Chiral Separations Performed by Enhanced-Fluidity Liquid

Shannon Phillips and Susan V. Olesik ... Steven H. Hoke, II, J. David Pinkston, Ruth E. Bailey, Suzanne L. Tanguay, and ... Martin J. Beres , Susan V...
3 downloads 0 Views 104KB Size
Download PDF
Anal. Chem. 1999, 71, 2139-2145

Chiral Separations Performed by Enhanced-Fluidity Liquid Chromatography on a Macrocyclic Antibiotic Chiral Stationary Phase Qian Sun and Susan V. Olesik*

Department of Chemistry, The Ohio State University, 100 West 18th Avenue, Columbus, Ohio 43210-1153

The use of enhanced-fluidity liquid chromatography (EFLC) for chiral separations was demonstrated on a macrocyclic antibiotic column, Chirobiotic-V. This technique was compared to high performance liquid chromatography (HPLC) and supercritical fluid chromatography (SFC) for the separation of chiral compounds in normal-phase mode. The highest resolution was always observed for EFLC condition. Higher efficiency and shorter retention time were also observed for most separations with portions of CO2 in the range of 0-50 mol %. Larger amounts of CO2 caused efficiency to decrease and retention time to be prolonged. For some separations, the temperature was elevated to bring the mobile phase to the supercritical condition. Improved efficiency was obtained in SFC, whereas resolution and selectivity were worse. The use of EFLC in reversed-phase chiral separations was also tested. Enantiomer resolution improved under the EFLC condition. For the tested methanol/H2O mixture, fluoroform provided more significant improvements in chromatographic performance than CO2 when used as a fluidity enhancing liquid. The use of EFLC instead of HPLC also caused a markedly lower pressure drop across the column for commonly used flow rates. The lowpressure drop will allow the use of longer columns or multiple columns to increase the total efficiency of the separation. Since chiral columns are often inefficient, this attribute may be very important for chiral separations. The separation of enantiomers is one of the most challenging tasks in separation science. With the development of a large number of chiral stationary phases, liquid chromatography is presently predominately used for chiral separations. There is also continued interest in the use of super- and subcritical fluid chromatography for chiral separations because higher efficiencies and shorter retention times are typically obtained with SFC compared to HPLC.1-3 However, the low solvent strength of a supercritical fluid often limits the range of compounds that can be separated by SFC. (1) Camel, V.; Thie´baut, D.; Caude, M. J. Chromatogr. 1992, 605, 95. (2) Macaudie`re, P.; Caude, M.; Rosset, R.; Tambute´, A. J. Chromatogr. Sci. 1989, 27, 583. (3) Anton, K.; Eppinger, J.; Frederiksen, L.; Francotte, E. J. Chromatogr. A 1994, 666, 395. 10.1021/ac981134m CCC: $18.00 Published on Web 04/28/1999

© 1999 American Chemical Society

Figure 1. Structure of vancomycin. It has three fused macrocyclic rings, 18 stereogenic centers (/), 9 hydroxyl groups, and 2 amine groups.

Enhanced-fluidity liquid chromatography involves using liquified gases, such as CO2 or CHF3, combined with polar liquids, such as methanol, as the mobile phase for HPLC. The solvent strength of these mixtures is typically high, while their viscosities and diffusivities approach those of a supercritical fluid.4 In many ways enhanced-fluidity liquid mixtures possess the advantages of both common liquids and supercritical fluids. Previous applications in reversed-phase and normal-phase HPLC, as well as SEC, demonstrated these advantages.4-8 Here, we demonstrate the use of EFLC for chiral separations under normal- and reversed-phase conditions with a packed analytical-scale column. Results are compared to those obtained under HPLC and SFC conditions. The chiral stationary phase (CSP) characterized in this study was a Chirobiotic-V column which was packed with 5 µm silica gel that has the macrolide antibiotic, vancomycin, chemically bonded to the surface through a spacer arm.9 The structure of vancomycin is shown in Figure 1. Vancomycin was first applied to HPLC as a (4) Cui, Y.; Olesik, S. V. Anal. Chem. 1991, 63, 1812. (5) Yuan, H.; Olesik, S. V. J. Chromatogr. A 1997, 785, 35. (6) Yuan, H.; Olesik, S. V. Anal. Chem. 1998, 70, 1595. (7) Lee, S. T.; Olesik, S. V. Anal. Chem. 1994, 66, 4498. (8) Lee, S. T.; Olesik, S. V. J. Chromatogr. A 1995, 707, 217. (9) Chirobiotic Handbook, 2nd ed.; Advanced Separation Technologies Inc.: NJ, 1997.

