Anal. Chem. 1998, 70, 4023-4029
Surface-Confined Living Radical Polymerization for Coatings in Capillary Electrophoresis Xueying Huang, Leon J. Doneski, and Mary J. Wirth*
Department of Chemistry & Biochemistry, University of Delaware, Newark, Delaware 19716
Surface-confined living radical polymerization is shown to be a controlled means of covalently bonding both linear and cross-linked polymer films on silica. CuCl/bipyridine initiates radical formation through atom transfer with a self-assembled monolayer of benzyl chloride, onto which polymer then grows. The polymerization is intrinsically confined to the surface, avoiding problems associated with polymer formed in the solution. The surface-confined polymerization scheme is generally applicable to radical polymerization of vinyl monomers and was studied here for the case of acrylamide. Infrared spectroscopy shows that the film growth is controllable, and atomic force microscopy reveals that smooth films are prepared. The surface-confinement polymerization scheme was tested for both linear and cross-linked polyacrylamide. Capillary electrophoresis of strongly basic proteins confirms that the coated capillaries provide the high efficiency expected for polyacrylamide. The cross-linked coating exhibits higher reproducibility with respect to migration time than does the linear coating. Surface-confined living radical polymerization prepares linear and cross-linked polymer films without danger of clogging narrow capillaries and will ultimately facilitate cross-laboratory comparisons by enabling control of film thickness. Capillary electrophoresis is becoming an increasingly powerful tool for biomolecule separations.1,2 Its applicability is limited by the negative surface charge of the silica capillaries, which leads to zone broadening and tailing as well as electroosmotic flow. Coatings must be applied to silica to allow its use in the separation of proteins. Basic electrostatics explains how the negative surface charge affects protein separations and how the problems can be minimized. A protein of charge z will have an electrostatic interaction energy, Ees, depending on its distance, x, from a surface that has an electrostatic potential, φ0, where κ-1 is the Debye length.
Ees ) zFφ0 exp(-κx)
(1)
increase the distance between the protein and the silica surface. Coatings are tested between pH 4 and 5 because Ees for silica surfaces and basic proteins typically maximizes in this range. In addition to adsorption, the surface potential also gives rise to electroosmotic flow, for which the mobility, µeo, is proportional to the potential at the boundary layer, ζ. Hjerte´n derived an equation for electroosmotic flow for a coated capillary, integrating over the viscosity with respect to potential.3
∫ η1 dφ
µeo ) 0
ζ
0
(2)
This shows that thick polymer coatings have the additional advantage of reducing electroosmotic flow because the high viscosity of the polymer film prevents flow in the region of high potential. Variation in electroosmotic flow is presently an even more severe problem than adsorption. Any additional potential introduced from hydrolysis of the polymer, which occurs for polyacrylamide, would give an additive contribution to µeo due to the superposition principle of electric fields. A thick coating of neutral, inert polymer would prevent both protein adsorption and electroosmotic flow. Hjerte´n reported the first covalent bonding of a polymer film for capillary electrophoresis in 1985.3 The chemistry for this pioneering work relies on a monolayer with polymerizable functional groups (carbon-carbon double bond in methacrylate) that is chemically attached to the silica surface. The monomer (acrylamide), initiator (potassium persulfate), and solvent (water) are then introduced into the capillary, and at elevated temperature the initiator dissociates to form radicals to initiate the polymerization, which occurs both in solution and on the silica surface. Since 1985, numerous papers have been published about covalently bonded polymers that give improved capillary electrophoresis of proteins. Some of these papers have introduced polymerizable functional groups covalently bonded to the silica surface, including methacryloxypropyl through SiO2-O-Si-C linkage,3,4 methacryloxypropyl through SiO2-Si-C linkage,5 and vinyl through SiO2-Si-C linkage,6 and a mixed methyl- and allylsilane monolayer.7 Others have introduced new monomers
