Molded Rigid Polymer Monoliths as Separation Media for Capillary

Molded Rigid Polymer Monoliths as Separation Media for Capillary ...... Classification of extra virgin olive oils according to their geographical orig...
0 downloads 0 Views 132KB Size
Download PDF
Anal. Chem. 1998, 70, 2288-2295

Molded Rigid Polymer Monoliths as Separation Media for Capillary Electrochromatography. 1. Fine Control of Porous Properties and Surface Chemistry Eric C. Peters, Miroslav Petro, Frantisek Svec, and Jean M. J. Fre´chet*

Department of Chemistry, University of California, Berkeley, California 94720-1460

Monolithic columns for capillary electrochromatography have been prepared within the confines of untreated fused-silica capillaries in a single step by a simple copolymerization of mixtures of butyl methacrylate, ethylene dimethacrylate, and 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS) in the presence of a porogenic solvent. The use of these novel macroporous monoliths eliminates the need for frits, the difficulties encountered with packed capillaries, and capillary surface functionalization. Since the porous properties of the monolithic materials can be easily tailored through changes in the composition of the ternary porogenic solvent, the effects of both pore size and the percentage of sulfonic acid monomer on the efficiency and the electroosmotic flow velocity of the capillary columns could be studied independently over a broad range. A simple increase in the content of charged functionalities within the monolith leads to an expected acceleration of the flow velocity. However, increasing the pore size leads to a substantial deterioration of the efficiency of the separation. In contrast, monoliths with increasing levels of AMPS in which the pore size remains fixed due to adjustments in the composition of the porogenic solvent show no deterioration in efficiency while maintaining the same increase in flow velocity, thus producing a significant reduction in separation time. Additionally, measurements on monoliths with constant levels of AMPS but different pore sizes suggest that flow velocity may be affected by the flow resistance within the capillary column. Capillary electrochromatography (CEC) is an emerging “hybrid” separation method which employs the electrically driven flow characteristic of electrophoretic separation methods within capillary columns packed with a solid stationary phase typical of HPLC techniques. In contrast to electrophoresis, CEC using columns packed with C18-modified silica particles allows the separation of neutral molecules by a reversed-phase HPLC mechanism.1 Unlike pressure-driven flow through a tube, which is characterized by a parabolic profile of flow velocities along its diameter, the flow profile resulting from electroosmotic force is claimed to be almost flat, since the dragging effect of friction along the wall is minimized. As a result, the axial dispersion of analyte compounds 2288 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

is substantially lower, and therefore, much higher plate numbers can be achieved in CEC compared to traditional HPLC.1 The concept of CEC was first described by Pretorius in 1974,2 and its potential for high-efficiency separations was demonstrated in the 1980s after fused-silica capillaries became readily available.3 However, serious technical problems have slowed its rapid development and widespread implementation.4 These problems include the difficult fabrication of frits within a capillary, the laborious packing of micrometer-sized beads into narrow-diameter tubes, and the limited stability of the resulting packed columns. Open tubular capillary columns for CEC avoid the major technical problems of packed capillaries. However, their sample capacities are low, thus making the detection of peaks difficult.5 An alternative solution employs the technology of continuous monoliths that recently have been prepared from both silica6 and organic polymers7 and successfully used in HPLC separations. For example, polymeric monolithic capillary columns for CEC (1) Dittmann, M.; Wienand, K.; Bek, F.; Rozing, G. P. LC-GC 1995, 13, 800813. Ross, G.; Dittmann, M.; Bek, F.; Rozing, G. P. Am. Lab. 1996, (March), 34-38. Colo´n, L. A.; Guo, Y.; Fermier, A. Anal. Chem. 1997, 69, 461A467A. Robson, M. M.; Cikalo, M. G.; Myers, P.; Euerby, M. R.; Bartle, K. D. J. Microcolumn Sep. 1997, 9, 357-372. Rathore, A. S.; Horva´th, C. J. Chromatogr. A 1997, 781, 185-195. (2) Pretorius, V.; Hopkins, B. J.; Schieke, J. J. Chromatogr. 1974, 99, 23-30. (3) Jorgenson, J. W.; Lukacs, K. D. J. Chromatogr. 1981, 218, 209-216. Tsuda, T.; Nomura, K.; Nakagawa, G. J. Chromatogr. 1982, 248, 241-247. Stevens, T. S.; Cortes, H. J. Anal. Chem. 1983, 55, 1365-1370. Knox, J. H.; Grant, I. H. Chromatographia 1987, 24, 135-143. (4) Boughtflower, B.; Underwood, T.; Paterson, C. J. Chromatographia 1995, 40, 329-343. Yan, C.; Dadoo, P.; Zhao, H.; Zare, R. N.; Rakestraw, D. J. Anal. Chem. 1995, 67, 2026-2029. Schmeer, K.; Behnke, B.; Bayer, E. Anal. Chem. 1995, 67, 3656-3658. Pesek, J. J.; Matyska, M. T. J. Chromatogr. A 1996, 736, 255-264. Robson, M. M.; Roulin, S.; Shariff, S. M.; Raynor, M. W.; Bartle, K. D.; Clifford, A. A.; Myers, P.; Euerby, M. R.; Johnson, C. M. Chromatographia 1996, 43, 313-321. (5) Bruin, G. J. M.; Tock, P. P. H.; Kraak, J. C.; Poppe, H. J. Chromatogr. 1990, 517, 557-572. Pfeffer, W. D.; Yeung, E. S. Anal. Chem. 1990, 62, 21782182. Pfeffer, W. D.; Yeung, E. S. J. Chromatogr. 1991, 557, 125-136. Mayer, S.; Schurig, V. J. High Resolut. Chromatogr. 1992, 15, 129-131. Garner, F. W.; Yeung, E. S. J. Chromatogr. 1993, 640, 397-402. Mayer, S.; Schurig, V. J. Liq. Chromatogr. 1993, 16, 915-931. Armstrong, D. W.; Tang, Y.; Ward, T.; Nichols, M. Anal. Chem. 1993, 65, 1114-1117. Mayer, S.; Schurig, V. Electrophoresis 1994, 15, 835-841. Meyer, S.; Schleimer, M.; Schurig, V. J. Microcolumn Sep. 1994, 6, 43-48. Guo, Y.; Colon, L. A. Anal. Chem. 1995, 67, 2511-2516. Tan, Z. J.; Remcho, V. T. Anal. Chem. 1997, 69, 581-586. (6) Fields, S. M. Anal. Chem. 1996, 68, 2709-2712; Minakuchi, H.; Nakanishi, K.,; Soga, N.; Ishizuka, N.; Tanaka, N. Anal. Chem. 1996, 68, 3498-3501. Minakuchi, H.; Nakanishi, K.; Soga, N.; Ishizuka, N.; Tanaka, N. J. Chromatogr. A 1997, 762, 135-146. S0003-2700(97)01351-6 CCC: $15.00

