Carbon Dioxide Separation with Novel Solvents as Liquid Membranes

Department of Chemical Engineering, New Jersey Institute of Technology, Newark, New Jersey 07102. Glycerol carbonate is studied as a new physical solv...
0 downloads 0 Views 102KB Size
Download PDF
Ind. Eng. Chem. Res. 2002, 41, 2287-2295

2287

Carbon Dioxide Separation with Novel Solvents as Liquid Membranes A. Sarma Kovvali† and Kamalesh K. Sirkar* Department of Chemical Engineering, New Jersey Institute of Technology, Newark, New Jersey 07102

Glycerol carbonate is studied as a new physical solvent for carbon dioxide separation from CO2/ N2 mixtures. CO2-selective behavior of glycerol carbonate is explored in an immobilized liquid membrane (ILM) configuration. Pure glycerol carbonate ILMs retained their CO2/N2 selectivity around 80-130 over a large range of CO2 partial pressures. Its CO2 selectivity was not affected by the absence of humidity in feed and sweep streams. When exposed to dry feed gas conditions, the CO2 permeability through pure glycerol carbonate ILMs was about 100 barrer, which increased to about 350 barrer in the presence of humidified feed streams. Addition of small amounts of facilitating carriers such as poly(amidoamine) dendrimer (generation zero) and sodium glycinate appears to significantly help CO2 facilitation at low CO2 partial pressures. However, at high CO2 feed partial pressures, there was a loss of selectivity with the addition of the carriers. The potential for glycerol carbonate based ILMs in CO2 separation from different CO2-containing gas mixtures is discussed and compared with existing solvent/membrane configurations. 1. Introduction Polymeric membrane-based gas separation has achieved major technical and commercial breakthroughs in removing CO2 from industrial gas streams.1-3 Nonporous and glassy polymer membranes employed in such commercialized processes possess, however, rather low CO2/N2, CO2/O2, or CO2/CH4 selectivities, around 15-30. The CO2 flux through such membranes, JCO2, is a product of the CO2 permeance (QCO2/tm) and the difference between the CO2 partial pressure on the feed side (pCO2,f) and that on the permeate side (pCO2,p):

JCO2 ) (QCO2/tm)(pCO2,f - pCO2,p)

(1)

Here, QCO2 is the permeability of CO2 through the membrane of thickness tm. For a mixture of CO2 and N2, the ideal selectivity of CO2 over N2 for the system is defined as

R*CO2/N2 ) QCO2/QN2 )

|

yCO2(1 - xCO2) xCO2(1 - yCO2)

(2)

pCO

2,pf0

where yCO2 is the mole fraction of CO2 in the gas appearing in the permeate/product side and xCO2 is the mole fraction of CO2 in the feed gas to be treated and exposed to the membrane. Consider, for example, removal of CO2 from wet flue gases for sequestration, for which one desires a reasonably pure CO2 permeate stream. Assuming that the wet feed gas will contain, say, 14% CO2 (i.e., xCO2 ) 0.14), a polymeric membrane having an ideal selectivity R*CO2/N2 of, say, 20 will, on the basis of relation (2), produce a permeate gas having a CO2 mole fraction of only 0.77. On the other hand, a membrane of selectivity R*CO2/N2 ) 100 will produce a permeate mole fraction, yCO2, of 0.94, a stream substan* To whom correspondence should be addressed. Tel.: (973) 596-8447. Fax: (973) 642-4854. E-mail: [email protected]. † Present address: Baker Hughes, Houston, TX.

tially purer in CO2 and useful for sequestration. In addition, from eq 1, the higher is the value of QCO2/tm for a given pCO2,f - pCO2,p, the higher is the value of JCO2 and, correspondingly, the lower is the membrane area required to recover a given fraction of CO2 from the feed gas. This reduces the capital cost needed for purification. It has been reported that the economic separation of CO2 from flue gases is achieved via membranes if RCO2/N2 exceeds 70 with a minimum QCO2 of 100 barrer.4 The CO2 separation problem becomes particularly acute for nonporous polymeric membranes at lower CO2 feed partial pressures; the CO2 flux through the membrane becomes quite low. If extraction of CO2 from the ambient atmosphere is to be carried out, then xCO2 ) 0.000 35 (corresponding to 350 ppm). The membrane has to have a very high selectivity for CO2 to produce a permeate highly enriched in CO2. Another application where this situation arises is in highly selective CO2 removal from enclosed breathing atmospheres as in space walk and space cabin atmospheres. In these applications, one would like to have as little permeation of O2 as possible and as high a CO2 removal as possible. Desirable CO2/O2 selectivities are in the range of 10002000. Facilitated transport membranes (FTMs) are likely to be the only viable approach available for this goal. FTMs containing species reversibly reactive with CO2 have been studied in a variety of forms: fixed carrier membranes,5-7 contained liquid membranes,8 polyelectrolyte membranes,9-11 solvent-swollen polymer films,12 and immobilized liquid membranes (ILMs).13-18 Of these, ILMs can potentially yield high CO2 permeances and CO2/N2 selectivities at low CO2 concentrations. The problem with most past studies of the ILMs was that they were based on water as the solvent. So, unless the feed and sweep streams are completely humidified, the membrane is unstable. We have recently adopted the approach of using the nonvolatile solvent (glycerol) and nonvolatile carriers

10.1021/ie010757e CCC: $22.00 © 2002 American Chemical Society Published on Web 04/04/2002

