Electrokinetic Chromatography and Mass Spectrometric Detection

Anal. Chem. 2010, 82, 4046–4054

Electrokinetic Chromatography and Mass Spectrometric Detection Using Latex Nanoparticles as a Pseudostationary Phase Christopher P. Palmer,*,†,‡ Emily F. Hilder,‡ Joselito P. Quirino,‡ and Paul R. Haddad‡ Department of Chemistry and Biochemistry, University of Montana, Missoula, Montana 59803, and Australian Centre for Research on Separation Science, School of Chemistry, University of Tasmania, Hobart, TAS, Australia The utility of novel latex nanoparticles as pseudostationary phases for electrokinetic chromatography with UV and mass spectrometric detection is demonstrated. The nanoparticles are synthesized using ab initio RAFT (reversible addition-fragmentation chain transfer) in emulsion polymerization, which yields small (63 nm) particles with a narrow size distribution, a hydrophobic core, and an ionic shell. The nanoparticles are shown to provide efficient and selective separations, with retention and separation selectivity dominated by hydrophobic interactions. The nanoparticles are highly retentive, such that they are effective at relatively low concentrations. Addition of the nanoparticles to the background electrolyte at these concentrations has a minor effect on the noise with UV detection, no measurable effect on the separation current, and minor effects on analyte ionization efficiency during electrospray ionization. The nanoparticles do not cause fouling or degradation of the electrospray-mass spectrometer interface even after several weeks of use. The combination of online sample preconcentration via sweeping and selective mass spectrometric detection yields low detection limits (10-16 ppb), particularly for more hydrophobic compounds. Electrokinetic chromatography (EKC) was introduced by Terabe et al. in 19841 and has proven to be a powerful technique for separation of a variety of nonionic and ionic analytes.2-8 The technique differs from capillary zone electrophoresis in that it utilizes an ionic pseudostationary phase (PSP) dissolved or suspended in the background electrolyte to effect the separation * To whom correspondence should be addressed. E-mail: christopher.palmer@ umontana.edu. Fax: 1-406-243-4227. † University of Montana. ‡ University of Tasmania. (1) Terabe, S.; Otsuka, K.; Ichikawa, A.; Tsuchiya, A.; Ando, T. Anal. Chem. 1984, 56, 111–113. (2) Kenndler, E.; Rizzi, A. J. Chromatogr. Libr. 2004, 69A, 297–318. (3) Otsuka, K.; Terabe, S. Methods Mol. Biol. (Totowa, NJ, U. S.) 2004, 243, 355–363. (4) Pappas, T. J.; Gayton-Ely, M.; Holland, L. A. Electrophoresis 2005, 26, 719– 734. (5) Pyell, U., Ed. Electrokinetic Chromatography. Theory, Instrumentation and Applications; John Wiley and Sons Ltd.: West Sussex, England, 2006. (6) Silva, M. Electrophoresis 2007, 28, 174–192. (7) Terabe, S. Anal. Chem. 2004, 76, 240A–246A. (8) Vindevogel, J.; Sandra, P., Eds. Introduction to Micellar Electrokinetic Chromatography; Huthig: Heidelberg, 1992.

