Anal. Chem. 2004, 76, 4684-4689
FT-IR Studies of Acetylacetonates in Supercritical CO2 Using a Capillary Cell at Pressures up to 3.1 kbar Matthew C. Henry and Clement R. Yonker*
Chemical Sciences Division, Pacific Northwest National Laboratory, P.O. Box 999, MS P8-19, Richland, Washington 99352
Supercritical fluids in general and supercritical carbon dioxide in particular are the subjects of intensive research as less toxic “green” solvents. Supercritical CO2 (SC CO2) is used in catalysis,1,2 polymer processing,3,4 separations,5-7 and microelectronics8 according to recent reviews. One of the advantages of SC CO2 as solvent is the fact that its solvating power may be “tuned” by changing the pressure and temperature. Current spectroscopic techniques are able to explore a wide range of temperatures in SC CO2, but the pressures accessible in conventional highpressure IR cells are typically limited to less than 1 kbar.9-13 Higher pressures can be obtained in diamond anvil cells14 and
using hydraulic compression in “pill-box” type cells,15 but these techniques are difficult to apply to highly compressible fluids such as CO2. Yonker and co-workers have used fused-silica capillary cells in NMR to study pressures up to 4 kbar.16 In this work, we extend the use of fused-silica capillaries to Fourier transform infrared (FT-IR) spectroscopy. Typically quartz has a lowfrequency cutoff of 3000 cm-1 in the infrared,17 but the high-purity fused silica used in manufacturing the capillary is transparent to 2000 cm-1, with some transmittance to 1500 cm-1 (vide infra). This increased transparency permits spectra to be acquired in the CH and OH stretch region. With the fused-silica capillary mounted on the stage of a FT-IR microscope, IR spectra of SC CO2 solutions over a wide range of temperatures and pressures may be acquired. The capillary cells used are quite inexpensive (∼$10/m). The capillary cell is disposable, making it attractive for studying corrosive or contaminated fluids that would quickly render a conventional IR cell unusable. The small volume of the capillary also reduces sample size requirements and reduces the risks associated with large volumes of fluid under high pressure. In this work, the capillary cell and FT-IR microscope were used to study β-diketones as neat liquids and in SC CO2 solution. The β-diketone 2,4-pentanedione (AcAc) and its trifluoro (trifluoroacetylacetone, TFA) and hexafluoro (hexafluoroacetylacetone, HFA) analogues shown in Figure 1 are commonly used chelating agents for solubilizing metal ions in supercritical CO2.18-21 The β-diketones are interesting because they exist in tautomeric equilibrium between keto and enol forms. Keto-enol tautomerism has been studied in AcAc using NMR techniques (in liquid,22-26
* To whom correspondence should be addressed. E-mail: clement.yonker@ pnl.gov. (1) Schneider, M. S.; Grunwaldt, J. D.; Burgi, T.; Baiker, A. Rev. Sci. Instrum. 2003, 74, 4121-4128. (2) Hyde, J. R.; Licence, P.; Carter, D.; Poliakoff, M. Appl. Catal., A 2001, 222, 119-131. (3) Ajzenberg, N.; Trabelsi, F.; Recasens, F. Chem. Eng. Technol. 2000, 23, 829-839. (4) Hay, J. N.; Kahn, A. J. Mater. Sci. 2002, 37, 4841-4850. (5) Lang, Q. Y.; Wai, C. M. Talanta 2001, 53, 771-782. (6) Sarrade, S.; Guizard, C.; Rios, G. M. Sep. Purif. Technol. 2003, 32, 57-63. (7) Erkey, C. J. Supercrit. Fluids 2000, 17, 259-287. (8) Weibel, G. L.; Ober, C. K. Microelectron. Eng. 2003, 65, 145-152. (9) Fulton, J. L.; Yee, G. G.; Smith, R. D. J. Am. Chem. Soc. 1991, 113, 83278334. (10) Blitz, J. P.; Fulton, J. L.; Smith, R. D. Appl. Spectrosc. 1989, 43, 812-815. (11) Hoffmann, M. M.; Addleman, R. S.; Fulton, J. L. Rev. Sci. Instrum. 2000, 71, 1552-1556. (12) Andersen, W. C.; Sievers, R. E.; Lagalante, A. F.; Bruno, T. J. J. Chem. Eng. Data 2001, 46, 1045-1049.