2288

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002

Table 1. Important Characteristics of Glycerol Carbonatea

a Aldrich Catalog. b Measured in a Kruss K-8 interfacial tensiometer.22

(sodium carbonate and sodium glycinate) for improving the stability of the ILMs.16-18 Glycerol, when used as the solvent, improved the stability of ILMs tremendously for different relative humidity (RH) conditions and carbon dioxide partial pressures. Glycerol has high viscosity and vanishingly low vapor pressure and is highly hygroscopic. However, glycerol is not highly selective for CO2 over N2, with the selectivity being only 1.5. Recently, a novel solvent/carrier with excellent facilitation characteristics for CO2 was also studied in a stable ILM configuration using dendrimers.19,20 As part of the continuing efforts to identify such solvents, new solvents/carriers are being explored for carbon dioxide separation from gas mixtures. One such solvent is glycerol carbonate, which is likely to have a high solubility for CO2 unlike glycerol. There are several industrial processes for separating CO2 based on physical/chemical solvents.21 In order for the solvent-based process to be practical, the solvents should have higher solubility for CO2 than in water and must have extremely low vapor pressure, low viscosity, and low or moderate hygroscopicity. Some of the solvents used are methanol at low temperatures (Rectisol process), propylene carbonate (Fluor process), N-methyl2-pyrrolidone (Purisol process), dimethyl ether of poly(ethylene glycol) (Selexol process), tributyl phosphate (Estasolvan process), and a mixture of diisopropanolamine, sulfolane, and water (Sulfinol process). Most of these solvents have higher solubility for H2S than for CO2. Another common feature of these processes is that they are used in contactor-stripper mode, requiring two separate steps in CO2 separation. None of these systems appear to have used glycerol carbonate for absorption or in a liquid membrane configuration. Glycerol carbonate is explored in the present study as a physical solvent in an ILM configuration. Glycerol carbonate (also referred to as glycerine carbonate) is an experimental compound. Some of the important characteristics of glycerol carbonate are provided in Table 1. It should be noted that most of the alkaline carbonates (glycerol carbonate is one of them) are being developed as alternative “safe and environmentally friendly” solvents to conventional aromatic

solvents. Glycerol carbonate satisfies most of the requirements for an ideal solvent for carbon dioxide separation. It is nontoxic and nonvolatile (vapor pressure is reported to be extremely low). It will be of interest to study if glycerol carbonate can be an effective physical solvent for the separation of carbon dioxide or an alternative solvent to either water or glycerol for dissolving various CO2-facilitating carriers such as sodium glycinate, sodium carbonate, dendrimers, etc. In the present study, glycerol carbonate in its pure form and in the presence of a few carriers has been studied in an ILM configuration for the separation of CO2/N2 mixtures. The main questions the present exploratory work attempts to address are as follows: (1) Can glycerol carbonate be used effectively as a solvent in its pure form for selective separation of carbon dioxide from gas mixtures? (2) Does the carbon dioxide concentration in the feed gas mixture (and hence the carbon dioxide partial pressure difference across the membrane) have any effect on the carbon dioxide permeation across glycerol carbonate based liquid membranes? (3) Is the observed selectivity for glycerol carbonate dependent on the presence of moisture in the feed stream? (4) Does the thickness of the substrate (and that of the ILM) have any effect on the selectivity for and permeability of carbon dioxide? (5) What are the effects of the addition of carriers, such as sodium glycinate and dendrimers, on the separation behavior of the ILMs? 2. Experimental Details Glycerol carbonate as obtained (Huntsman Corp., Houston, TX) was used in all experiments without any further purification. Solutions of 0.06 and 0.127 M of poly(amidoamine) (PAMAM) dendrimer of generation zero (Aldrich, St. Louis, MO) in glycerol carbonate solutions were prepared by dissolving pure dendrimer in glycerol carbonate. The preparation of liquid membranes having higher concentrations of dendrimer in glycerol carbonate was not attempted. Liquid membranes containing 0.1 M sodium glycinate in glycerol carbonate were prepared by dissolving the required amount of sodium glycinate (Fisher Scientific, Springfield, NJ) in glycerol carbonate and stirring it overnight. Higher concentrations of sodium glycinate (0.5 M) tended to precipitate in the solution. It was difficult to dissolve sodium carbonate in glycerol carbonate in useful concentrations. The method of preparation of ILMs was similar to those for other ILMs studied earlier16-20 and posed no significant problems. Hydrophilized poly(vinylidene fluoride) (PVDF) of 100 µm thickness (0.1 µm, Millipore, Bedford, MA) and hydrophilized Celgard 2500 of 25 µm thickness (Celgard LLC, Charlotte, NC) films were employed. The wetting of both films was almost instantaneous. Excess solution from the surfaces of the substrate was removed with kimwipes. All ILMs formed in this work filled the void region of the substrates completely. Their thicknesses were assumed to be equal to that of the substrate. Two 100 µm thick hydrophilized PVDF films having the ILMs were put one on top of the other to prepare a 200 µm ILM and study the effect of ILM thickness. The experimental setup used in CO2 separation studies with a dendrimer liquid membrane19 was also

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002 2289

used for the present work. All experiments employed dry helium as the sweep gas, while the feed gas was introduced after complete humidification. A number of experiments were done with dry feed gas to study the effect of the feed-side RH on the overall performance of glycerol carbonate based liquid membranes. The CO2 concentration in the feed gas was varied from 0.5% to 25%, with the balance being nitrogen. The transmembrane pressure difference was maintained around 6 psi in all experiments. The experiments were conducted at a room temperature of 23 ( 2 °C. The calculation procedures for determining the permeance, the permeability of each gas species (CO2 and N2), and the separation factor (RCO2/N2) of the glycerol carbonate based ILMs are the same as those used for dendrimer liquid membranes. 3. Calculation Procedure for Gas Permeances and the Separation Factor Based on the calibration in the gas chromatograph and the peak areas obtained, the mole fractions of the species in the sweep gas stream can be obtained. When multiplied by the sweep-side gas volumetric flow rate, one obtains the volumetric permeation rate of the species through the membrane. The effective permeance of a species through the membrane was calculated by