4046

Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

of nonionic analytes and/or to alter the separation selectivity of ionic analytes. It differs from capillary electrochromatography and conventional liquid chromatography in that analytes are separated via selective interactions with the dispersed PSP rather than interactions with a stationary phase immobilized on a packed bed or monolith. The method offers relatively fast, selective, and efficient analytical separations in condensed media, as well as the convenience of easily replaceable PSPs. A variety of materials have been introduced and characterized as PSPs for EKC. The original report utilized micelles of sodium dodecyl sulfate,1 as has most of the work since that time. Additional micellar PSPs have also been introduced and shown to provide varied separation selectivity.9-15 Polymeric PSPs were introduced in the 1990s,16,17 and extensive studies have shown that these materials offer significant advantages and selectivity differences relative to micellar PSPs.18-24 More recently, polymeric nanoparticle (NP) PSPs have been introduced, and their separation performance has been characterized.25-31 (9) Fuguet, E.; Rafols, C.; Bosch, E.; Abraham, M. H.; Roses, M. J. Chromatogr., A 2002, 942, 237–248. (10) Fuguet, E.; Rafols, C.; Bosch, E.; Abraham, M. H.; Roses, M. Electrophoresis 2006, 27, 1900–1914. (11) Poole, S. K.; Poole, C. F. Analyst 1997, 122, 267–274. (12) Trone, M. D.; Khaledi, M. G. Anal. Chem. 1999, 71, 1270–1277. (13) Trone, M. D.; Khaledi, M. G. J. Microcolumn Sep. 2000, 12, 433–441. (14) Trone, M. D.; Khaledi, M. G. Electrophoresis 2000, 21, 2390–2396. (15) Trone, M. D.; Mack, J. P.; Goodell, H. P.; Khaledi, M. G. J. Chromatogr., A 2000, 888, 229–240. (16) Palmer, C. P.; Khaled, M. Y.; Mcnair, H. M. J. High Resolution Chromatogr. 1992, 15, 756–762. (17) Palmer, C. P.; Mcnair, H. M. J. Microcolumn Sep. 1992, 4, 509–514. (18) Palmer, C. P. Electrophoresis 2000, 21, 4054–4072. (19) Palmer, C. P. Electrophoresis 2002, 23, 3993–4004. (20) Palmer, C. P. In Electrokinetic Chromatography. Theory, Instrumentation and Applications; Pyell, U., Ed.; John Wiley and Sons Ltd.: West Sussex, England, 2006. (21) Palmer, C. P. Electrophoresis 2007, 28, 164–173. (22) Palmer, C. P. J. Sep. Sci. 2008, 31, 783–793. (23) Palmer, C. P.; McCarney, J. P. Electrophoresis 2004, 25, 4086–4094. (24) Palmer, C. P. Electrophoresis 2009, 30, 163–168. (25) Nilsson, C.; Birnbaum, S.; Nilsson, S. J. Chromatogr., A 2007, 1168, 212– 224. (26) Nilsson, C.; Nilsson, S. Electrophoresis 2006, 27, 76–83. (27) Priego-Capote, F.; Ye, L.; Shakil, S.; Shamsi, S. A.; Nilsson, S. Anal. Chem. (Washington, DC, U. S.) 2008, 80, 2881–2887. (28) Spegel, P.; Viberg, P.; Carlstedt, J.; Petersson, P.; Joernten-Karlsson, M. J. Chromatogr., A 2007, 1154, 379–385. (29) Viberg, P.; Spegel, P.; Carlstedt, J.; Joernten-Karlsson, M.; Petersson, P. J. Chromatogr., A 2007, 1154, 386–389. (30) Viberg, P.; Jornten-Karlsson, M.; Petersson, P.; Spegel, P.; Nilsson, S. Anal. Chem. 2002, 74, 4595–4601. 10.1021/ac902922u  2010 American Chemical Society Published on Web 04/19/2010

