- Email: [email protected]
Membrane separations using reverse micelles in nearcritical and supercritical fluid solvents
J. of Supercritical Fluids 25 (2003) 225 /231 www.elsevier.com/locate/supflu
Membrane separations using reverse micelles in nearcritical and supercritical fluid solvents Clement R. Yonker a,*, John L. Fulton a, Max R. Phelps b, Larry E. Bowman a a
b
Fundamental Science Division, Pacific Northwest National Laboratory, Richland, WA 99352, USA Environmental Technology Division, Pacific Northwest National Laboratory, Richland WA 99352, USA Received 11 December 2001; received in revised form 27 February 2002; accepted 30 April 2002
Abstract The use of reverse micelles coupled with ultrafiltration membranes for the separation of macromolecules dissolved in the cores of the reverse micelles using nearcritical and supercritical fluid solvents is described. This methodology allows one to address the separation of a wide range of polar molecules greatly extending the type of molecules that can be separated using only pure supercritical fluids. The solutes to be separated are initially dissolved in the reverse micellar solution and introduce into the pressure vessel containing the membrane. The surfactant and water core are passed through the membrane while the macromolecule selectivity is based on size and molecular weight. The ability for continuous recycle in an extraction system is discussed. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Supercritical; Membrane; Separation; Proteins; Polymers
1. Introduction The use of supercritical fluids for routine separations has been described in numerous published works over the past two decades. The solute recovery step in these separations entails the thermodynamic manipulation of the solution to affect the solute precipitation. Refinements of this approach such as density ‘tuning’ can be used to affect the separation of one compound from another. By their nature, these separations are
* Corresponding author. Tel.: /1-509-372-4748; fax: /1509-375-6660 E-mail address: [email protected] (C.R. Yonker).
often limited to batch processes. In addition, a great deal of energy is required to recycle the solvent to its original thermodynamic state for further processing. Over the last 5/7 years, research has been focused on membrane separations in supercritical fluids, particularly supercritical CO2, in which the solute can be removed from the fluid solvent without a significant change in the thermodynamic state of the fluid [1 /12]. While this has great potential to enhance supercritical fluid extraction and separation processes, it is still limited to relatively non-polar molecules with demonstrated solubility in supercritical fluids. Supercritical fluids such as CO2 are nonpolar solvents based on their physicochemical properties and this inherently limits their use in extraction to
0896-8446/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 8 9 6 - 8 4 4 6 ( 0 2 ) 0 0 0 9 8 - 0
226
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
low molecular weight, nonpolar solute molecules. While, a few research groups have focused on the combination of membrane separations with supercritical fluids as the solvent, these have been limited to the molecular weight regime of / 300 /1000 g/mol. Most of these previous studies were interested in determining the permeability of mixtures of small hydrocarbon molecules in CO2 through different membrane structures. Using only pure CO2 or small amounts of binary modifiers limits the range (polarity) of molecules that can be seperated through membrane process. In this research, we describe a membrane process using nearcritical and supercritical fluids for the separation of polar, water-soluble macromolecules in these solvents. This technique is based upon the use of reverse micelles in the fluid solvents in which the macromolecule is soluble in the polar core of the reverse micelle and not in the fluid itself. There are numerous studies of reverse micelles or microemulsions in supercritical fluids starting from the late 1980s [13 /18]. Reverse microemulsion (and implicitly, reverse micelle) solutions are clear, thermodynamically stable colloidal systems in which submicroscopic aggregates of water and surfactant are dispersed in a low polarity solvent, in this case a supercritical fluid. Reverse micelles, consisting of from /10 to 50 surfactant molecules per aggregate, form spontaneously upon addition of certain surfactants to these fluids. Adding water swells the micelle core, but, because these larger droplets are still smaller than about 100 nm, the solution remains clear. Such systems are commonly referred to as waterin-oil (w/o) microemulsions. These systems have potential for applications exploiting the readily variable properties of supercritical fluids in conjunction with the unique solvent environments of the reverse micelle or microemulsion phase. Thus, reverse micelles allow highly polar or polarizable compounds to be dispersed in these non-polar fluids. We will discuss membrane based separation in two different solvent systems, one a microemulsion in nearcritical propane and the second in supercritical ethane. The separations involve removing Cytochrome c from the microemulsion surfactant and the separation of two dextrans differing in
their molecular weight. Neither the protein nor the dextrans are directly soluble in the fluids. This separation technology can be used to recover a wide range of polar solute molecules from supercritical fluid solutions and can be used in fermentation broth or natural product extractions.