() Qi tm

)

eff

Vi At∆pi

(3)

where Vi is the volumetric permeation rate of species i, ∆pi is the partial pressure difference of species i across the membrane, and At is the total membrane area. The true permeance of the species (Qi/tm)true is obtained from

() () Qi tm

)

true

Qi tm

τm eff m

(4)

The permeability Qi of a particular gas species through the ILM is obtained by multiplying its true permeance through the ILM and the membrane thickness:

Qi )

() Qi tm

tm

(5)

true

The separation factor of species i with respect to species j, Ri/j, is defined by

Ri/j )

Qi (Qi/tm)eff (Qi/tm)true ) ) Qj (Qj /tm)eff (Qj /tm)true

(6)

Gas permeances are reported in units of cm3(STP)/cm2‚ s‚cmHg. Wherever convenient, gas permeances are reported in the units of GPU, which is equivalent to 10-6 cm3(STP)/cm2‚s‚cmHg. Gas permeabilities are reported in units of barrer, which is 10-10 cm3(STP)/cm2‚s‚cmHg. The toruosity (τm) and porosity (m) of the PVDF substrate (2.58 and 0.7) and of the Celgard 2500 substrate (2.54 and 0.45) were obtained from Chen et al.16 and Prasad et al.,23 respectively. 4. Results and Discussion 4.1. Effect of the Carbon Dioxide Partial Pressure Difference on the Performance of Pure Glycerol Carbonate ILM. Figure 1 presents the variation

Figure 1. Performance of a pure glycerol carbonate ILM in a PVDF substrate.

of the permeability of carbon dioxide and the selectivity of the pure glycerol carbonate ILM for carbon dioxide over nitrogen against the carbon dioxide partial pressure difference across the liquid membrane. The substrate used was a hydrophilic PVDF substrate of 100 µm thickness having an average pore size of 0.1 µm. The carbon dioxide partial pressure difference across the membrane was varied from 0.56 to 28.1 cmHg to observe any dependence of the ILM permeances/permeabilities on ∆pCO2. Under humidified feed conditions, the carbon dioxide permeability ranged between 380 and 280 barrer for the range of CO2 partial pressure differences studied with no clear signs of any significant facilitation at lower ∆pCO2’s. The selectivity (eq 6) of the ILM (RCO2/N2) ranged between 90 and 110. RCO2/N2 and QCO2 of glycerol carbonate ILM do not appear to be affected much by the variation of ∆pCO2 across the membrane, which is significant. Pure glycerol carbonate liquid membranes were also studied in a 25 µm thick hydrophilized Celgard 2500 substrate. The behavior is similar to that observed in the PVDF-based ILMs and shown in Figure 2. For the range of carbon dioxide partial pressure differences studied between 0.55 and 28.1 cmHg, there is no significant variation in carbon dioxide permeability. QCO2 varied between 350 and 260 barrer, with the range of selectivity being 50 and 60. The somewhat lower selectivities observed in the case of hydrophilized Celgard substrates compared to the PVDF substrates are expected based on our earlier studies. As we have identified in our previous studies of dendrimer liquid membranes20 and glycerol-based ILMs,17 the hydrophilization of the particular Celgard substrate does not appear to be uniform. Consequently, there is a significant increase in N2 permeability through the nonwetted corridors of the membrane. 4.2. Effect of the Feed Gas RH. The pure glycerol carbonate ILMs were also exposed to dry feed gas to study the effect of moisture on the ILM performance. For the PVDF-based ILMs under dry feed gas conditions, the QCO2’s were reduced to about one-third of the humidified QCO2 values. For the same ∆pCO2 range, the QCO2 values under dry feed gas conditions ranged from 80 to 110 barrer, with selectivities ranging from 90 to

2290

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002

Table 2. Performance of Pure Glycerol Carbonate ILMs in PVDF and Hydrophilized Celgard Substrates ∆pCO2, cmHg

feed condition

(QCO2/tm)eff, cm3/cm2‚s‚cmHg

(QN2/tm)eff, cm3/cm2‚s‚cmHg

QCO2, barrer

QN2, barrer

RCO2/N2

380 280 320 110 80 100

3.6 2.8 3.7 0.86 0.88 1.1

110 100 90 130 90 90

350 270 260 160

6.3 5.9 4.4 3.5

60 50 60 50

Two PVDF Substrates Sandwiched (Thickness 200 µm) 3.77 × 10-7 4.32 × 10-9 280 3.00 × 10-7 4.23 × 10-9 220

3.2 3.1

90 70

2.14 5.34 28.1 2.14 5.34 28.1

humid humid humid dry dry dry

0.1 µm PVDF Substrate (Thickness 100 µm) 1.02 × 10-6 9.80 × 10-9 7.40 × 10-7 7.70 × 10-9 8.73 × 10-7 1.01 × 10-8 2.95 × 10-7 2.33 × 10-9 2.10 × 10-7 2.40 × 10-9 2.63 × 10-7 2.99 × 10-9

0.55 2.32 28.10 28.1

humid humid humid dry

Hydrophilized Celgard 2500 (Thickness 25 µm) 2.46 × 10-6 4.47 × 10-8 1.88 × 10-6 4.20 × 10-8 1.86 × 10-6 3.08 × 10-8 1.15 × 10-6 2.47 × 10-8

2.14 26.43

humid humid

Figure 2. Performance of a pure glycerol carbonate ILM in a hydrophilized Celgard 2500 substrate.

