from the feed
means the residue gas concentration is 2.1% CO2. The membrane
area required to achieve the same fractional removal is then
slightly less, at 4800 m2 rather than 5000 m2. Finally, some oxygen
from the sweep side permeates into the feed side and some
nitrogen from the feed permeates to the sweep. This effect is
described below.

Fig. 10. Calculated oxygen concentration of sweep gas in a non-ideal membrane

contactor used to remove 80% CO2 from ue gas with a countercurrent air sweep
stream. The performance obtained with different hypothetical membranes is
plotted as a function of the ue gas feed pressure. The sweep ratio is maintained
at 1.0 in all the calculations shown.

3. The effect of membrane selectivity

The calculations for the ideal contactor described in the rst
section of this paper assumed the contactor membrane was
permeable to CO2 and impermeable to all other gases. Fig. 9 shows
what might be expected when a membrane with more realistic
properties is used. The membrane still achieves the required
removal of CO2 from the feed, but a signicant amount of nitrogen
also permeates with the CO2 to the sweep, and some oxygen back
permeates from the sweep into the feed. A consequence of these
additional ows is that when this contactor is used in the CO2
separation scheme illustrated in Fig. 1, the oxygen concentration in
the air sweep sent to the coal boiler decreases from 21% to
approximately 18%. Fortunately, an oxygen level of 18% in the
boiler air stream will still provide efcient combustion, although
minor changes to the boiler burner may be required [10].
Fig. 10 shows a plot of the oxygen concentration in the sweep
gas leaving the contactor on its way to the boiler as a function of
feed gas pressure. The performance proles of several different
hypothetical membranes are shown. The CO2 ue gas feed pressure is varied from a pressure of 1 bar (the base case shown in
Fig. 9) to a feed pressure of 3 bar. As described earlier in the

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

discussion of Fig. 8, increasing the feed pressure has a very

signicant impact on the driving force for CO2 permeation and
contactor performance. This means the membrane area required
to perform the target separation decreases from 4800 m2 at a feed
gas pressure of 1 bar (the base case) to 850 m2 at a feed gas
pressure of 3 bar.
The top line in Fig. 10 is a reference line showing the normal
oxygen concentration of air (21%). The next line down, marked
CO2 1000 gpu, N2 0 gpu, O2 0 gpu, is the calculated performance
of a contactor tted with a membrane that permeates CO2, but is
impermeable to nitrogen and oxygen. The oxygen concentration in
the sweep air leaving a contactor tted with this membrane is 19.3%.
This decrease in oxygen concentration from 21% to 19.3% is due to the
dilution effect of CO2 permeating the membrane into the sweep gas,
thus increasing its volume. The next line in Fig. 10, marked
CO2 1000 gpu, N2 25 gpu, O2 0 gpu, shows the dilution effect
due to permeation of both CO2 and nitrogen from the feed into the
sweep gas. The nitrogen contribution to dilution increases slightly as
the feed pressure increases, reecting the effect of pressure on the
driving force for nitrogen permeation. Finally, the bottom line in
Fig. 10, marked CO2 1000 gpu, N2 25 gpu, O2 50 gpu, shows the
sweep gas concentration when the base case membrane from Fig. 9
is used. The difference between this line and the one immediately
above it is the contribution of oxygen loss from the sweep gas to the
feed. This loss is small compared to the combined dilution effects of
CO2 and nitrogen, and decreases as the feed pressure increases
because the membrane area needed to perform the target separation
decreases with increasing feed pressure.

4. Conclusions
In this paper, we have shown the effect of operating parameters
on the performance of a gas/gas membrane contactor used for CO2
removal. These devices are not in common use today, but could
nd future use in CO2 capture processes. The most important
operating parameters affecting the gas/gas contactor performance
are the sweep gas-to-feed gas volume ratio, the relative pressures
of the feed and sweep gases, and the permeance and selectivity of
the membranes used.

