Journal of Minerals & Materials Characterization & Engineering, Vol. 7, No.2, pp 175-191, 2008
jmmce.org Printed in the USA. All rights reserved
175
Towards the Development of CO
2
Separation Membranes
Gerard Caneba*, Michael Renier, and Brian Ott
Department of Chemical Engineering and
Center for Environmentally Benign Functional Materials (CEBFM)
Michigan Technological University, Houghton, MI 49931
*Email: caneba@mtu.edu
Abstract
This work pertains to current results in the development of CO
2
separation membranes
from flue gas streams typically found in coal-fired power plants. A versatile free-radical-
based polymerization method is employed for the development of a multifunctional block
copolymer that has good affinity to CO
2
, is processable into and applicable as gas
separation polymer membranes. In order to validate the resulting materials, thin slabs of
the polymer were melt-processed, and then sorbed with CO
2
and N
2
in a pressure cell.
When the pressure is released, foaming tendencies at the outer regions of the samples
were observed. A quantitative model involving measurements of unfoamed regions is
used to correlate with permeability ratios as well as CO
2
-polymer mutual diffusivities.
One particular optimized material, called RB1-215, is shown to possess a good CO
2
relative permeability to N2. Thus, the experimental methodology has been shown to
possibly be able to develop the next generation of CO
2
separation polymer membranes
for carbon sequestration applications.
Keywords
Carbon sequestration, polymer membranes, CO
2
separation, flue gases, FRRPP process
1. INTRODUCTION
The separation of CO
2
from flue gases in coal power plants has been known to be an
attractive method of sequestering potential global warming/climate change gases. In this
regard, the use of membrane separation of CO
2
is attractive because of energy savings
that can potentially be realized. For example, conventional methods rely on cryogenic
conditions in order to absorb the CO
2
even with the use of favorable fluids, such as
ammonia, amines, glycols, etc. The problem of CO
2
sequestration from flue gases
reduces to its separation from N
2
gas, because other components are either minor, easily
176 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
removable, or can be included with the CO
2
-rich product stream. Thus, good CO
2
separation factor relative to N
2
in polymer membranes translates to a viable CO
2
sequestration operation.
Gas membrane separation in this paper involves the solution-diffusion mechanism of the
penetrant through a polymer film. Penetrant is introduced on one side of the film and the
other side is maintained at a lower penetrant concentration. The rate per unit film area at
which the penetrant traverses the thickness of the film is related to the diffusivity. Details
of the calculation of the diffusivity depend on the type of experiment used.
1.1 Steady-State Experiment
In the steady-state experiment, the film with thickness L has no penetrant at the
beginning. At time t=0, the upstream face of the film is exposed to a constant penetrant
concentration C
1
, while the downstream face is exposed to a lower constant penetrant
concentration, C
2
. If the system is at steady-state and there is no chemical reaction in the
film, the flux F of the penetrant is constant. Fick’s first law of diffusion results in
l
CCD
F)(
21
= (1),
where D is the mutual diffusivity.
1
Note that Equation 1 applies to constant diffusivity
for a binary system.
For a multicomponent penetrant system, it is customary to apply Equation 1 for each of
the penetrant components, provided these components do not interact with one another.
Also, it is customary to express the penetrant i concentration in terms of its partial
pressure on the corresponding face of the membrane (Henry’s Law), such as
iii
pSC
=
(2).
Thus, for a noninteracting multicomponent penetrant system following Equation 2
(Henry’s Law) Equation 1 becomes
)(
21 iiii
ppPF
−=
(3),
Where P
i
is the permeability of component i. If are comparing relative fluxes of
penetrants A (such as CO
2
) and B (such as N
2
) through the polymer membrane, then the
permeability ratio or separation factor between A and B is obtained as
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
177
BB
AA
B
A
AB
SD
SD
P
P==
α
(4).
Values for permeability ratios for CO
2
relative to N
2
are 17 in polysulfones and 6 in
PDMS, and these membranes have N
2
permeabilities of 0.5-5 L(STP)/min-m
2
-atm and 5-
50 L(STP)/min-m
2
-atm, respectively.
2
Thus, the task of membrane design is to increase
both selectivity and flux of CO
2
compared to N
2
at various prevailing temperatures and
pressures.
