Supercritical Fluid Generated Stationary Phases for Liquid

for Liquid Chromatography and Capillary. Electrochromatography. Liam O. Healy, Vincent P. Owens, Tom O'Mahony, Supalax Srijaranai,† Justin D. Holmes...
0 downloads 0 Views 178KB Size
Download PDF
Anal. Chem. 2003, 75, 5860-5869

Supercritical Fluid Generated Stationary Phases for Liquid Chromatography and Capillary Electrochromatography Liam O. Healy, Vincent P. Owens, Tom O’Mahony, Supalax Srijaranai,† Justin D. Holmes, and Jeremy D. Glennon*

Analytical Chemistry, Department of Chemistry, University College Cork, Cork, Ireland Gerd Fischer and Klaus Albert

Institut fu¨r Organische Chemie, Universita¨t Tu¨bingen, D-7207 Tu¨bingen, Germany

Chromatographic silica-bonded stationary phases have been prepared using supercritical CO2 as the reaction medium. 29Si solid-state NMR spectra of the generated bonded silica phases show unreacted silica species Q3 and Q4, alongside important resonances for surfacebound ligands, T1, T2, and T3. Initially, a fluorinated octyl silica (C8) phase was produced, by reacting 1H,1H,2H,2Hperfluorooctyltriethoxysilane with silica particles (3 µm) in sc-CO2 at 60 °C and 450 atm for 3 h. In-house-packed LC columns of this fluorinated sc-C8 silica phase yielded typical reversed-phase behavior when a standard test mix of benzamide (k′ ) 1.03), benzophenone (k′ ) 8.11), and biphenyl (k′ ) 14.92) was eluted. When packed into fused-silica capillaries for CEC, this fluorinated sc-C8 silica phase gave linear plots of log k′ versus percentage MeOH for benzophenone and biphenyl and, in contrast to octyl or octadecyl silica phases, displayed selectivity for aromatic thioureas when chromatographed among a series of synthetic organic thiourea test solutes. Similarily, an octadecyl silica phase (sc-C18 silica) was prepared by reaction of n-octadecyltriethoxysilane in sc-CO2, packed at 9500 psi and examined by LC. The sc-C18 silica LC column yielded high column efficiency (up to 141 000 N/m (fluorene)) and excellent asymmetry factors (1.06, fluorene) without resource to end-capping. Following a second silylating or end-capping step using hexamethyldisilazane in sc-CO2, sc-end-capped sc-C18 silica phases elute N,N-DMA before toluene and the toluidine isomers as a single peak, indicating lowered silanol activity according to the Engelhardt test. A rapid separation of the important pharmaceutical substances, ketoprofen, naproxen, fentoprofen, and ibuprofen, on an sc-end-capped scC18 silica phase is also shown. Most chromatographic stationary phases today are composed of two distinct partssa support and an attached chemical moiety or ligand. Stationary-phase materials include silica,1-3 alumina,4 * Corresponding author. E-mail: [email protected]. † On leave from the Department of Chemistry, Faculty of Science, Khon Kaen University, Thailand. (1) Nawrocki, J. J. Chromatogr., A 1997, 779, 29. (2) Zhuravlez, L. T. Colloids Surf. 2000, 173, 1.

5860 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

polystyrene-divinylbenzene,5 and porous graphitic carbon.6 Of these, silica is the most widely used due to its favorable properties, including the relative ease with which it can be functionalized.7 A wide range of ligands has been successfully immobilized on silica, ranging from straight-chain hydrocarbons, of which C8 and C18 chain lengths are the most popular,8 to complex macrocycles such as cyclodextrins, calixarenes, and antibiotics.9 The usual manner in which these phases are prepared is to introduce a reactive form of the ligand to the support, forming covalent bonds to ensure a stable bonded phase. There are a number of well-known synthetic routes to silica bonded phases. Silica can first be treated with thionyl chloride and then reacted with an amine, with the elimination of HCl, to form an Si(s)-NR linkage (where Si(s) denotes a surface silicon atom). Two distinct synthetic routes to the formation of stable Si(s)-C linkages can also be taken. The first route involves reacting the silica with thionyl chloride and then drying at 350 °C to remove any water, followed by the addition of a Grignard.10 The second alternative route involves the hydrosilation of unsaturated hydrocarbons on silica hydride surfaces. This approach was used by Sandoval and Pesek to synthesize a number of bonded phases.11 It was also used by our laboratory to successfully immobilize calixarene macrocycles.12 Pesek and co-workers have recently shown that C18 phases prepared by this silanization/hydrosilation method and subsequently end-capped show very good peak asymmetry and long lifetimes, even when exposed to acidic and alkaline mobile eluents. Pesek et al. also used a similar methodology to react an alkyne with a silica hydride intermediate, forming a bidentate ligand.13 (3) Stella, C.; Rudaz, S.; Veuthey, J. L.; Tchapla, A. Chromatographia 2001, 53, 113. (4) Braithwaite, A.; Cooper, M. Chromatographia 1996, 42, 77. (5) Edwards, B. R.; Giauque, A. P.; Lamb, J. D. J. Chromatogr., A 1995, 706, 69. (6) Karlsson, A.; Karlsson, O. J. Chromatogr., A 2001, 905, 329. (7) Scott, R. P. W. Silica Gel and Bonded Phases; John Wiley & Sons Ltd.: West Sussex, England, 1993. (8) Pursch, M.; Brindle, R.; Ellwanger, A.; Sander, L. C.; Bell, C. M.; Ha¨ndel, H.; Albert, K. Solid State Nucl. Magn. Reson. 1997, 9, 191. (9) Armstrong, D. W.; Tang, Y. B.; Chen, S. S.; Zhou, Y. W.; Bagwill, C.; Chen, J. R. Anal. Chem. 1994, 66, 1473-1484. (10) Yamamoto, K.; Tatsumi, T. Microporous Mesoporous Mater. 2001, 44, 459. (11) Sandoval, J.; Pesek, J. J. Anal. Chem. 1991, 63, 2634. (12) Brindle, R.; Albert, K.; Harris, S. J.; Troltzsch, C.; Horne, E.; Glennon, J.D. J. Chromatogr., A 1996, 731, 41. (13) Pesek, J. J.; Matyska, M. T.; Oliva, M.; Evanchic, M. J. Chromatogr., A 1998, 818, 145. 10.1021/ac034511q CCC: $25.00