One significant and potentially negative consequence of the use of a dissolved or dispersed PSP is that the PSP itself enters or migrates through the detector, unless special and often complicated partial filling32 techniques are utilized. NP PSPs can also cause interference with the commonly used UV detection due to light scattering by the dispersed particles. Detector interference has also proven to be a hindrance for the potentially very useful hyphenation of EKC with mass spectrometric (MS) detection. Utilizing micellar PSPs with MS detection has proven to be particularly problematic because introduction of the low molecular weight surfactants into the electrospray ionization (ESI) interface and the mass spectrometer can lead to suppression of analyte ionization, fouling of the interface, and interference with the detection of low molecular weight analytes.33-38 Initial efforts to interface EKC with online ESI-MS detection included partial filling techniques,39-41 reversed migration of the micellar PSP away from the interface,42 the use of volatile surfactant micelles,43,44 and the use of high molecular weight polymeric PSPs.34,45,46 Due to complications with partial filling and reversed migration techniques, and the limited scope of volatile surfactants, the use of polymeric PSPs has significant advantages and has perhaps been the most successful for coupling with ESI. Promising results have also been reported with conventional micellar PSPs using chemical ionization47 or photoionization48,49 techniques. Combination of the polymeric PSP approach with the use of orthogonal ESI interfaces has recently been shown to provide sensitive, selective, and robust EKC-ESI-MS.50-54 In a similar approach, polymeric NP PSPs have also been shown to be compatible with EKC-ESI-MS in partial filling or continuous full-filling modes.30,55,56 (31) Nilsson, C.; Becker, K.; Harwigsson, I.; Bulow, L.; Birnbaum, S.; Nilsson, S. Anal. Chem. (Washington, DC, U. S.) 2009, 81, 315–321. (32) Amini, A.; Paulsen-Sorman, U.; Westerlund, D. Chromatographia 1999, 50, 497–506. (33) Lu, W.; Poon, G. K.; Carmichael, P. L.; Cole, R. B. Anal. Chem. 1996, 68, 668–674. (34) Lu, W. Z.; Shamsi, S. A.; Mccarley, T. D.; Warner, I. M. Electrophoresis 1998, 19, 2193–2199. (35) Rundlett, K. L.; Armstrong, D. W. Anal. Chem. 1996, 68, 3493–3497. (36) Varghese, J.; Cole, R. B. J. Chromatogr., A 1993, 652, 369–376. (37) Somsen, G. W.; Mol, R.; de Jong, G. J. J. Chromatogr., A 2003, 1000, 953– 961. (38) Somsen, G. W.; Mol, R.; Jong, G. J. Anal. Bioanal. Chem. 2006, 384, 31– 33. (39) Muijselaar, P. G.; Otsuka, K.; Terabe, S. J. Chromatogr., A 1998, 802, 3– 15. (40) Nelson, W. M.; Lee, C. S. Anal. Chem. 1996, 68, 3265–3269. (41) Nelson, W. M.; Tang, Q.; Harrata, A. K.; Lee, C. S. J. Chromatogr., A 1996, 749, 219–226. (42) Yang, L. Y.; Harrata, A. K.; Lee, C. S. Anal. Chem. 1997, 69, 1820–1826. (43) Ishihama, Y.; Katayama, H.; Asakawa, N. Anal. Biochem. 2000, 287, 45– 54. (44) Petersson, P.; Jornten-Karlsson, M.; Stalebro, M. Electrophoresis 2003, 24, 999–1007. (45) Ozaki, H.; Itou, N.; Terabe, S.; Takada, Y.; Sakairi, M.; Koizumi, H. J. Chromatogr., A 1995, 716, 69–79. (46) Ozaki, H.; Terabe, S. J. Chromatogr., A 1998, 794, 317–325. (47) Takada, Y.; Sakairi, M.; Koizumi, H. Rapid Commun. Mass Spectrom. 1995, 9, 488–490. (48) Mol, R.; de Jong, G. J.; Somsen, G. W. Anal. Chem. 2005, 77, 5277–5282. (49) Mol, R.; de Jong, G. J.; Somsen, G. W. Electrophoresis 2005, 26, 146–154. (50) Akbay, C.; Rizvi, S. A. A.; Shamsi, S. A. Anal. Chem. 2005, 77, 1672–1683. (51) Hou, J.; Zheng, J.; Rizvi, S. A. A.; Shamsi, S. A. Electrophoresis 2007, 28, 1352–1363. (52) Hou, J.; Zheng, J.; Shamsi, S. A. J. Chromatogr., A 2007, 1159, 208–216. (53) Hou, J.; Zheng, J.; Shamsi, S. A. Electrophoresis 2007, 28, 1426–1434. (54) Rizvi, S. A. A.; Zheng, J.; Apkarian, R. P.; Dublin, S. N.; Shamsi, S. A. Anal. Chem. 2007, 79, 879–898.

