J . Phys. Chem. 1991, 95, 1403-1409
1403
FT-IR Investigation of the Partitioning of Sodium Bis( 2-ethylhexyi) Sulfosuccinate between an Aqueous and a Propane Phase Geary G. Ye, John L. Fulton, Jonathan P. Blitz, and Richard D. Smith* Chemical Methods & Separations Group, Chemical Sciences Department, Battelle. Pacific Northwest Laboratories, Richland, Washington 99352 (Received: March 16. 1990; In Final Form: June 29, 1990)
The partitioning of the surfactant sodium bis(2-ethylhexyl) sulfosuccinate (AOT) from a buffered aqueous phase into a near-critical propane phase was investigated by using Fourier transform infrared spectroscopy. The partitioning of AOT is shown to be dependent upon the fluid pressure as well as upon the molar water-to-surfactant ratio, W. The uptake of water into the propane microemulsion phase coincides with the partitioning of AOT. The phase behavior of this system appears to be controlled by the attractive interactions between droplets in the microemulsion phase, as well as from limitations upon the curvature of the interfacial surfactant layer. Potential applications of these systems for separations are discussed.
Introduction
Water-in-oil (wlo) microemulsions formed with the surfactant sodium bis(2-ethylhexyl) sulfosuccinate (AOT) in liquid alkanes contain spherical droplets of surfactant-encapsulated water whose diameters are typically 5-50 nm.’ In addition to being thermodynamically stable, these solutions are optically clear since structures of this size are poor scatterers of light. Water-in-oil microemulsions formed in the liquid alkanes have high affinities for hydrophilic biomolecules, and hence these solutions are good candidates for extracting these substances from bulk aqueous solutions.*-5 The selectivity of such a separation can be controlled by adjusting the pH or the ionic strength of the aqueous phase. Reverse micelle and w/o microemulsion phases are also known to readily form in supercritical and near-critical and these solutions have properties that may enhance the selectivity and efficiency of separations. Supercritical fluids are gaslike substances above their critical temperature (T,)and critical pressure (PC)?q9 whereas we define a near-critical fluid as a liquid that is at a temperature below its critical temperature, T, but above a reduced temperature ( TR = T / T J of approximately 0.75. Due to the proximity to the critical point, near-critical fluids are in a liquid state that still exhibits a great deal of compressibility in contrast to normal liquids which are far below their critical point. The density, dielectric constant, and viscosity, as well as other physical properties of a supercritical fluid, can be continuously varied between the gas- and liquid-phase limits simply by adjusting the pressure. The solvent properties of supercritical fluids, and to a lesser but still significant extent near-critical fluids, are strongly dependent upon the pressure of the system. We have previously shown that w/o microemulsions formed in ethane and propane can readily solubilize polar molecules such as dyes and high molecular weight protein^.^*^.'^ In particular, liquid propane at 25 ‘C (which may be considered a near-critical fluid) is capable of dissolving a wide range of high molecular weight biomolecules such as cytochrome c and hemoglobin.’ The transport properties of supercritical and near-critical fluids make them an ideal medium for a wide range of separation processes. For instance, the viscosities of liquid ethane and propane are 5-10 times lower than those of isooctane, resulting in micellar diffusion rates which are 5-10 times higher than their liquid analogues.I0 A preliminary study showing the potential advantages of liquid propane microemulsions for the extraction of proteins from bulk aqueous phases has been r e p ~ r t e d . ~These studies showed an abrupt increase in the partitioning of the polar solute into the liquid propane microemulsion phase occurring at around 200 bar. Small changes in the system’s pressure were utilized to control the partitioning of the solute between the propane microemulsion and a lower bulk aqueous phase. This behavior was unexpected since liquid propane is only a moderately compressible fluid. An approach to understanding the partitioning behavior of these systems