Analytical Chemistry, Vol. 71, No. 11, June 1, 1999 2139

chiral selector by Armstrong et al. in 1994.10 Numerous chiral compounds have been separated with this column under both normal-phase and reversed-phase conditions.10-13 EXPERIMENTAL SECTION Materials. Hexane, methanol, ethanol, and tetrahydrofuran (THF) were HPLC-grade solvents purchased from Fisher Scientific (Pittsburgh, PA) and were used as received. Distilled water was purified using a Nanopure II filtration system (SYBRON/Barnstead, Boston, MA) with a resistivity of 18.3 MΩ. Ammonium nitrate (Aldrich Chemical Co., Milwaukee, WI) was used to prepare buffers. Chiral test analytes with pKa’s ranging from 1.0 to 9.5 were chosen for study (Table 1). The samples analyzed in the normal-phase study (obtained from Aldrich Chemical Co.) include an ethanol solution of 5-methyl-5-phenyl hydantoin (4 mg/ mL), 1,1′-bi-2-naphthol (2 mg/mL), R-methyl-R-phenyl succinimide (8 mg/mL), ftorafur (8 mg/mL), and a mixture of (R,R)-(+)- and (S,S)-(-)-N,N′-bis(R-methyl-benzyl) sulfamide (7 mg/mL). Test molecules used in reversed-phase investigation include warfarin (Aldrich Chemical Co., Milwaukee, WI), 3-(R-acetonyl-4-chlorobenzyl)- 4-hydroxy coumarin, and an enantiomer mixture of (R)(-)- and (S)-(+)-1,1′-binaphthyl-2,2′-diyl-hydrogen phosphate (Sigma Chemical Co., St. Louis, MO) dissolved in methanol with a concentration of 2 mg/mL. SFE/SFC-grade CO2 (99.9999%) and CHF3 (99.999%) were obtained from Air Products and Chemicals (Allentown, PA). Instrumentation. The experimental setup was similar to the one described previously.6 Briefly, the chromatographic system consisted of an Isco LC-260D syringe pump (ISCO, Lincoln, NE), a Valco-W-series high-pressure injection valve with a 200-nL internal injection loop (Valco Instruments, Houston, TX), a Chirobiotic-V, 150 mm long × 4.6 mm i.d. HPLC column packed with 5-µm diameter silica particles that were chemically bonded with the macrolide antibiotic, vancomycin (Advanced Separation Technologies, Whippany, NJ), and a Spectra-100 variable wavelength UV/vis absorbance detector (Thermo Separation, Fremont, CA). The detection cell was created by removing the polyimide coating from a piece of 250-µm i.d. fused silica tubing (Polymicro Technologies, Phoenix, AZ). All of the experiments were done at the maximum absorbance wavelength of the tested compound which falls between 210 and 308 nm. For the reversed-phase separations, a BetaBasic (silica) column (Keystone Scientific, Bellefonte, PA) was installed between the injector and the column to protect the chiral column. A Perkin-Elmer Sigma oven (PerkinElmer Co., Norwalk, CT) was used to elevate the column temperature. A fused silica capillary of a certain length was used as a restrictor to control the flow rate of the mobile phase. The flow rate was read from the syringe pump directly ((0.001 mL/ min). A Setra, model 204, pressure transducer (Setra Systems, Acton, MA) was placed after the detector and before the restrictor to make certain the operating pressure of the mobile phase was above the minimum pressure necessary to maintain single-phase conditions. Mobile-Phase Preparation and Data Analysis. The method used to prepare the enhanced-fluidity liquid mobile phase was the (10) Armstrong, D. W.; Tang, Y.; Chen, S.; Zhou, Y. Anal. Chem. 1994, 66, 1473. (11) Armstrong, D. W.; Rundlett, K.; Reid, G. L. Anal. Chem. 1994, 66, 1690. (12) Armstrong, D. W.; Zhou, Y. J. Liq. Chromatogr. 1994, 17, 1695. (13) Medvedovici, A.; Sandra, P.; Toribio, L.; David, F. J. Chromatogr. A 1997, 785, 159.