Poor concentration sensitivity is another significant limitation of EKC, especially when UV absorbance detection is utilized. Various online preconcentration techniques, which allow for the injection of large volumes of sample followed by focusing and concentration of the analytes by a factor of several thousand or more, have been introduced to address this problem.57-65 These techniques include sweeping, field-enhanced, or amplified sample stacking, focusing with photopolymerized porous monoliths, the use of solvent gradients, and focusing by micellar collapse. To the authors’ knowledge, sweeping with a polymeric pseudostationary phase has only been reported once,66 and NP PSPs have never been evaluated for sample preconcentration. A significant advantage of the use of online preconcentration techniques with polymeric or NP PSPs would be the ability to detect and analyze very low concentration samples using MS detection. In the current work, a novel polymeric NP PSP is introduced and is shown to provide good separation performance with continuous full filling EKC with UV and ESI-MS detection. The PSP is also employed for sample preconcentration via sweeping, allowing very sensitive analysis with ESI-MS detection. The NPs are synthesized by a different approach and have significant advantages over those reported earlier. The synthetic approach, ab initio RAFT (reversible addition-fragmentation chain transfer) in emulsion polymerization,67,68 generates uniformly sized NPs with a poly(butyl acrylate) core and poly(acrylic acid) exterior. The synthetic approach affords unprecedented control over the size and chemical composition of the NPs. No complicated postsynthetic purification procedures are required with this approach, because no additional surfactants or organic solvents are utilized for the synthesis. The NPs are smaller in diameter (60 nm) than previously reported NP PSPs, which reduces light scattering interference with UV detection and may enhance separation performance.55 The NPs form stable colloidal suspensions in aqueous and organic solvent-modified aqueous buffers. Also, these materials can be utilized directly as PSPs without postsynthetic chemical modification, unlike many previously reported NP PSPs.55,56 The new NPs have high chemical affinity for certain compounds and can, thus, be used to achieve separations or alter separation selectivity at relatively low concentrations where interference with UV and MS (55) Nilsson, C.; Viberg, P.; Spegel, P.; Joernten-Karlsson, M.; Petersson, P.; Nilsson, S. Anal. Chem. 2006, 78, 6088–6095. (56) Viberg, P.; Spegel, P.; Nilsson, J.; Petersson, P.; Joernten-Karlsson, M.; Nilsson, S. Chromatographia 2007, 65, 291–297. (57) Aranas, A. T.; Guidote, A. M., Jr.; Quirino, J. P. Anal. Bioanal. Chem. 2009, 394, 175–185. (58) Breadmore, M. C.; Thabano, J. R. E.; Dawod, M.; Kazarian, A. A.; Quirino, J. P.; Guijt, R. M. Electrophoresis 2009, 30, 230–248. (59) Liu, Z.; Sam, P.; Sirimanne, S. R.; McClure, P. C.; Grainger, J.; Patterson, D. G., Jr. J. Chromatogr., A 1994, 673, 125–132. (60) Quirino, J. P.; Haddad, P. R. Anal. Chem. 2008, 80, 6824–6829. (61) Quirino, J. P.; Dulay, M. T.; Bennett, B. D.; Zare, R. N. Anal. Chem. 2001, 73, 5539–5543. (62) Quirino, J. P.; Dulay, M. T.; Zare, R. N. Anal. Chem. 2001, 73, 5557–5563. (63) Quirino, J. P.; Terabe, S. Science (Washington, D. C.) 1998, 282, 465–468. (64) Quirino, J. P.; Terabe, S. Anal. Chem. 1998, 70, 1893–1901. (65) Simpson, S. L.; Quirino, J. P.; Terabe, S. J. Chromatogr., A 2008, 1184, 504–541. (66) Shi, W.; Palmer, C. P. J. Sep. Sci. 2002, 25, 215–221. (67) Ferguson, C. J.; Hughes, R. J.; Nguyen, D.; Pham, B. T. T.; Gilbert, R. G.; Serelis, A. K.; Such, C. H.; Hawkett, B. S. Macromolecules 2005, 38, 2191– 2204. (68) Sprong, E.; Leswin, J. S. K.; Lamb, D. J.; Ferguson, C. J.; Hawkett, B. S.; Pham, B. T. T.; Nguyen, D.; Such, C. H.; Serelis, A. K.; Gilbert, R. G. Macromol. Symp. 2006, 231, 84–93.