© 1998 American Chemical Society Published on Web 05/01/1998

were obtained by an in situ polymerization of water-soluble monomers within the confines of a fused-silica capillary whose inner wall had been first vinylized.8,9 These monoliths eliminate the technical problems of frit formation and column packing. However, they are often restricted to the use of monomers with limited hydrophobicity.8 Alternatively, more hydrophobic media for reversed-phase separations have been produced using either aqueous solutions of hydrophilic monomers that contained emulsified hydrophobic monomer or sophisticated multistep preparation methods.9 Such processes may make the control over the entire process difficult and less reproducible. Although the literature involving CEC is growing exponentially, the overwhelming majority of reports to date detail specific applications and technical issues. By contrast, only a few reports address the effects of the stationary phase on the chromatographic process.10-12 This is especially surprising in light of the dual role CEC stationary phases must perform. Specifically, in addition to effecting the separation like traditional HPLC packing, the CEC stationary phase is also the vehicle that enables the electroosmotic flow. Despite this important difference, the majority of papers investigating the effect of the stationary phase simply compare commercially available HPLC phases and, not surprisingly, show that they possess the expected trends in selectivity and/or retentivity. A similar comparison was reported for continuous hydrogel columns.7 However, such studies give little insight into the factors important for CEC separations or methods for their control. Such fundamental research is of particular interest, since it has been recently shown that traditional HPLC media are not optimized for CEC applications.11 Studies that directly relate the physical morphology and/or properties of a stationary phase to its performance in CEC are scarce. A recent report correlated properties such as residual silanol activity and carbon load of various commercial C18 stationary phases with their electroosmotic flow rate.11 However, no significant correlations were found, giving little hope that these parameters could be used to optimize performance in CEC. Li and Remcho demonstrated the effect of pore size on the ability to support perfusive flow in packed capillaries.12 Unfortunately, the restricted choice of commercially available silica packings does not allow a thorough investigation and control of the factors critical for CEC. Choudhary and Horva´th have studied the electroosmotic flow and conductance in capillaries packed with both charged and uncharged beads prepared from silica and styrene-divinylbenzene copolymers.13 (7) Hjerte´n, S.; Li. Y.-M.; Liao, J. L.; Nakazato, K.; Mohammad, J.; Pettersson, G. Nature 1992, 356, 810-811. Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1992, 54, 820-822. Svec F.; Fre´chet J. M. J. Science 1996, 273, 205-211. (8) Fujimoto, C. Anal. Chem. 1995, 67, 2050-2053. Fujimoto, C.; Kino, J.; Sawada, H. J. Chromatogr. A 1995, 716, 107-113. Fujimoto, C.; Fujise, Y.; Matsuzawa, E. Anal. Chem. 1996, 68, 2753-2757. Schweitz, L.; Andersson, L. I.; Nilsson, S. Anal. Chem. 1997, 69, 1179-1183. Nilsson, S.; Schweitz, L.; Petersson, M. Electrophoresis 1997, 18, 884-890. Palm, A.; Novotny, M. V. Anal. Chem. 1997, 69, 4499-4507. (9) Hjerte´n, S.; Eaker, D.; Elenbring, K.; Ericson, C.; Kubo, K.; Liao, J.-L.; Zeng, C.-M.; Lindstro¨m, P.-A.; Lindh, C.; Palm, A.; Srichiayo, T.; Valtcheva, L.; Zhang, R. Jpn. J. Electrophor. 1995, 39, 105-118. Liao, J.-L.; Chen, N.; Ericson, C.; Hjerte´n, S. Anal. Chem. 1996, 68, 3468-3472. Ericson, C.; Liao, J.-L., Nakazato, K.; Hjerte´n, S. J. Chromatogr. A 1997, 767, 33-41. (10) Dittmann, M. N.; Rozing, G. P. J. Chromatogr. 1996, 744, 63-71. Dittmann, M. N.; Rozing, G. P. J. Microcolumn Sep. 1997, 9, 399-408. (11) Zimina, T. M.; Smith, R. M.; Myers, P. J. Chromatogr. A 1997, 758, 191197. (12) Li, D. M.; Remcho, V. T. J. Microcolumn Sep. 1997, 9, 389-397.

