Silica Xerogel as a Continuous Column Support for High-Performance

A preliminary study of the chromatographic performance and permeability of a continuous silica xerogel column under reversed-phase HPLC conditions was...
2 downloads 12 Views 176KB Size
Anal. Chem. 1996, 68, 2709-2712

Silica Xerogel as a Continuous Column Support for High-Performance Liquid Chromatography Steven M. Fields*

Hoechst Marion Roussel, Analytical and Structural Sciences, 2110 East Galbraith Road, Cincinnati, Ohio 45215

A preliminary study of the chromatographic performance and permeability of a continuous silica xerogel column under reversed-phase HPLC conditions was performed. A porous chromatographic support was synthesized inside a 0.32 mm i.d. × 13 cm fused silica tube from potassium silicate solution and derivatized with dimethyloctadecylchlorosilane. The plate height at 0.01 cm/s (0.5 µL/min), near the apparent optimum linear velocity, was about 65 µm. The column efficiencies in terms of numbers of plates per meter were 5000 and 13 000 for ethyl benzoate (k ) 0.8) and naphthalene (k ) 2.0), respectively, at 0.5 µL/min. The major parameter affecting column efficiency was the heterogeneous morphology of the xerogel, modifications to which are expected to improve chromatographic performance. The column provided efficiencies comparable to those reported for continuous polymeric columns but less than that previously reported for a continuous silica column. Gradient elution mode was demonstrated with a mixture of polycyclic aromatic hydrocarbons. The column was highly permeable, exhibiting a linear dependence of pressure to flow rate and a back pressure of only 632 psi at 10 µL/min when a 95% aqueous mobile phase was used. Open tubes and tubes filled with discrete particles are the primary formats of current chromatographic columns. In highperformance liquid chromatography (HPLC), microporous particles with diameters of 10 µm or less provide rapid, efficient separations. However, as the particle diameter decreases below 5 µm, two adverse affects are that the back pressure increases rapidly to the equipment limiting point and the packing of the column is more difficult. The latter issue is even more important in preparative separations and is, in part, an impetus for the development of perfusive particles,1 which are porous polymers containing throughpores that effectively decrease the mean particle size and diffusion distance and also increase column permeability. In contrast, surface-coated pellicular packings eliminate the stagnant mobile phase within the particles and, in the case of 1.5 µm diameter particles, provide very short analysis times.2 Continuous support columns, which consisted primarily of polyurethane foams cast in situ, were first evaluated for gas3-5 and liquid3,6,7 chromatography. This format, sometimes referred * Present address: Alza Corp., 950 Page Mill Rd., Palo Alto, CA 94303. (1) Afeyan, N. B.; Gordon, N. F.; Masaroff, I.; Varady, L.; Fulton, S. P.; Tang, Y. B.; Regnier, F. E. J. Chromatogr. 1990, 519, 1-29. (2) Glesche, H.; Unger, K. J. Chromatogr. 1989, 465, 39-57. (3) Ross, W. D.; Jefferson, R. T. J. Chromatogr. Sci. 1970, 8, 386-389. (4) Schnecks, H.; Bieber, O. Chromatographia 1971, 4, 109-112. S0003-2700(95)01247-9 CCC: $12.00

