Anal. Chem. 1999, 71, 3013-3021
Cellulose Tris(3,5-dimethylphenylcarbamate)Coated Zirconia as a Chiral Stationary Phase for HPLC Cecilia B. Castells and Peter W. Carr*
Department of Chemistry, Kolthoff and Smith Halls, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455
In this work we explore the use of microparticulate porous zirconia coated with cellulose tris(3,5-dimethylphenylcarbamate) (CDMPC) as a support for separation of chiral compounds by HPLC. The surface of zirconia, previously sintered but not rehydroxylated, provides a stable surface for depositing the chiral polymer. Zirconia’s surface prior to coating was investigated by diffuse reflectance FT-IR. The spectra indicate the presence of residual hydroxyl groups even after treatment at 750 °C for 5 h. The amount of chiral polymer deposited was systematically varied, and the pore structure of the resulting particles was assessed by nitrogen sorptometry. Dynamic studies of columns packed with these stationary phases were also conducted. We found that columns packed with about 3-4% (w/w) CDMPC coated on 2.5-µm zirconia particles provide an excellent compromise between loading need to impart good chiral recognition ability to the stationary phase and column’s chromatographic efficiency. Preliminary results show resolutions higher than 1 for 9 out of 16 racemic mixtures in packed 5-cm columns. The use of shorter columns combined with reduced particle size to provide sufficient resolution has the advantage of decreasing the analysis times and reducing eluent volumes. CDMPCcoated zirconia columns exhibit high stability under normal-phase conditions at relatively high linear velocities. It is now well recognized that the biological activity of most drugs and agrochemicals containing stereogenic centers is significantly related to the absolute configuration of the molecule.1-5 Due to the promulgation of regulatory guidelines, during the past decade a growing body of chemical research has focused on stereoselective synthesis,6 which ultimately calls for the development of efficient chiral assays to evaluate the enantiomeric purity of the synthesized stereoisomers of biologically active compounds. (1) Simonyi, M. Med. Res. Rev. 1984, 4, 359-413. (2) Ariens, E. J. Med. Res. Rev. 1986, 6, 451-460. (3) Ariens, E. J.; van Rensen, J. J.; Welling, W. Stereoselectivity of Pesticides; Elsevier: Amsterdam, The Netherlands, 1988. (4) Powell, J. R.; Ambre, J. T.; Ruo, T. I. In Drug Stereochemistry: Analytical Methods and Pharmacology; Wainer, I. W., Drayer, D., Eds.; Marcel Dekker: New York, 1988; Chapter 10. (5) Wainer, I. W. In Drug Stereochemistry: Analytical Methods and Pharmacology; Wainer, I. W., Ed.; Marcel Dekker: New York, 1993; Chapter 6. (6) Stinton, S. C. Chem. Eng. News 1993, 73, (Sept 27), 38. 10.1021/ac990021f CCC: $18.00 Published on Web 06/25/1999
© 1999 American Chemical Society
The high sensitivity requirements needed to determine optical purity have displaced classical polarimetric techniques and led to the development of chiral methods in HPLC, GC, SFC, NMR, and more recently CE. The introduction of HPLC separation methods based on chiral stationary phases (CSPs) has become one of the most attractive approaches to chiral separations, mainly due to their simplicity for determining optical purity but also because they are easily extended to the semipreparative and preparative scales.7 Starting from a broad variety of natural and synthetic chiral building blocks, an almost unlimited number of optically active stationary phases are possible. However, one of the major problems in using many CSPs is their narrow range of analyte applicability; they can only discriminate a limited number of specific types of chemical entities, and it is frequently necessary to derivatize the compounds of interest to achieve separation.8 On the other hand, the polysaccharide-based CSPs developed by Okamoto and co-workers9-11 have proven to be highly versatile and rugged. Okamoto12 reported the resolution of 64% of 483 racemic mixtures on cellulose tris(3,5-dimethylphenyl carbamate) (CDMPC) and 80% were successfully resolved on either the cellulose or the corresponding amylose carbamate. Fast method development, high