Coated capillaries can both reduce the surface potential and (1) Hjerte´n, S. In High-Resolution Separation and Analysis of Biological Macromolecules, Part A: Fundamentals; Karger, B. L., Hancock, W. S., Eds.; Methods in Enzymology 270; Academic Press: New York, 1996; p 296319. (2) Capillary Electrophoresis in Analytical Biotechnology; Righetti, P. G., Ed.; CRC Series in Analytical Biotechnology; CRC Press: Boca Raton, FL, 1996. S0003-2700(98)00231-5 CCC: $15.00 Published on Web 08/19/1998
© 1998 American Chemical Society
(3) Hjerte´n, S. J. Chromatogr. 1985, 347, 191. (4) Tan, Z. J.; Remcho, V. T. Anal. Chem. 1997, 69, 581. (5) Chiari, M.; Nesi, M.; Sandoval, J. E.; Pesek, J. J. J. Chromatogr. A 1995, 717, 1. (6) Cobb, K. A.; Dolnik, V.; Novotny, M. Anal. Chem. 1990, 62, 2478. (7) Huang, M.; Pubrovcakova-Schneiderman, E.; Novotny, M. V.; Fatunmbi, H. O.; Wirth, M. J. J. Microcolumn Sep. 1994, 6, 571.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4023
to form hydrophilic polymers attached to the surface, including polyacrylamide,3,6,8,9 poly(N-acryloxyamino)ethoxyl ethanol,5 poly(N-acryloylamino) ethyl-β-D-glucopyranose,10 hydrogel polymer,11 epoxy polymer,12,13 and poly(ethylene glycol).14,15 In all of these papers, polymerization schemes are employed that produce polymer both in solution and on the surface, and the film thickness is not predictable. The lack of knowledge of the film thickness hinders fundamental study, as discussed by Righetti et al.16 Without knowledge of the film thickness, eqs 1 and 2 are only useful qualitatively, and interlaboratory comparisons of polymer films are difficult. Whenever a chemical synthesis is performed, it is advantageous to know what has been made, yet it is presently difficult to analyze the film thickness inside the capillary. Also, from a practical standpoint, the polymer formed in solution can clog the capillary, raising manufacturing costs and impeding the development of cross-linked coatings. For example, in the polymerization of acrylamide, even 0.2% cross-linker (N,N′-methylenebisacrylamide) added in the solution clogs the capillary column to make the column useless.17 Better control of the polymerization would be valuable. In this paper, we describe surface-confined living radical polymerization (SCLRP) as a method for covalently bonding polymers to capillary surfaces. The goal is illustrated in Scheme 1, which shows that SCLRP yields only polymer chains covalently attached to the surface, while solution polymerization yields both attached and free polymers. The method is expected to be widely applicable to vinyl monomers, and in this paper it is applied to the polymerization of acrylamide, which is chosen because it is a common material used for coatings in capillary electrophoresis. This polymerization has been demonstrated for the imposing challenge of applying a thin polyacrylamide film to porous silica gel, where the pores would easily be clogged if there were any free polymer formed in solution.18 Successful size-exclusion separation of proteins established that the pores remained unblocked after polymerization. The SCLRP reaction is illustrated in Scheme 2, where benzyl chloride groups bonded to the surface form a controlled number of radicals by atom transfer to a Cu(bpy)2Cl catalyst. An equilibrium is maintained between the dormant and active chain ends, RCl and R•, respectively, with the equilibrium favoring the dormant species, minimizing termination and radical-transfer reactions. The minimal termination and transfer reactions give rise to living polymerization, thus enabling polymer chain length to be controlled by monomer concentration. The purpose of this work is to test whether film thickness can be controlled by monomer concentration, to investigate whether the surface confinement is sufficient to enable controlled prepara(8) Engelhardt, H.; Cunat-Walter, M. A. J. Chromatogr. 1995, 716, 27. (9) Huang, M.; Vorkink, W. P.; Lee, M. L. J. Microcolumn Sep. 1992, 4, 233. (10) Chiari, M.; Dell’Orto, N.; Gelain, A. Anal. Chem. 1996, 683, 3. (11) Huang, M.; Lee, M. L. J. Microcolumn Sep. 1992, 4, 491. (12) Towns, J. K.; Bao, J.; Regnier, F. R. J. Chromatogr. 1992, 599, 227. (13) Liu, Y.; Fu, R.; Gu, J. J. Chromatogr. A 1995, 694, 498. (14) Bruin, G. J. M.; Chang, J. P.; Kuhlman, R. H.; Zegers, K.; Kraak, J. C.; Poppe, H. J. Chromatogr. 1989, 471, 429. (15) Nashabeh, W.; El Rassi, Z. J. Chromatogr. 1991, 559, 367. (16) Barberi, R.; Bonvent, J. J.; Bartolino, R.; Roeraade, J.; Capelli, L.; Righetti, P. G. J. Chromatogr. B 1996, 683, 3. (17) Schmalzing, D.; Piggee, C. A.; Foret, F.; Carrilho, E.; Karger, B. L. J. Chromatogr. A 1993, 652, 149. (18) Huang, X.; Wirth, M. J. Anal. Chem. 1997, 69, 4577.