130. The data are shown in Figure 1. The CO2 permeabilities of dry glycerol carbonate are almost 2 orders of magnitude larger than those of glycerol-based ILMs. The retention of selectivity for CO2 at the level of 90130 even under dry feed conditions is highly significant. The ILMs were found to be completely stable in the absence of water vapor in the feed gas mixture. Glycerol carbonate is partially miscible with water. In the presence of water in the feed gas, because of the sorption of water, the viscosity of glycerol carbonate ILM would be reduced, increasing the diffusion coefficients of both nitrogen and carbon dioxide in the liquid. This will increase the permeabilities of both gas species through the physical solvent-based ILM. When the Celgard-based pure glycerol carbonate ILM was exposed to dry feed gas, the ILM was not stable over a long time. The ILM was exposed to dry feed gas conditions for one carbon dioxide partial pressure and is shown in Figure 2. Reintroduction of humidity on the feed side did not revive the ILM performance, indicating the possible loss of hydrophilicity of at least part of the substrate Celgard membrane. The behaviors of pure glycerol carbonate liquid membranes under both humidified and dry feed gas conditions, particularly with the PVDF substrates, indicate that the transports of both carbon dioxide and nitrogen are not substantially affected by the concentration of carbon dioxide in the feed gas mixture (and therefore

∆pCO2). Any trend observed in the variation of RCO2/N2 with ∆pCO2 is mainly the result of the variation of QN2 from experiment to experiment and membrane to membrane rather than any particular trend in the variation of QCO2. The corresponding effective permeances of carbon dioxide and nitrogen for these conditions are provided in Table 2. This table indicates that glycerol carbonate in its pure form as the liquid membrane does not react with carbon dioxide reversibly as any facilitated transport carrier does and behaves like a physical solvent for carbon dioxide. The observed selectivity of glycerol carbonate for carbon dioxide over nitrogen could only be achieved because of considerably higher solubility of carbon dioxide compared to that of N2 in glycerol carbonate. This is a significant observation. The diffusion coefficients of CO2 and N2 are likely to be similar because of their molecular sizes. We can now compare this behavior with those of other solvents conventionally used. Water has a RCO2/N2 of about 40,14 while pure glycerol without any sorbed moisture has a RCO2/N2 of only about 1.5.16 Values of solubility of CO2 and N2 in either glycerol or glycerol carbonate are not available, making it difficult to compare the corresponding solubility behaviors. It can be inferred from Figure 1 that glycerol carbonate has either a much higher solubility for carbon dioxide or a much reduced solubility for nitrogen, resulting in higher selectivities; further, such a behavior is not affected by the presence of water. The presence of water vapor in the feed gas (and in the liquid membrane) has the effect of increasing the permeabilities of both CO2 and N2 proportionally. Compared to this behavior, the presence of water in a glycerol liquid membrane dominates its behavior by increasing QCO2 drastically without a corresponding increase in nitrogen permeation, resulting in an increase in RCO2/N2 from 2 to 40 under humidified feed conditions.16 It appears that glycerol carbonate is a highly CO2-selective physical solvent whose CO2-N2 selectivity appears to be independent of CO2 concentrations and feed gas RHs. 4.3. Effect of the Liquid Membrane Thickness. An important feature of a physical solvent for membranebased carbon dioxide separation would be that its performance is not affected by the carbon dioxide partial pressure and the thickness of the substrate. For a facilitated transport liquid membrane, a higher thickness of the liquid membrane generally results in a higher value of RCO2/N2 and a lower QCO2/tm. It has already been shown that the carbon dioxide partial pressure does not affect the CO2-selective behavior of

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002 2291

Figure 3. Effect of the substrate thickness on pure glycerol carbonate liquid membranes.

pure glycerol carbonate full ILMs in both hydrophilic PVDF and hydrophilized Celgard substrates. To verify that the thickness of the substrate (and that of the ILM) does not alter the performance of glycerol carbonate liquid membranes, two 100 µm PVDF films filled with pure glycerol carbonate were placed one on top of the other to provide a total thickness of 200 µm. Figure 3 compares the behavior of the three liquid membranes having thicknesses of 25 µm (Celgard), 100 µm (PVDF), and 200 µm (two PVDF membrane composite) for two different CO2 partial pressures. The permeability of carbon dioxide and RCO2/N2 are plotted against two widely different partial pressure differences of carbon dioxide for a humidified feed. The data are also presented in Table 2. There does not appear to be any strong effect of liquid membrane thickness on QCO2 for both ∆pCO2’s studied. The RCO2/N2 values for 100 and 200 µm thick liquid membranes are quite similar for both ∆pCO2’s studied. The selectivities observed in hydophilized Celgard are in reasonable agreement with those of PVDF substrates. The reasons for this behavior of Celgard substrates were discussed earlier. It appears that the CO2 permeability of glycerol carbonate membranes is essentially independent of the substrate thickness and the CO2-N2 selectivity can be retained, irrespective of the substrate thickness. Any apparent effect in terms of a small reduction in the permeability of the gases is to be sought in the way the films were prepared. The 200 µm ILM was formed by putting one 100 µm ILM over another. Any residual solvent on the surfaces will form an additional liquid film between the two polymeric supports. (It is difficult to remove liquids completely from the surfaces of highly porous membranes. Such residual liquids do not reduce the Qi values for a single ILM.) These results suggest that if a glycerol carbonate ILM can be formed with a thickness of less than 25 µm, the values of QCO2 and RCO2/N2 can be expected to be similar to those of a 200 µm thick liquid membrane. However, because permeability is the product of the true permeance and the membrane thickness, a reduction in liquid membrane thickness from 200 µm to, say, 10 µm would increase the effective permeance of carbon dioxide by up to 20 times while maintaining the same selectivity. In essence, with glycerol carbonate, it would be possible to form ultrathin liquid membranes with high permeances without compromising on its selectivity for

Figure 4. Performance of glycerol carbonate ILMs with various carriers in a PVDF substrate. Carriers: 0.06 and 0.127 M PAMAM dendrimer and 0.1 M sodium glycinate.