1.2 Concentration-Distance Measurements
The idea behind this approach is the measurement of the concentration profile of the
penetrant, with subsequent calculation of the diffusivity. Methods of measurement of
penetrant concentration include the use of various means of establishing contrast. A
simple way is to use differences in light intensities if the penetrant includes a dye in it.
Another method is based on refractive index differences with penetrant concentration. In
the interference technique,
3,4
Fabry-Perot fringes corresponding to refractive index
contours are obtained when viewing the polymer perpendicular to the diffusion direction
through semi-silvered glass plates with about 10 % light transmission. If the refractive
index is known as a function of concentration in the monotonic sense, then concentration-
distance curves can be easily obtained from the fringe pattern. If the refractive index has
a maximum with concentration, it is still possible to calculate the concentration-distance
curve with some difficulty. Another way of obtaining the refractive index profile is
through the Lamm scale method developed by Hutcheon and coworkers as a
microtechnique.
5
They used a 1-m focal length at unit magnification to record the
“distorted” scale photographically. The scale displacements on the plate are related to the
refractive index gradient.
Other detection methods for the concentration profile that have been used in the past
include the use of radiotracers.
6-9
The simplest way of using concentration profile measurements is to cut the film into
parallel strips along the direction of diffusion after steady state is reached. From
concentration measurements of the strips of swollen polymer and the flux of penetrant,
Equation 1 can be used to calculate the diffusivity. This approach was first used by
Gillespie and Williams to obtain the diffusivity of water in cellulose acetate.
10
From the measured concentration profile of the unsteady-state system, the error-curve
method has also been introduced to determine constant diffuvities.
11
For calculation of
178 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
concentration-dependent diffusivities, the Matano method has been used to determine
diffusivities of the whole concentration range of the polymer and penetrant.
12
It is also possible to estimate the penetration distance of a penetrant by subsequently
dropping the penetrant pressure to cause it to form a foam in the polymer. An
examination of the foamed front provides an idea of the penetration depth and thus, the
penetrant diffusivity within the polymer. This penetration depth is also related to the
solubility of the penetrant, since a threshold amount of solubility is needed for a foam to
form within the polymer. Details can be quantitatively resolved by considering the
problem as that of a semi-infinite polymer medium for the penetrant. Since it takes about
6 wt % of blowing agent to form a foam, then the lowest penetrant concentration of the
foam layer can be considered to be at 6 wt %.
2. EXPERIMENTAL
2.1 Polymer Design
Among all the polymerization methods, only the free-radical-based mechanism offers a
robust technique in implementing the formation of block copolymer materials with the
widest variety of monomer types. The reason is that reactive sites (acid, base, hydrogen
bonds, ions) from some of the more useful monomers normally interfere with the clean
formation of the polymer chain. Whereas, free radicals do not normally interact with
other reactive groups in various monomers and polymer segments. Also, solvents used in
the polymerization process normally do not interfere with free-radical reactivities, while
other polymerization mechanisms can be sensitive to the solvent used. Thus, selection of
solvent(s) in a mixture of monomers to impart multifunctionality is very difficult, unless
a free-radical-based polymerization mechanism is employed. It should therefore be no
surprise that commercially sold polymer membranes are made from premade resins or
their modified forms. Even for these materials, they are sold with almost no flexibility in
terms of the functionalities of their segments during polymerization; functional flexibility
is imparted through functionalization reactions of side groups, which is limited at best.
In this work, a collection of monomers, oligomers, and/or polymers possessing various
chemical groups are incorporated within a set of polymer chains that make up the main
component of the composition. These functional groups may be naturally present in the
raw materials used to produce the multifunctional multipolymeric membranes, or may be
purposely or naturally altered by modifying reactions. A subpolymer (equivalent to a
number of monomer segments) representation of these chemical functionalities in a
portion of the entire polymer is shown in Figure 1, which is depicted by the A,B,C,D,…-
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
179
functionalities emanating from various polymer segments within the subpolymers. To
form block copolymers, there will be a set of adjacent segments that have the same
functional groups.
Figure 1.
Subpolymer representation of proposed multifunctional multipolymeric
membrane materials with various functional groups (A, B, C, D, …)
Table 1.
Possible functionalities (A,B,C,D,…) for the above structural representation of
the multifunctional multipolymeric membranes. Examples of commercial polymers with
various functionality classes as well as their corresponding application features are given.