4024 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Scheme 1. Conventional Solution Polymerization Yields Both Attached and Free Polymer, while Surface-Confined Living Radical Polymerization Yields Only Attached Polymer
tion of cross-linked films, and to determine whether the high performance of polyacrylamide in capillary electrophoresis is maintained using this new method of preparation. EXPERIMENTAL SECTION Reagents and Materials. Acrylamide (electrophoresis grade, 99.9% minimum) and N,N′-dimethylformamide were purchased from Fisher scientific. Tris(hydroxymethyl)aminomethane (Tris, 99.9+%, ultrapure grade), 2,2′-dipyridyl (99+%), copper(I) chloride (98+%), N,N′-methylenebisacrylamide, 3-(trimethoxysilyl)propyl methacrylate, ammonium peroxydisulfate (98+%), and N,N,N′,N′tetramethylenediamine (TEMED, 99+%) were from Aldrich. 2-(NCyclohexylamino)ethanesulfonic acid (CHES, minimum 99%, titration), 2-amino-2-methyl-1,3-propanediol (SigmaUltra, minimum 99%), N-tris(hydroxymethy)methyl-3-aminopropanesulfonic acid (TAPS, SigmaUltra, minimum 99.5%), and all the proteins were bought from Sigma. 1-Trichlorosilyl-2-(m-p-chloromethylphenyl)ethane was from United Chemical Technology, Inc. Pure water (18 MΩ‚m) was obtained using a Barnstead E-pure system. Bare fused silica capillary (5 m, 363 µm o.d., 75 µm i.d.) was obtained from Supelco, Inc. The silicon wafers (3 in. diameter, 20.5-21.5 mm thickness, orientation (100), both sides polished) were from Semiconductor Processing Co. (Boston, MA). Preparation of Polyacrylamide Coating. A 1-m section of bare silica capillary was rinsed with a flowing solution of 1.0 M KOH aqueous solution for 1.0 h and then rinsed with pure water for 3.0 h. The capillary was dried by flowing N2 through it overnight at room temperature. The self-assembled benzyl chloride monolayer was prepared by first adding 0.3 mL of 1-trichlorosilyl-2-(m-p-chloromethylphenyl)ethane to 5.7 mL of dried toluene by passing it through an Al2O3:SiO2 packed column, sonicating for 30 s, and then im-
Scheme 2. Preparation of the Polymer Film by SCLRPa
a Step 1: a self-assembled benzyl chloride monolayer on silica contains the reactive chloro groups. Step 2: the CuCl/Bpy auxiliary initiator and acrylamide monomer are introduced, and the polyacrylamide film grows from the self-assembled monolayer.
mediately flowing the solution into the capillary column. After 3 drops of solution came out of the other end of the capillary, each end of the capillary was sealed with a silicone rubber septum. The reaction was allowed to proceed at room temperature for 18 h. The capillary was then rinsed with toluene for 30 min and acetone for 30 min. Finally, the capillary was put into an oven at 110 °C and baked for 1.0 h. The surface living radical polymerization was carried out by first adding 0.75 g of acrylamide, 0.0316 g of 2,2′-dipyridyl, and 0.0067 g of copper(I) chloride to 0.75 mL of DMF. The mixture was sonicated for 1 min to accelerate dissolution into DMF and then immediately flowed into the capillary. After 3 drops of mixture flowed out of the other end, each end of the capillary was sealed with a high-temperature silicone rubber septum. The capillary was placed in an oven at 130 °C for 40.0 h. After the reaction, the capillary was rinsed with pure water using an LC pump. For the case of cross-linked polymerization, 2% N,N′methylenebisacrylamide (0.015 g) was added into the above mixture, and the rest of the procedures were same as those for linear polymerization. Spin Coating. The spin caster was made in-house. Aqueous solutions of polyacrylamide were prepared by dissolving pure polyacrylamide into water, with concentrations varying from 5% to 25% (w/w). The clean silicon wafer was mounted on the spin caster. In each case, three drops of polyacrylamide solution were put on the wafer, and the spin caster was immediately rotated at 3000 rpm for 30 s. The color of the film changed during spinning, indicating the evaporation of water. Ellipsometry. The thickness of polyacrylamide films, spincoated on silicon wafers, was measured by a null-ellipsometer (Rudolph Auto EL-II, Fairfield, NJ). The wavelength of the laser beam was 632.8 nm, and the angle of incidence was 70°. The refractive index of polyacrylamide was estimated to be 1.54.19 Each sample was measured for 10 spots, and the average value was taken as the thickness of the film. The oxide layer (SiO2) on the bare silicon wafer was determined to be 18.2 Å. The thickness of polyacrylamide film was obtained by subtracting the contribution of the oxide layer. (19) Polymer handbook; CRC Press: Boca Raton, FL, 1995.