© 2003 American Chemical Society Published on Web 10/03/2003

Existence of the bidentate structure was confirmed using solidstate NMR studies and DRIFT spectra. Despite the enhanced hydrolytic stability of the Si(s)-C linkage, it is not as commonly utilized as the formation of Si(s)-O-Si-R linkages, which has become the industry’s choice for the production of liquid chromatographic phases today.14 The general form of this reaction is

Si(s)-OH + SiRnX4-n f Si(s)OSiRnX3-n + HX

(n ) 1-3)

where X is usually chloro or an alkoxy group.15 The nature of the reagent along with the presence of water on the silica surface determines the eventual form of the phase.16 Three different types of phases can be produced, namely, monomeric or “brush”-type phases, oligomeric phases, and polymeric or “bulk”-type phases. These reactions of alkoxysilanes and chlorosilanes with silica have been extensively studied.17 The most favorable method of synthesis is to pass a gaseous stream of an organosilane at high temperatures (>300 °C) over the silica.18 The chloro or alkoxy group (X) reacts with the surface hydroxyl group on the silica oxide leaving the organo group extending from the surface. Alternatively, if a nonvolatile organosilane is employed, it is reacted with the silica in a nonaqueous liquid solution below 100 °C.19 The organosilane reacts with trace amounts of water (present either on the silica or in the solution) to form an organosilanol which, in turn, reacts with the surface silanol groups.20

RnSiX4-n + (4 - n)H2O f RnSi(OH)4-n + (4 - n)HX (n ) 1-3) Si(s)OH + RnSi(OH)4-n f Si(s)O-Si(OH)3-nRn + H2O Bonded phases are characterizable using both spectroscopic and chromatographic means. Solid-state NMR21 and IR22 are the principal spectroscopic tools employed. When Pesek et al. used silanization/hydrosilation reactions previously described, to modify silica with 4-phenyl-1-butene, DRIFT spectra of the silica product confirmed that first silanization and then hydrosilation did occur.23 Garbasi et al. similarly used DRIFT spectral analysis to monitor the reaction between trichlorooctadecylsilane and silica24 while Pe´re´ et al. demonstrated that DRIFT could be used to quantify the amount of organic compound immobilized on the silica surface.25 Solid-state NMR has the advantage of being better able to differentiate between the different forms of silanols and can give (14) Stella, C.; Rudaz, S.; Veuthey, J.-L.; Tchapla, A. Chromatographia 2001, 53, S113. (15) Nakamura, H.; Matsui, Y. J. Am. Chem. Soc. 1995, 117, 2651. (16) Barrett, D. A.; Brown, V. A.; Shaw, P. N.; Davies, M. C.; Ritchie, H.; Ross, P. J. Chromatogr. Sci. 1996, 34, 146. (17) Blu ¨ mel, J. J. Am. Chem. Soc., 1995, 117, 2112. (18) Tripp, C. P.; Hair, M. L. Langmuir 1991, 7, 923. (19) Tripp, C. P.; Hair, M. L. Langmuir 1992, 8, 1120. (20) Sagiv, J. J. Am. Chem. Soc. 1980, 102, 92. (21) Albert, K.; Bayer, E. J. Chromatogr. 1991, 544, 345. (22) Combes, J. R.; White, L. D.; Tripp, C. P. Langmuir 1999, 15, 7870. (23) Pesek, J. J.; Matyska, M. T.; Soczewinski, E.; Christensen, P. Chromatographia 1994, 9, 520. (24) Garbassi, F.; Balducci, L.; Chiurlo, P.; Deiana, L. Appl. Surf. Sci. 1995, 84, 145. (25) Pe´re´, E.; Cardy, H.; Cairon, O.; Simon, M.; Lacombe, S. Vib. Spectrosc. 2001, 25, 163.