may be based upon a thermodynamic analysis of the type and structure of various possible equilibrium phases. The large number and variety of possible microstructures that AOT forms in mixtures of water and oil can be challenging from the experimentalist’s point of view and daunting from the thermodynamicist’s. A simpje binary mixture of AOT and water can form normal micelle, lamellar liquid crystal, and hexagonal liquid crystal phases.” Small additions of an oil to these phases increase the characteristic dimensions of the structures by being absorbed into the hydrocarbon tail region of the surfactant interfacial layer. On the other hand, oil-rich ternary mixtures with AOT and water form reverse micelles and water droplet-in-oil microemulsions. These dispersed droplet phases can mimic the properties of a pure liquid, having a well-defined critical point and liquid-gas phase The stability of the w/o microemulsion phases formed in near-critical and supercritical fluids is governed to a large extent by strong interdroplet attractive interactions. The magnitude of a major component of these attractive interactions, the London-van der Waals type of interdroplet force, has recently been calculated for a range of near-critical and supercritical fluids.14 Of equal importance, however, for a thermodynamic description of the equilibrium between microemulsion and aqueous phases is a description of the structural and phase changes occurring in the water continuous phase. The aim of this work is to establish the equilibrium concentrations of water and AOT in propane as a function of the pressure of the system. This information is the first step toward determining the structure of the various equilibrium phases and ultimately a thermodynamic model for prediction of phase behavior. We report here a study in which Fourier transform-infrared (FT-IR) ( I ) Luisi, P. L.; Straub, B. E., Eds. In Reverse Micelles; Plenum Press: New York, 1984. (2) Fletcher, P. D. I.; Parrott, D. J . Chem. SOC.,Faraday Trans. I 1988. 84, 1131-1 144. (3) Luisi, P. L. Angew. Chem., Int. Ed. Engl. 1985, 24, 439-450. (4) Goklen, K. E.; Hatton, T. A. Biotechnol. frog. 1985, I , 69-74. (5) Shev, E.; Goklen, K. E.; Hatton, T. A.; Chen, S.-H. Biotechnol. Prog. 1986, 2. 175-186. (6) Gale, R. W.; Fulton, J. L.; Smith, R. D. J . Am. Chem. Soc. 1987, 109, 920-921. (7) Smith, R. D.; Fulton, J. L.;Blitz, J. P.; Tingey, J. M. J . Phys. Chem. 1990, 94, 781. (8) Paulaitis, M. E.; Penninger, J. M. L.; Gray, R. D.; Davidson, P., Eds.
In Chemical Engineering at Supercritical Fluid Conditions; Ann Arbor Science: Ann Arbor, MI, 1983. (9) Penninger, J. M. L.;Radosz, M.; McHugh, M. A.; Krukonis, V. J., Eds. In Supercritical Fluid Technology; Elsevier: Amsterdam, 1985. (IO) Smith, R. D.; Blitz, J. P.; Fulton, J. L. In Supercritical NuidScience ond Technology; Johnston, K. P., Penninger, J. M. L., Eds.; ACS Symposium Series 406; American Chemical Society: Washington, DC, 1989. ( I I ) Kunieda, H.; Shinoda, K.J . Colloid InterfaceSci. 1979, 70,577-583. (12) Kotlarchyk. M.; Chen. S.-H.; Huang, J. S.;Kim, M. W . Phys. Reu.
‘To whom correspondence should be addressed.
0022-365419 112095-1403$02.50/0
0 1991 American Chemical Society
Yee et al.
1404 The Journal of Physical Chemistry, Vol. 95, No. 3, I991 TABLE I: Surfactant and Aqueous Buffer Concentrations of the AOT Partitioning Experiments" expts
4s
a
1.95 X IOm2 1.91X10-2 1.97 X 1.87 X 1.62 X 1.34 X 7.24 X
b c d e
f g
4w
W
[AOT], mM
1.32X10-2 2.25 X 6.67 X 2.00 X IO-' 3.33 X IO-' 2.83 X IO-I
2.0 15.6 25.7 79.9 278 560 87.9
48.2 47.9 49.8 49.7 50.0 49.6 249.9
0
f--
-2
beam
s
f1
__-___ I
r-----
" 6 , is the volume fraction of the surfactant, and bWis the volume fraction of the 0.1 N sodium phosphate buffer (pH = 7). W is the resulting ovcrall system water-to-surfactant molar ratio for the total amount or buffer and surfactant added to the two-phase system. [AOT] is the surfactant concentration expressed in terms of the moles of surfactant added per liter of propane. For all experiments, T = 25 O C and P = 10-400 bar. V
spectroscopy is used to study the partitioning process. The unusual pressure-dependent phase behavior of this system can be attributed to changes in the microstructure of surfactant aggregates in both the aqueous and propane phases, as well as to changes in the interaggregate attractive forces arising from the variable solvent properties of the near-critical propane phase.