carbon dioxide over nitrogen. Our current laboratory investigations are directed to this goal of reducing the ILM thickness; our future communications will focus on it. 4.4. Effect of the Addition of Carriers to Glycerol Carbonate. As is seen above, pure glycerol carbonate liquid membranes provide a good framework for preparing thinner liquid membranes for providing high CO2 permeances. It would be of practical importance to see if the carbon dioxide permeances and selectivity can be enhanced further by adding facilitating carriers. A few carriers such as sodium carbonate, sodium glycinate, and dendrimers were studied earlier16,17,20 by dissolving them in glycerol. The objective was to study the effect of these carriers by dissolving them in glycerol carbonate. However, sodium carbonate could not be dissolved in glycerol carbonate in any concentration, whereas sodium glycinate was found to be somewhat soluble in glycerol carbonate. In the present work, a 0.1 M sodium glycinate in glycerol carbonate solution was immobilized in a PVDF substrate. However, 0.06 and 0.127 M solutions of PAMAM dendrimer of generation zero in glycerol carbonate could be easily prepared. Liquid membranes having higher concentrations of dendrimer in glycerol carbonate were not studied. Figure 4 and Table 3 present the behavior of such glycerol carbonate based ILMs for various carbon dioxide partial pressure differences. The effect of feed-side RH on these liquid membranes was also studied by subjecting the liquid membranes to dry feed gas conditions. The presence of small amounts of dendrimer or sodium glycinate in glycerol carbonate introduces the facilitated transport mechanism to carbon dioxide transport across the liquid membrane in the presence of feed RH. For example, the addition of 0.06 M dendrimer to glycerol carbonate increased the carbon dioxide effective permeance by a factor of about 18 times at a low ∆pCO2 of 0.52 cmHg, which is relevant to space suit applications.17 However, the average nitrogen permeance also increased marginally by a factor of 2, resulting in a RCO2/N2 of about 1000. Similarly, at a ∆pCO2 of about 5-6 cmHg, there is an increase in the effective permeance of carbon dioxide by a factor of 2 when either dendrimer

2292

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002

Table 3. Performance of Glycerol Carbonate based ILMs Containing PAMAM Dendrimer and Sodium Glycinatea ∆pCO2, cmHg

(QCO2/tm)eff, cm3/cm2‚s‚cmHg

(QN2/tm)eff, cm3/cm2‚s‚cmHg

QCO2, barrer

QN2, barrer

RCO2/N2

humid humid humid humid humid dry dry

0.06 M PAMAM Dendrimer-Glycerol Carbonate ILM 1.78 × 10-5 1.77 × 10-8 6600 7.87 × 10-6 1.74 × 10-8 2900 1.58 × 10-6 1.53 × 10-8 580 6.52 × 10-7 2.34 × 10-8 240 -7 -8 6.64 × 10 2.27 × 10 250 1.74 × 10-8 9.30 × 10-8 2.05 × 10-8 34

6.53 6.4 5.64 8.62 8.4 6.4 7.56

1000 450 100 30 30

0.52 5.35 26.79

humid humid humid

0.127 M PAMAM Dendrimer-Glycerol Carbonate ILM 7.18 × 10-6 7.15 × 10-9 2600 6.38 × 10-7 6.96 × 10-9 240 3.11 × 10-7 7.94 × 10-9 120

2.64 2.56 2.93

1000 90 40

0.52 1.64 5.35 17.78 27.43

humid humid humid humid humid

0.1 M Sodium Glycinate-Glycerol Carbonate ILM 5.30 × 10-6 1.11 × 10-8 2000 2.80 × 10-6 2.82 × 10-8 1000 1.76 × 10-6 3.08 × 10-8 650 1.27 × 10-6 3.18 × 10-8 470 6.78 × 10-7 1.73 × 10-8 250

4.1 10.38 11.3 11.7 6.4

480 100 60 40 40

0.52 0.62 6.13 26.43 30.27 0.62 30.27

a

feed condition

4.5

A 100 µm hydrophilized PVDF substrate.

or sodium glycinate is added to glycerol carbonate. Similar behavior was observed with 0.127 M dendrimer in glycerol carbonate ILMs. The addition of small amounts of dendrimer appears to significantly help the facilitation of CO2 at low concentrations by substantially increasing the CO2 permeances. However, at high carbon dioxide partial pressure differences, there appears to be a significant decrease in the carbon dioxide effective permeance for dendrimer-glycerol carbonate ILMs. Combined with somewhat higher nitrogen permeances, RCO2/N2 of such an ILM was reduced to 30 in the case of a 0.06 M dendrimer ILM and 40 in the case of a 0.127 M dendrimer ILM. This behavior was reproducible. Although a similar trend could be seen with sodium glycinate-glycerol ILM also, the glycerol-based ILMs had higher selectivities in the range of 100-150.17 Pure dendrimer liquid membranes had a selectivity of about 720 at the high CO2 partial pressures.20 On the addition of facilitating carriers, the solubility of a gas species will decrease because of the salting out effect. In Table 3, the N2 permeability and permeance decrease by a factor of 2.2-2.5 at any given ∆pCO2 when the dendrimer concentration changes from 0.06 to 0.127 M. The CO2 permeability and permeance also appear to decrease by a similar factor with an increase in the dendrimer concentration (compare ∆pCO2 ) 26.43 cmHg in Table 3 for the two dendrimer concentrations). However, the CO2 permeability decrease appears to be quite steep at higher ∆pCO2’s. It (250 barrer) has gotten reduced to the level of pure glycerol carbonate ILMs (see Table 2) at 0.06 M dendrimer. This is expected because carrier saturation occurs rapidly at this low dendrimer concentration. Salting out behavior reduces the CO2 permeability to 120 at a 0.127 M dendrimer concentration. When the dendrimer-glycerol carbonate ILM was subjected to dry feed gas conditions, the CO2 effective permeance was reduced drastically by a factor of 8 compared to the humidified feed gas condition. This observation lends credence to the following explanation. It was observed earlier20 that the dendrimer liquid requires a highly humid environment for effective facilitation. So, in the absence of feed side humidity, any reduction in CO2 effective permeance in 0.06 M dendrimer-glycerol carbonate ILM should mean a dryness-