efficiency, rapid resolution of enantiomers, and robustness are clearly the main criteria for chiral separation methods, especially in the pharmaceutical industry. These priorities require stable CSPs capable of achieving baseline resolution in the shortest possible time, which ultimately means high selectivity and efficiency. Here we report preliminary results obtained with cellulose tris(3,5-dimethylphenylcarbamate) coated on 2.5-µm zirconia particles. Our main goal is to develop a new family of chiral columns, different from traditional silica, on which more efficient, robust, (7) Francotte, E. In Chiral Separations: Applications and Technology; Ahuja, S., Ed.; American Chemical Society: Washington, 1997; Chapter 5. (8) Dingenen, J. In A Practical Approach to Chiral Separations by Liquid Chromatography; Subramanian, G., Ed.; VCH: New York, 1994; Chapter 6. (9) Okamoto, Y.; Kawashima, M.; Yamamoto, K.; Hatada, K. Chem. Lett. 1984, 739-742. (10) Okamoto, Y.; Kawashima, M.; Hatada, K. J. Am. Chem. Soc. 1984, 106, 5357-5359. (11) Okamoto, O.; Kawashima, M.; Hatada, K. J. Chromatogr. 1986, 363, 173186. (12) Okamoto, Y.; Kaida, Y.; Aburatani, R.; Hatada, K. In Chiral Separations by Liquid Chromatography; Ahuja, S., Ed.; ACS Symposium Series 471; American Chemical Society: Washington, DC, 1991; pp 101-113.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999 3013
and faster chiral analytical methods can be carried out. We are also concerned with the role played by the base substrate (silica, zirconia) on the enantioseparation. Traditionally, cellulose and amylose derivatives are coated on 7-10 µm very large pore (1000 Å) aminopropyl-derivatized silica. The usual polymer loading is 20% (w/w).11 Earlier studies of Shibata et al. demonstrated that the resolving properties of cellulose acetate and cellulose benzoate coated on silica were affected by the solvent from which the chiral polymer is deposited.13 Since then, several other experimental variables have been extensively studied; significant changes in chiral selectivity result by changing the conditions under which polysaccharide-based CSPs are deposited. In brief, the pore size of the support, the solvating agents, the amount of polymer loaded,14 the particle size, and the use of different chemically modified silica surfaces15-17 have been investigated. However, most of the results are solutedependent and general conclusions are difficult to draw. Francotte and Zhang reported18 a dramatic effect on the discriminating ability depending on the procedure used to deposit m-methylbenzoylcellulose on aminopropylsilica. They found that precipitation of the polymer onto the surface produced a stationary phase different from that obtained by evaporation of the solvent. The optical resolving power was better for most racemates with the former phase. Moreover, they observed inversion of elution order for some enantiomers from these two materials and attributed these observations to changes in the supramolecular structure of the CSPs.18,19 Clearly molecular modeling, which does not incorporate the higher order structures of the cellulosic CSPs, is not going to provide much insight.20 While these CSPs provided quite good enantioselectivity for a variety of racemates, the column efficiency is typically in the range of only a few thousand theoretical plates, which is much less than that observed for achiral columns of similar dimensions.21,22 Although the attainment of high separation factors between the pair of enantiomers is the main objective, resolution also depends on column efficiency. Higher column efficiencies would allow reduction of column lengths and thus decrease analysis time while adequate resolution is maintained. Moreover, improvements in plate counts should, in principle, apply to all types of enantiomeric pairs. Matlin et al. pioneered the use of small silica particles (2.5 µm) to support cellulose tris(arylcarbamate) phases and reduce the total analysis time.17 They coated 120 Å pore diameter silica particles with 15% (w/w) chiral polymer. To our knowledge there have not been systematic studies of the influence of polymer (13) Shibata, T.; Sei, T.; Nishimura, H.; Deguchi, K. Chromatographia 