A,B,C,D,…-
Functionality
Class
Examples of A,B,C,D,…-Functionality found
in Commercial Polymers
Application features of
functional groups
Hydroxyl Poly(vinyl alcohol) Biodegradability, adhesion
to polar surfaces (wood,
masonry, cement),
colloidal properties
Ketone Poly(methyl vinyl ketone), Poly(isopropyl vinyl
ketone), Ethylene-Carbon Monoxide
Copolymers
Carbon dioxide
absorption, UV
degradability, distinct UV
absorbance
Ester Poly(vinyl acetate), Poly(methyl acrylate),
Poly(ethyl acrylate), Poly(methyl
methacrylate), poly(butyl acrylate), Ethylene-
Vinyl Acetate Copolymers
Good interactions with
carbon dioxide, adhesion
to polar surfaces and for
composition adjustment
based on application
temperature
Carboxylic acid Poly(acrylic acid), Poly(methacrylic acid), Reactivity with metal and
A
B
C
D
A
C
E B
F
G C
B
H
D
I
B
J
E
180 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
Ethylene-Acrylic Acid Copolymers, Ethylene-
Methacrylic Acid Copolymers
non-metal bases with and
without additional
functionalities, Water
dispersibility, adhesion to
metals
Amine Poly(acrylamide), Poly(isopropyl acrylamide) Good interactions with
carbon dioxide, reactivity
with acids with and
without additional
functionalities, Water
dispersibility
Epoxy Glycidyl methacrylate copolymers Self-crosslinking,
reactivity with amines,
isocyanates, alcohols
Aromatic
Hydrocarbon
Polystyrene Source of chromophores,
and capability of
modification of the
aromatic ring (such as
sulfonation)
Aliphatic
Hydrocarbon
Polyolefins Affinity to hydrocarbon
oils and moisture
resistance
Halogens Poly(vinylidene chloride), poly(vinylidene
fluoride), polychlorotrifluoroethylene,
polytetrafluoroethylene, Zonyl™ copolymers
Superior strength,
chemical, thermal, fire
resistances, Zonyl™
copolymers have good
carbon dioxide
interactions
Note that the above table shows functionality classes that could impart carbon dioxide
permeability in the overall polymer.
A two-plus stage reactor procedure is proposed using the design-of-experiment guidelines
(DOE) in Table 2 below, using the so-called free-radical retrograde-precipitation
polymerization process.
13-14
The development of an optimized polymer as membrane material for CO
2
separation
could first be done through a selection of basic features, such as CO
2
absorption, strength,
reactive and/or nonreactive adhesion/crosslinker, and chemical/thermal resistances. In
addition, processability aspects have been looked into through low temperature elasticity
segments and use of various possible block copolymer architectures, such as diblock and
triblock-type of copolymers. A halogenated RB1-215 triblock (A-B-C) material with
strength A-group, reactive B-group, and CO
2
-interacting C-group was developed.
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
181
The development of the RB1-215 material was done through a design-of-experiment
approach as shown in the Table 2. Its proposed morphological structure is shown in
Figure 3 below.
Table 2.
DOE matrix and responses expected from the block copolymer under
development.
# !st
Stage
2
nd
Stage
Arc. Responses
V
C Z
G
B
D,T Strength CO
2
Absorp
tion
Chem
Resis.
Thermal
Resis.
Adh.
1
H
L L
L
L
D L L TBD TBD TBD
2
H
L L
H
L
D H L TBD TBD H
3
H
L H
H
L
D TBD H TBD TBD H
4
H
L H
L
L
D TBD H TBD TBD TBD
5
H
L L
L
L
T H L TBD TBD TBD
6
H
L L
H
L
T H L TBD TBD H
7
H
L H
H
L
T H H TBD TBD H
8
H
L H
L
L
T H TBD TBD TBD TBD
V – for strength
C – comonomer for reaction facilitation/processability of strength monomer (V)
Z – for CO
2
absorption
G – for adhesion/crosslinking promoter
B – for low temperature elasticity
D – Diblock-based molecular architecture
T – Triblock-based molecular architecture
First Stage Compositions (based on first stage polymer):
L – 5-15%
H – 50-90%
Second Stage Compositions (based on overall polymer unless otherwise noted):
L – 0-20% (B based on 2
nd
stage polymer), 10-20% (Z,G)
H – 20-40%
Both high pressure (4,500 psig) metal and low pressure (45 psig) glass stirred-tank
reactor systems have been used in this work (Figure 2 below), recipes and conditions
were generated using high throughput experimentation methods.