4047

detection is minimized and where the PSP does not have any measurable effect on the separation current. METHODS AND MATERIALS Nanoparticle Synthesis. RAFT agent, 2-{[(butylsulfanyl)carbonothioyl]sulfanyl}propanoic acid was synthesized according to a published procedure.67 The RAFT agent was then reacted with acrylic acid in the presence of initiator to give a macroRAFT agent expected to contain, on average, five acrylic acid units. RAFT agent (3.2910 g, 1.39 × 10-2 mol), 0.3883 g (1.39 × 10-3 mol) of V-501, 0.555 g (1.39 × 10-2 mol) of NaOH, 4.9954 g (6.94 × 10-2 mol) of acrylic acid, and 8.3080 g of water were added to a round bottomed flask. The flask was sealed with a rubber septum and deoxygenated by bubbling nitrogen through the solution. The polymerization was then allowed to proceed at 60 °C for 2 h with stirring. Ab initio RAFT in emulsion polymerization was carried out according to the following procedure: 0.5863 g (8.25 × 10-4 mol of RAFT/g) of the macro-RAFT agent, 0.0763 g (2.4 × 10-4 mol) of V-501, 0.1220 g (3.05 × 10-3 mol) of NaOH, and 80.62 g of water were added to a round bottomed flask. The flask was sealed with a rubber septum and deoxygenated by bubbling nitrogen through the solution. Butyl acrylate (20.00 g, 0.1560 mol) was separately deoxygenated with nitrogen and then drawn into a 25 mL gastight Hamilton syringe. The flask was then immersed in an oil bath at 60 °C, and monomer addition was begun. Butyl acrylate was added at a feed rate of 1.0 g/h for the first 2 h, which was then increased to 6.0 g/h for the next 3 h. The reaction was then allowed to proceed for a further 1 h after the completion of the monomer feed to allow it to reach high conversion. Vigorous magnetic stirring was used throughout to stop the monomer pooling on the top of the reaction mixture. Following the completion of the reaction, the mixture containing the formed nanoparticles was cooled to room temperature and then stored at 4 °C until use. Nanoparticle Characterization. The macro-RAFT agent was characterized by direct infusion electrospray ionization mass spectrometry (ESI-MS) using a Finnigan Mat LCQ MS. The macro-RAFT agent was dissolved in 1:1 methanol/water and infused into the MS at 0.2 mL/min. The electrospray voltage was 5 kV, nitrogen sheath gas pressure was 7 kPa, and the source was heated to 200 °C. Particle size measurements were conducted by dynamic light scattering using a Malvern Zetasizer Series ZS instrument. Measurements were performed at 25 °C on 0.1 wt %/vol nanoparticle suspensions in distilled water and in 27 mM, pH 9.2 borate buffer. The Malvern Zetasizer software conducts a statistical moments analysis of the scattering data and provides a value for the polydispersity index (PDI). The relative polydispersity (coefficient of variation in the particle size distribution) can be calculated from the PDI using the equation 100 × (PDI)1/2. Electrokinetic Chromatography. All EKC-UV experiments were conducted on an Agilent 3DCE instrument controlled with Agilent Chemstation software. Fused silica capillary (Polymicro Technologies) with internal diameter of 50 µm was used for all studies. Separation capillaries were 48 cm in total length and 39.5 cm in effective length. All new capillaries were flushed for 45 min with 0.1 N sodium hydroxide, 30 min with distilled 4048