an accurate picture of both the functionalized and underivatized silanol groups.26,27 29Si spectra are particularly useful at determining whether a monofunctional, bifunctional, or trifunctional ligand is used since each reacts differently with the silica surface, producing characteristic resonances.28 A phase that has been endcapped can similarly be characterized using 13C NMR, since the methyl groups bound to the silicon atom give unique resonances not observed in non-end-capped bonded silica phases. In addition, a significant body of work also exists detailing how various characteristics of stationary phases, such as silanol activity, hydrophobicity, and column efficiency, can be determined chromatographically.29-36 In 1991, Engelhardt et al. formulated what is today one of the most widely used chromatographic tests.34 Through a series of elutions, an LC column is classified as “good” or “bad”, depending on its performance in the tests. The tests employ seven test probessaniline, phenol, N,N-dimethylaniline (DMA), toluene. and p-, o-, and m-toluidine. The mobile-phase conditions are MeOH-H2O (55:45, v/v). The test decrees that aniline should elute before phenol, the reason being that the basic aniline would be more susceptible to undesirable interaction with surface silanol groups. If it elutes before phenol, structurally very similar but not prone to silanol interaction, then the effects of silanol activity are minimal. This same reasoning also dictates that DMA should elute before toluene. Furthermore, any peak tailing observed for these solutes, corresponding to interaction with residual silanols, is undesirable. The ratio of peak asymmetries for aniline and phenol should be smaller than 1.3. In addition, a phase exhibiting very little silanol activity should not be able to separate isomeric toluidines, which only differ in their pKa values and not in their hydrophobicities. Walters chose two test solutes that differ in their primary mode of retention on a chromatographic stationary phase.37 Anthracene is assumed to be retained solely due to hydrophobic interactions, while N,N-diethyltoluamide is thought to be highly dependent on silanophilic interactions. Silanol activity is determined by the ratio of capacity factors of the two probes, while hydrophobicity is calculated by observing the ratio of capacity factors of anthracene and benzene. Acetonitrile is used as the organic component of the mobile phase to ensure good solvation of the alkyl chains. Galushko proposed a calculatory model, which could predict selected column characteristics as a result of the elution of four test probessaniline, phenol, benzene and toluene.38 Scholten et al. doubted the reliability of calculating the strength of aminesilanol interactions from just one basic analyte, noting better (26) Schaefer, J.; Stejskal, E. O. J. Am. Chem. Soc. 1976, 98, 1031. (27) Albert, K.; Brindle, R.; Martin, P.; Wilson, I. D. J. Chromatogr., A 1994, 655, 253. (28) Pursch, M.; Sander, L. C.; Albert, K. Anal. Chem. 1996, 68, 4107. (29) Claessens, H. A.; van Straten, M. A.; Cramers, C. A.; Jezierska, M.; Buszewski B. J. Chromatogr., A 1998, 826, 135. (30) Rogers, S. D.; Dorsey, J. G. J. Chromatogr., A 2000, 892, 57. (31) Vervoort, R. J. M.; Debets, A. J. J.; Claessens, H. A.; Cramers, C. A.; de Jong, G. J. J. Chromatogr., A 2000, 897, 1. (32) Kimata, K. K.; Iwaguchi, K.; Onishi, S.; Jinno, K.; Eksteen, R.; Hosoya, K.; Araki, M.; Tanaka, N. J. Chromatogr. Sci. 1989, 27, 721. (33) Buszewski, B.; Gadzala-Kopciuich, R. M.; Markuszewski, M.; Kaliszan, R. Anal. Chem. 1997, 69, 3277. (34) Engelhardt, H.; Low, H.; Go ¨tzinger, W. J. Chromatogr., A 1991, 544, 371. (35) Grossmann, F.; Ehwald, V.; du Fresne von Hohenesche, C.; Unger, K. K. J. Chromatogr., A 2001, 910, 223. (36) Engelhardt, H.; Lobert, T. Anal. Chem. 1999, 71, 1885. (37) Walters, M. J. J. Assoc. Off. Anal. Chem. 1987, 70, 465. (38) Galushko, S. V. Chromatographia 1993, 36, 93.

Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

5861

results for the Engelhardt test.39 However, Jiskra et al. successfully used the Galushko test for electrochromatography as well as for pressure-driven systems.40 A recent paper by Berek and Tarbajovska outlines the use of macrocycles to determine silanol activity on typical reversed-phase stationary phases such as C18 silica.41,42 Supercritical CO2 has certain properties that make it an attractive solvent compared with traditional organic solvents. It is nonflammable, more environmentally friendly, and considerably less hazardous than most organic solvents. The supercritical state can be achieved at comparably low temperatures (31.2 °C) and pressures (1058 psi) when compared to other substances. The density and solvating power can be “tuned” by varying temperature and pressure. High diffusivity and mass-transfer kinetics in sc-CO2 should provide improved access to silanol groups in porous silica and faster reactions than in conventional solvents. These advantageous properties of sc-CO2 have yet to be tested for producing bonded silica chromatography phases. Organosilanes have been used before to modify materials using supercritical fluids as the reaction solvent. Combes et al. monitored a surface reaction between various silanes and silica in supercritical and liquid CO2, monitoring the results by IR.43 Shin et al. used supercritical CO2 to successfully modify a commercial zeolite with mercaptopropyl silane.44 Silica-based phases can experience difficulties with residual surface silanols interacting with certain analytes.45 This is especially pronounced for basic compounds.46 To overcome this problem, a phase is end-capped after the ligand is attached.47 This silylation process uses a silylating agent, such as trimethylchlorosilane or hexamethyldisilizane to react with these residual surface silanols, thereby inhibiting unwanted attractions to analytes. Yarita et al. successfully employed supercritical CO2 to end-cap an octadecyl silica (C18) chromatographic stationary phase.48 This research examines the functionalization of silica with organosilanes in supercritical CO2 to generate chemical bonded phases for liquid chromatography and capillary electrochromatography. Solid-state NMR is chosen for surface characterization. The fluoroalkyl and alkyl bonded silicas produced are evaluated chromatographically using test solutes to assess silanol activity and column efficiency. A further supercritical end-capping reaction is also investigated. Applications in reversed-phase LC and CEC separations of organic thioureas and pharmaceutical drugs are examined. EXPERIMENTAL SECTION Materials. Silica (Hypersil, 3 µm) was obtained from Thermo Hypersil-Keystone (Runcorn, Cheshire, U.K.). HPLC grade acetonitrile, methanol, and isopropyl alcohol were obtained from (39) Scholten, A. B.; Claessens, H. A.; de Haan, J. W.; Cramers, C. A. J. Chromatogr., A 1997, 759, 37. (40) Jiskra, J.; Cramers, C. A.; Byelik, M.; Claessens, H. A. J. Chromatogr., A 1999, 862, 121. (41) Berek, D. J. Chromatogr., A 2002, 950, 75. (42) Berek, D.; Tarbajovska, J. J. Chromatogr. A 2002, 976, 27. (43) Combes, J. R.; White, L. D.; Tripp, C. P. Langmuir 1999, 15, 7870. (44) Shin, Y.; Zemaniam, T. S.; Fryxell, G. E.; Wang, L. Q.; Liu, J. Microporous Mesoporous Mater. 2000, 37, 49. (45) Scholten, A. B.; Claessens, H. A.; de Haan, J. W.; Cramers, C. A. J. Chromatogr., A 1997, 759, 37. (46) Buszewski, B.; Schimd, J.; Albert, K.; Bayer, E. J. Chromatogr., A 1991, 552, 415. (47) McCalley, D. V.; Brereton, R. G. J. Chromatogr., A 1998, 828, 407. (48) Yarita, T.; Nomura, A.; Horimoto, Y. J. Chromatogr., A 1996, 724, 373.

5862 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

Figure 1. Structure of the synthetic organic thiourea test solutes.

Labscan (Dublin, Ireland). All water used was distilled and deionized to a resitivity of 18.2 M Ω‚cm. Aniline, N,N-dimethylaniline, fluorene, uracil, biphenyl, benzamide, dimethyl phthalate, anisole, and diphenylamine were obtained from Fluka (Dublin, Ireland). p-, m-, and o- toluidine were obtained from Riedel-de Hae¨n (Hanover, Germany). Phenol was obtained from BDH Chemicals (Poole, England). Benzophenone was obtained from Aldrich (Steinheim, Germany). CO2 was obtained from BOC gases. 1H,1H,2H,2H-perfluorooctyltriethoxysilane was obtained from Lancaster Synthesis (Morecambe, England). n-Octadecyldimethylmethoxysilane, n-octadecyltriethoxysilane, and hexamethyldisilazane were obtained from Fluka. The synthesis of the organic thiourea derivatives, as test solutes, was carried out in-house and will be reported elsewhere. Included were T1, CF3(CF2)2CH2NHCSNHCH3; T2, CH3(CH2)7NHCSNHCH3; T3, PhNHCSNHCH2CF3; T4, PhNHCSNHCH2(CF2)2CF3, and T5, CH3(CH2)11NHCSNHCH3 (Figure 1). Reaction Procedure. Fluorinated sc-C8 Phase. The first trial reaction was carried out using a fluorinated ligand. The reaction was performed using an Isco model 260D syringe pump with an external stainless steel reaction cell. The reaction cell (internal volume 50 mL, i.d. of 2.5 cm) was custom-made using stainless steel by High-Pressure Limited. Silica (2.21 g, Hypersil, 5 µm) was added, along with 1H,1H,2H,2H-perfluorooctyltriethoxysilane (0.359 mL) and a magnetic stirrer bar. The cell was filled with 50 mL of CO2, the temperature raised to 60 °C, and the pressure raised to 450 atm. The stirrer plate was switched on, ensuring agitation of the reactants in the supercritical CO2, and the reaction allowed to proceed for 3 h. The system was then allowed to cool by removing the cell from the water bath, depressurized, and the

Figure 2.