2140 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

Table 1. Structure and PKa Value of Test Molecules

1 Serjeant, E. P.; Dempsey, B. Ionisation Constants of Organic Acids in Aqueous Solution; Pergamon Press: Oxford, New York, 1979. 2 pKa value is not available.

same as described previously.6 Pressurized liquid CO2 or CHF3 was transferred from an Isco syringe pump into another Isco syringe pump that contained the other components of the desired mobile phase. The amount of CO2 or CHF3 added to the second pump was determined from the density of the liquified gas and the volume difference before and after the transfer. All mixtures containing CO2 or CHF3 were allowed to equilibrate at least 12 h before use. The inlet pressure of the chromatographic system was maintained at 170 atm for both normal-phase and reversed-phase experiments so that all of the mobile-phase mixtures (ethanol/ hexane/CO2, methanol/CO2, or methanol/H2O/CO2 (CHF3)) were one phase at room temperature. For the separation of

Table 2. Enantiomeric Resolution, Selectivity, Retention Factor, Band Dispersion, Asymmetry Factor (10% Height) for the Second Eluting Peak, and Pressure Drop Across the Column for the Separation of N,N′-Bis(r-methylbenzyl) Sulfamide Enantiomers in Ethanol/Hexane (20/80 mol %)/CO2 Mobile Phase with Different Amounts of CO2 in the Mixture mol %, CO2

resolution, Rs

selectivity, R

0 10 20 30 40 50 60 70 80 85

0.79 0.81 0.85 0.90 0.93 0.96 0.97 0.99 1.05 1.09

1.07 1.07 1.08 1.08 1.09 1.09 1.09 1.10 1.10 1.11

retention factor, k′ R,R S,S 5.45 4.96 4.76 4.69 4.66 5.06 5.77 8.49 12.3 17.4

5.83 5.32 5.14 5.07 5.07 5.51 6.31 9.33 13.6 19.3

compound 6 in the reversed-phase mode, 60 mM ammonium nitrate aqueous buffer solution was always used instead of pure water to improve peak shape and maintain elution strength. The chromatographic data were collected and analyzed using EZChrom data system (Scientific Software, San Ramon, CA) and Peakfit 4.0 (Jandel Scientific, San Rafael, CA). The second statistical moment was used to provide band dispersion information. Each reported data point is an average of two or three measurements. RESULTS AND DISCUSSION A decrease in resolution (Rs) and retention time (tr), together with an increase in reduced plate height (h) was observed when the flow rate was increased in both the normal- and the reversedphase studies. In addition, neither the selectivity factor (R) nor the retention factor (k′) was affected by flow rate variation. The retention time of the injection solvent was used to determine t0, the retention time of unretained compounds. All experiments were undertaken at room temperature and an inlet pressure of 170 atm, unless otherwise specified. Normal-Phase Separations. Ethanol/hexane mixtures were chosen as the organic components of the eluent for the initial separations. This mixture provided higher efficiency and better resolution for the test molecules compared to that obtained with for example, isopropyl alcohol/hexane mixtures. N,N′-Bis(Rmethyl-benzyl) sulfamide, compound 1, was the first test system for the normal-phase studies. An ethanol/hexane mixture with a volume ratio of 10/90, which corresponds to a 20/80 mole ratio, was employed as the initial mobile phase for LC separation. Next, CO2 was added into the mobile phase by starting with 10 mol % CO2 and increasing the proportion of CO2 in 10 mol % increments. The highest CO2 composition studied was 85 instead of 90 mol %, due to the long retention time obtained when using 90 mol % CO2. The flow rate was maintained at 1 mL/min. Table 2 shows the experimental results. Both resolution and selectivity increased with increasing proportions of CO2. Resolution increased by 40% for the addition of 85 mol % CO2. The retention factor decreased with increasing proportions of CO2 until the proportion exceeded 60 mol % CO2. Further amounts of added CO2 caused significant increases in the retention factor. The reduced plate height, h, as measured by the second statistical moment, remained relatively constant until more than 60 mol %