deionized water, and 10 min with separation background electrolyte (BGE) before being used for separations. Capillaries were flushed for 3 min with the BGE between injections. Analyte solutions were injected by pressure at 3 kPa for 3 s. Analytes were obtained in the highest purity available (g98%) from Sigma-Aldrich, except dipropyl phthalate (>99%), which was obtained from Fluka. Stock solutions (1000 ppm) of the analytes were prepared in water, acetone, or methanol, and analysis samples were prepared by dilution of the stock solutions to the reported concentration in the separation electrolyte, unless noted otherwise. Analyte peaks were identified by injection of individual standards or by spectral matching with an in-house generated spectral library. EKC buffers were prepared in distilled deionized water and analytical reagent grade acetonitrile (Ajax Finechem). Borate buffers were prepared by dissolving analytical reagent grade sodium tetraborate (Ajax Chemical) in water. The pH was measured (using a LabChem pH meter with glass electrode) and adjusted if necessary with sodium hydroxide (analytical reagent, BDH) before dilution to the reported concentration with water and acetonitrile. Ammonium carbonate buffers were prepared in water using laboratory reagent grade ammonium hydrogen carbonate (Ajax Chemical), and the pH was adjusted as necessary with concentrated aqueous ammonium hydroxide (analytical reagent, 28%, Ajax Chemical) before dilution to the final reported carbonate concentration with water and acetonitrile. Nanoparticle PSP solutions were prepared by dilution of a 22.7 wt %/vol aqueous suspension of NP in the reported BGE. The stock NP solution was refrigerated when not in use. All buffer solutions had a pH of 9 or above to ensure ionization of the polyacrylic acid surface groups. The migration time and the effective electrophoretic mobility of the NPs was estimated using the iterative method presented by Bushey and Jorgenson69 using acetone as a marker of electroosmotic flow and the migration times of five alkyl phenyl ketone homologues: acetophenone, propiophenone, butyrophenone, valerophenone, and hexanophenone. The Excel application solver was used to find a value for the migration time of the NPs that gave a maximum R2 for the plot of log retention factor (k) vs carbon number, with all R2 values being greater than 0.995. The migration times of the NP and acetone were then used to calculate the effective electrophoretic mobility of the NP. The methylene selectivity was calculated as 10m, where m is the slope of the plot of log(k) vs carbon number. Migration times and plate numbers were taken from Chemstation after manual peak integration. EKC with Mass Spectrometric Detection. All experiments were conducted on an Agilent 3DCE system interfaced with an Agilent 6320 ion trap MS system using an Agilent G1607A ESI interface and controlled with Agilent Chemstation software. A single 94 cm, 50 µm ID fused silica capillary was used for all experiments. The capillary was rinsed as described above before use, for 10 min following each change in separation buffer composition, and for 4 min with separation buffer between injections. Analyte solutions were injected by pressure at 5 kPa for 3 s. Separations were run in acetonitrile-modified ammonium carbonate buffers prepared as described above with an applied potential of 25 (69) Bushey, M. M.; Jorgenson, J. W. Anal. Chem. 1989, 61, 491–493.