29Si

CPMAS solid-state NMR of the fluorinated sc-C8 silica phase.

modified silica was recovered. Elemental analysis of the fluorinated sc-C8 phase gave C 5.54% and H 0.78%. Non-End-Capped sc-C18 Phase. A C18 silica phase was also synthesized using the same apparatus. Silica (2.24 g, Hypersil, 3 µm) was added along with 0.387 g of n-octadecyltriethoxysilane. The reaction was allowed to proceed for 3 h, under the same conditions as above, again with stirring. The system was then allowed to cool, depressurized, and the modified silica was recovered. Elemental analysis of the sc-C18 silica gave C 20.58% amd H 1.44%. sc-End-Capped sc-C18 Phase. To produce an end-capped sc-C18 silica phase using sc-CO2, an sc-C18 silica batch was prepared as described above. After the 3-h reaction, 1.5 mL of hexamethyldisilazane was added to the reacted silica. The system was again pressurized to 450 atm and the temperature raised to 60 °C. The end-capping reaction was allowed to proceed for 3 h, before recovering the end-capped sc-C18 phase. When reactions are completed, stirring is stopped, and the headspace above the silica is dynamically removed under supercritical fluid conditions. The stationary phases also receive a washing in 2-propanol. 29Si and 13C CPMAS NMR Spectra. Solid-state NMR analysis was carried out, as recently reviewed elsewhere.49 13C CPMAS NMR spectra were obtained using a Bruker DSX 200 instrument. Magic angle spinning was carried out with 4-mm double bearing rotors of ZrO2 and spinning rates of 10 000 Hz. The proton 90° (49) Albert, K. J. Sep. Sci. 2003, 3-4, 215.

pulse length was 7 µs; the contact time and delay time were 2 ms and 1s, respectively. The line broadening used was 30 Hz, and all chemical shifts were referenced to glycine. 29Si CPMAS NMR spectra were also obtained using a Bruker DSX 200 instrument. Magic angle spinning was carried out with 7-mm double bearing rotors of ZrO2 and spinning rates of 4000 Hz. The proton 90° pulse length was 5.5 µs; the contact time and delay time were 5 ms and 1 s, respectively. The line broadening used was 50 Hz, and all chemical shifts were referenced to Q8M8. Column Packing for Liquid Chromatography. The fluorinated sc-C8 silica phase was packed in-house on a Shandon column packer (Shandon, U.K.). Isopropyl alcohol (HPLC grade, Merck, Darmstadt, Germany) was used as a packing solvent and 50:50 methanol-water used as a conditioning solvent. Packing was performed at 6000 psi, employing 100 mL of packing solvent followed by inversion of the column and packing with 100 mL of conditioning solvent. All chromatography columns were made of stainless steel, were of length 150 mm and internal diameter 4.6 mm, and were obtained from Jones Chromatography (Glamorgan, U.K.). The sc-C18 phases were packed to a standard employed in the production of commercial phases, with 2-propanol as the packing solvent and packing pressures of 9500 psi. Slurry volume was 100 mL with a packing flow of 100 mL min-1. All chromatography was performed on an Agilent 1100 series LC system, with recorded back pressures typically 1100-1300 psi at 1.0 mL min-1. All mobile phases were filtered and degassed before use. Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

5863

Figure 3. phase.

13C

CPMAS solid-state NMR of the fluorinated sc-C8 silica

Capillary Electrochromatography. All CEC chromatograms were obtained using a Beckman P/ACE MDQ instrument (Beckman, Fullerton, CA). The system comprised a 0-30-kV highvoltage power supply, a diode array detector, and the P/ACE software (version 1.6) for system control and data processing.

Figure 4. side.

29Si

Fused-silica capillary of 75 µm i.d., total length of 30 cm (21 cm to the detector), was used. The temperature was controlled at 25 °C using a fluorocarbon-based cooling system. All separations were carried out under normal polarity conditions (EOF toward cathode). Samples were introduced into the packed capillary by voltage (5 kV for 2 s), unless stated otherwise. Preparation of Packed Capillaries. For frit preparation, a small amount of silica powder (Hypersil, 3 µm) was moistened with a dilute solution of sodium silicate to form a paste. This was introduced into the capillary by pushing the end of the capillary into the paste, against a hard surface, until 1.0-2.0 mm of the capillary was packed. The frit was then sintered with gentle heating over the filament of the frit former (InnovaTech, Hertfordshire, U.K.). The frit was tested by applying a pressure of 7000 psi (Shimadzu LC 8A pump) with methanol, whereupon a fine spray of methanol indicated the formation of a successful frit with minimum flow resistance and good permeability to allow maximum packing material flow. The stationary phase (∼100 mg) was initially slurried in 1.5 mL of methanol and placed in an ultrasonic bath for ∼30 min. The slurry was removed by syringe and transferred to the slurry chamber, which consisted of a 150mm-length, 4.6-mm-i.d. stainless steel tubing. The slurry reservoir used was an empty stainless steel liquid chromatography column (Shandon). The fused-silica capillary (35 cm × 75 µm i.d.) protruded ∼30 mm into the reservoir, effectively upside down, and the slurry pumped at 8000 psi (Shimadzu LC 8A pump operated at constant pressure mode). The packing solvent chosen

CPMAS solid-state NMR of the sc-C18 silica phase. Known silicon resonances for surface species are quoted at the left-hand

5864 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

Figure 7. Log k′ for derivatized thioureas versus the percent acetonitrile content of the mobile phase. Conditions: 30 cm × 75 µm i.d. (21-packed length) fluorinated sc-C8 phase; mobile phase, acetonitrile-5 mM sodium tetraborate pH 8.0; UV detection, 214 nm.