Recently, we described the first molded rigid polymer monoliths designed as separation media for reversed-phase CEC.14 Our specifically designed ternary porogen system allowed the direct incorporation of both hydrophobic (butyl methacrylate and ethylene dimethacrylate) and charged (2-acrylamido-2-methyl-1propanesulfonic acid) monomers into a single homogeneous mixture, which was then polymerized in a single step within the confines of an untreated fused-silica capillary. Using this extremely simple preparation method, monolithic capillary columns with high efficiencies were easily obtained. Although this simplicity of preparation is appealing, the true advantage of our approach lies in the excellent control that can easily be exerted over the properties of the polymer. This control is very important, since properties such as the pore diameter of the flow-through channels and the level of charged functionalities were found to have dramatic effects on the chromatographic properties of the molded capillary column.14 This paper reports the first fundamental study of the effects of polymer morphology and composition on the overall CEC process. The excellent control that can be easily exerted over the properties of the polymer monoliths also allows the deconvolution of the effects of several variables changing simultaneously. EXPERIMENTAL SECTION Preparation of the Monolithic Polymer Capillaries. The procedure for the production of the monolithic polymer capillaries, along with the materials and instrumentation used, was detailed previously.14 Briefly, azobisisobutyronitrile (0.024 g, 1 wt % with respect to the monomers) was dissolved in 2.4 g of mixtures consisting of 40 wt % ethylene dimethacrylate and 60 wt % of combined butyl methacrylate and 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS) in various ratios. Ternary porogenic solvent (3.6 g) consisting of 10 wt % water and 90 wt % of mixtures of 1-propanol and 1,4-butanediol were slowly admixed to the monomers. AMPS charges of below 10 mg were added as an aqueous solution, such that the total water charge remained constant. The homogeneous mixtures were deaerated by purging with nitrogen for 10 min. A small part of the polymerization mixture was removed for capillary preparation using a syringe, and the remainder was sealed within a glass vial. A 50-cm length of unmodified capillary was attached to the syringe inlet and filled with the polymerization mixture to a total length of 30 cm. The ends of the capillary were plugged with a piece of rubber tubing, and both the capillary and the vial with bulk polymerization solution were submerged in a 60 °C bath for 20 h. Caution: AMPS monomer has been reported as a suspected carcinogen, and several methacrylates are known sensitizing agents. Proper precautions should be taken during the physical handling of these materials. The resulting monolith within the capillary was washed for 5 h with the mobile phase using a simple kdScientific (model KDS100) infusion syringe pump. A detection window was then created at the end of the continuous polymer bed by removing the polyimide coating using a razor blade. The glass vial containing the monolith formed from the bulk polymerization (13) Choudhary, G.; Horva´th, C. J. Chromatogr. A 1997, 781, 161-183. (14) Peters, E. C.; Petro, M.; Svec, F.; Fre´chet J. M. J. Anal. Chem. 1997, 69, 3646-3649.

Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2289

Figure 1. Effect of percentage of 1-propanol in the porogenic mixture on porous properties of monolithic polymers. Reaction conditions: polymerization mixture, ethylene dimethacrylate 16.00 wt %, butyl methacrylate 23.88 wt %, 2-acrylamido-2-methyl-1-propanesulfonic acid 0.12 wt %, ternary porogen solvent 60.00 wt % (consisting of 10 wt % water and 90 wt % of mixtures of 1-propanol and 1,4-butanediol), azobisisobutyronitrile 1 wt % (with respect to monomers); polymerization time 20 h at 60 °C.

mixture was carefully crushed, and the polymer was cut into small pieces, Soxhlet extracted with methanol for 12 h to remove any soluble compounds, and vacuum-dried at 60 °C overnight. This polymer was used for the porosimetric measurements, which were performed using a custom-made combined BET-sorptometer and mercury intrusion porosimeter (Porous Materials Inc., Ithaca, NY). Electrochromatography. Electrochromatographic experiments were carried out using a Hewlett-Packard 3DCE system equipped with a DAD 1050 UV detector and upgraded with an external pressure device for its use in capillary electrochromatography. Data acquisition and processing were done using HP ChemStation software (Hewlett-Packard). The chromatographic evaluation of the capillaries was performed as described previously.14 RESULTS AND DISCUSSION In our preliminary report,14 we clearly demonstrated an effect of both the size of the flow-through pores and the content of charged monomer on the overall performance of the monolithic capillaries in CEC. Since our initial study covered only three pore sizes within a very broad range of 670-4000 nm, we prepared a larger series of capillary columns in order to obtain a better understanding of the relationship between the properties of the monoliths and their chromatographic behavior. Control of Porous Properties. The ability of a liquid to flow through the network of large canallike pores that traverse the length of the macroporous monoliths is essential to all their applications. The ternary porogenic system developed earlier14 for the preparation of monolithic capillaries for CEC allows the fine control of porous properties over a broad range. For example, Figure 1 shows the mode pore diameter (pore diameter at the maximum of the distribution curve) as a function of 1-propanol content in the porogenic solvent for a polymer containing 0.3 wt % AMPS. Using this relationship, monoliths of any pore size in the range of 250-1300 nm can easily be produced simply by changing the ratio of 1-propanol to 1,4-butanediol in the porogen mixture. It should be noted that the relatively small difference of 2.5% in the weight percentage of 1-propanol that brackets this 2290 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

Figure 2. Differential pore size distribution profiles of porous polymer of monolithic capillary columns with mode pore diameters of 255 (1), 465 (2), 690 (3), and 1000 nm (4).

broad range of pore sizes is still sufficiently large to obtain polymers of any desired pore size with an accuracy of (25 nm with respect to a targeted value. Despite the fact that these monoliths are prepared from a polymerization mixture containing monomers of very different polarities, the mercury porosimetry profiles seen in Figure 2 all exhibit distribution curves similar to those found for polymers prepared from mixtures of fully miscible monomers.15 Effect of Pore Size on Efficiency of Electrochromatographic Separation. The major advantage of CEC compared to classical HPLC is that much higher column efficiencies can be achieved using the identical separation medium. For columns packed with beads, the efficiency in both of these methods is particle size dependent and increases as the size of the packing decreases.11 Since the monolithic columns are not packed but rather molded, issues of particle size become irrelevant, and instead, the size of the pores within the monolithic material is the variable that is expected to effect the chromatographic efficiency. Figure 3 shows the effect of the mode pore diameter of the monolithic capillaries on their efficiencies for the separation of a mixture of thiourea and three different alkylbenzenes. The effect is complex, with the curves exhibiting two distinct maximums: one at a pore size of 450 nm and the other in the range 550-750 nm. While the first maximum is very sharp and centered at the same pore size for all of the analytes, both the position and height of the second maximum depend on the retention of the compound. The existence of this pattern was confirmed by preparing capillaries at intermediate pore sizes, all of which were subsequently found to conform to the observed curve shape. The origin of these dual maximum curves is currently not completely clear. Since the polymer matrix, and thus the chemical functionality of all of the monoliths, are identical, the observed changes in the profiles of the individual analytes are not likely due to differences in their polarity. Rather, the unusual shape of these curves seems to reflect changes in the overall pore size distributions within the monoliths. The mode pore sizes used for the (15) Viklund, C.; Svec, F.; Fre´chet, J. M. J.; Irgum, K. Chem. Mater. 1996, 8, 744-750.