137

Starting from the expression for gas permeation of individual

components1:
Ji

Pi f
p  psi
i

A1

we obtain the following equation for the partial pressure gradients:

!
f
dJ i P i dpi dpsi


A2
d d d
The partial pressure gradients can also be obtained from the
mass balance in each differential element:
J i da 

J i da 

dpfi
pf
f
dp
or
 Ji t
i
f
d

Ff
pt

A3

dps
Fs s
ps
dp or i  J i t
ps i
d
Fs

A4

Ff

Combining Eqs. (A2)(A4) gives

!
dJ i P i pst pft
J

d i F s F f

A5

Integrating Eq. (A5) over the membrane area, the expression for
the permeate ux as a function of the membrane area then gives
f

J i;a J i;0 eP i =pt =F s  pt =F f a J i;0 eba

s

with
P ps pf
b i t t
Fs Ff

A6

!
A7

Inserting Eq. (A6) into Eqs. (A3) and (A4) and integrating over
the membrane area then give the following expression for the
partial pressures in the two exit streams:
pfi;A pfi;0 J i;0

psi;0

pft 1 ebA
Ff
b

psi;A  pfi;0 P i =pst =F s 1  ebA =b

1  P i =pst =F s 1  ebA =b

A8

A9

Combining with following equation,

Pi f
p  ps i;o
i;0

Acknowledgments

J i;0

This work was performed as part of a research program

supported by the U.S. Department of Energy Project no. DE-FE0007553.

we obtain for the partial pressure in the sweep outlet stream:

psi;0

psi;A  pfi;0 P i =pst =F s 1  ebA =b

1 P i =pst =F s 1  ebA =b

A10

A11

and for the partial pressure in the feed outlet stream:

Appendix 1
Equations used to calculate the partial pressure difference
between the gases on either side of a gas/gas membrane contactor
(Fig. A1).

pfi;A pi;0 f 

pft =F f s
p  psi;A
pst =F s i;0

A12

An Excel le based on these equations, which calculates the

pressure proles and local permeate uxes, is attached in the
supplementary material to this paper. The graphics in the le will
auto-adjust if the inputs are changed.

Appendix 2
The limiting (maximum possible) removal of CO2 is linked to
the contactor membrane permeance, membrane area and the
Fig. A1. Model for the countercurrent sweep module. Pressure drops in the feed
and sweep channels are ignored. The permeating compound is assumed to be
present at a low concentration, which means that the feed and sweep ow rates
can be assumed to be constant throughout the module.

1
In the equations that follow the superscripts f and s represent the feed and
sweep side of the membrane, and the terms 0 and A represents positions along the
membrane module from the feed entrance (0) to the feed exit (A).

138

P. Hao et al. / Journal of Membrane Science 462 (2014) 131138

operating conditions of the device. The CO2 removal (x) is the ratio
of the amount of CO2 permeating the membrane divided by the
amount of the CO2 that enters the contactor; that is,
RA
J CO2 dA
A13
x 0
F f nCO2
where J CO2 is the ux through the membrane at any point in the
contactor area A. Ff is the volumetric feed ow of gas into the
contactor at standard temperature and pressure conditions (STP),
and nCO2 is the molar fraction of CO2 in the feed gas.
The maximum CO2 removed (xmax) is obtained at innite sweep
ratio, in which case the membrane ux at any point is proportional
to membrane permeance P CO2 =l and the feed side CO2 partial
pressure pfCO2 (the permeate side CO2 partial pressure is close to
zero and can be ignored). Eq. (A13) can then be written2
R0
f
A P CO2 =lpCO2 dA
xmax
A14
F f nCO2

Jv
J CO2
i
nCO2
P CO2 =l
pfCO2 ;0
pfCO2 ;A
pft
xmax

and since
nCO2

pfCO2 ;0

A15

pft

where pft is the total pressure on the feed side of the module,
R0
pft P CO2 =l A pfCO2 dA
xmax
A16
F f pCO2 ;0 f
R0
The integral of the partial pressure driving force A pfCO2 dA has
the familiar form for the log mean and Eq. (A16) can become
xmax

pft P CO2 =lA

F f pfCO2 ;0

pfCO2 ;0  pfCO2 ;A
lnpfCO2 ;0 =ppfCO2 ;A

A17

Because the fractional removal (xmax) can be also be written as

xmax

pfCO2 ;0  pfCO2 ;A

A18

pfCO2 ;0

Combining Eqs. (A17) and (A18) gives

0
1 

pfCO2 ;0
P CO2 =l pft A
@
A

 ln1  xmax
ln f
Ff
p

A19

CO2 ;A

which can be rearranged to

xmax 1  exp

P CO2 =l U pft UA
Ff

!
A20

This expression shows the dependence of the limiting value for

CO2 removal on the permeance, pressure, area and feed ow rate
of an ideal contactor.