2. Experimental 2.1. Reagents and materials The compressed gases propane and ethane were commercial purity and were obtained from Matheson and Scott Specialty Gases, respectively. The surfactant AOT (Dioctyl sulfosuccinate; sodium salt; MW, 444.5 g/mol) was used as received from Aldrich. The protein Cytochrome c (horse heart; MW, 12 384 g/mol) was purchased from Sigma. Dextrans having narrow nominal molecular weights (nMW) were purchased from Molecular Probes Inc. (Eugene, OR). Each particular molecular weight dextran has a specific chromophore label (Cascade Blue-lmax, 400 nm; Lucifer Yellowlmax, 428 nm or Texas Red-lmax, 595 nm). This strategy allows the permeate and the retentate to be characterized via UV/Vis absorption spectrometry. A high-pressure system was assembled from various commercially available and in-house fabricated components. These include syringe pumps (Isco), high-pressure fittings and adapters (High Pressure Equipment), a high-pressure membrane holder (Millipore), a view cell (in-house), and two small autoclaves (in-house). The membranes (Millipore) were of ultrafiltration type (regenerated cellulose and polyethersulfone) that allow dissolved species having a molecular weight below a given cutoff to pass through. The membrane nominal molecular weight cut-off is listed in the specific examples shown in Figs. 2 /4. A schematic showing the experimental apparatus is given in Fig. 1. Dissolutions of the microemulsions in the fluids were prepared in the view cell, which allows visual verification of complete solution of all components. Autoclave 1 (AC1) was filled with the same surfactant solution as the view cell (without the dextran or protein) and provides a feed stream of
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
the microemulsion solution into the view cell. AC2 is a collection vessel for the permeate fluid. UV/Vis absorption spectra were acquired with a Varian/ Cary 2200 UV/Vis spectrometer.
227
depressurized. Permeate and retentate were collected and dissolved in hexane. UV/Vis absorption spectra were acquired of these solutions using a 1 cm pathlength cuvette.
2.2. Procedure 3. Results and discussion Measured amounts of the dextrans or protein, AOT, and water were added to the view cell. The view cell was then sealed, connected to the pressure system and pressurized with either ethane or propane. It was stirred using a magnetic stir bar located in the cell. The view cell was heated when the separation was to be carried out in the supercritical regime. Once the contents of the view cell were observed to be in solution, valves were adjusted to isolate the view cell while the rest of the apparatus was pressurized. For the separation of the dextrans in supercritical ethane, AC1 was prepared with the same composition of AOT and water as in the view cell. For separations in the supercritical regime, the view cell and the other indicated portions of the apparatus were immersed in a thermostatted water bath. Flow through the system was initiated by adjusting the valves appropriately and setting both pumps to maintain a constant differential pressure. That is pump 1 delivered fluid into the apparatus while pump 2 filled with fluid from AC2 via displacement. The fluid fluxes for these membranes were typically in the range of /0.006 /0.015 ml/min cm2. The pressure drops across the membrane were in the range from /10 to 50 psi. Once enough fluid solution had flowed through the apparatus, the experiment was stopped and the apparatus was
Supercritical fluid extractions have been demonstrated as an effective means for the removal of soluble nonpolar molecules from complex matrices. These extractions involve the thermodynamic manipulation of the solvent characteristics (reducing pressure or temperature) to affect the separation of the fluid from the target compound. Membranes could eliminate the need for depressurization of the fluid solution for the removal of the target compound, but further fundamental effort is needed to understand the membrane retention process. Currently as practiced, membrane separations in nearcritical or supercritical CO2 are limited to small, nonpolar molecules. There is a wide range of potential applications for membrane separations in compressible solvent solutions if polar macromolecules could be addressed. A reverse micelle supercritical fluid solution allows one to solubilize polar macromolecules in the water core of the micelle contained in the solvent. A membrane separation of these macromolecular compounds can then be accomplished via a simple size exclusion process through the membrane pore based on molecular weight. The selection of the microemulsion systems used here is based on previous studies of supercritical and nearcritical alkane systems. The structure and
Fig. 1. Schematic of the experimental apparatus. Pumps 1 and 2 are syringe pumps; RD, rupture disks; PX, pressure transducer, MH, membrane holder; AC1 and AC2, autoclaves 1 and 2.
228
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
phase behavior of microemulsions formed in supercritical or near-critical fluids is strongly dependent on the pressure of the system in contrast to normal liquid solvents in which the effect of pressure has little or no effect. Liquid propane at 25 8C is a fluid that is not far below its critical point (Tc /96.7 8C, Pc /42.5 bar). Hence moderate changes in the system pressure cause appreciable changes in the properties of the fluid. These unique properties of a propane continuous phase can be continuously adjusted without changing the chemical composition of the microemulsion. The second system that we studied was a supercritical ethane microemulsion. The pressure and temperature for both of these systems must be selected to assure that a single homogeneous solution exists. Detailed studies of the phase behavior of liquid propane and supercritical ethane have been previously reported [19,20] and these provide the basis for the pressure and temperature conditions used in this study. In particular, the cloud point pressure is strongly a function of W , the molar water-to-surfactant ratio, and only weakly dependant on the overall surfactant concentration. The surfactant AOT is very soluble in ethane and propane and the solubilization properties of the aqueous cores of these micelles are nearly identical to AOT micelles in liquid alkanes that have been the subject of numerous studies. It is now well established that the aqueous core regions of these AOT micelles are capable of dissolving a broad range of highly polar, ionic and/or high molecular weight species. It is for these reasons that we selected the AOT/ alkane system to demonstrate these membrane separations. The critical micelle concentrations (CMC B/1 mM) in propane and ethane are typically below those of liquid alkanes. Thus the CMC is only a small fraction the total surfactant used in this study. Fig. 2, demonstrates a membrane separation of Cytochrome c from the AOT surfactant used to form reverse micelles in liquid propane at 20.2 8C and 3750 psi. In this case, the overall surfactant concentration is [AOT] /100 mM and the water content is W /[H2O]/[AOT] /20. Cytochrome c has a molecular weight of 12 384 g/mol and is solubilized in the water core of the micelle.