(13) Rack, J. J.; Polyakov, O. G.; Gaudinski, C. M.; Hammel, J. W.; Kasperbauer, P.; Hochheimer, H. D.; Strauss, S. H. Appl. Spectrosc. 1998, 52, 10351038. (14) Schettino, V.; Bini, R. Phys. Chem. Chem. Phys. 2003, 5, 1951-1965. (15) Zahl, A.; Igel, P.; Weller, M.; Khoshtariya, D. E.; Hamza, M. S. A.; van Eldik, R. Rev. Sci. Instrum. 2003, 74, 3758-3762. (16) Wallen, S. L.; Yonker, C. R.; Phelps, C. L.; Wai, C. M. J. Chem. Soc., Faraday Trans. 1997, 93, 2391-2394. (17) Grunwaldt, J. D.; Wandeler, R.; Baiker, A. Catal. Rev.-Sci. Eng. 2003, 45, 1-96. (18) Laintz, K. E.; Wai, C. M.; Yonker, C. R.; Smith, R. D. Anal. Chem. 1992, 64, 2875-2878. (19) Laintz, K. E.; Hale, C. D.; Stark, P.; Roquette, C. L.; Wilkinson, J. Anal. Chem. 1998, 70, 400-404. (20) Lin, Y. H.; Wai, C. M. Anal. Chem. 1994, 66, 1971-1975. (21) Wai, C. M.; Wang, S. F. J. Chromatogr., A 1997, 785, 369-383. (22) Burdett, J. L.; Rogers, M. T. J. Phys. Chem. 1966, 70, 939-941. (23) Burdett, J. L.; Rogers, M. T. J. Am. Chem. Soc. 1964, 86, 2105. (24) Allen, G.; Dwek, R. J. Chem. Soc. B 1966, 161-163. (25) Jarrett, H. S.; Sadler, M. S.; Shoolery, J. N. J. Chem. Phys. 1953, 21, 2092.
The keto-enol equilibria of the β-diketones acetylacetone, trifluoroacetylacetone, and hexafluoroacetylacetone were determined using Fourier transform infrared spectroscopy in a novel high-pressure capillary cell. Acetylacetone and its fluorinated analogues were studied as neat liquid and as supercritical CO2 solutions at pressures up to 3.1 kbar. The keto form was found to be favored at high pressure and low temperature. The change in partial molar volume and enthalpy between the keto and enol forms was determined for the acetylacetone and trifluoroacetylacetone. Under all conditions studied, only the enol form of hexafluoroacetylacetone was observed. Based on the thermodynamic data obtained, there appears to be no advantage gained in conducting metal extractions at high pressures and low temperatures using acetylacetone or trifluoroacetylacetone.
4684 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
10.1021/ac049451i CCC: $27.50
© 2004 American Chemical Society Published on Web 07/14/2004
Figure 1. β-Diketones used in this study. AcAc ) acetylacetone, TFA ) trifluoroacetylacetone, and HFA ) hexafluoroacetylacetone. Note that for AcAc and HFA forms II and III are equivalent.