1987, 24, 552-554. (14) Yashima, E.; Sahavattanapong, P.; Okamoto, Y. Chirality 1996, 8, 446451. (15) Grieb, S. J.; Matlin, S. A.; Belenguer, A. M.; Ritchie, H. J. J. Chromatogr., A 1995, 1995, 271-278. (16) Felix, G.; Zhang, T. J. Chromatogr., A 1993, 639, 141-149. (17) Grieb, S. J.; Matlin, S. A.; Phillips, J. G.; Belenguer, A. M.; Ritchie, H. J. Chirality 1994, 6, 129-134. (18) Francotte, E.; Zhang, T. J. Chromatogr., A 1995, 718, 257-266. (19) Francotte, E.; Wolf, R. M.; Lohmann, D.; Mueller, R. J. Chromatogr., A 1985, 347, 25-37. (20) O’Brien, T.; Crocker, L.; Thompson, R.; Thompson, K.; Toma, P. H.; Conlon, D. A.; Feibusch, B.; Moeder, C.; Bicker, G.; Grinberg, N. Anal. Chem. 1997, 69, 1999-2007. (21) Kirkland, K. M. J. Chromatogr., A 1995, 718, 9-26. (22) Application Guide for chiral column selection, 2nd ed.; Daicel Chemical Industries, L., Tokyo, 1993.
3014 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
loading and coating procedure on the pore structure characteristics and the chromatographic consequences. In the present study, we explore porous zirconia particles as a support for cellulose tris(3,5-dimethylphenylcarbamate). Various amounts of polymer were deposited, and the phases were physically and chemically characterized. The chromatographic performance was evaluated by measuring the column efficiency and chiral discrimination of several racemates on all columns studied. We also examine the effect of coating conditions on resolution. Finally the phase stability at relatively high flow rates was investigated. EXPERIMENTAL SECTION Reagents. All reagents used for the synthesis of the stationary phase were reagent grade or better. Microcrystalline cellulose (Avicel, Merck, Darmstadt, Germany) was purchased from Fluka. 3,5-Dimethylphenyl isocyanate and pyridine were obtained from Aldrich (Milwaukee, WI). Methanol, acetone, and 2-propanol were HPLC grade (Mallinckrodt, ChromAR). n-Hexane and tetrahydrofuran (THF) were purchased from EM Sciences (Gibbstown, NJ). The racemic compounds studied are shown in Figure 1. All are commercially available. Solutions at a concentration of 0.51.0 mg/mL were prepared by dissolving the enantiomeric mixture in the mobile phase. Synthesis and Characterization of CDMPC. CDMPC was synthesized as previously reported.11 Dried cellulose was refluxed in dry pyridine for 12 h. An excess of 3,5-dimethylphenyl isocyanate was added dropwise to the cellulose suspension under magnetic stirring and the reaction was continued for 24 h under reflux. The final product was isolated as the methanol-insoluble fraction and purified by reprecipitation from an acetone solution. The solid was filtered, washed several times with methanol, and dried. The yield was 67%. The derivative was characterized by elemental analysis (MicroAnalysis, Inc., Wilmington, DE), infrared spectroscopy (Nicolet Magna 750 FT-IR instrument), and 1H NMR spectroscopy (Varian VAC-200, Palo Alto, CA). Elemental Anal. (C33H37N3O8)n Found: C, 63.63; H, 6.12; N, 6.59. Calcd: C, 65.66; H, 6.18; N, 6.96. IR analysis (KBr pellets) ν(urethane CO) 1750 cm-1, ν(uretane NH) 3295 cm-1, ν(NH) 1560 cm-1, substantially no absorption near 3500 cm-1 due to hydroxyl groups of cellulose was observed. 1H NMR (200 MHz) spectrum of CDMPC dissolved in acetone-d6 at room temperature showed the following characteristic absorptions: 7.98.4 (H of amide groups), 2.8-5 (H of cellulose ring and methylene in position 6), 2-2.3 ppm (H of CH3-aryl groups). The ratios of H groups were satisfactory. Zirconia Particles and Coating Conditions. Microparticulate zirconia particles (batch PICA-7) were synthesized in this laboratory by the polymerization-induced colloid aggregation method.23 To dehydroxylate zirconia’s surface, the particles were heated at 750 °C for 5 h and cooled over phosphorus pentoxide before use. Typically, 3.0 g of particles was suspended in 20 mL of THF and sonicated under vacuum for 15 min to eliminate the air from the pores. Several ratios of polymer to support were prepared. The corresponding amount of CDMPC was dissolved in 10 mL of THF (23) Sun, L.; Annen, M.; Lorenzano-Porras, F.; Carr, P. W.; McCormick, A. V. J. Colloid Interface Sci. 1994, 163, 464-473.