Fig. 2. Cytochrome c separated from the surfactant AOT in a liquid propane microemulsion at 20.2 8C and 3750 psi using a 10 000 nMW regenerated cellulose ultrafiltration membrane. The majority of the protein (MW, 12 384 g/mol; lmax, 418 nm) is retained as evidenced by the peak in the spectrum of the retentate and the lack of absorbance in the permeate spectrum.
Cytochrome c has no demonstrated solubility in pure liquid propane under these conditions. The experiment was initiated by dissolving the Cytochrome c in the propane reverse micelle solution in a view cell to visually confirm that the macromolecule was dissolved in the optically clear microemulsion phase. Flow under constant pressure was established through the membrane holder, see Fig. 1, and the permeate was captured in AC2 while the retentate was captured on the upstream side of the membrane. After depressurization, the permeate and retentate were then dissolved in hexane and the UV/Vis spectra were used to determine the protein concentrations in the permeate and retentate. As shown in Fig. 2, the retentate contains the Cytochrome c with a small amount of surfactant (shoulder at 250 nm), while the permeate is free of Cytochrome c, containing only surfactant. A simple size exclusion process is demonstrated in this separation, in which, the Cytochrome c is too large to penetrate the pores of the regenerated cellulose ultrafiltration membrane having a nominal molecular weight cut-off of 10 000 nMW. The AOT surfactant having a molecular weight of 444.5 g/mol passes through the membrane. Thus the reverse micelle structures reassemble on the down-stream side of the membrane after the surfactant passes through the membrane. The water in the micelle core also is transported through the membrane via a similar
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
Fig. 3. Two labeled dextrans in an AOT microemulsion in nearcritical propane at 22 8C, 2000 psi using a 5000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that virtually all of the 40 000 nMW dextran (Texas Red-lmax, 595 nm) is retained while most of a 3000 nMW dextran (Cascade Blue-lmax, 400 nm) passed through the membrane.
mechanism. More experimental effort is needed to determine the exact transport mechanism for both the surfactant and water through the membrane under these conditions. To demonstrate a separation of two polar macromolecules under nearcritical and supercritical fluid conditions, two different dextrans labeled with specific chromophores were chosen due to their solubility in the water core of the reverse micelle. In Fig. 3, a 3000 nMW dextran labeled
229
with Cascade Blue (lmax, 400 nm) was separated from a 40 000 nMW dextran labeled with Texas Red (lmax, 595 nm). The two dextrans were first dissolved in the AOT microemulsion ([AOT] /100 mM, W /10) in nearcritical propane at 22 8C, 2000 psi in the view cell before initiating flow through the membrane. The membrane used for this separation was a 5000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that virtually all of the 40 000 nMW dextran is retained, while most of the 3000 nMW dextran passes through the membrane. There is a small amount of the 3000 nMW dextran in the retentate, which is most likely due to slow transport of the dextran through the small pores of the membrane that are only slightly larger than the diameter of the dextran. Once again, the separation mechanism is one of a simple size exclusion process. In Fig. 4, a supercritical fluid reverse micelle membrane separation is demonstrated. The two dextrans were 3000 nMW labeled with Texas Red (lmax, 595 nm) and 40 000 nMW labeled with Lucifer Yellow (lmax, 428 nm). The AOT microemulsion ([AOT] /100 mM, W /10) was in supercritical ethane at 40 8C and 7250 psi and the polyethersulfone ultrafiltration membrane had a 10 000 nMW cut-off. This pressure was selected to be above the cloud point. At lower pressures
Fig. 4. Two labeled dextrans in an AOT microemulsion in supercritical ethane at 40 8C, 7250 psi using a 10 000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that the 3000 nMW dextran (Texas Red-lmax, 595 nm) preferentially passed through the membrane while most of the 40 000 nMW dextran (Lucifer Yellow-lmax, 428 nm) was retained.
230
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
this AOT microemulsion phase would separate into a liquid and fluid component destroying the microemulsion. As seen in the figure the 3000 nMW dextran passed through the membrane, while the 40 000 nMW dextran was retained by the membrane. Once again the presence of some residual 3000 nMW dextran in the retentate means that an insufficient total volume of solution has been passed through the membrane to achieve 100% separation. This study did not directly address the long-term viability of the membrane during exposure to nearcritical and supercritical fluids, but over the course of the 2-h separation there was no degradation of performance in either the regenerated cellulose or the polyethersulfone membrane in either fluid solvent.
4. Conclusions We have demonstrated the separation of dextrans and a protein dissolved in an AOT microemulsion in nearcritical and supercritical fluids using ultrafiltration membranes. Neither the dextrans nor the protein are directly soluble in the pure fluids. However, they are soluble in the water core of the reverse micelle in the fluid/microemulsion solution. This methodology directly extends the capabilities of membrane separations in supercritical fluids to include both nonpolar and polar molecules. Also, high molecular weight biomacromolecule/micelle solutions are readily amenable for processing by this simple size exclusion separation process. In the future, one could envision a combination of polar and nonpolar macromolecules being selectively removed from either a nearcritical or supercritical fluid extraction process based exclusively on the target compounds molecular weight. Similarly a cascade series of different molecular weight cut-off membranes could be used to have a continuous extraction, separation, and solvent re-cycle process for polar macromolecules. Further effort needs to be directed in the fundamental understanding of the membrane transport and separation process in both nearcritical and supercritical fluid solvents.