gas phase,27 and solution28) and using ab initio calculations.29 The tautomeric equilibrium of AcAc and its analogues has also been studied in SC CO2 solution at low pressures using IR spectroscopy30 and at high pressures using NMR spectroscopy.16 The equilibrium between keto and enol forms is of interest in predicting the chelating ability of acetylacetonates dissolved in SC CO2. The formation of an enolate anion is an important step in the chelation of metal cations.20 The enol form of the tautomer is favored by increasing fluorination16 (HFA > TFA > AcAc) and decreasing temperature.22 The role of pressure in shifting the tautomeric equilibrium is uncertain. In a study at low total pressures, the keto form was reported to be favored as pressure was increased, due to the preferential solvation of the less polar keto form by CO2 as the density was increased.30 However, other NMR studies show no detectable change in equilibrium with increased pressure.16 The pressure dependence of the equilibrium can be related to the change in partial molar volume between the keto and enol forms. The equilibrium constant Keq is given by eq 1, where A
Keq )
[keto] Aketo/keto ) [enol] Aenol/enol
(1)
and are the absorbance and molar absorptivity of the keto and enol species. The equilibrium constant can be related to the change in partial molar volume (∆V) by
δ ln Keq/δP ) -∆V/RT
(2)
where R is the ideal gas constant, T is the absolute temperature, and P is the absolute pressure. By measuring changes in the relative concentration between the keto and enol species (i.e., the mole fraction) instead of absolute concentration, the compressibility of the fluid does not need to be taken into account, assuming that the ratio of the molar absorptivities of the two species remains constant. By keeping the pressure of the system constant and (26) Jouanne, J.; Heidberg, J. J. Magn. Reson. 1972, 7, 1-4. (27) Folkendt, M. M.; Weiss-Lopez, B. E.; Chauvel, J. P.; True, N. S. J. Phys. Chem. 1985, 89, 3347-3352. (28) Rogers, M. T.; Burdett, J. L. Can. J. Chem. 1965, 43, 1516-1526. (29) Ishida, T.; Hirata, F.; Kato, S. J. Chem. Phys. 1999, 110, 3938-3945. (30) Yagi, Y.; Saito, S.; Inomata, H. J. Chem. Eng. Jpn. 1993, 26, 116-118.
Figure 2. Experimental apparatus. V1-4 valves, SP1-2 syringe pumps, RD1-3 rupture disks, and PX1-2 pressure transducers.
varying the temperature, the enthalpic and entropic contributions to the free energy change during tautomerization can be determined. The well-known Van’t Hoff equation
ln Keq )
-∆H 1 ∆S + R T R
()
(3)
can be used to determine the change in enthalpy (∆H) and entropy (∆S) from a plot of the natural log of the equilibrium constant (Keq) versus 1/T, where T is the absolute temperature and R is the ideal gas constant. In this work, we used the capillary cell to obtain thermodynamic information about the equilibrium between the keto and enol forms of AcAc and its fluorinated analogues. Since the enthalpy is determined from the slope of the ln Keq plotted as a function of temperature, it is not necessary to know the absolute concentration of each species, only their relative concentrations. The entropy term requires calculation of absolute concentrations, since it is dependent on the magnitude of the intercept of a plot of ln Keq versus temperature. In the capillary cell, it is difficult to ensure thorough mixing of the sample with SC CO2 to obtain a precisely known concentration throughout the length of the capillary, so entropy values could not be calculated. The concentration within the window region of the capillary was stable for over 24 h, with no evidence of concentration changes due to diffusional mixing. EXPERIMENTAL SECTION Reagents. Acetylacetone (99%), carbon tetrachloride (99.9%), and trifluoroacetylacetone (99%) were used as received from Aldrich. Hexafluoroacetylacetone was used as received from Lancaster Synthetics (Pelham, NH). The carbon dioxide used was SFC grade (Scott Specialty Gases, Plumsteadville, PA). Instrumentation. The experimental apparatus is shown in Figure 2. All tubing between the tank and V3 is 1/16-in. stainless steel; all tubing between V3 and the capillary is 1/8-in. stainless steel. All valves and fittings between the CO2 tank and V3 are Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
4685