Infrared Spectroscopy. Infrared spectra were collected in the transmission mode using a Mattson Galaxy 5000 series FTIR. The unpolarized beam passed through the sample at normal incidence. A total of 2048 scans were made with a resolution 2 cm-1 for each sample, which required a scanning time of 8.68 min. A bare silicon wafer was used as the blank. The net absorbance of CdO at 1663 cm-1 was used to calibrate the thickness of polyacrylamide films. Preparation of Polyacrylamide Film on Silicon Wafer. The silicon wafers were cut into smaller pieces (2.5 cm × 2.5 cm) and boiled in H2SO4:H2O2 (70:30 V/V) solution (“piranha solution”) for 4.0 h. (Caution: Piranha soltuion is a very strong oxidizing reagent. Handle carefully. Double gloves were wore for protection.) The wafers were then rinsed copiously with pure water. The procedures for self-assembling of benzyl chloride monolayer and surface living radical polymerization were the same as those for the capillary columns. Atomic Force Microscopy. A TopoMetrix Explorer AFM was employed to obtain the topographic images of the samples in air. The “high amplitude resonance-amplitude detection” mode was employed, using a silicon cantilever having a 100-µm length and a 250-kHz resonant frequency. Images were analyzed using TopoMetrix version 3.06.06 software and were leveled horizontally and vertically using a first-order algorithm. Rootmean-square roughness was calculated using the TopoMetrix software. Electrophoresis. The electropherograph was an Isco model 3850 (Lincoln, NE). The dimension of the capillary column used here was 30 × 53 cm, 75 µm. A 1-cm-long window for detection was made by burning the capillary column with a match flame and then rinsing with acetone. Mesityl oxide was used as neutral marker at a concentration of 1 µL/mL in water. All samples were hydrodynamically injected under 0.5 psi pressure for 2.0 s. Proteins and the neutral marker were detected at 214 and 254 nm, respectively. For the basic proteins, two different buffers were used: 50 mM Tris-HCl at pH 4.7 and 50 mM acetate at pH 4.5; the protein concentration was 100 ppm. The voltage applied to the capillary column was 15 kV. Between runs, the column was rinsed with buffer solution. After every five runs, the buffer solution was replaced with fresh buffer. For the acidic proteins, Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4025
Figure 1. Dependence of film thickness, d, on monomer concentration, C, for (a) linear polyacrylamide and (b) cross-linked polyacrylamide, both made by SCLRP.
the buffer solution was 20 mM TAPS-AMPD, which was made by adding AMPD to 20 mM TAPS until the pH reached 8.8. Three proteins, R-lactalbumin, β-lactoglobulin B, and trypsin inhibitor, were dissolved in buffer to achieve protein concentrations of 100, 350, and 350 ppm, respectively. The voltage applied to the capillary column was -20 kV. Between runs, the column was rinsed with buffer solution. After every 20 runs, the buffer solution in the reservoir was replaced with fresh one. Microcal Origin 4.0 software was used for data processing and analysis. RESULTS AND DISCUSSION I. Film Preparation and Characterization. Polyacrylamide films were first prepared on silicon wafers to study the film thickness and topography. Silicon was used because it is transmissive in the infrared, yet its native oxide layer has the same surface as fused silica.20 The thickness of the polyacrylamide film was determined from the infrared absorbance of the CdO stretch. This was calibrated from a series of standard polyacrylamide films prepared by spin-coating, which gives a predictable thickness that was confirmed by ellipsometry. A linear calibration curve was then obtained for the infrared absorbance of the standards,
Figure 2. Atomic force micrographs for polyacrylamide films prepared by linear SCLRP, cross-linked SCLRP, and the conventional method of solution polymerization.
thickness, where d is in angstroms and C is in molarity.