Acknowledgements Work at the Pacific Northwest National Laboratory (PNNL) was supported by the Office of Science, Office of Basic Energy Sciences, Chemical Sciences Division of the US Department of Energy, under Contract DE-AC076RLO 1830.
References [1] S.I. Semenova, H. Ohya, T. Higashijima, Y. Negishi, Dependence of permeability through polyimide membranes on state of gas, vapor, liquid and supercritical fluid at high temperature, J. Membr. Sci. 67 (1992) 29 /37. [2] S.I. Semenova, H. Ohya, T. Higashijima, Y. Negishi, Separation of supercritical CO2 and ethanol mixtures with an asymmetric polyimide membrane, J. Membr. Sci. 74 (1992) 131 /139. [3] H. Ohya, T. Higashijima, Y. Tsuchiya, H. Tokunaga, Y. Negishi, Separation of supercritical CO2 and iso-octane mixtures with an asymmetric polyimide membrane, J. Membr. Sci. 84 (1993) 185 /189. [4] J.-H. Hsu, C.-S. Tan, Separation of ethanol from aqueous solution by a method incorporating supercritical CO2 with reverse osmosis, J. Membr. Sci. 81 (1993) 273 /285. [5] T. Higashijima, H. Ohya, Y. Tsuchiya, H. Tokunaga, M. Aihara, Y. Negishi, Separation of supercritical fluid mixtures of CO2 and petroleum components with an asymmetric polyimide membrane, J. Membr. Sci. 93 (1994) 165 /173. [6] T. Fujii, Y. Tokunaga, K. Nakamura, Effect of solute adsorption properties on its separation from supercritical carbon dioxide with a thin porous silica membrane, Biosci. Biotech. Biochem. 60 (1996) 1945 /1949. [7] Y. Tokunaga, T. Fujii, K. Nakamura, Separation of caffeine from supercritical carbon dioxide with a zeolite membrane, Biosci. Biotech. Biochem. 61 (1997) 1024 / 1026. [8] S. Sarrade, G.M. Rios, M. Carle´s, Nanofiltration membrane behavior in a supercritical medium, J. Membr. Sci. 114 (1996) 81 /91. [9] G. Afrane, E.H. Chimowitz, Experimental investigation of a new supercritical fluid-inorganic membrane separation process, J. Membr. Sci. 116 (1996) 293 /299. [10] S. Sarrade, G.M. Rios, M. Carle´s, Supercritical CO2 extraction coupled with nanofiltration separation applications to natural products, Sep. Pur. Technol. 14 (1998) 19 / 25. [11] S. Sarrade, C. Perre, G.M. Rios, M. Carle´s, Performance of supercritical carbon dioxide extraction coupled with nanofiltration, in: G. Brunner, M. Perrut (Eds.), Proceeding of 3rd International Symposium On Supercritical Fluids, vol. 2, Strasbourg, Institut National Polytechnique de Lorraine. 17 /19 October 1994, pp. 71 /76.
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231 [12] M.G.E. Goemans, F.M. Tiller, L. Li, E.F. Gloyna, Separation of metal oxides from supercritical water by crossflow microfiltration, J. Membr. Sci. 124 (1997) 129 / 145. [13] R.W. Gale, J.L. Fulton, R.D. Smith, Organized molecular assemblies in the gas phase: reverse micelles and microemulsions in supercritical fluids, J. Phys. Chem. 109 (1987) 920 /921. [14] R.D. Smith, J.L. Fulton, J.P. Blitz, J.M. Tingey, Reverse micelles and microemulsions in near-critical and supercritical fluids, J. Phys. Chem. 94 (1990) 781 / 787. [15] J.L. Fulton, D.M. Pfund, J.B. McClain, T.J. Romack, E.E. Maury, J.R. Combes, E.T. Samulski, J.M. DeSimone, M. Capel, Aggregation of amphiphilic molecules in supercritical carbon dioxide: a small angle X-ray scattering study, Langmuir 11 (1995) 4241 /4249.
231
[16] K.A. Consani, R.D. Smith, Observations on the solubility of surfactants and related molecules in carbon dioxide at 50 8C, J. Supercrit. Fluids 3 (1990) 51 /65. [17] K.P. Johnston, K.L. Harrison, M.J. Clarke, S.M. Howdle, M.P. Heitz, F.V. Bright, C. Carlier, T.W. Randolph, Water-in-carbon dioxide microemulsions: an environment for hydrophiles including proteins, Science 271 (1996) 624 /626. [18] K.A. Bartscherer, M. Minier, H. Renon, Microemulsions in compressible fluids */a review, Fluid Phase Equilibria 107 (1995) 93 /150. [19] J.M. Tingey, J.L. Fulton, R.D. Smith, Interdroplet attractive forces in AOT water-in-oil microemulsions formed in subcritical and supercritical solvents, J. Phys. Chem. 94 (1990) 1997 /2004. [20] R.D. Smith, J.L. Fulton, J.P. Blitz, J.M. Tingey, Reverse micelles and microemulsions in near-critical and supercritical fluids, J. Phys. Chem. 94 (1990) 781 /787.