Figure 3. Effect of porosity of monolithic capillaries on their chromatographic efficiencies for thiourea (1), benzene (2), propylbenzene (3), and amylbenzene (4). Conditions: capillary columns, 100 µm i.d. × 30 cm active length; stationary phase with 0.3 wt % 2-acrylamido-2-methyl-1-propanesulfonic acid.

plot indicate only the positions of the maximums (modes) of the pore size distribution curves rather than the frequency of occurrence of all of the pores. However, Figure 2 clearly shows that the pore size distribution profiles for monoliths with smaller mode pore size are more asymmetric. For example, the distribution curve for monolith 1 is clearly bimodal, with a number of pores in the size range of 400-800 nm in addition to the mode at 250 nm. Additionally, deviations from a pluglike flow profile are known to exist in the smaller transport channels,12,16 which leads to decreased efficiencies. These effects would be expected to be more prevalent for the smaller mode pore diameter capillaries. It has to be emphasized that the measurement of porous properties using mercury intrusion methods is always performed in the dry state, while the capillary columns operate in a solvent mixture. As a result, the data shown in this study do not represent the actual operational pore size of the capillaries during the chromatographic process. Unfortunately, there is no simple method to determine the porous properties of the monolithic capillaries in their “swollen” state. For example, inverse size exclusion chromatography cannot be applied readily to materials with pore sizes in the range of hundreds of nanometers to well over 1 µm. However, the strong correlations that exist between the measured (dry) porous properties of the capillaries and their chromatographic performance suggest that the polymer matrix swells reproducibly and that “dry” porosity measurements can be used to tailor column performance when mobile phases of similar thermodynamic solvating quality are used. Although no simple correlation between mode pore diameter and column efficiency can be drawn, the curves yield other valuable information. The efficiency of an unretained peak is often presented in the literature as a measure of column quality. On the basis of the unretained marker peak of thiourea, the monolithic capillaries with mode pore diameters of 750 and 460 nm afford similar efficiencies of 110 000 and 99 000 plates/m, respectively. However, the real value of a separation medium is doubtlessly more accurately assessed using the efficiency of a retained peak. (16) Knox, J. H. J. Chromatogr. A 1994, 680, 3-13.

Figure 4. Electrochromatographic separation of benzene derivatives on monolithic capillary column. Conditions: capillary column, 100 µm i.d. × 25 cm active length; stationary phase with 0.3 wt % 2-acrylamido-2-methyl-1-propanesulfonic acid; pore size, 465 nm; mobile phase, 80:20 (v/v) mixture of acetonitrile and 5 mmol/L phosphate buffer pH 7; UV detection at 215 nm; voltage, 25 kV; pressure in vials, 0.2 MPa; sample concentration, 2 mg/mL of each compound; injection, 5 kV for 3 s. Peaks: thiourea (1), benzyl alcohol (2), benzaldehyde (3), benzene (4), toluene (5), ethylbenzene (6), propylbenzene (7), butylbenzene (8), and amylbenzene (9)

Therefore, if one considers the most retained compound (amylbenzene), the two capillaries show very different efficiencies of 34 000 and 91 000 plates/m, respectively. While the former is only modest, the latter is rather high, and very close to that calculated for the unretained thiourea. Thus, the ability to finely control the pore size of the capillaries allowed the identification of a polymer morphology that provides almost equal efficiency for both unretained and retained analytes. Figure 4 shows a representative electrochromatographic separation of a model mixture obtained on the 460-nm monolithic capillary. Reproducibility of the Preparation. Since the preparation of our monolithic capillaries is a truly single-step process, the number of effects that have a deleterious influence on the columnto-column reproducibility is substantially reduced. For example, the relative standard deviation for the flow velocities in nine different columns prepared from the identical polymerization mixture was found to be 2.5%, while the variations in retention factors k′ for all of the solutes shown in Figure 4 were in the range 4.5-5.1% for these nine columns. These data are essentially equal to those published recently by Palm and Novotny8 for acrylamidebased monolithic CEC capillaries. A more extensive study of the reproducibility of our monolithic columns is currently in progress. Control of Surface Charge. The electroosmotic flow of silicabased separation media for CEC is due to the presence of surface silanol functionalities that remain after incomplete surface functionalization. Therefore, accurate control over their level is difficult. Further, since these silanol moieties have widely varying acidities, the number of ionized groups that support the electroosmotic flow varies with the pH of the mobile phase, with mobile phases of lower pH showing lower flow rates. Ideally, this issue of varying acidity might be addressed by using a mobile phase with a high pH, to ionize all the silanol groups. However, at pH values higher than 8, hydrolysis of the silica occurs. Initially, this leads to an increase in the electrooosmotic flow as new surface Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2291