A ) -2.62 × 10-4 + (4.3 × 10-5)d
(3)
where d is the thickness of polyacrylamide film in units of angstroms and A is the net absorbance of CdO stretch. For the SCLRP, the relationships between the film thickness and monomer concentration are shown in Figure 1 for the linear and cross-linked polymerizations. For the linear polymerization, the relationship is linear, a hallmark of living polymerization. The relationship between the thickness of the linear polyacrylamide film, dlinear, and monomer concentration, C, allows control of film (20) Zhuravelev, L. T. Langmuir 1987, 3, 316.
4026 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
dlinear ) 4.6 +18.8C
(4)
For the cross-linked polymer, the percentage of bisacrylamide was kept constant, and the total concentration of acrylamide plus bisacrylamide was varied. The relationship between total monomer concentration and film thickness was found to be nonlinear. This is not surprising, because the amount of the more reactive cross-linker accumulates in the monolayer, increasing the reactivity of the monolayer with time. The relationship between the thickness of the cross-linked polyacrylamide film, dcross, and monomer concentration fits well to an exponential growth function.
dcross ) 8.9 exp(0.75C)
(5)
Figure 3. Electropherograms for three basic proteins in a 50 mM Tris/HCl buffer at pH 4.7 for the SCLRP films: (a) linear and (b) crosslinked polyacrylamide coatings. The proteins are (1) cytochrome c, N ) 8.2 × 105, (2) lysozyme, N ) 8.2 × 105, and (3) ribonuclease A, N ) 6.5 × 105.
The data show that the thickness of the cross-linked film is predictable from the monomer concentration out to at least 700 Å. II. Atomic Force Microscopy. Figure 2 shows atomic force micrographs for the two living polymer films and a film prepared by conventional polymerization.6 Both the linear and cross-linked
films prepared by living polymerization are significantly smoother than that prepared by conventional solution polymerization. The film prepared by solution polymerization shows evidence for attachment of oligomer or polymer in solution. The cross-linked film is rougher than the linear film, which is not surprising due to the random branching caused by the cross-linker. The rootmean-square roughness values are 5.1 and 20 Å for the linear and cross-linked films, respectively. The film prepared by solution polymerization was heterogeneous, with a wide distribution of rootmean-square roughness values: 30 ( 20 Å. The uniformity of the films prepared by SCLRP, combined with surface confinement of the polymerization, prevents clogging of capillaries for electrophoresis and also allows coating nanoporous silica.18 III. Capillary Electrophoresis. The two types of capillary coatings were used for electrophoresis: a 140-Å linear polyacrylamide film and a 300-Å cross-linked polyacrylamide film. Proteins with pI > 9.0 (cytochrome c, lysozyme, and ribonuclease A) were separated at pH 4.7 to test the ability of the polyacrylamide coating to reduce electrostatic interaction. A Tris-HCl buffer was used to match the conditions of previously reported electropherograms for these same proteins. Figure 3 shows the electropherograms for a mixture of these three proteins. For the case of linear polyacrylamide, the separation efficiencies are comparable to those of previous reports.6,7 The efficiency is slightly lower for the crosslinked column. Given that the efficiency is still very high, whatever is causing this reduction in efficiency must be small, and one can only speculate about its origin. Nonetheless, the efficiencies are very high for both types of films. The Tris-HCl buffer used for the separation of the basic proteins was chosen by