Membrane separations using reverse micelles in nearcritical and supercritical fluid solvents Clement R. Yonker a,*, John L. Fulton a, Max R. Phelps b, Larry E. Bowman a a
b
Fundamental Science Division, Pacific Northwest National Laboratory, Richland, WA 99352, USA Environmental Technology Division, Pacific Northwest National Laboratory, Richland WA 99352, USA Received 11 December 2001; received in revised form 27 February 2002; accepted 30 April 2002
Abstract The use of reverse micelles coupled with ultrafiltration membranes for the separation of macromolecules dissolved in the cores of the reverse micelles using nearcritical and supercritical fluid solvents is described. This methodology allows one to address the separation of a wide range of polar molecules greatly extending the type of molecules that can be separated using only pure supercritical fluids. The solutes to be separated are initially dissolved in the reverse micellar solution and introduce into the pressure vessel containing the membrane. The surfactant and water core are passed through the membrane while the macromolecule selectivity is based on size and molecular weight. The ability for continuous recycle in an extraction system is discussed. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Supercritical; Membrane; Separation; Proteins; Polymers
1. Introduction The use of supercritical fluids for routine separations has been described in numerous published works over the past two decades. The solute recovery step in these separations entails the thermodynamic manipulation of the solution to affect the solute precipitation. Refinements of this approach such as density ‘tuning’ can be used to affect the separation of one compound from another. By their nature, these separations are
* Corresponding author. Tel.: /1-509-372-4748; fax: /1509-375-6660 E-mail address: [email protected] (C.R. Yonker).
often limited to batch processes. In addition, a great deal of energy is required to recycle the solvent to its original thermodynamic state for further processing. Over the last 5/7 years, research has been focused on membrane separations in supercritical fluids, particularly supercritical CO2, in which the solute can be removed from the fluid solvent without a significant change in the thermodynamic state of the fluid [1 /12]. While this has great potential to enhance supercritical fluid extraction and separation processes, it is still limited to relatively non-polar molecules with demonstrated solubility in supercritical fluids. Supercritical fluids such as CO2 are nonpolar solvents based on their physicochemical properties and this inherently limits their use in extraction to
0896-8446/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 8 9 6 - 8 4 4 6 ( 0 2 ) 0 0 0 9 8 - 0
226
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
low molecular weight, nonpolar solute molecules. While, a few research groups have focused on the combination of membrane separations with supercritical fluids as the solvent, these have been limited to the molecular weight regime of / 300 /1000 g/mol. Most of these previous studies were interested in determining the permeability of mixtures of small hydrocarbon molecules in CO2 through different membrane structures. Using only pure CO2 or small amounts of binary modifiers limits the range (polarity) of molecules that can be seperated through membrane process. In this research, we describe a membrane process using nearcritical and supercritical fluids for the separation of polar, water-soluble macromolecules in these solvents. This technique is based upon the use of reverse micelles in the fluid solvents in which the macromolecule is soluble in the polar core of the reverse micelle and not in the fluid itself. There are numerous studies of reverse micelles or microemulsions in supercritical fluids starting from the late 1980s [13 /18]. Reverse microemulsion (and implicitly, reverse micelle) solutions are clear, thermodynamically stable colloidal systems in which submicroscopic aggregates of water and surfactant are dispersed in a low polarity solvent, in this case a supercritical fluid. Reverse micelles, consisting of from /10 to 50 surfactant molecules per aggregate, form spontaneously upon addition of certain surfactants to these fluids. Adding water swells the micelle core, but, because these larger droplets are still smaller than about 100 nm, the solution remains clear. Such systems are commonly referred to as waterin-oil (w/o) microemulsions. These systems have potential for applications exploiting the readily variable properties of supercritical fluids in conjunction with the unique solvent environments of the reverse micelle or microemulsion phase. Thus, reverse micelles allow highly polar or polarizable compounds to be dispersed in these non-polar fluids. We will discuss membrane based separation in two different solvent systems, one a microemulsion in nearcritical propane and the second in supercritical ethane. The separations involve removing Cytochrome c from the microemulsion surfactant and the separation of two dextrans differing in
their molecular weight. Neither the protein nor the dextrans are directly soluble in the fluids. This separation technology can be used to recover a wide range of polar solute molecules from supercritical fluid solutions and can be used in fermentation broth or natural product extractions.
2. Experimental 2.1. Reagents and materials The compressed gases propane and ethane were commercial purity and were obtained from Matheson and Scott Specialty Gases, respectively. The surfactant AOT (Dioctyl sulfosuccinate; sodium salt; MW, 444.5 g/mol) was used as received from Aldrich. The protein Cytochrome c (horse heart; MW, 12 384 g/mol) was purchased from Sigma. Dextrans having narrow nominal molecular weights (nMW) were purchased from Molecular Probes Inc. (Eugene, OR). Each particular molecular weight dextran has a specific chromophore label (Cascade Blue-lmax, 400 nm; Lucifer Yellowlmax, 428 nm or Texas Red-lmax, 595 nm). This strategy allows the permeate and the retentate to be characterized via UV/Vis absorption spectrometry. A high-pressure system was assembled from various commercially available and in-house fabricated components. These include syringe pumps (Isco), high-pressure fittings and adapters (High Pressure Equipment), a high-pressure membrane holder (Millipore), a view cell (in-house), and two small autoclaves (in-house). The membranes (Millipore) were of ultrafiltration type (regenerated cellulose and polyethersulfone) that allow dissolved species having a molecular weight below a given cutoff to pass through. The membrane nominal molecular weight cut-off is listed in the specific examples shown in Figs. 2 /4. A schematic showing the experimental apparatus is given in Fig. 1. Dissolutions of the microemulsions in the fluids were prepared in the view cell, which allows visual verification of complete solution of all components. Autoclave 1 (AC1) was filled with the same surfactant solution as the view cell (without the dextran or protein) and provides a feed stream of
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
the microemulsion solution into the view cell. AC2 is a collection vessel for the permeate fluid. UV/Vis absorption spectra were acquired with a Varian/ Cary 2200 UV/Vis spectrometer.