kV and 0.5 kPa of pressure applied to the inlet vial to improve migration efficiency by offseting apparent hydrodynamic flow caused by the ESI interface or difference in height between the inlet vial and interface. Sheath liquid flow was supplied by an Agilent 1200 pump fitted with a 1:100 flow splitter. The sheath liquid composition was selected to be similar to that for a previously reported optimized EKC-MS method for the same or similar analytes50 and consisted of 80/20 methanol/ water with 40 mM ammonium carbonate at pH 8.0. Sheath liquid flow rate was found to provide a good trade-off between electrospray stability and MS sensitivity at 4.5 µL/min, also similar to the previously reported optimum.50 Nebulizer pressure was set to a minimum to provide stable electrospray operation (48 kPa) and highest possible separation efficiency. Additional spray chamber parameters were set as follows: dry gas flow rate of 5.0 L/min., drying gas temperature of 250 °C, and electrospray voltage of +3.0 kV (end plate voltage of -3.0 kV). Sweeping EKC. All experiments were conducted on the Agilent CE-MS system described above with the same capillary dimensions, sheath liquid, and ionization conditions. All injections were made using pressure generated by the air pump on the Agilent CE instrument. For conventional injections, analytes were dissolved in the run buffer and injections were made in the standard Chemstation injection protocol. For sweeping injections, analytes were dissolved at reduced concentration in buffers without a NP or an acetonitrile modifier. Sweeping injections were performed by applying approximately 90 kPa of pressure to the sample vial for a fixed period, by selecting “flush” for a specified time on the Agilent CE Chemstation injection system. Enhancement factors were calculated using the peak heights of the observed peaks in the following equation, Hswp Ccon EF ) Hcon Cswp where H refers to the peak height, C refers to the analyte concentrations, and the subscripts swp and con refer to the sweeping results and conventional results, respectively. RESULTS AND DISCUSSION Nanoparticle Synthesis. The latex NPs are synthesized following the general scheme introduced by Ferguson et al.67 Briefly, amphiphilic diblock copolymers that can self-assemble into micelles are formed via RAFT control. This is done via the sequential polymerization of first a hydrophilic and then a hydrophobic monomer in the presence of a suitable RAFT agent. In this case, acrylic acid was used as the hydrophilic monomer and was polymerized in the presence of the RAFT agent to yield a macro RAFT agent expected to have an average of five acrylic acid units. Further polymerization with a hydrophobic monomer, butyl acrylate, yielded the diblock copolymer from which the micelles were formed in situ. A further controlled monomer (butyl acrylate) addition caused these micelles to evolve into latex particles, which also grow under RAFT control, thus with excellent control of polydispersity. Nanoparticle Characterization. Direct infusion electrospray mass spectrometry of the hydrophilic block/macro-RAFT agent

Figure 1. Latex nanoparticle size distribution as determined by dynamic light scattering.

Figure 2. Separation of acetone (2%) and alkyl-phenyl ketones (20 ppm): 1, acetophenone; 2, propiophenone; 3, butyrophenone; 4, valerophenone; and 5, hexanophenone using 0.12% NP in 20 mM aqueous borate buffer at pH 9.2. λ ) 240 nm.

gives an envelope of peaks of different m/z, with a single peak for each degree of polymerization. Identification and integration of these peaks allows analysis of the average and range of chain lengths. Using this approach, the macro-RAFT agent was determined to have an average degree of polymerization of 5.3, with a range of 1-10. Approximately 65% of the material had degree of polymerization between 4 and 7. Figure 1 shows the distribution of nanoparticle size as determined by dynamic light scattering. The nanoparticles have a Zavg diameter of 63 nm and a relative polydispersity (coefficient of variation in the particle size distribution) of 26%. EKC-UV Detection. A representative optimized separation of an homologous series of alkyl-phenyl ketones using the NP PSP is presented in Figure 2. These nonionic solutes are not separated in the absence of an ionic PSP. The result demonstrates that the NPs can be used to generate highly efficient and selective separations. The average plate number generated in this separation is 205 000 ± 39 000, which is somewhat less than or equivalent to that reported for SDS micelles under similar conditions.70-72 The plate number does decrease with retention, however, and while acetophenone and propiophenone both have plate numbers of (70) Palmer, C. P.; Terabe, S. Anal. Chem. 1997, 69, 1852–1860. (71) Peterson, D. S.; Palmer, C. P. Electrophoresis 2001, 22, 1314–1321. (72) Shi, W.; Watson, C. J.; Palmer, C. P. J. Chromatogr., A 2001, 905, 281– 290.