previous workers because it is matched to the mobility of the proteins, giving the very high efficiency needed to test for small electrostatic interactions. However, the buffer capacity of Tris-HCl is low at pH 4.7, so it is not as useful for sensitively testing the reproducibility of the separations. An acetate buffer at pH 4.5 was used to study reproducibility of the migration of the basic proteins. Figure 4 compares the electropherograms from the fourth run and the 150th runs of the three basic proteins using the linear polyacrylamide column. The analogous comparison is made in Figure 4 using the cross-linked polyacrylamide column. Slight deterioration is evident in the linear case, but none is evident in the cross-linked case. The average migration times and relative standard deviations for 20 replicates are shown in Table 1 for both capillaries. The
Table 1. Separation Efficiency, Migration Time, and Relative Standard Deviation of Migration Time for Basic Proteins, Initially and after 150 Runsa linear polyacrylamide coating peak
protein
1
lysozyme runs 1-20 runs 151-170 cytochrome c runs 1-20 runs 151-170 ribonuclease A runs 1-20 runs 151-170
2 3
a
N
(×105)
cross-linked polyacrylamide coating
tm (min)
rsd (%)
N (×105)
tm (min)
rsd (%)
3.5 2.5
6.70 6.57
0.31 0.64
4.2 4.2
7.29 7.26
0.36 0.61
7.8 7.4
6.96 6.93
0.27 0.51
4.9 5.0
7.57 7.54
0.36 0.64
4.4 4.2
9.26 9.06
0.63 0.91
4.0 4.2
9.78 9.77
0.38 0.71
The separation conditions were the same as those for Figure 4.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4027
Table 2. Separation Efficiency, Migration Time, and Relative Standard Deviation of Migration Time for Acidic Proteins, Initially and after 20 Runsa linear polyacrylamide coating peak
protein
1
trypsin inhibitor runs 1-20 runs 21-40 β-lactoglobulin B runs 1-20 runs 21-40 R-lactalbumin runs 1-20 runs 21-40
2 3
a
N
(×105)
cross-linked polyacrylamide coating
tm (min)
rsd (%)
N (×105)
tm (min)
rsd (%)
0.5 0.5
6.10 6.33
0.57 0.63
1.2 1.5
7.65 7.72
0.41 0.49
0.3 0.3
7.86 8.00
0.50 0.53
1.1 0.6
8.48 8.53
0.30 0.31
1.3 1.2
13.00 13.15
0.60 0.67
1.6 1.2
12.89 12.99
0.37 0.36
The separation conditions were the same as those for Figure 5.
Figure 4. Reproducibility of the separation of three basic proteins for the SCLRP films: (a) linear polyacrylamide, fourth run, (b) linear polyacrylamide, 150th run, (c) cross-linked polyacrylamide, fourth run, (d) cross-linked polyacrylamide, 150th run. The basic proteins are the same as those described in Figure 3, and the same separation conditions were used, except that a 50 mM acetate buffer at pH 4.5 was used.
reproducibility for 20 replicates is excellent, less than 1% in all cases. For the longer term, the migration times of the proteins are shorter after 150 runs for both capillaries, presumably due to an increase in electroosmotic flow resulting from carboxylate groups formed slowly upon hydrolysis of the polyacrylamide film. However, ∆tm is much smaller for the cross-linked polymer: the 4028 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Figure 5. Reproducibility of the separation of three acidic proteins for SCLRP film: (a) linear polyacrylamide, first run, (b) linear polyacrylamide, 40th run, (c) cross-linked polyacrylamide, first run, (d) cross-linked polyacrylamide, 40th run. The acidic proteins are (1) trypsin inhibitor, (2) β-lactoglobulin B, and (3) R-lactalbumin, separated at pH 8.8.
change in migration time after 150 runs (2 weeks) is only 2σ, where σ is the standard deviation of the zone. If the criterion for reproducibility is that the peak position must stay within the zone width, the cross-linked polyacrylamide film allows reproducible electrophoresis at pH 4.5 for at least 2 weeks.
