Facilitated transport of CO,. through highly swollen ion-exchange membranes: the effect of hot glycerine pretreatment* John
D. Noble t
National Bureau of Standards, Center for Chemical Engineering, 7 7 3 . 1 0 , 3 2 5 Broadway, Boulder, CO 8 0 3 0 3 , USA Received 7 July 1988 A pretreatment for perfluorosulphonic acid ion-exchange films, using glycerine and heating, causes the membranes to swell and imbibe greater (up to three-fold) amounts of water than obtained using standard membrane hydration techniques. The permeation of CO, CO2 and H2S has been tested for these membranes in the diffusive transport mode and with carrier-mediated (facilitated) transport, using ethylenediamine (EDA). When compared to "normally" hydrated membranes the results indicate that: 1, these membranes are up to 50% thicker; 2, have four to six times higher flux in both diffusive and facilitated transport; 3, maintain a high degree of facilitation for CO2 and H2S;and 4, maintain similar selectivity versus non-carrier reactive gases as previously observed. Experimental results show the effects of pretreatment temperature on the CO2 flux. Keywords: CO2 separation; ion-exchange membranes; facilitated transport
The facilitated transport of acid gases, CO2 and n2s, has been studied I-3 using a poly-perfluorosulphonic acid (PFSA) ion-exchange membrane for the production of H2 from coal gasification processes. Facilitated transport is a process whereby a complexing agent (carder), contained in a liquid membrane, can reversibly react with a solute of interest. A diagram of this process is shown in Figure 1 and additional details have been presented in review articles 4-6. The use of ion-exchange membranes as supports for facilitated transport was first described by LeBlanc et al. 7 for both CO2 and ethylene. Using ion-exchange membranes overcomes two major problems associated with immobilizing a liquid membrane in a porous support. The solvent and carder can be lost from a porous support leading to gas 'short-circuiting', but due to the non-porous nature of an ion-exchange film this cannot happen. Also in an ion-exchange film the carder is held in the membrane by electrostatic forces which protect the carder from oxidative deactivation and also allows a greater amount of carder to be loaded into the membrane. PFSA is a cation exchange membrane with a fluoropolymer backbone. Negatively charged sulphonic acid or sulphonate groups are permanently attached to the polymer backbone at random intervals. The ionic side chains account for the hydrophilic nature of the film. The prevailing view of the internal structure of PFSA films supports the existence of ionic clusters. The sulphonate side groups are concentrated in these cluster regions, surrounded by a predominantly fluorocarbon rich phase.
The clusters are of the order of 4 nm and contain the bulk of the water in the film. Channels, of the order of 1 nm, are envisioned as connecting the main clusters. Diffusion through the film is probably dominated by transport through the cluster/channel network 8. In facilitated transport the carder molecule is a cation which is probably associated with the ion-exchange sites of the membrane. The overall permeability of gaseous solute is controlled by its solubility and diffusion coefficient in the ionic (cluster/channel) network, as well as its reaction kinetics with the carder molecule and the resulting complex's diffusivity through the ionic network. Therefore, increases in gas permeance (P/l) may be accomplished by decreasing the film's thickness, increasing the solute-complex's diffusion coefficient and/ or increasing the volume fraction of the ionic network. Initial work in this area used commercial 200 lam
CO ..... ~
*Contribution of the National Bureau of Standards, not subject to copyright t University of Colorado, Department of Chemical Engineering, Boulder, CO 80309, USA Figure 1 0950-4214/88/030126-05503.00 © 1988 Butterworth Et Co (Publishers) Ltd. 1 2 6 Gas Separation Et Purification 1 9 8 8 Vol 2 September
CO . . . . . . . . . . . . . . . . . . - ~ C O
Schematic diagram of facilitated CO2 transport
Facilitated transport of C02: J.J. Pellegrino et al. PFSA films to study CO2 and H2S transport (see References 1 and 2, respectively) both as single components and as parts of mixtures containing non-carrier-reactive gases. In general, high selectivity for the acid gases existed and good agreement was obtained between the experimental results and theoretical predictions based on the model of Noble et al. 9. T h e main deficit of these m e m b r a n e s is their low overall permeance relative to some commercial gas separation m e m b r a n e devices. The facilitated transport PFSA membranes have an order of magnitude lower permeance primarily due to their thickness. The work has continued by studying acid gas separations using experimental PFSA m e m b r a n e s with a nominal 30 tam thickness s. Higher permeance was obtained with these membranes but it is still several-fold lower than commercially desirable. In recent work on PFSA membranes Heaney and Giugla (° reported increased ionic conductances through