rated to 1.0 kbar, and all fittings after V3 are rated to 4.13 kbar. Valves and fittings were obtained from High-Pressure Equipment Inc. (Erie, PA). Pressure was controlled using two syringe pumps. Pump SP1 was an Isco 260D syringe pump (Isco, Inc., Lincoln, NE) while pump SP2 was a High-Pressure Equipment 10-cm3 pressure generator. The pumps were operated in two stages to generate pressures up to 3.1 kbar. The system was pressurized using SP1 until the pressure reached the operation limit of 517 bar. Valve V3 was then closed and the system further pressurized using pump SP2. System pressure was monitored using pressure transducers PX1 and PX2. Although the volume of CO2 contained in the capillary represents a negligible hazard even at 3.1 kbar, the valves and piping contain enough CO2 to present a safety hazard. All closed spaces within the experimental system are protected by appropriately sized rupture disks to prevent uncontrolled ruptures from occurring. Rupture disks RD1 and RD2 are rated to 0.6 kbar, and RD 3 is rated to 4.0 kbar. The sample cell used is a 1-m length of 355-µm-o.d., 103-µmi.d. polyimide-coated fused-silica capillary (Polymicro Technology, Pheonix, AZ). One end of the capillary was flame sealed using a jeweler’s torch. The exposed fused silica was coated with poly(ethyl acrylate) to protect it from damage. A 0.5-cm window was burned in the polyimide coating ∼7 cm from the sealed end of the capillary and cleaned using 2-propanol. The open end of the capillary was fitted into a Vespel ferrule (Alltech) and glued in place using cyanoacrylate adhesive. All failures of capillaries during experiments resulted from failures at the ferrule rather than failure of the capillary itself. Best results were obtained when the adhesive on the ferrule was allowed to cure while the ferrule was compressed in the steel fitting. The capillary was inserted into an aluminum heating block and mounted on the microscope stage. The heating block was equipped with two 50-W cartridge heaters and two RTD temperature sensors. Temperature was maintained by a temperature controller (Watlow; Winona, MN). The temperature can be adjusted between ambient temperature and 270 °C. FT-IR spectra were recorded on a Bruker IFS-66V spectrophotometer using an IRScope-I FT-IR microscope equipped with a liquid nitrogen-cooled MCT detector. Data processing was carried out using the Bruker OPUS 3.2 software. All spectra are the average of 1024 scans. The IR scope was equipped with a 34×, 0.4 NA IR lens and a 0.75-mm aperture, producing an observed spot size of 50 µm. The optimal spot size was determined by acquiring blank spectra of a capillary containing only CO2 with successively smaller apertures until a flat baseline was achieved. Too large an aperture causes baseline artifacts due to focusing of the IR beam by the capillary. Samples were loaded into the capillary using a vacuum fill technique. The open end of the capillary was inserted through a rubber septum into a sample vial containing the sample. The open end of the capillary was held below the level of the liquid, while a syringe needle was inserted into the headspace above the sample. The needle was withdrawn to create a slight negative pressure within the sample container. This negative pressure was maintained until enough air had been expelled from the capillary to allow the sample to fill a section of the capillary when the vacuum was released. 4686
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
To assist in the assignment of peaks in the CH stretch region, spectra of AcAc and TFA diluted in CCl4 were collected using a 0.1-mm-path length cell equipped with NaCl windows. The AcAc concentration was varied between 14 and 100 mol %, and the concentration of TFA was varied between 16 and 100 mol %. The data acquisition conditions are as described for the capillary cell. RESULTS AND DISCUSSION Figure 3 shows the FT-IR spectra of liquid AcAc, TFA, and HFA and their CO2 solutions, along with the spectra of the empty capillary. The fused-silica capillary is reasonably transparent between 2000 and 4000 cm-1, allowing spectra to be obtained in the CH and OH stretch regions. The CO2 combination bands do not significantly overlap the OH and CH stretch bands of the acetylacetonates. The peaks of primary interest are those for keto CO stretch overtone (3415 cm-1), enol CH stretch (2965 cm-1), and keto CH stretch at (2934 cm-1). Peak assignments were made by comparing the relative intensities of the peaks in pure TFA or AcAc to the intensities of their CCl4 solutions (see Supporting Information). The equilibrium of AcAc is known to shift toward the enol form upon dilution in CCl4, and the trend is assumed to be similar for TFA.28 Peaks that increase in intensity upon dilution are assumed assigned to the keto form, while those that decrease are assigned to the enol form. The ratio of the CH stretch intensity to either the enol CO overtone or CH stretch intensity is proportional to the equilibrium constant, Keq, of the keto-enol equilibrium as shown in eq 1, assuming the ratio of