induced reduced permeability (by a factor of 3 in Table 2), reduced solubility due to the presence of dendrimer (by a factor of, say, 2.5), and the absence of any facilitation. However, the increase in nitrogen permeance in the presence of these carriers (compared to the values in Table 2 for dry gas) cannot be explained. The transport behavior of nitrogen is reproducible for both of these ILMs and seems constant for different experiments whether the feed is wet or dry. It can be postulated that the presence of these carriers with their limited solubilities in glycerol carbonate is producing a heterogeneous phase mixture at a microscopic level, resulting in potential interphase corridors of increased nitrogen transport. In the presence of moisture, CO2 permeance is reduced and N2 permeance is increased to yield an RCO2-N2 of 30. In a dry atmosphere, CO2 permeance is reduced even further to yield even lower selectivity. At this point, a more rigorous explanation of the observed behavior is not possible without studying these ILMs in greater detail in terms of their solution behavior and gas solubility behavior. 4.5. Relevance of Glycerol Carbonate for CO2 Separation. Glycerol carbonate, in its pure form, appears to be a suitable physical solvent for selective CO2 separation. Its selectivity for CO2 over N2 is far superior to that of glycerol and almost 2 times that of water. The significance of glycerol carbonate is that its selectivity is not dependent on the presence of water vapor in either the feed side or the sweep side. From the experimental data collected during the present work, glycerol carbonate liquid membranes were stable for more than a week of continuous operation. Further studies on the long-term stability of glycerol carbonate ILMs are necessary to confirm the promising trend observed in the present work. However, the vapor pressure is so low that we expect a highly stable behavior. Table 4 compares the performance of pure glycerol carbonate ILMs in a 100 µm thick PVDF substrate with various other solvents in membrane configurations studied in the literature. For comparison, the performances of some polymer blend membranes and polyimide membranes are also presented. Among various physical solvents and membranes, glycerol carbonate appears to have the highest CO2-N2 selectivity, around 90-130. The significant aspect of glycerol carbonate is

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002 2293 Table 4. Comparison of Pure Glycerol Carbonate ILMs with Other Solvents/Membrane Configurations solvent/membrane

feed condition

PEO-containing and cross-linked polymer films polyimide NMP-swollen polymer glycerold glycerol carbonated water PEI/PVA blend membrane glycerold glycerol carbonated

dry dry dry dry dry humid humid humid humid

(QCO2/tm)eff, cm3/cm2‚s‚cmHg

2.1 × 10-5 1.5 × 10-8 3.0 × 10-7 2.0 × 10-6 3.0 × 10-7 9.0 × 10-7

QCO2, barrer

RCO2/N2

ref

250 160 1000 5.6 80-110 2090 ∼440 111 280-380

60a 19b 13c 1.5 90-130 40 70 38.5 90-110

4 24 12 16 present work 14 25 16 present work

a Based on pure gas permeabilities. b The selectivity reported is that of CO -CH . c The selectivity reported is that of CO -CH ; the 2 4 2 4 permeability was calculated based on a prefilm solution thickness of 50 µm. Values are reported after 5 min of exposure to gas. d Sweep side: dry helium. Substrate: 100 µm thick hydrophilized PVDF.

Table 5. Solubility of CO2 in Various Physical Solventsa solvent water methanol propylene carbonate dimethyl ether of PEG NMP sulfolane glycerol carbonate

viscosity, cP 1 2.4 3.6 1.7 11.5 61.6

boiling point, °C 100 240 276 201 285 137 (at 0.5 mmHg)

mol wt 18 32 102 99 118

SCO2, cm3(gas)/cm3(liq)‚cmHg

ref

1.05 × 10-2 10.5 × 10-2 b 4.6 × 10-2 4.7 × 10-2 5.13 × 10-2 3.8 × 10-2 2.00 × 10-2 c

26 26 26 26 26 26 present work

a

The solubility of CO2, SCO2, is reported at 70-80 °F.26 b The solubility is measured at 263 K. c The solubility is calculated based on QCO2 under dry feed conditions, and the diffusivity (DCO2) is from the Wilke-Chang correlation.27

its retention of CO2-N2 selectivity, irrespective of feed gas RH. As discussed earlier, its permeation and selective behavior does not appear to be affected by the thickness of the substrate. In other words, if a 50 µm substrate is used, the CO2 permeance through pure glycerol carbonate ILM could be 2.0 × 10-6 cm3/cm2‚s‚ cmHg under humidified feed conditions. Solvent-swollen polymers based on N-methylpyrrolidone (NMP)12 with a prefilm thickness of 50 µm yield a CO2 permeance of 2.1 × 10-5 cm3/cm2‚s‚cmHg with a CO2-CH4 selectivity of only 13. Thinner liquid membranes of glycerol carbonate of a thickness of, say, 5 µm, could potentially result in a CO2 permeance of 2.0 × 10-5 cm3/cm2‚s‚cmHg with a selectivity of 100. Table 5 presents the solubility of CO2 in various physical solvents studied under ambient conditions. The solvents selected are used in commercial processes for CO2 separation. Amines are not included for comparison because CO2 separation with amines is due to chemical reaction and not physical absorption. Table 5 also presents the viscosity and boiling points of these various solvents. It should be noted that the boiling point of glycerol carbonate is reported at 0.5 mmHg pressure. Because data are not available for CO2 solubility in glycerol carbonate, an estimate is made from experimental permeability data and the predicted diffusivity of CO2. The diffusivity of CO2 in glycerol carbonate is estimated using the Wilke-Chang correlation.27 A CO2 permeability of 100 barrer under dry feed and sweep conditions was taken to calculate the CO2 solubility in pure glycerol carbonate. From these calculations, glycerol carbonate has a good potential to be a physical solvent for CO2. Given its relatively high viscosity, diffusivities of CO2 and N2 can be lower than those through other solvents by an order of magnitude or more. Data on the diffusivity or solubility of N2 through other solvents are not available in the literature. It should be noted that CO2 permeabilities through pure glycerol carbonate membranes increased by more than 3 times when the feed side was humidi-