Experimental Section The surfactant sodium bis(2-ethylhexyl) sulfosuccinate, AOT (Fluka, >98% "purum" grade), and propane (Linde, CP grade) were used as received. The lower aqueous phase contained a 0.1 N sodium phosphate buffer solution having a pH of 7. This solution was prepared from NaH2P0,.H20 (EM Science, EM Industries), Na2HP04.7H20(EM Science, EM Industries), and water which was distilled and filtered through a Millipore Milli-Q system. The sodium phosphate buffer was used to prevent the formation of a viscous propane emulsion which forms at the higher overall water contents. The use of a buffer also negates the need to further purify the AOT since the main goal of such purification steps is to remove salts. We did however repeat selected experiments (experiment c in Table 1) using AOT which was purified by using the method of K0t1archyk.l~ The results obtained with the purified AOT were within 15% of the results obtained with the "unpurified" AOT. Determination of the concentrations of three substances (propane, AOT, and water), varying greatly in volatility, polarity, and solubility, is a difficult analytical task, particularly when the systems are under high pressure. Our method of choice is an in situ, W-IR spectroscopic technique. A primary advantage of the technique is that a sample does not have to be removed from the system for analysis. Hence, there are no perturbations of the solution concentration, and potential systematic errors are greatly reduced. The concentration of each of the three components can be determined directly from the peak heights or areas of vibrational bands specific to each analyte of interest. We have constructed a reflectance cell for these measurements rather than use a mechanically simpler total internal reflectance cell so as to avoid surface-related solvation effects which are especially important for these highly compressible solutions. A Nicolet 740 FT-IR spectrometer (Nicolet Analytical Instruments) purged with dry nitrogen was used to obtain all infrared spectra. The spectrometer was equipped with a germanium-onKBr beam splitter and a mercury-cadmium-telluride detector. We co-added 128 scans at 4-cm-' wavenumber resolution to obtained the desired signal-to-noise ratio. The IR cell, schematically illustrated in Figure 1, was a high-pressure stainless steel reflectance-type cell containing two window parts and an IR mirror mounted on the stem of a high-pressure valve housing (High Pressure Equipment Co.). The IR window port contained a 1.27-cm-thick by 2.54-cm-diameter zinc sulfide window which was sealed to the metal block by a 1.27-cm-diameter gold-plated (15) Kotlarchyk, M.;Chen, S.; Huang, J. S.; Kim, 1984, 29, 2054-2069.
M.W.Phys. Reo. A
L
I
I
1
Figure 1. Schematic of the high-pressure IR reflectance cell containing the following components: ml, a stainless steel mirror with an optical finish; m2, beam steering mirrors; w,an IR and an observation window; s, sample chamber area; f l , sample fill port; f2, fluid fill port; v, highpressure valve housing. b
Ia I
r
-rr
I
Figure 2. Apparatus used for partitioning process: (a) magnetically coupled gear pump for fluid circulation, (b) IR reflectance cell, (c) view cell, (d) syringe pump, (e) pressure transducer, and (9 discharge valve. metal V-ring seal (Parker, No. 8812-2001-0050). The other window port contained a sapphire window with a 2.54-cm-diameter seal. This window provides a separate means of viewing the sample to determine the number of phases. A highly polished stainless steel mirror (8.47-mm diameter) was attached to the stem tip of the high-pressure valve. The mirror tip was gimbaled to the valve stem so as to facilitate proper alignment between the mirror and window surface. By moving the mirror up to contact the window the mirror was rotated so that it was parallel with the window. The mirror was then retracted to the desired path length. The valve housing allowed for adjustment of the path length from 0 to 8750 wm while the system was under pressure. The cell had a volume of 10.0 mL and had a maximum pressure rating of 400 bar. A 10 X 3 mm Teflon-coated, magnetic stir bar was placed directly in the IR cell so that the solution could be stirred while the cell was mounted in the FT-IR instrument. The experimental arrangement used in the study of the partitioning process is shown in Figure 2. Briefly, the apparatus consisted of (a) a magnetically coupled gear pump (Micropump, No. 182-346) for fluid circulation, (b) the high-pressure IR cell placed in the sample compartment of the FT-IR spectrometer, (c) a 44" high-pressure stainless steel view cell which allowed for mixing and visual determination of the phases of the sample, and (d) a high-pressure syringe pump (Varian 8500) for the