the molar absorptivity of the two species remains constant. Infrared spectra for the empty capillary, pure CO2, and solutions of AcAc, TFA, and HFA are shown in Figure 3A. Liquid AcAc, solid AcAc (frozen at 3.1 kbar), liquid TFA, and liquid HFA are shown in Figure 3A. The dashed line in the AcAc spectrum in Figure 3B represents AcAc at 25 °C and 3.1 kbar. Under these conditions, the sample was observed to freeze in the capillary. The spectrum of the frozen AcAc shows sharper peaks for all of the CH stretch peaks relative to the liquid, and the intensity of the enol CH is increased relative to the liquid state. This may indicate that the enol form is favored in the solid phase. It is interesting to note that the keto CO overtone peak at 3415 cm-1 is sharp in the liquid AcAc sample, quite broad in liquid TFA, and scarcely distinguishable as a peak in the liquid HFA sample. The broadening of the peak may indicate that the carbonyl group participates in hydrogen bonding with the more numerous enol OH groups, forming multimers. The same trend is observed for acetylacetonates in CO2 solution. The broadening of the CO peak in the TFA spectra may also be due to the existence of two nonequivalent carbonyls in the keto form of TFA (see Figure 1). Figure 4 shows the difference spectra obtained for AcAc as the neat liquid (A) and in SC CO2 solution (B) as the pressure is varied between 0.1 and 3.1 kbar. The intensity of the CO overtone and CH2 peaks increases with increasing pressure, while the CH stretch peak diminishes in intensity. This result supports the conclusion that the keto form of AcAc is more stable at high pressures. The trend is similar for both neat liquid and SC CO2 solution, indicating that the shift in equilibrium is due to pressure alone and not due to changes in CO2 properties with changes in density. Similar trends are observed for TFA, while HFA is exclusively in the keto form under all conditions studied (data not shown). It is interesting to note that the increase in the CO
Figure 4. Difference spectra of AcAc liquid (A) and AcAc CO2 solution (B). Spectra were calculated by subtracting the 0.1 kbar spectrum from each higher pressure spectrum. Pressure was increased from 0.1 to 3.1 kbar at a constant temperature of 36 °C.
Figure 3. (A) FT-IR spectra of the fused-silica capillary: SC CO2, AcAc in SC CO2, TFA in SC CO2, and HFA in SC CO2 (V). (B) Neat liquids in capillary: AcAc, TFA, and HFA. The dashed line in the AcAc spectrum is AcAc frozen in the capillary at 25 °C and 3.1 kbar. All other spectra were recorded at 35 °C and 0.1 kbar.
band intensity is commensurate with the increase in band intensity at 3160 cm-1, leading to the tentative assignment of this peak to the CH2 stretch. However, a strong peak occurs at approximately the same location in the TFA and HFA samples, possibly indicating that the CH2 peak is overlapped with the enol CO overtone and OH stretch in these samples. Figures 5 shows the equilibrium constant, Keq, plotted as a function of temperature for liquid AcAc (A) and AcAc in sub- and supercritical CO2 (B). The slope of these plots is related to the difference in partial molar volume between the keto and enol forms, as shown in eq 2. The values obtained are summarized in Table 1. For liquid AcAc, these values agree, within the limits of experimental error, with the value of -4.7 ( 0.7 cm3/mol obtained by NMR.26 For CO2 solutions of AcAc the values are ∼3-4 times larger than the values obtained for liquid AcAc, possibly indicating that the enol species exists in CO2 as dimer, trimer, or tetramer clusters, while the keto form exists as isolated molecules. An alternate explanation is that there is a local solvent density enhancement around the solute in SC CO2, and that the observed change in volume is a combination of the volume change due to tautomerization and the volume change due to local changes in solvent density. Local density enhancements have been observed Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
4687
Table 1. Change in Partial Molar Volume between Keto and Enol Tautomers of AcAc and TFA and Enthalpy of Tautomerization for AcAc in CO2 Solution Determined Using Eq 3 for Temperatures between 22 and 150 °C
AcAc ∆V (cm3/mol) TFA ∆V (cm3/mol)
AcAc ∆H(kcal/mol)
liquid CO2 soln liquid CO2 soln
CO2 soln
22 °C
36 °C
66 °C
-2.4 ( 1.8 -11.9 ( 1.8 -6.8 ( 4.5 -16.3 ( 4.8
-5.7 ( 1.6 -14.4 ( 1.3 -5.0 ( 2.4 -23.1 ( 8.7
-5.6 ( 1.4 -16.8 ( 1.5 -7.6 ( 2.7 -26.4 ( 7.8
0.1 kbar
0.17 kbar
0.35kbar
1.7 kbar
2.7 ( 0.3
2.8 ( 0.2
2.7 ( 0.3
2.5 ( 0.2