membranes as a function of the temperature of a hot solvent treatment. They developed a polymer processing scheme which increases the water weight fraction to 50% while maintaining the original mechanical properties of the polymer. They found that the ionic conductivity was a monotonic function of the treatment temperature (and the associated water content). An order of magnitude increase in conductivity was observed relative to the control. The increased flux characteristics were attributed to changes in the polymer morphology which allowed greater water content. This suggested to us that, at a minimum, the increased water content would increase the volume fraction of the ionic network and, therefore, that this treatment could significantly increase the membrane's gas permeability. The procedure was adapted and a group of 200 lam PFSA membranes were treated at different temperatures. There are a n u m b e r of treatment variables which can be modified but our initial screening study was limited to varying temperature. The objective of this study was to perform a screening test on the effects of the polymer morphology changes on the facilitated transport of acid gases. The modified membranes were tested for CO2 transport in both the diffusion (no carder present) and facilitated flux mode at three CO2 feed concentrations. A few samples were also tested with mixtures containing non-carder-reactive gases and with H2S. The details of the morphological changes in the polymer and the mechanisms of enhanced mass transfer observed in these experiments will be reported in a subsequent paper.
Membrane preparation Commercial poly-perfluorosulphonic acid ion-exchange membranes were used in this study. As obtained from the manufacturer the m e m b r a n e is nominally 200 jam thick; is in the H ÷ form and is at a low hydration level. The polymer has an 1100 equivalent weight (i.e. the weight of polymer, in grams, that will contain one equivalent of charged sites). These membranes are referred to as N117. Experimental m e m b r a n e s with a nominal 30 pm thickness and 1100 equivalent weight are referred to as NEIll. The N117 membranes were dried in a vacuum oven for 12 h at 343 K and 19 kPa absolute pressure. Their weight and thickness was then measured under anhydrous conditions. The samples were then put in 0.2 M N a O H solution and agitated for 30 min. The solution was
discarded and the step was repeated with a fresh solution. This solution was discarded, a fresh solution was added, and the m e m b r a n e sat in this solution for 48 h at 323 K with agitation. The ratio of Na ÷ in each solution to ionexchange sites was approximately 20 : 1. After this Na ÷ ion-exchange the membranes were again oven dried at 343 K and 19 kPa for more than 12 h. After drying, their weight and thickness was again recorded. After this step the control sample was removed and put into 0.2 M N a O H again. The other samples were refluxed in methanol for 24 h to further dehydrate them. Each m e m b r a n e was then placed in a glycerine bath and heated to a set temperature with mild agitation. When the bath reached the prescribed temperature it was removed from the heat source and allowed to slowly cool to room temperature. After cooling the samples were washed twice, for 30 rain each, with 0.2 M N a O H and then put in 0.2 M N a O H and allowed to agitate for more than 12 h (overnight). These samples were then weighed and their thickness was measured under humid conditions. The diffusion gas flux measurements were performed on the membranes in this form. After diffusive flux experiments the membranes were put into facilitated transport form using ethylenediamine (EDA) as the carrier. Monopositive EDA was created by adding one equivalent of HCI to the EDA solution. A 1.0 M EDA ÷ solution was used and the ratio o f E D A ÷ to ionexchange sites was approximately 100 : 1. The m e m b r a n e was immersed in the solution and agitated for more than 12 h. The facilitated flux experiments were performed on the membranes in this form.
Experimental description The apparatus and procedure used to measure the flux through the m e m b r a n e have been described in detail elsewhere" 1~. Mixtures of CO2 in He were flowed past the feed side of a fiat sheet membrane. A sweep gas of pure H E was flowed past the opposite side. The flow rates were measured with mass flow meters calibrated for the specific gases. The m e m b r a n e area was 44.9 cm 2. Both the feed and the sweep inlet streams were humidified by bubbling through deionized water before contacting the membrane. The outlet streams were chilled to 213 K, to remove moisture before they entered the gas chromatograph (GC) system. A thermal conductivity detector was used to measure both the feed and sweep side compositions. The G C was calibrated daily. The flux calculations were based on the measured flow rates and composition of the sweep gas, and the m e m b r a n e area. Periodic mass balances were performed using the feed gas composition exiting the membrane cell. All experiments reported here were done at ambient conditions of approximately 298 K and 84 kPa.