© 1996 American Chemical Society

to as a monolithic or rod column, was of interest from two perspectives: first, that it could yield a porous media with very high permeability and, second, that it could eliminate the tasks of particle synthesis and the sometimes difficult art of packing columns with discrete particles. Recently, Frechet and co-workers evaluated continuous polymethacrylate and polystyrene supports in a variety of HPLC modes, including reversed-phase partition,8,9 preparative ion-exchange separations of proteins,10,11 polymer precipitation-redissolution,12 and as a reactor bed for enzymatic digestion and affinity chromatography.13 Hjerten and co-workers have demonstrated that compressed polyacrylamide gels can be used to increase resolution of proteins in partition14-16 and ionexchange chromatography.17 Pretorius and co-workers18 synthesized continuous silica supports from silicate-surfactant solutions; however, no chromatographic data were presented, and the surfactant was not specified. Tanaka and co-workers19 synthesized a continuous silica column from the acid-catalyzed hydrolysis/condensation of tetraethoxysilane in the presence of an anionic surfactant.20 The material contained micropores of 10 nm and throughpores of 3 µm, producing plate heights of 45 µm for alkylbenzenes in 80% methanol. This report describes an investigation into the reversedphase HPLC chromatographic performance and permeability of continuous silica xerogels formed within a fused silica open tube from potassium silicate solution. EXPERIMENTAL SECTION The HPLC system used was a prototype pump on loan from Waters Corp., which provided a continuous mobile phase flow (5) Hileman, F. D.; Sievers, R. E.; Hess, G. G.; Ross, W. D. Anal. Chem. 1973, 45, 1126-1130. (6) Lynn, T. R.; Rushneck, D. R.; Cooper, A. R. J. Chromatogr. Sci. 1974, 12, 76-79. (7) Kubin, M.; Spacek, P.; Chromecek, R. Collect. Czech. Chem. Commun. 1967, 32, 3881-3887. (8) Svec, F.; Frechet, J. M. J. Anal. Chem. 1992, 64, 820-822. (9) Wang, Q. C.; Svec, F.; Frechet, J. M. J. J. Chromatogr., A 1994, 669, 230235. (10) Svec, F.; Frechet, J. M. J. J. Chromatogr., A 1995, 702, 89-95. (11) Svec, F.; Frechet, J. M. J. Biotechnol. Bioeng. 1995, 48, 476-480. (12) Petro, M.; Svec, F.; Gitsov, I.; Frechet, J. M. J. Anal. Chem. 1996, 68, 315-321. (13) Svec, F.; Frechet, J. M. J. Biotechnol. Bioeng. 1996, 49, 355-363. (14) Liao, J.-L.; Zhang, R.; Hjerten, S. J. Chromatogr. 1991, 586, 21-26. (15) Hjerten, S.; Li, Y.-M.; Liao, J.-L.; Mohammad, J.; Nakazato, K.; Pettersson, G. Nature 1992, 356, 810-811. (16) Hjerten, S.; Nakazato, K.; Mohammad, J.; Eaker, D. Chromatographia 1993, 23, 287-294. (17) Li, Y.-M.; Liao, J.-L.; Nakazato, K.; Mohammad, J.; Terenius, L.; Hjerten, S. Anal. Biochem. 1994, 223, 153-158. (18) Pretorius, V.; Davidtz, J. C.; Desty, D. H. J. High Resolut. Chromatogr. Chromatogr. Commun. 1979, 2, 583-584. (19) Tanaka, N.; Ishizuka, N.; Hosoya, K.; Kimata, K.; Minakuchi, H.; Nakanishi, K.; Soga, N. Kuromatogurafi 1993, 14, 50-51. (20) Nakanishi, K.; Soga, N. J. Am. Ceram. Soc. 1991, 74, 2518-2530.

Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2709

up to 10 µL/min, with full binary composition gradient capabilities. Composition gradient delays were only about 20 s at a flow rate of 5 µL/min. Solvent reservoir A contained 95/5 (v/v %) water/ acetonitrile, and solvent reservoir B contained 5/95 (v/v %) water/ acetonitrile. The analysis of the EPA 610 sample was performed with a composition gradient from 20 to 80% B after a 0.1 min initial hold. Injection was made with a high-pressure Cheminert valve (Valco) with internal sample loop of 0.1 µL, actuated using a Valco sequential valve timer and microelectronic actuator. Sample solutions were injected using a 1 s actuation. Detection was by UV absorption using a Linear 205 UV/visible detector (Thermoseparations, Inc.), equipped with a capillary UZ-View cell of 8 mm path length (LC Packings, Inc.). The UV cell and connecting tubing are formed from a single piece of fused silica. Connection of the column to the injector was made with a polyimide/graphite fused silica adapter (Valco) and to the UV cell tubing with a Teflon tubing connector (LC Packings). Electrochromatography was performed on a Prince capillary electrophoresis system (ATI Unicam). Detection was by UV absorption using a Unicam 4225, equipped with a capillary UZView cell of 3 mm path length (LC Packings). Connections were made in the same manner as with the HPLC system. The 0.32 mm i.d. fused silica (Polymicro Technologies) was not pretreated before the xerogel was cast. The xerogel was formed by a method similar to that used to cast column end frits in fused silica tubing for packed capillary HPLC.21 A 10 wt % solution of formamide in potassium silicate solution (Kasil No. 2130, PQ Corp.) was placed in a glass vial and vortexed until the formamide was completely dispersed. The fused silica column was filled with the solution and placed in a 100 °C gas chromatograph oven for 1 h. The column was then washed with water, 50/50 water/methanol, methanol, and tetrahydrofuran. The column was purged with dry helium for 24 h at 120 °C and then filled with a 10% solution of dimethyloctadecylchlorosilane (ODS) in dry toluene through an empty length of the same tubing (performed in an inflatable glovebag under nitrogen). The column and empty tubing ends were connected to form a circle and then heated at 70 °C for 5 h. The column was then washed with toluene, tetrahydrofuran, methanol, 50/50 water/methanol, and water. Short lengths of the column were cut off and vapordeposited with gold for observation with a scanning electron microscope (JEOL, Model JSM-840). For chromatographic evaluation, the column void volume was measured as the elution time of sodium nitrate, from which the mean linear velocity was calculated. Gravimetric determination of the column void volume using methanol and chloroform yielded a value about 10% lower than that obtained with the sodium nitrate method, a reasonable variation between the techniques.22 Chromatographic test analytes were obtained from Aldrich Chemical Co., and the EPA 610 sample was obtained from Supelco, Inc. Acetonitrile was HPLC-grade (Burdick and Jackson, Co.), and water was obtained from an in-house deionized source and further purified with a Barnstead Nanopure system. Anhydrous toluene and ODS were obtained from Fluka Chemical Co. RESULTS AND DISCUSSION The polymerization of the potassium silicate solution produced a continuous structure (Figure 1), which was chemically bonded to the fused silica column through surface silanols. The process for forming this xerogel differs from other methods23 in that the 2710 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996