Partitioning of AOT between Water and Propane introduction of pure propane. The cells and pumps were connected via stainless steel tubing. A high-pressure transducer (Setra Systems Inc., Model 3OOC) was used to monitor the fluid pressure. The total volume of the apparatus, excluding the syringe pump, was 60 mL. The temperature of the reflectance cell, the view cell, and the associated tubing was maintained at 25 OC using threemode controllers with platinum resistance probes (Omega, NO. N2001). The temperature was also monitored with a platinum resistive thermometer (Fluka, No. 2 108A). The sample preparation procedure was as follows. Known quantities of AOT and buffer solution were initially introduced into the view cell. Propane at 3-6 bar was then used to purge air from the apparatus. The system was then filled with propane to the desired pressure. The contents of the view cell were vigorously stirred with a magnetic stir bar for IO min and allowed to settle for I 5 min. After settling, the top propane phase was circulated to the IR reflectance cell by means of the magnetically coupled gear pump for 5 min. The fluid circulation speed was low enough that the bottom aqueoussolid phase was not disturbed. The stirringsettling-circulating process was performed twice at each pressure setting and was then followed by IR spectral measurements. The partitioning experiments for runs a-c listed in Table I were carried out using only the 1R reflectance cell since in these cases the volumes of the aqueous phases were below the level of the IR beam. Experimental methods were developed to obtain reliable spectra from the present reflectance cell arrangement. Since a primary reflection from the first surface of the IR window could not be isolated optically from the internal mirror reflection, a subtraction method was used to remove this unwanted reflection component from the sample spectrum. The absorbance spectra of the samples in the reflectance cell were determined from the expression where A is the absorbance, I, is the detector response with sample present in the cell, I , is the detector response with no sample and with the mirror removed from the cell, and Ib is the detector response with no sample present but with the mirror at the desired path length position. The factors a and @ correct for reflections off of the second window surface. The I, component is primarily due to a reflection off of the first surface of the IR window. It also contains a much smaller component (about 5%) from an internal reflection off of the inside window surface. The magnitude of this internal reflection is dependent upon the dielectric constant of the fluid in the cell. To maintain linearity above an absorbance of about 0.5, a scaling factor, a, must be included to correct for this secondary reflection. a can be determined from regions of total absorbance in the sample spectra. Furthermore, as the dielectric constant of the sample solution is increased, the transmission of the beam through the sample increases due to a reduction in the internal reflection of the IR window. A second scaling factor, @, determined from a spectral region with no absorbances, can be utilized to ensure that the spectra has the proper absolute scale. With two-phase systems there is an inherent danger with spectroscopic studies that a small film or droplets of the second phase may coat the internal surfaces of the windows or mirror, which would interfere with measurements of the upper phase. This is especially true for the surfactant systems, which, because of their low interfacial tensions, thin films can form on the surfaces which are difficult to detect by visual observation. A differential sampling technique was utilized to eliminate contributions from impurities on the window or mirror. This was done by acquiring IR spectra at two slightly different path lengths for each pressure and tempcrature studied. The difference between the spectral measurements resulted in the true absorbance spectra of the upper phase solution. In all the spectral subtractions, the bands under consideration were kept below 0.5 absorbance units to ensure linearity between absorbance and concentration.I6 This was (16) Griffiths, P. R.; de Haseth, J . A. In Fourier Transform Infrared Speclromelry; John Wiley & Sons: New York. 1985; Chapter IO.
The Journal of Physical Chemistry, Vol. 95, No. 3, 1991 1405
1
’
Wavrnumbrr
Figure 3. IR spectra at 150 bar, 25 OC of (A) pure propane and of (B) a 50 m M AOT, propane microemulsion with W,,, = 15. The absorption bands are assigned as follows: (a) propane combination bands, (b) 0-H stretch of water, (c) C-H stretches, (d) C=O stretch of AOT, (e) 0-H bend of water, (f) C-H deformations, (g) SOY band of AOT, and (h) the C-C skeletal modes.