fied. This would indicate that, in the presence of moisture, the CO2 diffusivity increases through the liquid membrane. Glycerol carbonate has potential in many current CO2 separation applications. Liquid membranes are currently not used in most of the applications where CO2 is encountered in low concentrations (e.g., atmosphere and space suit) or high concentrations (flue gas and biogas). The reasons are mainly the loss of the solvent water and/or the carrier. Glycerol was shown earlier16-18 to offer a major improvement in providing stable ILMs. Glycerol carbonate could be the next-generation solvent in the place of or in combination with glycerol in providing a much higher selectivity for CO2 even in the absence of moisture in the feed gas mixture. In the case of CO2 separation from a flue gas mixture, the CO2 concentration in the flue gas is usually between 10 and 15%. Most of the membrane processes currently have CO2 selectivities lower than 50.25 In poly(vinyl alcohol) (PVA)/poly(ether imide) (PEI) blend membranes, the CO2 effective permeance at 30 cmHg was about 2 GPU having a selectivity of about 50. In addition, these membranes show a dependence of the carbon dioxide partial pressure on their permeance and selectivity. This would indicate a dependence on the membrane thickness as well (the thickness of the membranes used in this study was 220 µm). On the other hand, pure glycerol carbonate liquid membranes in the present work did not show any effect of the membrane thickness or carbon dioxide partial pressure on their CO2-N2 selectivity and CO2 permeability. In this sense, if a 5 µm substrate is immobilized with a pure glycerol carbonate ILM, an effective CO2 permeance of 60 GPU having a CO2-N2 selectivity of 100 can be easily achieved. These values would place such a membrane within the realms of commercial success. Our future research is directed to such goals. Similarly, in situations where low concentrations of CO2 are encountered, such as a space walk, the glycerol carbonate membranes are poised for greater attention.

2294

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002

For example, the addition of a small amount of dendrimer (0.06 M) to pure glycerol carbonate results in an ILM in a 100 µm PVDF substrate with an effective CO2 permeance of 18 GPU and an RCO2/N2 of 1000. A thinner substrate, say, 5 µm thick and a modest increase in the dendrimer concentration may lead to a stable liquid membrane with an effective CO2 permeance of at least 100 GPU and an RCO2/N2 of 1000. To put this into perspective, the only other membrane with such a selectivity and CO2 permeance was an aqueous-based plasma grafted membrane impregnated with ethylenediamine.6 The glycerol carbonate based ILMs are expected to be environmentally friendly and inherently stable compared to the above liquid membranes containing ethylenediamine which will be lost with time even in an ion-exchange matrix.25 4.6. Concluding Remarks. From the above study on the utility of glycerol carbonate for carbon dioxide separation, the following concluding remarks can be made. (1) Glycerol carbonate shows great promise as an attractive physical solvent for carbon dioxide separation because of its low vapor pressure, high CO2-N2 selectivity, and reasonable CO2 permeability. It is nontoxic and environmentally friendly. (2) The performances of pure glycerol carbonate ILMs appear to be independent of the carbon dioxide partial pressure difference and substrate thickness. (3) The CO2-selective behavior of glycerol carbonate is not affected by the absence of moisture in the feed gas mixture. This makes glycerol carbonate an ideal solvent when RH values of less than 100% or variation of feed side humidities are expected. Glycerol carbonate ensures that RCO2/N2 remains constant (around 80-100) for any value of the feed-side RH. (4) It is worth exploring the formation of ultrathin liquid membranes using glycerol carbonate to achieve higher CO2 permeances and a selectivity of about 80100. (5) The potential of a glycerol carbonate based FTM based on dendrimers, etc., having high CO2-N2 selectivity and high CO2 permeance is considerable. (6) There is a need for studying the kinetic and physical parameters of glycerol carbonate with/without any carriers. The present study appears to be the first attempt in employing glycerol carbonate for any membrane gas separation application. For practical utilization of such a liquid as an ILM, it would be useful to use hollow fibers as a porous substrate, demonstrate ILM stability over 6 months to 1 year, illustrate recharging of the liquid in situ, and expose the ILM to real-life process streams containing a variety of contaminants. Further, the feed pressure has to be increased to at least 3-5 atm and the temperatures to 35-50 °C. Acknowledgment We thank Dr. Mark Posey of Huntsman Corp. for providing us with the research samples of glycerol carbonate. A.S.K. acknowledges support from the Membrane Separations and Biotechnology Program at New Jersey Institute of Technology. Nomenclature A ) membrane area Ji ) species flux through the membrane

pi ) partial pressure of species i P ) pressure Qi ) permeability coefficient of species i RH ) relative humidity tm ) membrane thickness Vi ) volumetric permeation rate of species i xi ) species mole fraction in the feed yi ) species mole fraction in the permeate Subscripts/Superscripts/Greek Letters eff ) effective f ) feed side i, j ) species in ) inlet condition out ) outlet condition p ) permeate side t ) total true ) true R*i/j ) ideal separation factor of species i over j, defined in eq 2 Ri/j ) separation factor of species i over j m ) porosity of the substrate ∆pCO2 ) difference between pCO2,f and pCO2,p τm ) tortuosity of the substrate