Nomenclature
A
Ff

Fs

actual gas velocity through the membrane support,

cm3/cm2 s
carbon dioxide ux through the membrane at any
point in the contactor, cm3(STP)/cm2 s
ith component in the feed or sweep side.
molar fraction of CO2 in the feed gas
CO2
permeance,
gpu
[1  10  6 cm3(STP)/
(cm2 s cmHg)]
CO2 partial pressure at the feed entrance of the
module (point 0), cmHg
CO2 partial pressure at the residue exit end of the
module (point A), cmHg
total feed pressure, cmHg
maximum possible fractional removal of a component from the feed gas, that is, the ratio of the
amount of a gas permeating the membrane divided
by the amount of the gas that enters the contactor.
For this paper, xmax is discussed and calculated only
for CO2.

area of the membrane contactor, m2

volumetric feed ow of gas into the contactor at
standard temperature and pressure conditions,
m3(STP)/s
volumetric sweep ow of gas into the contactor at
standard temperature and pressure conditions,
m3(STP)/s

2
In the equations that follow the superscripts f and s represent the feed and
sweep side of the membrane, and the terms 0 and A represents positions along the
membrane module from the feed entrance (0) to the feed exit (A).

References
[1] D. Chen, W. Li, H. Peng, An experimental study and model validation of a
membrane humidier for PEM fuel cell humidication control, J. Power
Sources 180 (2008) 461467.
[2] K. Mahmud, G.I. Mahmood, C.J. Simonson, R.W. Besant, Performance testing of
a counter-cross-ow run-around membrane energy exchanger (RAMEE)
system for HVAC applications, Energy Build. 42 (2010) 11391147.
[3] Y. Kusano, H. Shimanuki, T. Katagiri, M. Suziki, Humidiers, US patent
6,659,433, Dec 2003.
[4] T.C. Merkel, H. Lin, X. Wei, R.W. Baker, Power plant post-combustion carbon
dioxide capture: an opportunity for membranes, J. Membr. Sci. 359 (2010)
126139.
[5] T.C. Merkel, X. Wei, Z. He, L.S. White, J.G. Wijmans, R.W. Baker, Selective
exhaust gas recycle with membranes for CO2 capture from natural gas
combined cycle power plants, Ind. Eng. Chem. Res. 52 (2013) 11501159.
[6] K.L. Wang, S.H. McCray, D.D. Newbold, E.L. Cussler, Hollow ber air drying, J.
Membr. Sci. 72 (1992) 231244.
[7] H. Lin, S.M. Thompson, A. Serbanescu-Martin, J.G. Wijmans, K.D. Amo, K.
A. Lokhandwala, B.T. Low, T.C. Merkel, Dehydration of natural gas using
membranes, Part II: sweep/countercurrent design and eld test, J. Membr.
Sci. 432 (2013) 106114.
[8] P.L.T. Brian, in: U. Merten (Ed.), Mass Transport in Reverse Osmosis in
Desalination by Reverse Osmosis, MIT Press, Cambridge, MA, 1966, p. 161.
[9] R.W. Baker, J.G. Wijmans, Molecular separation of organic vapor from gas
streams, in: D.R. Paul, Y. Yampolskii (Eds.), Polymeric Gas Separation Membranes, CRC Press, Boca Raton, FL., 1994.
[10] T.C. Merkel, Pilot Testing of a Membrane System for Post-Combustion CO2
Capture, DOE NETL CO2 Capture Technology Meeting Presentation, Pittsburg,
PA, July 10, 2013.