Figure 1. Scanning electron micrograph of silica xerogel column end. The bar on the lower right represents 1 µm.

gelation time was shortened because the drying stage was initiated within a few minutes of introduction into the column. The morphology of the material was clearly inhomogeneous, but it does have a few identifiable features. First, the mean pore diameter appears to be on the order of 2 µm, with a distribution ranging from about 0.2 to about 3 µm. The bulk of the silica mass consists of sheets and globules interconnected by varying diameter rods. The silica occupies 12% of the column volume, as determined from the volume of the empty tube and the gravimetrically determined void volume. This is slightly less silica mass than that reported by Tanaka19 of 15%, and both values are similar to those for columns packed with porous silica.24 The only characterization of pore size was performed using SEM at a magnification of 100000×, the resolution limit of the instrument. Under these conditions, the sheet areas showed no indication of micropores, although pores smaller than about 100 Å would not have been visible. In contrast, the continuous silica made by Pretorius and co-workers18 was much more uniform in structure because of the use of emulsion and aerosol techniques during synthesis. The silica polymerization method20 used by Tanaka19 also incorporated a surfactant, which presumably produced a fairly uniform morphology similar to that shown in the indicated polymerization report. The polyurethane columns3,5 used for GC and HPLC consisted of solid, 1-10 µm diameter spheres, which coagulated to form large pore systems. The more recent polymethacrylate and polystyrene columns made by Frechet and co-workers10,12 contain bimodal pore size distributions, indicating the presence of both throughpores and micropores. Several silica columns were constructed, and all appeared very similar in terms of throughpore size and general morphology under SEM. All columns exhibited similar permeability, with (21) Cortes, H. J.; Pfeiffer, C. D.; Richter, B. E.; Stevens, T. S. J. High Resolut. Chromatogr. Chromatogr. Commun. 1987, 10, 446-448. (22) Simpson, C. F. In Techniques in Liquid Chromatography; Simpson, C. F., Ed.; Wiley: New York, Chapter 1, 1982. (23) Brinker, C. J.; Scherer, G. W. Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing; Academic Press: Boston, 1990. (24) Hancock, W. S. In Handbook of HPLC for the Separation of Amino Acids, Peptides, and Proteins; Hancock, W. S., Ed.; CRC Press: Boca Raton, FL, 1984; Chapter 1, Vol. 1.

Table 1. Continuous Silica Xerogel Column Permeability flow rate,a µL/min

back pressure, psi

flow rate,a µL/min

back pressure, psi

0.5 1 2 3

23 60 127 192

4 5 7 10

258 325 448 632

a

Mobile phase: 95/5 (v/v %) water/acetonitrile, room temperature.

back pressures varying by no more than 10-20%. However, columns longer than about 20 cm contained voids (0.1-1 mm long) that were visible without magnification. One of these columns was derivatized using an ODS concentration significantly higher than that estimated to be required for monomeric coverage.25 An evaluation of the reproducibility of the silanization process was not conducted, but with proper dedicated equipment it is expected that column variability should be reasonably low. The permeability of the column was evaluated under aqueous mobile phase conditions. The high permeability of the column is evident from the pressure drop at different flow rates using 95/5 (v/v %) water/acetonitrile (Table 1). The pressure was a linear function of flow rate over the range of 0.5-10 µL/min (r2 ) 0.9995, y-intercept ) -2.1 psi). In contrast, a 0.32 mm i.d. × 15 cm long capillary column packed with 5 µm spherical silica generates back pressures close to 1000 psi under the same conditions, and a 4.6 mm i.d. column back pressure is usually near 2000 psi. The back pressure is very similar to that reported for continuous polymeric columns.6,10 The column evaluated by Tanaka19 (6.8 mm i.d., 7.7 cm long) generated a significantly lower back pressure of only 15 psi at 1 mL/min with 80% methanol. A 13 cm long piece of the column was evaluated under reversed-phase HPLC conditions. The efficiency was measured over the flow rate range of 0.5-10 µL/min using ethyl benzoate and naphthalene; the chromatogram at 5 µL/min is shown in Figure 2. By comparison, a flow rate of 2-5 µL/min is typical for the same diameter column used in packed capillary HPLC. All analytes showed the typical H vs u curves of HPLC, with an apparent coupling26 of the eddy diffusion and mass transfer effects at higher flow rates (Figure 3). The optimum linear velocity was not determined but must be