accomplished by adjusting the path length (Le., adjusting the mirror position). The concentrations of AOT and water in the upper propane phase were determined by measurement of the peak heights of characteristic IR absorption peaks of the AOT and water. A multiple regression approach was used to fit a linear two-parameter model of the form yj = A, + BjxI + Cjx2 (2) where y I and y 2 are the concentrations of AOT or H 2 0 , respectively; Ai, B,, and Ci are the regressed coefficients; xIis the ratio of the absorbance at 1735 cm-I (the AOT carbonyl stretching mode) to the absorbance of an overtone band of a carbon-carbon skeletal vibration of propane” at 2235 cm-’; and x2 is the ratio of the absorbance at 1645 cm-I (the 0-H bending mode of H 2 0 ) to the propane overtone band at 2235 c d . The coefficients A,, B,, and C,of the regression equations were determined by p r e p aration of six different single-phase solutions with various nominal AOT concentrations ([AOT] = 10, 50, 250 mM) and molar water-to-surfactant ratios ( W, = 5 , 15, 25). W, is defined as the water-to-surfactant ratio in the propane continuous microemulsion phase and is only equal to the overall system waterto-surfactant ratio, W, for single-phase systems. Figure 3 shows the IR spectrum of both pure propane and a 50 mM AOT propane solution with W,,, = 15 and a pressure of 150 bar. The various bands in the region from 6000 to 480 cm-I are identified. Figure 4 shows an expanded portion of the spectra in Figure 3 in which the three bands used in the regression analysis are indicated. The measured AOT and water concentrations were corrected for the changes in the density of the propane continuous phase solvent, which increased about 12.5%from 10 to 400 bar. The total error for in situ measurements of AOT and water concentrations is approximately 10-1 5%. The precision of measurements by this method is somewhat better and is estimated to be *I%. Two other methods of analysis were attempted. They were a deconvolution process (Fourier self-deconvolution program, from Nicolet) and a spectral band subtraction method. Both involved resolving the 1735-cm-I AOT band and the 1645-cm-’ H 2 0 band. However, the results obtained from these methods were inferior to the simpler, regressed peak-height fits.
Results The phase behavior of propane/AOT/buffered water solutions is strongly dependent upon the system pressure and the overall ( 1 7) Szymanski, H. A. In Theory and Practice of Infrared Spectroscopy; Plenum Press: New York, 1964; Chapter 5.
Yee et ai.
1406 The Journal of Physical Chemistry, Vol. 95, No. 3, 1991 60
1I
A
b
50
40
b
Y 0
c
d
e
l
0
0
E
f:-
..c
9
30
ti
s 20
f 10
2i99
2k05
2111
1623 1629 Wavenumber
2617
1f35
16'tl 3 9 7
Figure 4. Expanded region of the IR spectra shown in Figure 3 for (A) pure propane and for (B) the 50 mM AOT propane microemulsion with W,,, = 15. The bands used in the multiple regression analysis for the determination of water and AOT concentrations are the (a) propane overtone band at 2235 cm-], (b) the AOT carbonyl stretch at 1735 cm-', and (c) the 0 - H bending mode of water a t 1645 cm-I.
water and surfactant concentrations. At the vapor pressure of propane where the density is 0.49 g/cm3 (about IO bar at 25 "C), AOT is soluble to greater than 30 vol %. Addition of small amounts of water ( W > 7) to a propane/AOT solution causes a phase separation. The maximum amount of water that can be dissolved into a single-phase propane microemulsion depends upon the system pressure. For instance, the maximum molar waterto-AOT ratio, Wo, increases from 7 at I O bar to 36 at 250 bar' and has been found to exceed 100 at 500 bar.I4 At intermediate water concentrations, 7 < W < 50, a two-phase system exists at low pressures but changes to a clear single-phase solution at higher pressures. Additions of large amounts of water eventually exceed the carrying capacity (Le., Wo) of the microemulsion phase so that a second aqueous phase is present at all pressures. For systems which contain more than 20 vol % water, an aqueous phase exists at all pressures, and the visual appearance of this aqueous phase was found to change with pressure. The phase behavior of these systems is summarized as follows. With a gaseous propane upper phase at 1 bar, the lower aqueous phase is an opaque, viscous solution containing the surfactant and buffer. At these pressures very little of the propane has partitioned into the lower aqueous phase. The lower aqueous phase under these conditions actually contains a biphasic mixture of water containing AOT normal micelles and an AOT liquid crystalline phase."~'*J9 At pressures above IO bar, AOT and water partition increasingly into the upper propane phase (as will be discussed in the following section) and the lower aqueous phase eventually becomes entirely clear at propane pressures above 350 bar. Under these conditions the lower aqueous phase may contain a low concentration (