182 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
Figure 2.
Stirred tank reactor systems (45-psig pressurized 1-L glass reactor on left and
4,500 psig Parr reactor systems on the right) that are available for this work.
Figure 3.
Depicted bulk morphology of RB1-215 copolymer
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
183
2.2 Postprocessing
After the polymerization runs, the RB1-215 polymer product was separated from the
reactor solvents , unreacted monomers, and unreacted initiator. First, the polymer was
coagulated with excess water (3:1 v/v or more) out of the reactor fluid into a gel or dough
(Figure 4 below).
Figure 4.
Coagulated RB1-215 product on a Petri dish
Then, solvent and unreacted reactants are extracted from the dough. The dough is dried
into a purified and safe-to-handle product, and melt-blended with other components, if
needed (Figure 5 below).
2.3 Solvent/Reactant Extraction
The residual solvent and reactants must be removed from the “dough”. Initial attempts to
purify the polymer included one vacuum oven drying step (4 hours at 80
o
C), used to
drive off the residual solvent, and followed by up to three supercritical CO
2
extraction
steps (8 hour soak at room temperature and 2000 psig), used to remove unreacted
monomers, oligomers and residual solvent.
184 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
Figure 5.
Dried and melt-blended RB1-215 product on a Petri dish
2.4 Membrane-related Foaming Experiments
Figure 6 below depicts the preliminary testing procedure used to screen various polymers
for gas sorption and diffusion, which translates to permeabilities. The polymer sheet of
1.61-2.34 mm in thickness (L
0
) is exposed to the gas (CO
2
or N
2
) at constant temperature
and pressure for a certain period of time. Then, the pressure is adiabatically reduced to
atmospheric for 1 minute or less. A portion of the sample adjacent to the surface will
form a foam, while the interior will be unfoamed. This unfoamed region will have a
thickness of L
u
. Thus, the penetration distance,
L, of the gas will be
2
0u
LL
L
=∆ (8).
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
185
Figure 6.
Procedure used to determine the penetration depth of a gas into the polymer.
Figure 7 below shows some of the details of the apparatus used by this project to generate
monolithic polymer foams in order to gauge their capabilities of absorbing CO
2
and N
2
.
(a)
L
0
Expose at
constant T,
P, for a
period of
time
Then, freeze-
fracture to show
Penetration
Depth,
L
L
u
Penetration Depth,
L = (L
0
-L
u
)/2
Foamed
Portion
Unfoamed
Portion
186 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
(b)
Figure 7.
Apparatus used to produce foam layers onto polymer samples: (a) is the cell
used to contain the polymer; and, (b) is the syringe pump and piping system used to
meter liquid carbon dioxide into the pressure cell where it turned into a gas or a
supercritical fluid.
3. RESULTS
3.1 Fractional Precipitation
The polymer RB1-222 (similar to RB1-215) was fractionated with 50% ethanol/water
solution. The polymer solution was diluted either 10:1 or 5:1 v/v with NMP as solvent.
Based on the fractogram (Figure 8), there is very little intermediate Stage 1 precipitate
(Block A product) at 0-15 ml precipitant added. A quantitative analysis of indicates that
the unreacted Stage 1 contamination amounts to less than 1 wt% of the product
copolymer. This result of relatively insignificant unreacted Stage 1 contamination seems
to be generally the case.
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
187
RB1-222 Fractional Precipitation
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
050100 150200 250
50/50 EtOH/Water Titrant Added, ml
Dry Weight Precipitate, g
Figure 8.
Fractional precipitation data for RB1-222 (similar to RB1-215) copolymer
product
3.2 Foam Formation
Results of the foaming experiment depicted in Figure 6 are shown in Table 3 below.
Table 3.
Results of 1-hr gas penetration experiments in various polymers.
Polymer T, °C P, psig L
0
, mm L
, mm L, mm
PMMA 47-48 700-750
CO
2
1.65 1.65
̴
0
PVMK 47-48 700-750
CO
2
2.34 1.48 0.43
RB1-215 47-48 930-933
CO
2
1.905
̴
0
̴
1.905
RB1-217
(Similar to
RB1-215)
36 740-750
CO
2
1.615
̴
0
̴
1.615
RB1-215 36 1500 N
2
1.905
̴
1.905
̴
0
RB1-217 36 1500 N
2
1.615
̴
1.615
̴
0
188 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
Some of the fractured surfaces are shown in Figures 9-12 below.
