Characterization of the Ability of CO2 To Act as an Antisolvent for Ionic

Feb 21, 2007 - Roberto I. Canales , Christoph Held , Michael J. Lubben , Joan F. Brennecke , and Gabriele .... Yuya Hiraga , Alif Duereh , Richard L. ...
0 downloads 0 Views 800KB Size
J. Phys. Chem. B 2007, 111, 4837-4843

4837

Characterization of the Ability of CO2 To Act as an Antisolvent for Ionic Liquid/Organic Mixtures† Berlyn R. Mellein and Joan F. Brennecke* Department of Chemical and Biomolecular Engineering, UniVersity of Notre Dame, Notre Dame, Indiana 46556 ReceiVed: October 31, 2006; In Final Form: January 9, 2007

This paper discusses the ability of CO2 to induce liquid/liquid-phase separation in mixtures of ionic liquids and organics. New data for the solubility of CO2 in the ionic liquid/organic mixtures and the volume expansion of the mixtures with added CO2 are used to analyze the results. Acetonitrile, 2-butanone, and 2,2,2trifluoroethanol are chosen to distinguish dipolar and hydrogen-bonding interactions. Likewise, 1-n-hexyl3-methylimidazolium bis(trifluoromethylsulfonyl)imide, 1-n-hexyl-2,3-dimethylimidazolium bis(trifluoromethylsulfonyl)imide, 1-n-hexyl-3-methylimidazolium triflate, and ethyl-dimethyl-propylammonium bis (trifluoromethylsulfonyl)imide were studied to vary hydrogen-bond-donating and -accepting abilities of the ionic liquids. Primarily, the ability of CO2 to act as an antisolvent depends on the solubility of the CO2 in the ionic liquid/organic mixture. Strong hydrogen bonding between the ionic liquid and the organic makes it more difficult for CO2 to induce a liquid/liquid-phase separation.

Introduction The combination of ionic liquids (ILs) and carbon dioxide leads to many design possibilities for reactions and separations. ILs have several positive qualities from an environmental, safety, and design viewpoint: no significant vapor pressure, large liquidus range, generally nonflammable, tunability through cation/anion combination, and ability to solvate many different types of compounds. ILs show potential for many different types of applications, including as reaction media for homogeneously catalyzed reactions by acting as a liquid support. CO2 is a nonflammable, nontoxic, and readily available compound that offers a unique complement to ILs for reactions and separations. Supercritical carbon dioxide (SCCO2) extraction for ILs has been previously reported1,2 and has led to the development of IL/SCCO2 biphasic systems.3-7 The use of supercritical CO2 to extract products from IL solutions is limited to compounds that have significant solubility in CO2. Thus, this method is not applicable to a variety of compounds, including many ionic species and catalysts used for homogeneously catalyzed reactions. However, CO2 can still be used for separationssas a gas antisolvent. In the gas antisolvent process, a dissolved gas alters the solvent strength of the liquid that leads to supersaturation of a dissolved solute. The solute may be liquid, for which the CO2 induces a liquidliquid-phase split, or a solid, for which the CO2 causes the solid to precipitate. This method is commonly used in organic-solventbased systems for thermally sensitive products and greater control of particle size and morphology.8,9 We and others have previously shown that CO2 can be used as an antisolvent for IL/organic liquid mixtures,10-12 as well as solid solutes dissolved in ILs.13,14 In organic systems, the dissolved gas lowers the solvent strength by causing the organic liquid to greatly expand. †

Part of the special issue “Physical Chemistry of Ionic Liquids”. * Corresponding author. Tel: 574-631-5847. Fax: 574-631-8366. E-mail: [email protected].

Therefore, volume expansion data are important when designing separation systems. Badilla et al. proposed a new method for reporting the volume expansion on the basis of molar volume rather than total volume.15 This definition allows for both positive and negative relative expansion. In other words, a negative expansion with gas pressure indicates that the liquid does not expand significantly to accommodate the gas; there are simply more moles (of liquid and gas) in the same volume. Many organics show little change in relative molar volume with added CO2 because the number of moles increases but the mixture occupies a significantly larger volume. This analysis method is sensitive to small differences in molar volume, which allows easier comparison for different liquid solvents with the same antisolvent. In addition, Badilla et al. observed a minimum in the volume expansion curves, which corresponded to the point where approximately 95% of the solids were precipitated from organic liquids using CO2. Thus, they found that this definition of volume expansion is better than others used for classification and prediction of the gas antisolvent process, and it allows easier optimization of the separation process. However, on a mole fraction basis ILs do not expand much at all when exposed to CO2 pressure,16 so the gas antisolvent process in ILs is not well understood. In using CO2 as an antisolvent to separate liquid solutes from IL reaction systems, vapor/liquid/liquid equilibrium (VLLE) is an important design criterion in order to achieve optimum selectivity and yield. In addition, the VLLE information, including the volume expansion of the liquid mixtures, will be of the utmost importance for designing all types of IL/liquid separation systems involving CO2. Therefore, a systematic investigation of the antisolvent ability of CO2 with different ILs and organics, concurrent with complementary solvent strength studies based on spectroscopic probes,17 will allow prediction of CO2 pressures required for separation of specific organic solutes. This paper discusses the ability of CO2 to act as an antisolvent for IL/organic mixtures. The organics were chosen to explore

10.1021/jp0671695 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/21/2007

4838 J. Phys. Chem. B, Vol. 111, No. 18, 2007

Mellein and Brennecke

TABLE 1: ILs Used in the Current Study with Structure, Nomenclature, Source, and Purity