2292 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

AU AU

silanol groups emerge; however, this process may ultimately lead to the complete dissolution of the chromatographic medium. Because the decrease in flow velocity that occurs at lower pH is undesirable, mixed-mode stationary phases containing both octadecyl chains and propylsulfonic acid functionalities (SCX/ODS) have been introduced. However, even in these improved phases, the residual silanol units make a substantial pH-dependent contribution to the generation of flow.10 In contrast to silica-based media for CEC, the ability to easily control the level of charged functionality that supports the electroosmotic flow is a major advantage of directly produced monolithic capillaries. The surface charge density in our monolithic capillary columns can be easily controlled simply by changing the percentage of AMPS in the polymerization mixture. Additionally, the polymeric media are stable even in strongly basic environment, and the sulfonic acid moieties remain completely ionized over a broad range of pH. Since the electroosmotic flow is directly proportional to the ζ-potential, which, in turn, is a function of the electrokinetic charge density (charge density at the surface of shear layer), an increase in the content of sulfonic acid groups within the monolith should increase the flow velocity and thus reduce the time of each chromatographic run. Therefore, a series of monolithic capillaries with increasing percentages of AMPS (0.3, 0.6, 1.2, and 1.8 wt %) were prepared, keeping both the percentage of cross-linker and the composition of the porogen constant. The capillaries were targeted to have mode pore diameters roughly at the second maximum of efficiency of 750 nm for thiourea, due to their faster flow rates and ease of washing with a syringe pump. Figure 5 shows the separations of a model mixture obtained with each of these capillaries. As expected, the flow velocity increases with increasing level of charged moiety. However, the quality of the separation deteriorates significantly for columns containing higher levels of AMPS. While the monolithic capillary with 0.3 wt % AMPS separates the test mixture well, very poor separation is observed with the column containing 1.8 wt % AMPS, and all of the alkylbenzenes elute as a single peak. Although initially puzzling, a thorough characterization of the polymer samples revealed the cause of this unexpected phenomenon. Table 1 summarizes the porous properties of these monolithic capillaries. Despite using identical porogen compositions for each sample, the data clearly show that increasing the percentage of AMPS in the polymerization mixture not only leads to the desired increase in the charge density but also results in a substantial increase in the mode pore diameter. Since both of these variables may affect the separation (vide supra), the accurate assessment of the effect of AMPS on the overall chromatographic properties of monolithic capillary columns requires the ability to deconvolute the individual contributions of both pore size and AMPS level. Deconvolution of Effects of Pore Size and Content of Charged Functionalities. This deconvolution of effects was made possible by the excellent level of control that can be easily exerted over the morphology of our polymer monoliths. Specifically, a second set of capillaries was produced with levels of AMPS increasing from 0.3 to 1.8 wt % as above, but with mode pore diameters all within the narrow range of 690-740 nm. As shown in Table 2, this homogeneity in the mode pore diameters was obtained by adjusting the percentage of 1-propanol in the porogen

Figure 5. Effect of percentage of 2-acrylamido-2-methyl-1-propanesulfonic acid on the electrochromatographic properties of monolithic capillaries with varying pore sizes. Conditions: capillary column, 100 µm i.d. × 30 cm active length; stationary phase with 0.3 (a), 0.6 (b), 1.2 (c), and 1.8 wt % (d) 2-acrylamido-2-methyl-1-propanesulfonic acid. For pore sizes see Table 1; for chromatographic conditions see Figure 4. Table 1. Porous Properties of Monolithic CEC Columns with Different Levels of AMPS and Unadjusted Pore Diameters AMPS,a wt %

1-ProOH,b wt %

Dp,mode,c nm

Vp,d mL/g

u,e cm/min

0.3 0.6 1.2 1.8

59.55 59.53 59.51 59.57

835 1295 2225 3865

1.97 2.05 2.09 2.03

8.20 13.46 15.36 15.55

a Percentage of AMPS in the monomer mixture. b Percentage of 1-propanol in the porogenic solvent. c Pore diameter at the peak of the distribution profile. d Pore volume. e Flow velocity.

mixture, with monoliths containing higher levels of AMPS requiring higher levels of 1-propanol. Figure 6 shows the separations obtained for the same model mixture using the monolithic columns with adjusted mode pore diameters. Although the expected increase in flow velocity with increasing level of charged moiety was observed, no concomitant decrease in efficiency was seen. As a result, separations of the same quality can be performed with the 1.8 wt % AMPS capillary in only half the time required for those performed with the 0.3 wt % AMPS capillary. Another benefit of the deconvolution experiments is the accurate assessment of the effect of AMPS level on flow rate. Figure 7 shows a slow increase in the flow rate with increasing AMPS percentage for the series of capillaries with similar pore size, which is in accord with theory.11 By contrast, a deceivingly fast increase in flow rate is observed for the unadjusted pore size

Table 2. Porous Properties of Monolithic CEC Columns with Different Levels of AMPS and Pore Diameters Adjusted to Similar Values AMPS,a wt %

1-ProOH,b wt %

Dp,mode,c nm

Vp,d mL/g

u,e cm/min

0.3 0.6 1.2 1.8

59.77 60.45 61.57 62.57

692 735 690 718

1.99 1.99 1.97 1.97

8.87 9.87 11.36 12.10

a Percentage of AMPS in the monomer mixture. b Percentage of 1-propanol in the porogenic solvent. c Pore diameter at the peak of the distribution profile. d Pore volume. e Flow velocity.