solution is again larger than the values obtained in the liquid state. Enhancement in local solvent density should be greater for TFA, due to the more favorable solvation of the CF3 group by CO2. However, the presence of the CF3 group may also increase intermolecular hydrogen bonding, so it is not possible to determine the relative contribution of conformational change and local solvent density to the observed change in partial molar volume. Van’t Hoff plots for AcAc in supercritical CO2 solution were used to determine the enthalpy of tautomerization (see Supporting Information for plots). As discussed previously, only the enthalpy for tautomerization is calculated since it depends on the relative concentrations of keto and enol species. Calculation of the entropy of tautomerization would require knowledge of the absolute concentration of keto and enol forms, which is difficult to determine due to presumed variations in concentration from incomplete mixing along the length of the capillary. The enthalpy of tautomerization for AcAc in CO2 is shown in Table 1. Yonker and co-workers reported a value of 3.2 ( 0.3 kcal/ mol for AcAc under similar conditions, in good agreement with the values reported above.16 The equilibrium data for TFA in CO2 solution were not of sufficient quality to allow the determination of enthalpy values. In the case of HFA, neither temperatures of 22-150 °C nor pressures up to 3.1 kbar produced any detectable change in the keto-enol equilibrium. The enol form was favored under all conditions studied.
Figure 5. Changes in Keq plotted as a function of pressure for liquid AcAc (A) and solutions of AcAc in sub- and supercritical CO2 (B). Y-axis error bars represent the 95% confidence interval, X-axis error bars are smaller than the symbols used.
in other systems in SC CO2.31,32 The current experiment cannot differentiate between the two effects, except to the extent that the volume change due to changes in tautomeric equilibrium is unlikely to be significantly different in CO2 than in the neat liquid. Similar experiments were conducted on TFA, and the graphical results are shown in the Supporting Information. The values obtained are given in Table 1. Although the equilibrium is shifted strongly in favor of the enol form by the CF3 group in TFA, Keq values could still be determined. For TFA, the value for CO2 (31) Umecky, T.; Kanakubo, M.; Ikushima, Y. J. Phys. Chem. B 2002, 106, 11114-11119. (32) Tachikawa, T.; Akiyama, K.; Yokoyama, C.; Tero-Kubota, S. Chem. Phys. Lett. 2003, 376, 350-357.
4688 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
CONCLUSIONS In this study, we have demonstrated the utility of capillary cells made of fused silica for FT-IR studies at pressures up to 3.1 kbar. The cells are inexpensive, disposable, and intrinsically safe. The operating pressure of the capillary cell is at least 3 times higher than the operating pressure of IR cells typically employed in SC CO2 work. The ability to work at high pressure allows FT-IR studies to be carried out in a regime where pressure has a large influence on reaction thermodynamics by shifting the equilibrium to favor the product with the smaller molar volume. This novel cell has been employed to study the keto-enol equilibrium of AcAc and its fluorinated analogues TFA and HFA as a function of temperature and pressure. The partial molar volume differences between the keto and enol forms were calculated by determining the change in equilibrium constant as a function of pressure at a constant temperature. The partial molar volume change was greater for AcAc and TFA in CO2 than the change for the corresponding pure liquid. In both pure liquid and CO2 solutions, TFA showed a greater partial molar volume change than AcAc, possibly due to differences in intermolecular hydrogen
bonding or to changes in local solvent density around the solutes. The keto form was favored at high pressure and low temperature as determined by our spectroscopic studies of the pure liquids and their CO2 solutions, indicating that there is no advantage to conducting metal extractions at high pressures and low temperatures. The enthalpy of tautomerization has been studied using Van’t Hoff plots, and values comparable to those in the literature were obtained. ACKNOWLEDGMENT This work was supported by Office of Basic Energy Sciences, U.S. Department of Energy. Pacific Northwest National Laboratory is operated for the U.S. Department of Energy by Battelle.
SUPPORTING INFORMATION AVAILABLE Peak assignments for AcAc and TFA, the partial molar volume change determination for TFA (neat and in SC CO2 soln.), and the Van’t Hoff plots for AcAc in SC CO2 solution. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review April 9, 2004. Accepted May 26, 2004. AC049451I
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004
4689