Figure 5. 13C CPMAS solid-state NMR spectrum of the sc-C18 phase. Known carbon resonances are given on the left-hand side with the experimental spectrum and resonances on the right.

Figure 8. Electrochromatogram of the separation of organic thiourea solutes on fluorinated sc-C8 silica. Conditions: 30 cm × 75 µm i.d. (21-cm packed length); mobile phase, acetonitrile-5 mM sodium tetraborate pH 8.0 (75:25); voltage, 15 kV; UV detection at 214 nm. Figure 6. Log k′ of benzophenone and biphenyl versus percent organic modifier in the mobile phase. Conditions: 30 cm × 75 µm i.d. (21-cm packed length) fluorinated sc-C8 phase; mobile phase, methanol-25 mM Tris pH 8.2; voltage, 10 kV; UV detection, 214 nm, thiourea as the EOF marker.

was methanol. When a sufficiently long piece of capillary had been packed (confirmed by placing the capillary in front of a black background), the pressure was slowly removed to prevent disturbance of the packed bed. Upon depressurization, the packing solvent was removed and replaced with water. The system was then repressurized (8000 psi) with a water flow for 1 h. An end frit and retaining frit were then formed using the frit former, to give a working length of 21 cm. Finally, the flow was inverted such that the direction of flow was toward the retaining frit. This inversion step was necessary for the removal of unwanted packing. The completed capillary was then mounted in the capillary cartridge holder and the detector window formed 1-2 mm beyond the retaining frit. Test Solute and Mobile-Phase Preparation. Test mixtures containing five thiourea derivatives (T1-T5, 5 mM each) were made up in 100% methanol. All eluents were filtered using an aqueous 0.45-µm Millipore filter membrane and sonicated (ULTRAsonik NEY) for 10 prior to use. A 0.025 M Tris buffer solution was prepared by dissolving 3.92 g of Tris-HCl with 3.025 g of Tris

base in 1 L of deionized water. A 0.005 M sodium tetraborate buffer was prepared by dissolving 1.9 g of the solid in 1 L of deionized water. The pH values of 8.0 and 8.2 were recorded respectively with a pH meter (Expandable Ion Analyser pH meter EA 920). To maintain constant ionic strength in the mobile phase, the appropriate volumes of organic modifier, water, and buffer were added together. Conditioning of Packed Capillary Columns. Upon preparation, capillary columns (75-µm i.d., 21-cm packed length, 30-cm total length) were flushed with mobile phase (acetonitrile-25 mM sodium Tris pH 8.2 (80:20)) for 1 h using an LC pump at 2000 psi (Shimadzu LC 8A) and then installed in the capillary cartridge. Both the inlet and outlet vials were pressurized, and the voltage was set to 20 kV for 40 min until the current stabilized. This procedure was employed whenever a new mobile phase was used. If bubble formation occurred, the capillary was connected to the LC pump and flushed with mobile phase for 15 min. RESULTS AND DISCUSSION 29Si and 13C CPMAS NMR. The reaction of 1H,1H,2H,2Hperfluorooctyltriethoxysilane with silica in sc-CO2 was chosen as a first experiment due to the expected solubility of the reactive ligand. Fluorinated phases can provide a unique separation selectivity for a range of solutes, particularly for the retention of Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

5865

Figure 9. Chromatogram showing a test mixture elution by LC on a non-end-capped sc-C18 silica column (150 mm × 4.6 mm i.d, 3-µm particles). Mobile phase used was 50:50 acetonitrile-water (v/v) pumped at a flow rate of 1.00 mL/min.

fluorinated compounds.50 In the fluorinated sc-C8 silica phase prepared here, while elemental analysis indicated ligand immobilization, detailed surface characterization is available from examination of the 29Si solid-state NMR spectrum. Alongside unreacted silica species Q3 and Q,4 are important resonances for surface bonded ligands, T2, T3, and T1, the latter in particular contributing to residual silanol content (Figure 2). 13C CPMAS NMR resonances for 1H,1H,2H,2H-perfluorooctane silica, prepared from silica hydride intermediate, show that methylene resonances are shifted to low field by the electron-withdrawing effect of the adjacent fluorine atoms.23 Electron-withdrawing effects on methylene resonances are also evident here for the stationary phase prepared in sc-CO2. The 13C spectrum shows two resonances shifted to a lower field because of the adjacent fluorine atoms (Figure 3). Similar 29Si solid-state NMR assignments can be made for the sc-C18 silica phase, as evident from Figure 4. Surface-bound species, T1, T2, and T3, alongside Q3 and Q4 silica resonances, confirm reaction of the triethoxy ligand. Partial resolution of the methylene groups is possible using 13C CPMAS solid-state NMR (Figure 5). The large peak at 32.81 ppm corresponds to the bulk of the carbon atoms in the hydrocarbon chain. Expected resonances are shown on the left in Figure 5 and are in good agreement with the values determined experimentally. These higher field methylene signals can be contrasted with those of the fluorinated sc-C8 silica phases shown earlier in Figure 3. Chromatographic Performance. Apart from specific test solutes used to examine silanol activity, the standard reversedphase test mixtures, such as benzamide, benzophenone, and biphenyl, alongside a series of synthetic organic thioureas, were used to examine the liquid chromatographic performances of the phases generated in sc-CO2. Using a mobile phase of 50:50 acetonitrile-water, the in-house-packed fluorinated sc-C8 phase yielded typical reversed-phase behavior when a standard test mix (50) Tzschucke, C. C.; Markert, C.; Bannwarth, W.; Roller, S.; Hebel, A.; Haag, R. Angew. Chem., Int. Ed. 2002, 41 (21), 3964-4000.