227
depressurized. Permeate and retentate were collected and dissolved in hexane. UV/Vis absorption spectra were acquired of these solutions using a 1 cm pathlength cuvette.
2.2. Procedure 3. Results and discussion Measured amounts of the dextrans or protein, AOT, and water were added to the view cell. The view cell was then sealed, connected to the pressure system and pressurized with either ethane or propane. It was stirred using a magnetic stir bar located in the cell. The view cell was heated when the separation was to be carried out in the supercritical regime. Once the contents of the view cell were observed to be in solution, valves were adjusted to isolate the view cell while the rest of the apparatus was pressurized. For the separation of the dextrans in supercritical ethane, AC1 was prepared with the same composition of AOT and water as in the view cell. For separations in the supercritical regime, the view cell and the other indicated portions of the apparatus were immersed in a thermostatted water bath. Flow through the system was initiated by adjusting the valves appropriately and setting both pumps to maintain a constant differential pressure. That is pump 1 delivered fluid into the apparatus while pump 2 filled with fluid from AC2 via displacement. The fluid fluxes for these membranes were typically in the range of /0.006 /0.015 ml/min cm2. The pressure drops across the membrane were in the range from /10 to 50 psi. Once enough fluid solution had flowed through the apparatus, the experiment was stopped and the apparatus was
Supercritical fluid extractions have been demonstrated as an effective means for the removal of soluble nonpolar molecules from complex matrices. These extractions involve the thermodynamic manipulation of the solvent characteristics (reducing pressure or temperature) to affect the separation of the fluid from the target compound. Membranes could eliminate the need for depressurization of the fluid solution for the removal of the target compound, but further fundamental effort is needed to understand the membrane retention process. Currently as practiced, membrane separations in nearcritical or supercritical CO2 are limited to small, nonpolar molecules. There is a wide range of potential applications for membrane separations in compressible solvent solutions if polar macromolecules could be addressed. A reverse micelle supercritical fluid solution allows one to solubilize polar macromolecules in the water core of the micelle contained in the solvent. A membrane separation of these macromolecular compounds can then be accomplished via a simple size exclusion process through the membrane pore based on molecular weight. The selection of the microemulsion systems used here is based on previous studies of supercritical and nearcritical alkane systems. The structure and
Fig. 1. Schematic of the experimental apparatus. Pumps 1 and 2 are syringe pumps; RD, rupture disks; PX, pressure transducer, MH, membrane holder; AC1 and AC2, autoclaves 1 and 2.
228
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
phase behavior of microemulsions formed in supercritical or near-critical fluids is strongly dependent on the pressure of the system in contrast to normal liquid solvents in which the effect of pressure has little or no effect. Liquid propane at 25 8C is a fluid that is not far below its critical point (Tc /96.7 8C, Pc /42.5 bar). Hence moderate changes in the system pressure cause appreciable changes in the properties of the fluid. These unique properties of a propane continuous phase can be continuously adjusted without changing the chemical composition of the microemulsion. The second system that we studied was a supercritical ethane microemulsion. The pressure and temperature for both of these systems must be selected to assure that a single homogeneous solution exists. Detailed studies of the phase behavior of liquid propane and supercritical ethane have been previously reported [19,20] and these provide the basis for the pressure and temperature conditions used in this study. In particular, the cloud point pressure is strongly a function of W , the molar water-to-surfactant ratio, and only weakly dependant on the overall surfactant concentration. The surfactant AOT is very soluble in ethane and propane and the solubilization properties of the aqueous cores of these micelles are nearly identical to AOT micelles in liquid alkanes that have been the subject of numerous studies. It is now well established that the aqueous core regions of these AOT micelles are capable of dissolving a broad range of highly polar, ionic and/or high molecular weight species. It is for these reasons that we selected the AOT/ alkane system to demonstrate these membrane separations. The critical micelle concentrations (CMC B/1 mM) in propane and ethane are typically below those of liquid alkanes. Thus the CMC is only a small fraction the total surfactant used in this study. Fig. 2, demonstrates a membrane separation of Cytochrome c from the AOT surfactant used to form reverse micelles in liquid propane at 20.2 8C and 3750 psi. In this case, the overall surfactant concentration is [AOT] /100 mM and the water content is W /[H2O]/[AOT] /20. Cytochrome c has a molecular weight of 12 384 g/mol and is solubilized in the water core of the micelle.