Figure 1. Structures of chiral solutes.
and the solution was added to the slurry of zirconia in THF at a rate of ∼0.03 mL/min (∼5 h). The suspension was stirred overnight and then the solvent was slowly removed in a rotary evaporator at room temperature. Finally, the particles were dried in vacuo at 50 °C. We will refer to the above-described as the “standard” coating procedure. The following variants from the standard conditions were examined: (a) The 10-mL polymeric solution was added to the zirconia suspension in a single aliquot, instead of by the slow incremental addition described above. A 3% (w/w) final ratio was used in this experiment. The solvent was then immediately evaporated under reduced pressure. This is denoted as preparation A. (b) The surface was coated by precipitating the CDMPC. The chiral polymer solution in THF was added as in the standard coating procedure. Then 100 mL of n-heptane as the precipitating agent was added dropwise at ∼0.4 mL/min at the reflux temperature. The particles were finally filtered and washed with nheptane. This is denoted as preparation B.
Physical Characterization of CDMPC-Coated Zirconia. A FT-IR spectroscopic study was carried out to compare the difference in surface chemical properties of zirconia particles after treatment at 750 °C. Spectra were collected with a Nicolet Magna 750 FT-IR spectrophotometer equipped with a diffuse reflectance accessory (DRIFT) and with a heatable vacuum cell to hold the solid. The spectra were acquired at 110 °C with 4-cm-1 resolution and 512 scans. The pore structure of the particles after coating was characterized by nitrogen adsorption measurements. Samples were heated at 60 °C under vacuum for 6 h to remove adsorbed gases. The measurements were conducted using a Micrometrics ASAP 2000 sorptometer (Micrometrics Instrument Corp., Norcross, GA). The data were processed according to both the BET24 and BJH methods.25 (24) Brunauer, S.; Emmet, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309. (25) Barret, E. P.; Joyner, L. G.; Halenda, P. P. J. Am. Chem. Soc. 1951, 73, 373-380.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
3015
Figure 2. Infrared spectra of zirconia heated at 120 °C for 16 h (A) and treated at 750 °C for 5 h (B). Acquired at 110 °C, with 4-cm-1 resolution.