reduced plate height, h R,R S,S 7.55 7.95 7.80 7.43 7.69 7.71 8.71 10.6 10.5 11.7

8.40 9.33 9.18 9.05 9.79 10.4 13.1 17.0 16.4 17.1

asymmetry, (10%) S,S

pressure drop, ∆P (MPa)

1.40 1.54 1.54 1.54 1.62 1.70 1.94 2.27 2.14 2.18

2.50 2.14 1.93 1.68 1.41 1.18 0.97 0.74 0.43 0.41

CO2 was added to the mobile phase. Similarly, the asymmetry factor remained constant until more than 60 mol % CO2 was added after which significant increases in asymmetry were observed. The decrease in efficiency as measured by the second statistical moment can be explained by the distortion of chromatography peaks. For the enhanced-fluidity liquid mixtures that have been studied to date, the solvent strength in terms of hydrogen-bond acidity and basicity typically decreases significantly for CO2 proportions >60 mol %.4 The diminished solvent strength can cause slow mass transfer kinetics between active sites on the chromatographic phase and polar groups of the analyte which could contribute to chromatographic band distortion. As described earlier, Table 2 shows that the resolution of this separation increased consistently as R increased, despite decreasing plate numbers and increasing k′ for increasing proportions of CO2. The pressure drop across the column is typically described using Darcy’s law:

∆P )

φηuL dp2

where φ is a dimensionless flow resistance parameter, L is column length, u is linear velocity, η is viscosity, and dp is particle diameter. The viscosity of liquid CO2 (0.09 at 25 °C, 170 atm)14 is approximately an order of magnitude lower than that of common liquids. According to Darcy’s law, the pressure drop should be lowered when using enhanced-fluidity liquid mixtures as mobile phase. Significantly lower pressure drops are observed with increasing amounts of CO2. For example, the addition of 50 mol % CO2 to the mobile phase caused the pressure drop across the column to decrease by a factor of 2. A lower pressure drop can be especially beneficial in chiral-HPLC separations. The lower pressure drop enables EFLC to be operated under higher flow rates than in HPLC, which increases the speed of analysis. In addition, for the same linear velocity used in HPLC, the total number of theoretical plates available can be increased by a factor of 2 by adding 50 mol % CO2 to the mobile phase and by placing two columns in series. (14) Handbook of Chemistry and Physics, 72nd ed.; CRC Press: Boca Raton, FL, 1991-1992.

Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

2141

Table 3. Enantiomeric Resolution, Selectivity, Retention Factor, and Band Dispersion for the Separation of 1,1′-Bi-2-naphthol, r-Methyl-r-phenyl Succinimide, and Ftorafur Enantiomers in Ethanol/ Hexane (36/64 mol %)/CO2 Mobile Phase with Different Amounts of CO2 in the Mixture

mol %, CO2

resolution, Rs

1,1′-bi-2-naphthol 0 0.91 20 1.06 40 1.30 60 1.55 80 1.81 90 1.90 R-methyl-R-phenyl succinimide 0 0.60 20 0.68 40 0.76 60 0.87 80 1.04 90 1.16 ftorafur 0 0 20 0 40 0.45 60 0.51

selectivity, R 1.10 1.12 1.14 1.16 1.19 1.22 1.06 1.07 1.07 1.08 1.09 1.11 1.00 1.00 1.05 1.06

retention factor, k′

reduced plate height, h

2.33, 2.56 8.27, 8.79 1.88, 2.10 6.77, 7.08 2.14, 2.43 6.70, 7.46 3.28, 3.81 7.91, 9.50 8.24, 9.78 10.7, 14.0 19.5, 23.7 15.2, 19.3 3.14, 3.32 2.40, 2.56 2.14, 2.30 2.37, 2.57 3.87, 4.23 8.11, 8.97