Figure 9.
Foamed and unfoamed interior portion of a poly(vinyl methyl ketone)
(PVMK) sample, showing the unfoamed region to be 1.48 mm thick.
Figure 10.
Foamed and unfoamed interior portion of an RB1-217 sample (similar to
RB1-215), showing the unfoamed region to be almost nonexistent.
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
189
Figure 11.
Foamed and unfoamed interior portion of an RB1-215 sample, showing the
unfoamed region to be almost nonexistent.
Figure 12.
Foamed and unfoamed interior portion of a PMMA sample, showing the
almost no foamed region.
190 Gerard Caneba, Michael Renier, and Brian Ott Vol.7, No.2
3.3 DSC/TGA Analysis
Results of thermal analyses that were done with intermediate and final products agree
with the thermal transitions depicted in Figure 3.
4. DISCUSSION OF RESULTS
From Table 3 and Figures 9-12, it is evident that the RB1-215 material has promise as
CO
2
separation membranes from N
2
. This is seen from the zero penetration depth for N
2
compared to CO
2
even at N
2
pressure of 1500 psig, compared to 750 psig for CO
2
. These
relative penetration depth values also indicate large relative permeability values of CO
2
to
N
2
for the RB1-215-type of polymer.
It is possible to determine an order-of-magnitude estimate of the minimum CO
2
diffusivity for the RB1-217 membrane through the following equation
( )
( )
t
L
D
L
Dt
2
2
1
(9)
Thus, for the penetration thickness of 1.614 mm and exposure time of 1 hr or 3600 s, the
diffusivity is at least 7.2x10
-6
cm
2
/s, which is a relatively large number for diffusion of
penetrants in polymer materials.
Thus, it is evident that the RB1-215-type material has been developed to be a promising
membrane for continuous separation of CO
2
from N
2
.
5. CONCLUSION
The development of multifunctional block copolymers has therefore been shown to
possibly result in the development of CO
2
-separating membranes for carbon sequestration
applications.
ACKNOWLEDGEMENTS
The authors wish to acknowledge partial support of the Michigan Technological
University, Center for Environmentally Benign Functional Materials (CEBFM), a unit
under the Sustainable Futures Institute (SFI). Acknowledgement is also given to
Raytheon Missile Systems (Tucson, AZ) and the Department of Defense (through
Vol.7, No.2 Towards the Development of CO
2
Separation Membranes
191
DARPA) for partial funding of work that yielded this manuscript. Finally, the authors
would like to acknowledge Dr. Munir Tasdemir of Marmara University (Istanbul) for
assistance in sample preparation and foaming studies.
REFERENCES
1. Crank, J., 1970, “Mathematics of Diffusion”, Oxford University Press, London.
2. McCabe, W.L, Smith, J.C., and Harriott, P., “Unit Operations of Chemical
Engineering”, 7
th
Edition, McGraw-Hill Book Co, 2005, Chapter 26.
3. Robinson, C., 1950, Proc. R. Soc., A204, 339.
4. Crank, J. and Robinson, C., 1951, Proc. R. Soc., A204, 549
5. Hutcheon, A.T., Kokes, R.J., Hoard, J.L., and Long, F.A., 1952, J. Chem. Phys.,
20, 1232.
6. Bueche, F., Cashin, W.M., and Debye, P.V., 1952, J. Chem. Phys., 20, 1956
7. Bressler, S.E., Zakharov, G.M., and Kirrilov, S.V., 1961, Vysokomolek. Seodin.,
3, 1072
8. Moore, R.S. and Ferry, J.D., 1962, J. Phys. Chem., Wash., 66, 2699
9. Auerbach, I. And Gehman, S.D., 1954, Analyt. Chem., 26, 658
10. Gillespie, T. and Williams, B.M., 1966, J. Polym. Sci. A-1, 4, 933
11. Gosting, L.J., 1956, Adv. Protein Chem., 11, 429
12. Matano, C., 1932-3, Jap. J. Phys., 8, 109
13. Caneba, G.T., "Free-Radical Retrograde-Precipitation Polymerization Process,"
U.S. Patent No. 5,173,551, December 22, 1992
14. Caneba, G.T., “Formation of Radicalized Vinylidene Chloride Copolymer
Particulates and Related Materials”, provisional U.S. patent application, June,
2007