Error analysis The mass flow meters were accurate to 1.7 x 10-2 cm3s, -I at 101.3 kPa and 273 K (1 seem), which represents a 2% error at our experimental flow rates. The G C was calibrated before each day's experiments and the composition of the calibration gas was reproducible within 0.5%. Independent material balances, based on gas mixtures set by flow meter control and subsequent G C analysis, close within 5%. This represents a possible
Gas Separation 8 Purification 1988 Vol 2 September
Facilitated transport of C02: J.J. Pellegrino e t
source of error in our assignment of the driving force, for a reported flux, with a bias to reporting too high a driving force. Only those experiments where the flux reached a n d maintained a steady state for at least 1 h before the experiment was terminated are reported. The 95% confidence intervals, for the flux measurements reported in this Paper were a function of the actual flux and range from 0.1 to 8 (10 -1° mol (cm 2 s)-l). Results
Table 1 presents a listing of the membranes' thickness and water content obtained from the treatment. The water content is defined as: [(Na form wet weight - H form dry weight)/(H form dry weight)]. The thickness was measured under water. The pretreatment greatly increased the amount of solvent in the membrane. The membrane is unconstrained during the treatment and the dimensional change, quantitatively reported in the thickness, was also observed in the length and width. Table 2 presents a tabulation of the flux results. For testing purposes we used CO2 as the permeating gas and ran Na ÷ and E D A ÷ forms of the membranes at feed mole percentages of 5, 50 and 100% CO2. For feed percentages less than 100% the diluent gas was He. All the reported fluxes were the result of a single experiment and, therefore, some deviation from monotonic behaviour with respect to treatment temperature is expected. Data for untreated NE111 and Nl17 membranes were listed in both of these tables for comparison. Several points are apparent from these permeation results: •
there seems to be permeability of the those treated at 473 observed for both
a major difference in the CO2 m e m b r a n e treated at 498 K versus K and below. This difference was Na + and E D A ÷ forms of the
Table1 Physical measuremems of gl~erine/heat pretre~ed memb~nes Membrane type
Treatment temperature (K)
Water content Na form (%)
NEll1 Nl17 Nl17 Nl17 Nl17 Nl17 N117 Nl17
untreated untmated 373 398 423 448 473 498
31.0 200.7 225.6 228.1 236.2 238.3 257.8 294.1
25.0 15.0 24.9 26.5 28.7 32.1 4t.4 55.6
Membrane Na form feed type/temperature (mol (cm2 s)-1 x lOl°)
In Table3 the CO2 diffusion coefficient in the Na + form of the m e m b r a n e and the experimental facilitation factors are listed. The diffusion coefficient is based on a linear fit of the Na + form flux versus feed fraction. This parameter normalizes the membrane's transport characteristics with respect to thickness and water content. Equation (1) is the form of the diffusive flux relation and the values of the physical constants used
J. = ( D . P . & Y,o)I(T.I) where: Jn = Da = P, = Y~0 = S, =
Na+ form diffusive flux (mol (cm 2 s)-l); solute diffusion coefficient in Na + form (cm 2 s-t); void volume (set equal to H20 fraction); feed fraction of solute; solute concentration in membrane at Y~0 = 1.0, 2.64 x t0 -5 (mol cm-3); Tn = tortuosity (set equal to 1); and l = m e m b r a n e thickness (cm).
The facilitation factor is defined as the ratio of the flux through the E D A + form to the flux through the Na + form at an equivalent driving force. The values listed here are from experiments performed at equal feed fractions. This will introduce some ambiguity since the mean driving force across the whole transport cell varies up to 8% in some experiments due to the high fluxes. The observed increase in the facilitation factor as the driving force (feed fraction) decreases is normal for facilitated transport. The diffusion coefficient data seem to suggest that there are at least two separate types of modification made to the membrane. One that occurs at heating below 473 K and one that occurs above that temperature. The CO2 diffusion coefficient in the Na + form of the membrane is three to four times higher than observed in untreated membranes.