8.15, 8.04 6.14, 6.89 5.52, 6.23 6.49, 6.36 7.37, 6.87 9.81, 10.2

12.9 10.4 9.39, 9.81 8.75, 18.2 10.3, 10.8 13.0, 21.1

Other chiral analytes were also separated using normal-phase EFLC. Ethanol/hexane mixtures with a volume ratio of 20/80, which corresponds to a 36/64 mole ratio, were used as the initial mobile phase for these separations to accelerate analysis speed while maintaining the same 1 mL/min flow rate. Table 3 shows the variation of resolution, selectivity, retention factor, and efficiency as a function of added CO2. The best separation was observed under EFLC conditions for all studied analytes. Baseline separation was achieved for compound 2 by using 60 mol % CO2/ ethanol/hexane (Figure 2). The co-eluted enantiomer pair of compound 4 was partially resolved under EFLC condition with shorter analysis time (Figure 3). 5-Methyl-5-phenyl hydantoin, compound 5, was another test molecule studied. It interacts strongly with the vancomycin stationary phase, and thus methanol was chosen as the mobile phase to shorten the retention time. CO2 was added to the mobile phase following the same procedure as described above. The results of the study are shown in Table 4. The pressure drop across the column decreased substantially when CO2 was added to the mobile phase. For example, the addition of 80 mol % CO2 to the mobile phase caused the pressure drop to decrease by 85%. The selectivity increased slowly with increasing CO2. For the mixture containing 80 mol % CO2, the resolution increased 130% relative to that in methanol. Unlike all other chiral compounds that were separated using EFLC under normal-phase conditions, the reduced plate height, h, and retention factor, k′ of compound 5 increased continuously with the addition of CO2. The cause of these differences will be studied further. Proportions of CO2 > 80 mol % are not recommended due to the long retention time. Comparison of Normal-Phase EFLC with SFC. The temperature of the column was elevated to bring the selected mobile phase into the supercritical state. Since the addition of CO2 lowers the critical temperature of the mixture, mixtures with high proportions of CO2 were chosen for studies of SFC to avoid 2142 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

Figure 2. Chromatograms of the separation of 1,1′-bi-2-naphthol at flow rate 1 mL/min using mobile phase (a) 36/64 mol % ethanol/ hexane and (b) 60/14/26 mol % CO2/ethanol/hexane.

possible damage to the column due to high temperatures. The critical temperature of the selected liquid mixture was estimated by using SF-Solver (ISCO, Lincoln, NE). The mobile phase compositions chosen were ethanol/CO2 (8.2/91.8, mole ratio) and methanol/CO2 (15/85, mole ratio), respectively. Instead of applying ethanol/hexane/CO2 mixture as the mobile phase, ethanol/ CO2 mixture was used because the SF-Solver program does not provide the critical temperatures for ternary mixtures. The transfer of solutes between mobile and stationary phases is generally faster at elevated temperature. Thus, higher efficiency and shorter analysis time are usually expected. When the ethanol/ CO2 mobile phase was used, the separations under SFC conditions had higher efficiencies, but the resolution, Rs, was lower (Table 5). Surprisingly, the retention time increased for these test molecules under SFC conditions compared to EFLC conditions. The change in enantioselectivity was similar for the separation of compound 5 in methanol/CO2 mobile phase (Table 4), but retention factor was larger for EFLC separation this time. The retention was rather long under both conditions and decreased S/N ratio made it impossible to calculate efficiency or resolution. In all, the change on the retention factor was not readily predictable when moving from the enhanced-fluidity liquid mixture to the supercritical state. An increase in temperature improved peak shape and separation efficiency, but chiral selectivity degraded. The chromatograms for the separation of compound 3 under EFLC and SFC conditions are shown in Figure 4. These preliminary data on normal-phase separations illustrate that resolution increases under EFLC conditions relative to that