Fig. 2. Cytochrome c separated from the surfactant AOT in a liquid propane microemulsion at 20.2 8C and 3750 psi using a 10 000 nMW regenerated cellulose ultrafiltration membrane. The majority of the protein (MW, 12 384 g/mol; lmax, 418 nm) is retained as evidenced by the peak in the spectrum of the retentate and the lack of absorbance in the permeate spectrum.
Cytochrome c has no demonstrated solubility in pure liquid propane under these conditions. The experiment was initiated by dissolving the Cytochrome c in the propane reverse micelle solution in a view cell to visually confirm that the macromolecule was dissolved in the optically clear microemulsion phase. Flow under constant pressure was established through the membrane holder, see Fig. 1, and the permeate was captured in AC2 while the retentate was captured on the upstream side of the membrane. After depressurization, the permeate and retentate were then dissolved in hexane and the UV/Vis spectra were used to determine the protein concentrations in the permeate and retentate. As shown in Fig. 2, the retentate contains the Cytochrome c with a small amount of surfactant (shoulder at 250 nm), while the permeate is free of Cytochrome c, containing only surfactant. A simple size exclusion process is demonstrated in this separation, in which, the Cytochrome c is too large to penetrate the pores of the regenerated cellulose ultrafiltration membrane having a nominal molecular weight cut-off of 10 000 nMW. The AOT surfactant having a molecular weight of 444.5 g/mol passes through the membrane. Thus the reverse micelle structures reassemble on the down-stream side of the membrane after the surfactant passes through the membrane. The water in the micelle core also is transported through the membrane via a similar
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
Fig. 3. Two labeled dextrans in an AOT microemulsion in nearcritical propane at 22 8C, 2000 psi using a 5000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that virtually all of the 40 000 nMW dextran (Texas Red-lmax, 595 nm) is retained while most of a 3000 nMW dextran (Cascade Blue-lmax, 400 nm) passed through the membrane.
mechanism. More experimental effort is needed to determine the exact transport mechanism for both the surfactant and water through the membrane under these conditions. To demonstrate a separation of two polar macromolecules under nearcritical and supercritical fluid conditions, two different dextrans labeled with specific chromophores were chosen due to their solubility in the water core of the reverse micelle. In Fig. 3, a 3000 nMW dextran labeled
229
with Cascade Blue (lmax, 400 nm) was separated from a 40 000 nMW dextran labeled with Texas Red (lmax, 595 nm). The two dextrans were first dissolved in the AOT microemulsion ([AOT] /100 mM, W /10) in nearcritical propane at 22 8C, 2000 psi in the view cell before initiating flow through the membrane. The membrane used for this separation was a 5000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that virtually all of the 40 000 nMW dextran is retained, while most of the 3000 nMW dextran passes through the membrane. There is a small amount of the 3000 nMW dextran in the retentate, which is most likely due to slow transport of the dextran through the small pores of the membrane that are only slightly larger than the diameter of the dextran. Once again, the separation mechanism is one of a simple size exclusion process. In Fig. 4, a supercritical fluid reverse micelle membrane separation is demonstrated. The two dextrans were 3000 nMW labeled with Texas Red (lmax, 595 nm) and 40 000 nMW labeled with Lucifer Yellow (lmax, 428 nm). The AOT microemulsion ([AOT] /100 mM, W /10) was in supercritical ethane at 40 8C and 7250 psi and the polyethersulfone ultrafiltration membrane had a 10 000 nMW cut-off. This pressure was selected to be above the cloud point. At lower pressures
Fig. 4. Two labeled dextrans in an AOT microemulsion in supercritical ethane at 40 8C, 7250 psi using a 10 000 nMW polyethersulfone ultrafiltration membrane. The spectra indicate that the 3000 nMW dextran (Texas Red-lmax, 595 nm) preferentially passed through the membrane while most of the 40 000 nMW dextran (Lucifer Yellow-lmax, 428 nm) was retained.
230
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231
this AOT microemulsion phase would separate into a liquid and fluid component destroying the microemulsion. As seen in the figure the 3000 nMW dextran passed through the membrane, while the 40 000 nMW dextran was retained by the membrane. Once again the presence of some residual 3000 nMW dextran in the retentate means that an insufficient total volume of solution has been passed through the membrane to achieve 100% separation. This study did not directly address the long-term viability of the membrane during exposure to nearcritical and supercritical fluids, but over the course of the 2-h separation there was no degradation of performance in either the regenerated cellulose or the polyethersulfone membrane in either fluid solvent.
4. Conclusions We have demonstrated the separation of dextrans and a protein dissolved in an AOT microemulsion in nearcritical and supercritical fluids using ultrafiltration membranes. Neither the dextrans nor the protein are directly soluble in the pure fluids. However, they are soluble in the water core of the reverse micelle in the fluid/microemulsion solution. This methodology directly extends the capabilities of membrane separations in supercritical fluids to include both nonpolar and polar molecules. Also, high molecular weight biomacromolecule/micelle solutions are readily amenable for processing by this simple size exclusion separation process. In the future, one could envision a combination of polar and nonpolar macromolecules being selectively removed from either a nearcritical or supercritical fluid extraction process based exclusively on the target compounds molecular weight. Similarly a cascade series of different molecular weight cut-off membranes could be used to have a continuous extraction, separation, and solvent re-cycle process for polar macromolecules. Further effort needs to be directed in the fundamental understanding of the membrane transport and separation process in both nearcritical and supercritical fluid solvents.