Absorbancy, mAU

Figure 7. Flow velocity in monolithic capillaries as a function of weight percentage of 2-acrylamido-2-methyl-1-propanesulfonic acid in a monomer mixture. Capillary columns with similar (9) and varying mode pore sizes (2). Table 3. Porous Properties of Monolithic CEC Columns with High Levels of AMPS AMPS,a wt %

1-ProOH,b wt %

Dp,mode,c nm

Vp,d mL/g

u,e cm/min

3.0 3.0 3.0 3.0

64.58 63.32 62.31 61.38

640 1295 2430 3710

2.08 2.15 2.13 2.14

9.56 9.27 11.23 11.96

5.0 5.0

66.67 64.44

1530 4890

2.22 2.22

4.24 5.33

a Percentage of AMPS in the monomer mixture. b Percentage of 1-propanol in the porogenic solvent. c Pore diameter at the peak of the distribution profile. d Pore volume. e Flow velocity.

Figure 6. Effect of percentage of 2-acrylamido-2-methyl-1-propanesulfonic acid on the electrochromatographic properties of monolithic capillaries with pores adjusted to similar sizes. Conditions: capillary column, 100 µm i.d. × 30 cm active length; stationary phase with 0.3 (a), 0.6 (b), 1.2 (c), and 1.8 wt % (d) 2-acrylamido-2-methyl1-propanesulfonic acid. For pore sizes see Table 2; for chromatographic conditions see Figure 4.

capillaries, in which the effects of both increased pore size and AMPS level are superimposed. Monolithic Capillaries with High Contents of AMPS. In light of the increased speed of separation seen with increasing level of AMPS, polymers with even higher levels of charged monomer were investigated. Thus, capillary columns with comparably adjusted mode pore diameters containing either 3 or 5 wt % AMPS were prepared. However, these capillaries possessed flow rates significantly lower than the values expected from an extrapolation of the data for the 0.3-1.8 wt % AMPS region. For example, the capillary containing 3 wt % AMPS exhibited a flow rate of 9.6 cm/min, as opposed to its extrapolated value of 14.5 cm/min. This discrepancy was even greater for the capillary containing 5 wt % AMPS, which exhibited a flow rate of only 4.0 cm/min compared to its extrapolated value of 19.2 cm/min. It must be emphasized that the reported porosity values are

measured with the porous polymers in their dry state, while they are operationally used in an aqueous environment. This suggests that the unexpected decrease in flow velocity might be the result of excessive swelling of the polymer matrix in the mobile phase, causing both an effective decrease in the pore diameter of the monolithic capillary and the corresponding reduction in flow rate. To test this hypothesis, capillaries with larger mode pore diameters were prepared for each level of AMPS, with the expectation that one of these capillaries would “swell down” to the desired effective pore size of approximately 700 nm and thus exhibit the expected extrapolated flow velocity. Table 3 summarizes both the porous properties and flow rates for these capillary columns. However, for the series containing 3 wt % AMPS, the extrapolated flow velocity was not achieved even with the capillary possessing a mode pore diameter of nearly 4 µm. This deviation was far worse for capillaries containing 5 wt % AMPS. Even flow-through channels of almost 5 µm were not sufficiently large to offset the swelling, supporting a flow velocity only a quarter of its extrapolated value. In fact, the flow velocity of 5.3 cm/min measured for this monolithic capillary is even slower than that found for a capillary with only 0.3 wt % AMPS and a mode pore diameter of just 250 nm (6.1 cm/min). Additionally, Figure 5 suggests that a substantial deterioration in chromatographic properties occurs in capillary columns with such extremely large flow-through pores. However, Figure 8 docuAnalytical Chemistry, Vol. 70, No. 11, June 1, 1998

2293

Figure 8. Electrochromatographic separation of benzene derivatives on monolithic capillary column with high percentage of the charged functionalities. Conditions: stationary phase with 5 wt % 2-acrylamido2-methyl-1-propanesulfonic acid; pore size in dry state, 4890 nm. For other conditions, see Figure 4.