Table 5. Comparison of Resolution, Selectivity, Retention Factor, and Band Dispersion for the Separation of N,N′-Bis(r-methylbenzyl) Sulfamide, 1,1′-Bi-2-naphthol and r-Methyl-r-phenyl Succinimide Enantiomers in EFLC and SFC Conditions Using Ethanol/CO2 (8.2/91.8 mol %) Mobile Phase

separation mode

resolution, Rs

selectivity, R

N,N′-Bis(R-methylbenzyl) sulfamide EFLC 0.85 1.08 SFCa 0.56 1.04 1,1′-Bi-2-naphthol EFLC 1.82 1.17 SFCa 1.60 1.12 R-methyl-R-phenyl succinimide EFLC 0.96 1.08 SFCa 0.82 1.06 a

retention factor, k′

reduced plate height, h

5.36, 5.76 7.36, 7.64

7.20, 9.55 4.32, 5.84

11.3, 13.2 13.4, 15.0 3.05, 3.30 4.73, 5.02

8.75, 12.6 5.54, 7.15 5.68, 5.89 5.20, 4.92

Temp ) 58 °C.

Figure 3. Chromatograms of the separation of ftorafur at flow rate 1 mL/min using mobile phase (a) 36/64 mol % ethanol/hexane and (b) 60/14/26 mol % CO2/ethanol/hexane. Table 4. Enantiomeric Resolution, Selectivity, Retention Factor, Band Dispersion, and Pressure Drop Across the Column for the Separation of 5-Methyl-5-phenyl-hydantoin Enantiomers in Methanol/ CO2 Mobile Phase with Different Amounts of CO2 in the Mixturea mol %, resolution, selectivity, CO2 Rs R 0 2.20 1.57 10 2.53 1.58 20 3.12 1.60 30 3.68 1.60 40 4.11 1.61 50 4.49 1.63 60 4.68 1.63 70 4.89 1.64 80 5.11 1.65 85 1.69 SFC condition temp ) 64 °C 85 1.50

retention factor, k′

reduced plate height, h

pressure drop, ∆P (MPa)

0.30, 0.47 4.20, 4.53 0.37, 0.58 4.37, 4.73 0.50, 0.81 4.74, 5.17 0.75, 1.20 5.39, 6.00 1.10, 1.77 6.72, 7.39 1.56, 2.54 8.13, 8.60 2.54, 4.14 10.9, 11.1 4.14, 6.79 12.8, 12.9 7.36, 12.2 15.5, 13.9 12.0, 20.3

1.82 1.60 1.45 1.24 1.06 0.66 0.43 0.36 0.27 0.52

10.9, 16.4

0.19

a Flow rate used is this study was 0.50 mL/min, except for separations using 85% CO2, which used 1.0 mL/min.

for standard HPLC or SFC. Variation in R with added CO2 was the major cause of the observed improvement on chiral resolution. For some enantiomers, higher Rs was observed even with the decrease in retention factor (k′) and the increase in the reduced plate height as in the separation of compound 1.

Figure 4. Chromatograms of the separation of R-methyl-R-phenyl succinimide enantiomers under (a) EFLC and (b) SFC (temperature 58 °C) conditions. Mobile phase composition is ethanol/CO2 (8.2/ 91.8 mol %). Flow rate 1 mL/min.

Reversed-Phase Separations. For highly polar compounds, chiral, reversed-phase HPLC is often the method of choice.15 An 80/20 volume ratio (64/36, mole ratio) methanol/H2O solvent was chosen here as the initial mobile phase for the reversed-phase studies. For this methanol/H2O mixture, the maximum amount of CO2 or CHF3 that is miscible at room temperature and 170 atm (15) Chiralcel OB-H Application Note; Chiral Technologies: Exton, PA, 1992.

Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

2143

Table 6. Enantiomeric Resolution, Selectivity, Retention Factor, and Band Dispersion for the Separation of 1,1′-Binaphthyl-2,2′-diyl-hydrogen Phosphate Enantiomers in Methanol/60 mM NH4NO3 (64/36 mol %)/Liquified Gas Mobile Phase with Different Amounts of CO2 (or CHF3) in the Mixture mol %, CO2 (or CHF3)

retention factor, k′ S R

reduced plate height, h S R

resolution, Rs

selectivity, R

0.46 0.70 0.79 0.86

1.08 1.07 1.07 1.07

0.62 2.44 3.30 4.83

0.67 2.61 3.54 5.19

5.57 6.36 6.53 6.27

6.97 7.37 7.18 6.96

0.68 0.95 1.16

1.09 1.10 1.10

1.21 2.52 3.36

1.32 2.77 5.87

6.29 6.58 5.61

8.13 7.71 7.28

CO2 0 10 20 27 CHF3 10 20 30

Table 7. Enantiomeric Resolution, Retention Time, and Band Dispersion for the Separation of Warfarin and 3-(r-Acetonyl-4-chlorobenzyl)-4-hydroxy Coumarin Enantiomers in Methanol/H2O (64/36 mol %)/Fluoroform Mobile Phase with Different Amounts of CHF3 in the Mixture mol %, CHF3

resolution, Rs

retention time, ta

warfarin 0 0.63 3.91, 4.06 10 0.72 4.12, 4.31 20 0.82 4.09, 4.30 30 0.87 4.14, 4.36 THF/H2O (20/80, V) 1.43 13.5, 16.3 3-(R-acetonyl-4-chlorobenzyl)-4-hydroxy coumarin 0 0.71 3.79, 3.98 10 1.00 4.17, 4.45 20 1.23 4.19, 4.52 30 1.34 4.29, 4.67 THF/H2O (20/80, V) 2.48 17.9, 24.3

reduced plate height, h 6.50, 4.21 4.86, 5.57 4.62, 5.22 5.26, 5.06 42.5, 31.0 7.02, 7.72 6.03, 5.79 5.98, 4.93 5.62, 5.75 27.6, 19.1

a Retention time is used instead of retention factor because the latter was rather small (0.1 for warfarin in HPLC) and a small variation on t0 measurement will induce a large deviation on k′.

is 27 or 35 mol %, respectively.16,17 Compound 6 was the test molecule used. It is a highly polar compound with pKa ) 1.0. In this case, 60 mM ammonium nitrate aqueous solution was used as the buffer to improve peak asymmetry. Previous studies showed that ammonium nitrate worked well as a buffer for vancomycin chiral stationary phase in maintaining efficient and reproducible separations.9 The variation of resolution, selectivity, retention factor, and efficiency as a function of added CO2 is shown in Table 6. Chiral resolution improved significantly with the addition of CO2. Rs almost doubled when 20 mol % CO2 was present in the mobile phase compared to using the methanol/H2O mixture as mobile phase. Separation efficiency also increased slightly when the proportion of CO2 increased. These improvements in chromatographic performance are similar to those found previously in achiral reversed-phase EFLC.18 However, different from the (16) Lee, S. T.; Reighard, T. S.; Olesik, S. V. Fluid Phase Equilib. 1996, 122, 223. (17) Zhao, J.; Olesik, S. V. Fluid Phase Equilib. 1999, 154, 261. (18) Cui, Y.; Olesik, S. V. J. Chromatogr. A 1995, 691, 151.