Acknowledgements Work at the Pacific Northwest National Laboratory (PNNL) was supported by the Office of Science, Office of Basic Energy Sciences, Chemical Sciences Division of the US Department of Energy, under Contract DE-AC076RLO 1830.
References [1] S.I. Semenova, H. Ohya, T. Higashijima, Y. Negishi, Dependence of permeability through polyimide membranes on state of gas, vapor, liquid and supercritical fluid at high temperature, J. Membr. Sci. 67 (1992) 29 /37. [2] S.I. Semenova, H. Ohya, T. Higashijima, Y. Negishi, Separation of supercritical CO2 and ethanol mixtures with an asymmetric polyimide membrane, J. Membr. Sci. 74 (1992) 131 /139. [3] H. Ohya, T. Higashijima, Y. Tsuchiya, H. Tokunaga, Y. Negishi, Separation of supercritical CO2 and iso-octane mixtures with an asymmetric polyimide membrane, J. Membr. Sci. 84 (1993) 185 /189. [4] J.-H. Hsu, C.-S. Tan, Separation of ethanol from aqueous solution by a method incorporating supercritical CO2 with reverse osmosis, J. Membr. Sci. 81 (1993) 273 /285. [5] T. Higashijima, H. Ohya, Y. Tsuchiya, H. Tokunaga, M. Aihara, Y. Negishi, Separation of supercritical fluid mixtures of CO2 and petroleum components with an asymmetric polyimide membrane, J. Membr. Sci. 93 (1994) 165 /173. [6] T. Fujii, Y. Tokunaga, K. Nakamura, Effect of solute adsorption properties on its separation from supercritical carbon dioxide with a thin porous silica membrane, Biosci. Biotech. Biochem. 60 (1996) 1945 /1949. [7] Y. Tokunaga, T. Fujii, K. Nakamura, Separation of caffeine from supercritical carbon dioxide with a zeolite membrane, Biosci. Biotech. Biochem. 61 (1997) 1024 / 1026. [8] S. Sarrade, G.M. Rios, M. Carle´s, Nanofiltration membrane behavior in a supercritical medium, J. Membr. Sci. 114 (1996) 81 /91. [9] G. Afrane, E.H. Chimowitz, Experimental investigation of a new supercritical fluid-inorganic membrane separation process, J. Membr. Sci. 116 (1996) 293 /299. [10] S. Sarrade, G.M. Rios, M. Carle´s, Supercritical CO2 extraction coupled with nanofiltration separation applications to natural products, Sep. Pur. Technol. 14 (1998) 19 / 25. [11] S. Sarrade, C. Perre, G.M. Rios, M. Carle´s, Performance of supercritical carbon dioxide extraction coupled with nanofiltration, in: G. Brunner, M. Perrut (Eds.), Proceeding of 3rd International Symposium On Supercritical Fluids, vol. 2, Strasbourg, Institut National Polytechnique de Lorraine. 17 /19 October 1994, pp. 71 /76.
C.R. Yonker et al. / J. of Supercritical Fluids 25 (2003) 225 /231 [12] M.G.E. Goemans, F.M. Tiller, L. Li, E.F. Gloyna, Separation of metal oxides from supercritical water by crossflow microfiltration, J. Membr. Sci. 124 (1997) 129 / 145. [13] R.W. Gale, J.L. Fulton, R.D. Smith, Organized molecular assemblies in the gas phase: reverse micelles and microemulsions in supercritical fluids, J. Phys. Chem. 109 (1987) 920 /921. [14] R.D. Smith, J.L. Fulton, J.P. Blitz, J.M. Tingey, Reverse micelles and microemulsions in near-critical and supercritical fluids, J. Phys. Chem. 94 (1990) 781 / 787. [15] J.L. Fulton, D.M. Pfund, J.B. McClain, T.J. Romack, E.E. Maury, J.R. Combes, E.T. Samulski, J.M. DeSimone, M. Capel, Aggregation of amphiphilic molecules in supercritical carbon dioxide: a small angle X-ray scattering study, Langmuir 11 (1995) 4241 /4249.
231
[16] K.A. Consani, R.D. Smith, Observations on the solubility of surfactants and related molecules in carbon dioxide at 50 8C, J. Supercrit. Fluids 3 (1990) 51 /65. [17] K.P. Johnston, K.L. Harrison, M.J. Clarke, S.M. Howdle, M.P. Heitz, F.V. Bright, C. Carlier, T.W. Randolph, Water-in-carbon dioxide microemulsions: an environment for hydrophiles including proteins, Science 271 (1996) 624 /626. [18] K.A. Bartscherer, M. Minier, H. Renon, Microemulsions in compressible fluids */a review, Fluid Phase Equilibria 107 (1995) 93 /150. [19] J.M. Tingey, J.L. Fulton, R.D. Smith, Interdroplet attractive forces in AOT water-in-oil microemulsions formed in subcritical and supercritical solvents, J. Phys. Chem. 94 (1990) 1997 /2004. [20] R.D. Smith, J.L. Fulton, J.P. Blitz, J.M. Tingey, Reverse micelles and microemulsions in near-critical and supercritical fluids, J. Phys. Chem. 94 (1990) 781 /787.