Table 3 CO2 diffusion coefficient (Na + form) and EDA form facilitation factors
CO2 flux measurements
NE111/untreated N117/untreated N117/373 K N117/398 K Nl17/423 K N117/448 K N117/473 K Nl17/498 K
membranes. This result suggests that there is a major morphology change in the m e m b r a n e treated in this way. The expected glass transition temperature of the fluoropolymer phase is expected at 479 K as reported by Kyu and Eisenbergl2; O there is a large change in the CO2 flux for treated versus untreated membranes at all treatment temperatures; and O the increase in flux for the EDA + form is much greater than that of the Na ÷ form. Apparently the morphology changes within the m e m b r a n e removed a significant constraint of the facilitated transport mechanism.
EDA form feed (mol (cm2 s)-1 x lO l°)
8.8 1.2 2.9 3.1 4.0 7.0 4.0 9.5
71.5 11.0 22.3 22.7 23.8 29.3 31.3 49.9
120.0 19.5 45.3 46.2 47.9 58.5 57.4 99.2
31.0 91.0 135.0 8.3 21.6 28.2 36.0 65.8 76.9 50.5 72.4 82,7 42.1 71.0 89.7 48.3 77.7 96.3 54.7 99,0 126,0 99.8 318.0 463.0
Na form Facilitation factor diffusion feed coefficient (cm 2 S-1 X 10 s) 5% 50% 100%
NE11 t/untreated N 117/untreated N117/373 K N 117/398 K N117/423 K N11 7 / 4 4 8 K Nl17/473 K N117/498 K
0,46 0.49 1.55 1,5 1.49 1.65 1,38 t ,99
Gas S e p a r a t i o n f t P u r i f i c a t i o n 1 9 8 8 V o l 2 S e p t e m b e r
3.52 6.9 12.4 16,3 10.5 6.9 13.7 10.5
1.27 1.96 2.95 3.2 2.98 4.03 3.16 6.37
1.13 1.45 1.7 1.8 1.87 1.65 2,2 4.67
Facilitated transport of C02: J.J. Pellegrino The experimental facilitation f a c t o r s at 5% feed are subject to the maximum error since they are derived by dividing by a smaller number. The trend seems to indicate that the CO2 transport is more facilitated in the treated membranes. This is probably due to greater mobility of the complex. Figure 2 is a plot of the membrane permeability as a function of the treatment temperature and the experimental conditions. The permeability has units of(cm 3cm) (cm 2 s kPa) -~ and therefore normalizes the membranes' permeance (P/l) for the increase in thickness that occurs during the treatment. It is a measure of the material property of the membrane to transport a solute excluding the consideration of how thin we are able to make the membrane. The trends suggest a gradual increase in permeability as the treatment temperature is increased, with a sharp increase above 473 K. As predicted by theory, the permeability is a function of the driving force in EDA ÷ facilitated transport. The fact that the data for the 5% CO2 in the Na ÷ form does not lie on top of the other Na ÷ data is probably due to experimental error. Figure 3 presents the permeance (P/l) through these membranes. The permeance is a measure of how good the membrane, as made, can transport a solute. A commercial gas separation membrane would have a permeance around 9 x 10-5 cm 3 (STP) (cm 2 s kPa) -I for CO2.
. ~ .¢n
E ~ i,.P~
Table 4 Flux results for mixed gas feeds with glycerine/heat ~retreated membranes Membrane form/temperature
Flux (mol (cm 2 s) -1 X 10 - l ° )
0.5 N 1 1 7 / E D A no treatment 50% CO 50% CO 1.1 N 11 7 / E D A 4 6 3 K N 117/EDA 463 K 5 0 / 5 0 , CO/CO2 CO2 = 97.2 CO = 0.9 N117/Na 423 K N 117/EDA 448 K
N 117/EDA 473 K
5 0 / 5 0 , CO/CO 2 CO2 = 40.7 CO = 0.8
CO2 = 41.9
CO, CO2, H2S
H2S = CO =
CO2= 81.9 H2S = < 0.01 CO = 1.46
C O , C O 2, H 2 S
Binary, systems It is important to make the membrane more permeable while maintaining the selectivity inherent in our facilitated transport system. In order to check this some tests were performed using CO as the non-reacting gas. These results are summarized in Table 4. The key observations from these experiments are: • the CO flux is governed by its solubility in H20 and is not significantly affected by the EDA carder; • the treatment maintains the membrane's integrity since the CO flux increase (versus the control) in the swollen membrane is proportional to the increase in water content; • CO2 facilitation is basically unaffected by the presence of CO; and • CO2 complexation is favoured relative to the H2S, as previously reported 2.
e t al.