5866 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

of benzamide (k′ ) 1.03), benzophenone (k′ ) 8.11), and biphenyl (k′ ) 14.92) was eluted. Selectivity values of Rbenzophenone/benzamide ) 7.87, Rbiphenyl/benzophenone ) 1.84, and Rbiphenyl/benzamide ) 14.49 were obtained. Similar selectivity was obtained using CEC in a 30 cm × 75 µm i.d. (21 cm packed) fluorinated sc-C8 capillary. Figure 6 shows a plot of logarithm of capacity factor, k′, versus organic content, showing linear behavior, characteristic of reversed-phase separations. Linear plots have also been obtained for the series of organic thiourea solutes, separated using a borate buffer (pH 8.0) against percentage acetonitrile (Figure 7). A typical electrochromatogram obtained for the separation of the organic thiourea test solutes is shown in Figure 8. The order of elution (with theoretical plate numbers per meter) T1 (154 000) < T2 (144 000) < T5 (142 000) < T3 (134 000) < T4 (131 000) contrasts with the order of elution by CEC on sc-C8 and Hypersil C18 phases, i.e., T1 < T2 ∼ T3 < T4 < T5. The decreased retention of T5 suggests an increased polarity of the fluorinated phase, resulting in increased selectivity for the aromatic thioureas T3 and T4. The liquid chromatographic and CEC performances of this first example of a bonded phase generated in a supercritical fluid lead to investigations of the preparation of the more widely used octadecyl silica phases. It has been estimated that over 60% of reversed-phase separations are performed on C8 and C18 phases, with C18 undoubtedly accounting for the bigger share.8 The nonend-capped sc-C18 silica phase prepared in sc-CO2 was packed at higher pressure and examined by LC using test solutes dimethyl phthalate, anisole, diphenylamine, and fluorene, to provide column efficiency and asymmetry values, with uracil included for t0 determination (Table 1). These theoretical plate numbers and asymmetry factors are surprisingly high, considering that the sc-C18 silica phase has not been end-capped. In fact, this phase passes standards set by commercial manufacturers who expect plate numbers in excess of 100 000 for a column of this length and asymmetry factors between 0.9 and 1.2. The chromatographic performance of this non-end-capped sc-C18 silica column for a reversed-phase separa-

Figure 10. Electrochromatogram of the separation of five organic thourea solutes on an sc-C18 silica phase. Conditions: 30 cm × 75 µm i.d. (21-cm packed length); mobile phase, acetonitrile-5 mM sodium tetraborate pH 8.0 (90:10); voltage, 20 kV. Table 1. Efficiency and Peak Asymmetry Factors for a Standard Test Solute Mixture on a Non-End-Capped sc-C18 Column name

efficiency (N/m at half-height

Asym at 10%

capacity factor (k′)

dimethyl phthalate anisole diphenylamine fluorene

64 000 82 000 108 000 141 000

1.13 1.04 1.10 1.06

0.53 0.89 1.46 3.07

tion is shown in Figure 9, with a 50:50 acetonitrile-water mobile phase. Similar non-end-capped C18 silica phases were also produced using different reactive ligands. Both octadecyldimethyltrichlorosilane and octadecyldimethylmethoxysilane were also successfully immobilized onto silica and the resultant phases, when packed, gave plate numbers of 106 000 and 101 000 for fluorene under the same conditions as outlined above. Electrochromatography was also performed on this non-end-capped C18 silica, packed in-house using organic thiourea test solutes in acetonitrile-5 mM sodium tetraborate buffer at pH 8.0. Linear plots of k′ versus percentage acetonitrile were again obtained, and the differing selectivity of the sc-C18 silica phase from that of the fluorinated sc-C8 silica phase is clearly evident (Figures 10 and 11). This order of elution is similar to that obtained for an inhouse-packed Hypersil ODS capillary. Engelhardt Silanol Activity Test. Silica bonded phases experience difficulties with residual surface silanols and with surface metal content. Both adversely affect chromatographic performance, with basic solutes being particularly susceptible to unwanted silanol attractions and solutes that have metal chelating ability affected by surface metal content. To minimize these silanol effects, a second silylating or end-capping step was performed again in sc-CO2 to yield an sc-end-capped sc-C18 silica phase. The end-capping agent used was hexamethyldisilazane.35,47,48 Other potential silylating agents such as trimethylchlorosilane have also been used, but these necessitate the use of an amine proton scavenger such as pyridine, since trace amounts of HCl are produced as a byproduct, which can cleave existing silica-ligand