2144 Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

Figure 5. Chromatograms of the separation of 1,1′-binaphthyl-2,2′diyl-hydrogen phosphate at flow rate 0.5 mL/min using mobile phase (a) 64/36 mol % methanol/60 mM NH4NO3, (b) 20/51/29 mol % CO2/ methanol/60 mM NH4NO3, and (c) 20/51/29 mol % CHF3/methanol/ 60 mM NH4NO3 (*unidentified peak).

previous results for achiral reversed-phase EFLC separations, retention time for this chiral separation increased under EFLC mode. The cause of this difference is presently under study. Comparison of CHF3 and CO2 as Fluidity-Enhancing Solvent for Reversed-Phase Chiral Separations. There are limitations to the use of CO2 as a fluidity-enhancing solvent. CO2 reacts with water to produce carbonic acid which changes the pH of mobile phase. CO2 also reacts with basic analytes including primary and secondary amine,19 to form complexes or insoluble (19) Fields, S. M.; Grolimund, K. H. C. J. High Resolut. Chromatogr. 1988, 11, 127.

carbamate salts. Fluoroform possesses all of the properties that make CO2 an attractive fluidity-enhancing solvent, but it does not have the disadvantages of CO2 that are listed above. For example, fluoroform, CHF3, has lower Tc, Pc, and viscosity than CO2.14 It is also a more polar solvent than CO2 and is therefore more suitable for the separation of highly polar molecules. Furthermore, CHF3 is chemically inert, nonflammable, and environmentally friendly.20 The applicability of CHF3 to the reversed-phase separation of achiral basic solutes was demonstrated previously.6 To investigate the applicability of fluoroform as fluidityenhancing solvent for chiral separations, compounds 6, 7, and 8 were tested using the methanol/H2O/fluoroform mixture as the mobile phase. An aqueous solution containing ammonium nitrate buffer was used again for compound 6 for the reason stated before. Results are shown in Tables 6 and 7. For compound 6, higher resolution, higher selectivity, and shorter retention time were observed for the mobile phase containing fluoroform instead of carbon dioxide. Also, in changing from 10 to 30 mol % fluoroform, the resolution and efficiency improved with the greater amount of fluoroform (Table 6). Addition of fluoroform to the mobile phase also allowed the enantiomers of compounds 7 and 8 to be separated (Table 7), whereas the addition of CO2 did not affect the resolution of the enantiomers of these compounds. Clearly, fluoroform was a better fluidity-enhancing solvent than carbon dioxide, and improved separation performance was observed for all test compounds when fluoroform was applied. Chromatograms for the separation of compound 6 under both LC and EFLC conditions are shown in Figure 5. Even though these results showed advantages of EFLC over LC, for compounds 7 and 8 the best enantiomeric resolution was obtained using a THF/H2O (20/ 80, volume ratio) mixture as the mobile phase. (20) Ravishankara, A. R.; Turnipseed, A. A.; Jenson, N. R. Science 1994, 263, 71.

CONCLUSIONS Fast method development and full control of solvent strength have made EFLC an attractive alternative to HPLC for many applications in separations and extractions. Our initial studies here on enhanced-fluidity liquid mixtures illustrate that EFLC is useful for chiral separations. For normal-phase separations, enantiomer resolution improved in EFLC compared to HPLC. Higher efficiency and faster separation were also observed for carbon dioxide proportion of 10-50 mol %. The optimum composition of the mobile phase mixture was different for different molecules, although the optimum often included CO2 proportions in the range of 60-70 mol %. The substantial decrease in pressure drop across the column in EFLC compared to HPLC will allow the use of long columns or columns that are placed in series. Because the chromatographic efficiency for chiral separations is often low, this attribute of EFLC is very important for practical applications. Also, the resolution and selectivity for enantiomeric pairs were greater using EFLC compared to those obtained in SFC. For reversed-phase chromatography, fluoroform was found to be a better choice for fluidity enhancement compared to carbon dioxide. Improved resolution and separation efficiency were observed for all test molecules when fluoroform was added to the methanol/H2O mobile phase. However, the observed increase in retention factor for the EFLC conditions compared to HPLC is not desirable. Further studies are underway to better understand the chiral recognition mechanism under EFLC conditions. ACKNOWLEDGMENT This work was supported by the National Science Foundation.

Received for review October 19, 1998. Accepted March 19, 1999. AC981134M

Analytical Chemistry, Vol. 71, No. 11, June 1, 1999

2145