Figure 2 CO2 Permeability versus glycerine treatment temperature for Na + (open symbols) and EDA + (solid symbols) forms of N 1 1 7 membranes at feed percents of: (3, e , 5%; El, e , 50%; A, & , 100% tO
APIog mean' k P a
Figure 3 COz permeance ( P / l ) versus the log mean pressure driving force for Na + and EDA + forms of N 1 1 7 membranes at glycerine treatment temperatures from 3 7 3 to 4 9 8 K
The glycerine/heat pretreatment of polyperfluorosulphonic acid membranes seems to be a very effective way of increasing their permeability during facilitated transport of CO~. There is an apparent major change in morphology which most dramatically affects the flux results for treatment above 473 K. This may be associated with an upper glass transition temperature of the polymer. The permeance (P/l) obtained for membranes prepared from commercially available PFSA are several fold higher than previously obtained but still lower than in a commercial gas permeator. Further work with thinner PFSA membranes should yield yet higher permeance. We are currently working to quantify some of the polymer structural changes using small angle X-ray scattering. This technique is useful for determining the size of the ionic clusters where most of the water presumably accumulates. We are seeking a correlation between the cluster size and/or spacing and the complex's apparent mobility. Acknowledgements
The authors acknowledge the support of the Department of Energy Morgantown Energy Technology Center for this work under DoE Interagency Agreement No. DEAI21-86MC23120, with special thanks to Lisa Jarr. We also acknowledge the support of Donald Brandt of
Gas Separation B Purification 1 9 8 8 Vol 2 September
Facilitated transport of COz: J.J. Pellegrino et al. E1 d u P o n t de N e m o u r s & Co., Inc., for supplying both c o m m e r c i a l a n d d e v e l o p m e n t a l forms of N a t i o n p e r f l u o r o s u l p h o n i c acid i o n o m e r films a n d m a n y helpful discussions.
Way, J.D., Noble, R.D., Reed, D.L., Ginley, G.M. and Jarr, L.A. Facilitated transport of CO 2 in ion exchange membranes AIChE J (1987) 33(3) 480-487 2 Way, J.D. and Noble, R.D. Hydrogen sulfide facilitated transport in perfluorosulphonic acid membranes in Liquid membranes: theory and applications ACS Symposium Series No. 347 (Eds. R.D. Noble and J.D. Way) (1987) 123-137 3 Noble, R.D., Pellegrino, J.J., Grosgogeat, E., Sperry, D. and Way, J.D. CO2 separation using facilitated transport ion exchange membranes Sep Sci Technol 0988) 23 12 4 Ward, W.J. III Analytical and experimental studies of facilitated transport AIChE J (1970) 16 405-410 5 Smith, D.R., Lander, RJ. and Quinn, J.A. Carrier-Mediated Transport in Synthetic Membranes in Recent Developments in Separation Science Vol 3 (Ed. N.N. Li) CRC Press, Cleveland, USA (1977) 225-242
Way, J.D. and Noble, R.D. Liquid membrane technology: an overview in Liquid membranes: theory and applications ACS Symposium Series No. 347 (Eds. R.D. Noble and J.D. Way) (1987) 1-27 7 LeBlane, O.H., Ward, WJ. III, Matson, S.L. and Kimura, S.G. Facilitatcd transport in ion exchangc membranes JMem Sci (1980) 6 339-343 8 Rieke, P.C. and Vanderborgh, N.E. Temperature dependence of water content and proton conductivity in polyperfluorosulphonic acid membranes J Mere Sci (1987) 32 313-328 9 Noble, R.D., Way, J.D. and Powers, LA. Effect of external mass transfer resistance on facilitated transport I & EC Fund (1986) 25 450-452 10 Heaney, M.D. and Glugla, P.G. Increased membrane ionic conductivity resulting from morphological modification of perfluorinated ionomers (in preparation) 11 Bateman, B.R., Way, J.D. and Larson, K.M. An apparatus for the measurement of gas fluxes through immobilized liquid membranes Sep Sci Technol (1984) 19 21-32 12 Kyu, T. and Eisenberg, A. Mechanical relaxations in perfluorosulphonate ionomer membranes in Perfluorinated ionomer membranes ACS Symposium Series No. 180 (Eds, Eisenberg, A. and Yeager, H.L.) (1982) 311
Gas Separation Et Purification 1988 Vol 2 September