Carbon content on samples coated with different polymer loading was confirmed by elemental analysis. Chromatography. Packing materials were suspended in a (1: 1) hexane/2-propanol mixture and packed into 50 × 4.6 mm (i.d.) columns using the upward stirred slurry method. Packing pressure used was 5000 psi, and 2-propanol was employed as the displacing solvent. Chromatographic studies were performed on an HP 1100 liquid chromatograph (Hewlett-Packard S. A., Palo Alto, CA) equipped with vacuum degasser, quaternary pump, autosampler, thermostated-column device, variable-wavelength UV detector, and computer-based HP Chemstation. Chromatographic studies were performed at 30 °C, using 2-propanol/hexane mixtures as mobile phase. Flow rate was 1 mL/min. The dead time was estimated by using 1,3,5-tri-tert-butylbenzene as unretained compound.26 The injection volume was 0.5-1 µL. Detection was carried out at 254 nm. RESULTS AND DISCUSSION Zirconia Surface Characteristics. To remove the strongly polar hydroxyl groups from the zirconia surface,27 we treated the particles at 750 °C for 5 h, prior to coating with CDMPC. The spectra shown in Figure 2 were acquired after heating at 120 °C under vacuum (A) and at 750 °C (B), respectively. Hydroxyl band frequencies for monoclinic zirconia have been reported and discussed in detail.28-30 Generally from two to four discrete bands are observed in the range 3780-3400 cm-1 as well as a waterbending band at 1620 cm-1. Hertl31 ascribed the latter to H-bonded OH groups rather than to water. The spectrum taken at 120 °C qualitatively agrees with the reported hydroxyl bands. Surprisingly, the hydroxyl bands do not completely disappear after (26) Koller, H.; Rimbock, K. H.; Mannschreck, A. J. Chromatogr., A 1983, 282, 89-96. (27) Nawrocki, J.; Rigney, M. P.; McCormick, A.; Carr, P. W. J. Chromatogr., A 1993, 657, 229-282. (28) Tsyganenko, A. A.; Filimonov, V. N. Spectrosc. Lett. 1972, 5, 477. (29) Erkelens, J.; Rijnten, H. T.; Eggink-DuBurck, S. H. Recueil 1972, 91, 14261432. (30) Argon, P. A.; Fuller, E. L.; Holmes, H. F. J. Colloid Interface Sci. 1975, 52, 553. (31) Hertl, W. Langmuir 1989, 5, 96-100.
3016 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
Figure 3. Chromatograms of trans-stilbene oxide (A) and 1-(9anthryl)-2,2,2-trifluoroethanol (B) eluted from 3.9% CDMPC-coated zirconia: mobile phase, (90:10) n-hexane/2-propanol; flow rate, 1 mL/ min; column temperature, 30 °C.
heating at 750 °C. Spectrum B shows a quantitative difference in the intensity of the wide band at 3750-3400 cm-1 and in the band at 1620 cm-1, probably indicating incomplete dehydroxylation of the surface. This observation is in agreement with the thermogravimetric analysis performed by Nawrocki et al.32 using monoclinic zirconia washed sequentially with acid and base. The thermal history reported by Nawrocki indicates nearly complete removal of hydroxyl groups at 900 °C. Chromatography of Chiral Compounds. Figure 3 shows chromatograms of two enantiomeric mixtures (solutes 2 and 7) eluted with (90:10) n-hexane/2-propanol mixture from a 5.0 × 0.46 cm column packed with 3.9% CDMPC-coated zirconia. Even though the separation was not optimized, the peaks shown in the chromatograms are quite well resolved in a reasonable period of time. Also Table 1 displays capacity factors, selectivity factors, resolution, and plate counts of the racemic mixtures shown in Figure 1 obtained with the same chiral column. Most of the racemates could be successfully separated using n-hexane/2propanol as a mobile phase. Selectivity factors range from 1.04 to 2.45, depending on the solute. Plate counts larger than 2000 were obtained for most racemates. These are not as good as they might be for a 5-cm column packed with 2.5-µm particle size. We point out that plate counts (HETP) are quite generally poorer in most types of CSPs used in HPLC compared to those obtained in RPLC. (32) Nawrocki, J.; Carr, P. W.; Annen, M. J.; Froelicher, S. Anal. Chim. Acta 1996, 327, 261-266.
Table 1. Chromatographic Results of 16 Enantiomeric Mixtures on CDMPC-Coated Zirconia racemate no.
k′1a
selectivity
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
0.92 0.42 0.85 2.31 3.67 0.50 1.69 6.27 3.81 2.63 2.64 0.75 2.41 1.60 4.93 1.32
1.20 1.88 1.52 1.0 1.04 1.0 2.45 1.21 1.18 1.15 1.07 1.0 1.10 1.29 1.15 1.51
resolutionb 0.9 3.2 2.8