Figure 11. Log k′ for organic thiourea test solutes versus the percent acetonitrile content. Conditions: 30 cm × 75 µm i.d. (21-cm packed length) packed sc-C18 phase; mobile phase, acetonitrile-5 mM sodium tetraborate at pH 8.0; UV detection at 214 nm.

bonds. The non-end-capped sc-C18 phase had theoretical plate numbers as high as 141 000 (fluorene) and eluted the test mixture with peak asymmetries of ∼1.05. However, when the basic solutes of the Engelhardt test, aniline and dimethylaniline, were eluted, peak asymmetries were high. The ratio of the peak asymmetries of aniline and phenol was 1.8. The sc-end-capped sc-C18 silica phase was then subjected to the Engelhardt test to ascertain whether the end-capping step had reduced the effect of residual silanols. Initially, aniline and phenol were chosen as a test pair of solutes, but their capacity factors were almost identical, with both peaks eluting before 2 minu. As shown in Figure 12a, the chromatogram of N,N-DMA and toluene shows the preferred order of elution with N,N-DMA eluting first. As this compound is basic and hence susceptible to surface silanol activity, its elution before toluene, and its peak asymmetry of 1.29, are indicative of low silanol activity. The ratio of peak asymmetries of N,N-DMA and toluene was 1.19. In addition, the isomers of toluidine eluted as a single peak in LC on this sc-end-capped sc-C18 phase, highlighting low silanol activity (Figure 12b). The toluidine isomers differ only in their acidity and not in their hydrophobicity. Hence if the primary mode of retention is hydrophobic in nature, the peaks will not be separated. The results of these chromatographic tests confirm that chemically bonded silica phases prepared in sc-CO2 can also be Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

5867

Figure 12. (a) Chromatogram of N,N-DMA and toluene on an sc-end-capped sc-C18 silica phase. (b). Chromatogram obtained on injection of mixture of isomeric toluidines (ortho, meta, and para) on an sc-end-capped sc-C18 silica phase.

further reacted in a second step in the same reaction cell, to endcap the phase and improve chromatographic performance. Pharmaceutical Analysis on sc-End-Capped sc-C18 LC Column. Liquid chromatographic methods based on isocratic reversedphase conditions continue to play a leading role in the analysis of pharmaceutical compounds. Many separations are carried out using octadecyl columns with either binary or tertiary hydroorganic eluents and phosphate buffers to control eluent pH.51 The use of CE and CEC techniques is also growing in the pharmaceutical industry, for example, in enantiomeric separations and in the analysis of biopharmaceuticals. To demonstrate the application of supercritical fluid generated octadecyl silica in this area, well-known analgesics and β-blockers were chromatographed as shown in Figures 13 and 14. In the latter, high flow rate was used to show a rapid separation of ketoprofen, naproxen, fentoprofen, and ibuprofen, in less than 3 min on the sc-end-capped sc-C18 column. (51) Gilpin, R. K.; Pachla, L. A. Anal. Chem. 1999, 71, 217R.

5868 Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

CONCLUSIONS In summary, it has been demonstrated that supercritical carbon dioxide is a viable medium in the generation of chromatographic stationary phases. Indeed, it has considerable advantages over traditional solvents. The use of CO2 has obvious environmental advantages over organic solvents such as toluene and dichloromethane, which are traditionally employed. Significantly, the resulting bonded phase silicas yielded high theoretical plate numbers and good asymmetry factors, even as non-end-capped phases. With future work focusing on adjusting the parameters of the functionalization and end-capping reactions, such as temperature, pressure, and duration, and on assessing different silicas and reactive ligands, further improvements in chromatographic performance and selectivity are expected. Supercritical fluids may yet represent a viable commercial “green” alternative for the synthesis and deposition of chromatographic stationary phases and films.

Figure 13. Liquid chromatogram of four β-blockers on an sc-end-capped sc-C18 column (100 mm × 4.6 mm i.d, 3-µm particles). Mobile phase used was MeOH/KH2PO4 buffer at pH 4, flow rate of 1.00 mL/min. Proterenol, tr ) 1.19 min; pronethalol, tr ) 5.70 min; labetalol, tr ) 8.07 min; propranolol, tr ) 11.96 min.

Figure 14. Rapid elution of a mixture of four analgesics by LC on an sc-end-capped sc-C18 column (100 mm × 4.6 mm i.d, 3-µm particles). Mobile phase used was AcN-KH2PO4 (25:75, v/v), with a flow rate of 2.00 mL/min. Ketoprofen, tr ) 0.94 min; naproxen, tr ) 1.11 min; fentoprofen, tr ) 1.62 min; ibuprofen, tr ) 2.56 min.

ACKNOWLEDGMENT We thank Mr. Pat Curtis of Waters Technologies Ireland, Drinagh, Co., Wexford, Ireland, who assisted in packing both scC18 phases. We thank Dr. Norman Smith of Imperial College London for his assistance in CEC training for T. O’Mahony. We thank Thermo Hypersil-Keystone (Runcorn, Cheshire, U.K.) for

supplying 3-µm silica. We also thank Enterprise Ireland for Grant IF/2001/342 from their Research Innovation Fund. Received for review May 14, 2003. Accepted August 6, 2003. AC034511Q

Analytical Chemistry, Vol. 75, No. 21, November 1, 2003

5869