Literature Cited (1) Zolandz, R. R.; Fleming, G. K. In Membrane Handbook; Ho, W. S. W., Sirkar, K. K., Eds.; Chapman and Hall: New York, 1992. (2) Koros, W. J.; Fleming, G. K. Membrane Based Gas Separation. J. Membr. Sci. 1993, 83, 1. (3) Stern, S. A. Polymers for Gas Separation: The Next Decade. J. Membr. Sci. 1994, 94, 1. (4) Hirayama, Y.; Kase, Y.; Tanihara, N.; Sumiyama, Y.; Kusuki, Y.; Haraya, K. Permeation Properties to CO2 and N2 of poly(ethylene oxide)-containing and Crosslinked Polymer Films. J. Membr. Sci. 1999, 160, 87. (5) LeBlanc, O. H.; Ward, W. J.; Matson, S. L.; Kimura, S. G. Facilitated Transport of CO2 in Ion-exchange Membranes. J. Membr. Sci. 1980, 6, 339. (6) Matsuyama, H.; Teramoto, M. Facilitated Transport of Carbon Dioxide through Functional Membranes Prepared by Plasma Graft Polymerization using Amines as Carrier. In Chemical Separations with Liquid Membranes; Bartsch, R. A., Way, J. D., Eds.; ACS Symposium Series 642; American Chemical Society: Washington, DC, 1996. (7) Matsuyama, H.; Teramoto, M.; Sakakura, H.; Iwai, K. Facilitated Transport of CO2 through Various Ion Exchange Membranes Prepared by Plasma Graft Polymerization. J. Membr. Sci. 1996, 117, 251. (8) Majumdar, S.; Guha, A. K.; Sirkar, K. K. A New Liquid Membrane Technique for Gas Separation. AIChE J. 1988, 34, 1135. (9) Laciak, D. V.; Quinn, R.; Pez, G. P.; Applyby, J. B.; Puri, P. S. Selective Permeation of Ammonia and Carbon Dioxide by Novel Membranes. Sep. Sci. Technol. 1990, 25, 1295. (10) Quinn, R.; Appleby, J. B.; Pez, G. P. New Facilitated Transport Membranes for the Separation of Carbon Dioxide from Hydrogen and Methane. J. Membr. Sci. 1995, 104, 139. (11) Quinn, R.; Laciak, D. V. Polyelectrolyte Membranes for Acid Gas Separations. J. Membr. Sci. 1997, 131, 49. (12) Matson, S. L.; Lee, E. K. L.; Friesen, D. T.; Kelly, D. J. Acid Gas Scrubbing by Composite Solvent-swollen Membranes. U.S. Patent 4,737,166, 1988. (13) Ward, W. J.; Robb, W. L. Carbon Dioxide-Oxygen Separation: Facilitated Transport of Carbon Dioxide across a Liquid Film. Science 1967, 156, 1481. (14) Bhave, R. R.; Sirkar, K. K. Gas Permeation and Separation by Aqueous Membranes Immobilized across the Whole Thickness or in a Thin Section of Hydrophobic Microporous Celgard Films. J. Membr. Sci. 1986, 27, 41. (15) Meldon, J. H.; Paboojian, A.; Rajangam, G. Selective CO2 Permeation in Immobilized Liquid Membranes. AIChE Symp. Ser. 1986, 248, 114.

Ind. Eng. Chem. Res., Vol. 41, No. 9, 2002 2295 (16) Chen, H.; Kovvali, A. S.; Majumdar, S.; Sirkar, K. K. Selective CO2 Separation from CO2-N2 Mixtures by Immobilized Carbonate-Glycerol Membranes. Ind. Eng. Chem. Res. 1999, 38, 3489. (17) Chen, H.; Kovvali, A. S.; Sirkar, K. K. Selective CO2 Separation from CO2-N2 Mixtures by Immobilized Glycine-NaGlycerol Membranes. Ind. Eng. Chem. Res. 2000, 39, 2447. (18) Chen, H.; Obuskovic, G.; Majumdar, S.; Sirkar, K. K. Immobilized Glycerol-based Liquid Membranes in Hollow Fibers for Selective CO2 Separation from CO2-N2 Mixtures. J. Membr. Sci. 2001, 183, 75. (19) Kovvali, A. S.; Chen, H.; Sirkar, K. K. Dendrimer membranes: A CO2-selective Molecular Gate. J. Am. Chem. Soc. 2000, 122, 7594. (20) Kovvali, A. S.; Sirkar, K. K. Dendrimer Liquid Membranes: CO2 Separation from Gas Mixtures. Ind. Eng. Chem. Res. 2001, 40, 2502. (21) Kohl, A. L.; Riesenfeld, F. C. Gas Purification; Gulf Publishing Co.: Houston, TX, 1979. (22) Kovvali, A. S. Immobilized Liquid Membranes for Facilitated Transport and Gas Separation. Ph.D. Dissertation, New Jersey Institute of Technology, Newark, NJ, 2001.

(23) Prasad, R.; Kiani, A.; Bhave, R. R.; Sirkar, K. K. Further Studies on Solvent Extraction with Immobilized Interfaces in a Microporous Hydrophobic Membrane. J. Membr. Sci. 1986, 26, 79. (24) White, L. S.; Blinka, T. A.; Kloczzewski, H. A.; Wang, I.-f. Properties of Polyimide Gas Separation Membrane in Natural Gas Streams. J. Membr. Sci. 1995, 103, 73. (25) Matsuyama, H.; Terada, A.; Nakagawara, T.; Kitamura, Y.; Teramoto, M. Facilitated Transport of CO2 through polyethyleneimine/poly(vinyl alcohol) Blend Membrane. J. Membr. Sci. 1999, 163, 221. (26) Astarita, G.; Savage, D. W.; Bisio, A. Gas Treating with Chemical Solvents; John Wiley & Sons: New York, 1983. (27) Wilke, C. R.; Chang, P. Correlation of Diffusion Coefficients in Dilute Solutions. AIChE J. 1955, 1, 264.

Received for review September 10, 2001 Revised manuscript received February 1, 2002 Accepted February 11, 2002 IE010757E