Electroosmotic flow is an important issue in capillary electrophoresis because changes in this quantity are believed to be the origin of irreproducibility. The electroosmotic mobility, µeo, was measured by using mesityl oxide as a neutral maker in 20 mM, pH 8.0 Tris buffer,
µeo ) Ld/Et
(6)
where E is the electric field and Ld is the length over which the marker migrates in time, t. Initially, at an electric field of 600 V/cm, the column was run for 3.0 h, and no peak was seen for either the linear or the cross-linked capillaries for the 30-cm detection length. This means the electroosmotic mobility must be less than 4.6 × 10-6 cm2/(V‚s) in both cases. After 150 runs, the electroosmotic flow remained unobservable for the crosslinked capillary but increased to 7 × 10-6 cm2/(V‚s) for the linear capillary, indicating that some hydrolysis of the polyacrylamide occurred for the linear polymer film. The electroosmotic mobility after 150 runs quantitatively explains ∆tm for the linear film, arguing that the initial electroosmotic flow is negligible compared to 7 × 10-6 cm2/(V‚s). Cross-linking apparently increases the hydrolytic stability of polyacrylamide; hence, no measurable electroosmotic flow develops, and no significant change in migration time occurs. Both polymer films effectively eliminate electroosmotic flow resulting from the silica surface charge, but hydrolysis of the linear polymer introduces a new source of surface charge. A set of acidic proteins, trypsin inhibitor, β-lactoglobulin B, and R-lactalbumin, were separated at pH 8.8. All three proteins are negatively charged at this pH, requiring a reverse voltage. The electropherograms for both the linear and cross-linked (21) Kato, M.; Kamigaito, M.; Sawamoto, M.; Higashimura, T. Macromolecules 1995, 28, 1721. (22) Wang, J.; Matyjaszewski, K. J. Am. Chem. Soc. 1995, 117, 5614. (23) Pattern, T. E.; Xia, J.; Abernathy, T.; Matyjaszewski, K. Science 1996, 272, 866. (24) Wang, J.; Matyjaszewski, K. Macromolecules 1995, 28, 7901. (25) Percec, V.; Barboiu, B. Macromolecules 1995, 28, 7970. (26) Percec, V.; Barboiu, B.; Neumann, A.; Ronda, J. C.; Zhao, M. Macromolecules 1996, 29, 3665. (27) Granel, C.; Dubois, Ph.; Jerome, R.; Teyssie, Ph. Macromolecules 1996, 29, 8576. (28) Hawker, C.; Barclay, G. G.; Dao, J. J. Am. Chem. Soc. 1996, 118, 11467. (29) Georges, M. K.; Veregin, R. P. N.; Kazmaier, P. M.; Hamer, G. L. Macromolecules 1993, 26, 2987. (30) Greszta, D.; Matyjaszewski, K. Macromolecules 1996, 29, 7661.
capillaries are shown in Figure 5. The number of theoretical plates for R-lactalbumin is 1.3 × 105, which is comparable to the literature reports.6 The separations of trypsin inhibitor and β-lactoglobulin B do not have high calculated efficiencies. The lower efficiency for trypsin inhibitor is probably attributable to an impure sample, which is indicated in its electropherogram in Figure 5. For some unknown reason, the β-lactoglobulin B zone does not have high efficiency, and this has also been reported by others.10,16 The reproducibility data are given in Table 2, which show that replicates of 20 runs agree closely with one another. The coating degrades faster at this higher pH due to base-catalyzed hydrolysis, as expected. In contrast to the basic proteins, hydrolysis of the coating causes the migration times of the acidic proteins to increase because the sign of the electric field is reversed. The results show that the coating can be used for separations of acidic proteins, but the scheme of surface-confined bonding does not avoid the intrinsic instability of polyacrylamide. The results establish that polyacrylamide films of controlled thickness can be prepared by surface-confined living polymerization and that the capillaries have high efficiencies for protein separations. The polymerization method enables films of known thicknesses to be prepared, which will enable future studies of the relationship between film thickness and electrophoretic parameters. The method also allows easy preparation of crosslinked films to enable both application and fundamental study of cross-linked films in electrophoresis. Finally, SCLRP is expected to be generally applicable to radical polymerization because polymerization has been demonstrated in solution for numerous types of vinyl monomers.21-30 Use of SCLRP would facilitate the systematic investigation of new polymers in capillary electrophoresis. ACKNOWLEDGMENT This work was supported by the Department of Energy under Grant DE-FG02-91ER14187. We greatly appreciate the helpful discussions with Dr. Mingxian Huang at Supelco, Inc. and Dr. Wenbao Li at Du Pont. We thank Professor Andrew Zydney of the UD Department of Chemical Engineering for use of his capillary electropherograph. The atomic force microscopy was performed in the UD William Keck Laboratory. Received for review February 27, 1998. Accepted July 1, 1998. AC980231C
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4029