ments that all of the components of the test mixture including the alkylbenzene derivatives can be separated using the 5 wt % AMPS capillary with 5 µm mode pore diameter. This behavioral evidence suggests that, despite its large pore diameter in the dry state, the pores within this 5 wt % AMPS capillary behave as if they were filled with swollen polymer chains, and thus they may well mimic the behavior of Fujimoto’s soft gel beds.7 In addition, the separation efficiency is rather poor because the actual pore size under the chromatographic conditions cannot be controlled properly and is most likely far from its optimal value. Effect of Pore Size on Flow Velocity. The experiments with higher AMPS levels as well as the deconvolution study indicate that the electroosmotic flow velocity depends not only on the level of charged surface functionalities but also on the size of the transport channels. This finding contradicts the general perception seen in the CEC literature that the electroosmotic flow is independent of the size of the packing and, therefore, the size of the interstitial voids between the particles, unless the channel size is so small that the electrical double layers overlap.16 Yet, it has been recognized that the velocity of electroosmotic flow in a packed capillary is up to 60% slower than that in an open tube.13,17 This phenomenon has been attributed to the existence of column porosity in a packed bed, as well as the lack of parallel orientation of these voids with the axis of the capillary.2 Although the issue of flow resistance is often completely ignored, our results suggest that flow resistance effects may play a role even in capillary electrochromatography, albeit a significantly reduced one compared to traditional HPLC. One of the reasons why the effect of flow resistance in CEC has not been previously reported relates to the fact that this phenomenon may be difficult to observe using traditional nonporous or small-pore silica-based media. Obviously, a decrease in the particle size of a stationary phase results in an increase in the total outer surface area of the packed beads and, therefore, an increase in the overall amount of surface silanol functionalities. (17) Knox, J. H.; Grant, I. H. Chromatographia 1991, 32, 317-328. Yamamoto, H.; Baumann, J.; Erni, F. J. Chromatogr. 1992, 593, 313-319.

2294 Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

Figure 9. Effect of mode pore diameter on flow velocity of the mobile phase through monolithic capillary columns. Conditions: capillary column, 100 µm i.d. × 30 cm active length; stationary phase with 0.3 wt % 2-acrylamido-2-methyl-1-propanesulfonic acid; mobile phase, 80:20 (v/v) mixture of acetonitrile and 5 mmol/L phosphate buffer, pH 7.

Since the electroosmotic flow is directly proportional to the number of surface silanol groups in the system,11 the slowing effect of increasing flow resistance with decreasing particle size might be offset by an increase in the driving force for the electroosmotic flow. In contrast, our ability to easily control the mode pore diameters of the monolithic capillaries allows the direct investigation of the net effect of transport channel size on flow velocity. Figure 9 clearly demonstrates a 2-fold increase in the flow velocity through the monolithic capillaries as the pore size increases from 250 to 1300 nm. A similar increase in flow velocity with an increase in pore size up to 4 µm also was seen in our previous report.14 These pore sizes significantly exceed the thickness of a few nanometers at which the electrical double layers might overlap in a system utilizing our mobile phase.16 If it is assumed that the observed decrease in flow rate with decreasing mode pore diameter is simply due to the increasing percentage of pores in which the electric double layer reduces the flow, then the flow velocity should reach a maximum for monoliths with sufficiently large pores and thereafter remain constant, since the number of pores where the flow is affected by the electric double layer decreases rapidly as the pore size increases (Figure 2). In practice, however, this phenomenon is not observed. The fact that the overall flow velocity increases linearly over a broad range of pore sizes strongly supports the contention that this increase in flow rate is macroscopically related to a decrease in resistance to flow through the channels. An additional effect may result from variations in the strength of the electrical field on the microscopical level in both small and large pores. The effects of tortuosity of the packed structure and variations in cross sectional area on the conductance of CEC capillaries packed with beads and, consequently, on the chromatographic performance, have been discussed recently.13 CONCLUSION “Molded” monoliths prepared in a single polymerization step within the confines of untreated capillaries are an attractive option

as separation media for capillary electrochromatography. A major advantage of these materials is their ease of preparation by a simple “molding”, which avoids the fabrication of frits, the tedious packing of very small particles into narrow-diameter capillaries, or any initial chemical modification or “treatment” of the inner walls of the capillary. In addition, both the porous properties and the percentage of charged functionalities can be independently and precisely controlled through fine changes in the composition of the polymerization mixture, allowing the preparation of capillaries with tailored properties. Since the pore size can be changed while keeping the concentration of charged functionalities representing the driving force for the electroosmotic flow constant, the effect of pore size on overall flow velocity can also be observed.

ACKNOWLEDGMENT Support of this research by a grant of the National Institute of General Medical Sciences, National Institutes of Health (GM48364) is gratefully acknowledged. E.C.P. also thanks the Rohm and Haas Co. for its financial sponsorship through gift funding to University of California.

Received for review December 17, 1997. Accepted March 18, 1998. AC9713518

Analytical Chemistry, Vol. 70, No. 11, June 1, 1998

2295