Figure 1. Diagram for the LCEP and K-point of an IL/organic mixture with CO2 as the antisolvent.10

the influence of different nonspecific and specific molecular interactions: acetonitrile is polar but aprotic, 2-butanone is a hydrogen-bond acceptor, and 2,2,2-trifluoroethanol is a strong hydrogen-bond donor. The main IL studied is 1-n-hexyl-3methylimidazolium bis(trifluoromethylsulfonyl)imide ([hmim][Tf2N]), the IUPAC standard IL. To further examine the importance of specific interactions between the IL and organic, three other ILs were studied. Two contain different cations, 1-nhexyl-2,3-dimethylimidazolium ([hmmim]+) and ethyl-dimethylpropylammonium ([N2113]+). [hmmim][Tf2N] has a methyl group substituted for the acidic hydrogen attached to the C2 carbon and was chosen to show the importance of cation hydrogen-bond donation. [N2113][Tf2N] is an ammonium-based cation, which is devoid of an aromatic ring and contains no particularly acidic hydrogens. The final IL has the [Tf2N]- anion replaced with triflate ([TfO]-), which is a stronger hydrogenbond acceptor than [Tf2N]- and is less coordinated with the cation than [Tf2N]-. The structures, nomenclature, and purity of ILs are shown in Table 1. We use Kamlet-Taft parameters to discuss some of the results and report new values for [N2113][Tf2N]. Kamlet-Taft parameters are measures of the dipolarity/ polarizability, hydrogen-bond-donating ability, and hydrogenbond-accepting ability of solvents. Experimental Section Here we quantify the ability of CO2 to act as an antisolvent for IL/organic systems by its ability to induce a liquid-liquidphase split. The dissolved CO2 expands the IL/organic liquid phase, until a phase split occurs. The lower critical end point (LCEP) is the temperature and pressure at which this occurs for a given initial IL/organic composition (Figure 1). At the LCEP, the homogeneous liquid phase splits into an IL-rich liquid phase and an organic-rich liquid phase, in contact with a CO2-rich vapor phase. As the CO2 pressure is increased beyond the LCEP, the organic phase continues to expand until the

organic-rich liquid phase and the CO2-rich vapor phase merge in the presence of an IL-rich liquid phase at a pressure which we show is identical to the “critical-point” of the binary organic/ CO2 mixture.10,12 In general, the “K-point” is the temperature and pressure at which the organic-rich liquid phase and the CO2-rich vapor phase merge into a single supercritical fluid phase. The pressure at which the organic-rich liquid phase and the CO2-rich vapor phase merge in the presence of an IL-rich liquid phase is independent of IL or initial IL concentration. This indicates that the concentration of IL in the organic-rich phase approaches zero as conditions approach this pressure. It is also why we use the term “K-point” throughout this manuscript to describe this pressure, even though we are fully aware that there is an IL-rich liquid phase present, as well. For each IL/organic mixture, the LCEP was measured for a mixture initially composed of approximately 10 mol % IL and 90 mol % organic. The K-point was measured once for an IL/organic combination to ensure that the K-point matched the organic/CO2 mixture critical point. This was done for some, but not all, of the mixtures to verify that there was no significant amount of IL in the vapor phase, as has been shown previously.12 The organic/CO2 binary vapor/liquid equilibria (VLE) was also measured for each organic and compared with literature data, where available. In addition, the solubility of the CO2 was recorded for each mixture measured as a function of applied pressure. When mixtures other than ∼10 mol % IL were measured, the initial concentrations of IL in the IL/organic mixture were increased in 5-10 mol % increments until no LCEP was observed. All solubility and LCEP measurements were carried out in a stoichiometric phase equilibrium apparatus, the details of which have been described elsewhere.18-20 The Span and Wagner equation of state was used to calculate the density of pure CO2 at operating conditions.21 This assumes no significant organic or IL in the vapor phase, which should not introduce any significant error in the calculation of liquid-phase compositions. A sapphire cell is used, which allows visual determination of the LCEP and K-point. In each experiment, a known amount of sample (either pure organic or IL/organic mixture) was added to the cell. As CO2 pressure was increased, the contents of the cell were manually stirred and allowed to fully equilibrate. The LCEP, where the liquid/liquid-phase split occurs, was observed visually. The mixture becomes cloudy at a pressure immediately below the LCEP, signaling the incipient phase split. Pressure was increased slightly, and the presence of a second (liquid/ liquid) meniscus verified the LCEP. The K-point was noted by the disappearance of the vapor/liquid meniscus. The Kamlet-Taft parameters were measured in the following manner. A stock solution of dichloromethane with each of the Kamlet-Taft probes was prepared. The stock solution was added dropwise to pure IL until the absorbance was ∼1. The sample

CO2 as an Antisolvent for IL/Organic Mixtures

J. Phys. Chem. B, Vol. 111, No. 18, 2007 4839

was put under house vacuum (∼25 mbar) at 40-60 °C for 1-2 days to remove the dichloromethane. The IL was then added to a cuvette, and the absorption spectra were measured. The Kamlet-Taft parameters were calculated as in previous papers.17 The organics acetonitrile (Fisher Scientific, 99.9%, or optima, 99.9+%), 2-butanone (Aldrich, 99.5+% HPLC grade), and 2,2,2-trifluoroethanol (Sigma-Aldrich, 99.5+% NMR grade) were stored over molecular sieves and filtered prior to use. The ILs [hmim][Tf2N], [hmmim][Tf2N], and [hmim][TfO] were synthesized as previously reported.16,17 [N2113][Tf2N] was used as received from EMD Chemicals. All ILs were dried under vacuum (