4049

Table 1. Comparison of Electrophoretic and Chromatographic Properties of the NPs with SDS Micelles pseudostationary phase

electrophoretic mobility (cm2V-1s-1)

methylene selectivity

SDS micelles54 latex nanoparticles

-4.05 ± 0.02 × 10-4 -4.06 ± 0.05 × 10-4

2.33 ± 0.04 3.25 ± 0.03

312 000 ± 11 000, hexanophenone has a plate number of 114 000 ± 12 000. In most cases, separation efficiencies in EKC are limited by instrumental factors and longitudinal diffusion.73-75 However, when this is the case, the plate numbers should be higher for later eluting peaks. This is not observed in this case, implying that additional factors are contributing to peak dispersion. The peaks for butyrophenone and valerophenone show some fronting (symmetry ) 2.63 and 3.25, respectively) implying dispersion caused by a nonlinear, anti-Langmuir type isotherm.76 Similar losses in efficiency and peak symmetry for longer chain homologues have been observed previously with polymeric PSPs.71,72 The efficiency and symmetry were improved for copolymers of 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS) and lauryl methacrylamide (LMAm) as the mole ratio of lipophilic to hydrophilic monomer was increased,72 implying that the observed loss in efficiency and symmetry may result from insufficient volume or accessibility of the lipophilic region of the PSP. The homologous series separation also allows for estimation of the NP mobility and methylene selectivity (the chromatographic selectivity between adjacent pairs of the homologous series) using the linearization method of Bushey and Jorgenson.69 The five ketones separated in Figure 2 cover over 70% of the migration range, and the method generates linear plots with consistent values for mobility and methylene selectivity. These results are presented in Table 1 along with results for SDS micelles obtained by the same approach under similar separation conditions.71 The NPs have mobility that is not significantly different from SDS micelles and, thus, provide a similar EKC migration range. Polymers of sodium undecylenate and sodium undecenyl sulfate have electrophoretic mobilities about 5% greater than SDS,70 while the mobilities of AMPS-LMAm copolymers vary from 5% less to five times greater than the mobility of SDS micelles depending on the mole ratio of AMPS/LMAm.72 The NPs have much higher methylene selectivity relative to SDS micelles, indicating that they provide stronger hydrophobic interactions. The methylene selectivity for the NPs is also much higher than that for most polymeric PSPs, which generally have lower methylene selectivity than SDS micelles.70,72 In fact, the NPs are highly retentive toward solutes with pendant alkyl chains like the alkyl-phenyl ketones and dialkyl phthalates (see also Figure 4A) and can, thus, be used at relatively low concentrations (0.12 wt %/vol in Figure 2) and with relatively high organic modifier content (20% acetonitrile in Figure 3B,C) to achieve high resolution separations. Figure 3 presents optimized separations for two mixtures of representative pharmaceutical compounds under different separa(73) (74) (75) (76)

Davis, J. M. J. Microcolumn Sep. 1998, 10, 479–489. Davis, J. M. Chem. Anal. (N. Y.) 1998, 146, 141–184. Davis, J. M. Adv. Chromatogr. (N. Y.) 2000, 40, 115–157. Williamson, Y.; Davis, J. M. Electrophoresis 2005, 26, 4026–4042.

4050


Figure 3. Separation of representative pharmaceutical compounds without (upper traces) and with 0.16% NP (lower traces). All solutes at 20 ppm. (A) β-blockers: 1, atenolol; 2, metoprolol; 3, alprenolol; and 4, propranolol in 8% acetonitrile, 20 mM borate buffer, pH 10.1, λ ) 202 nm; (B) 5, nortriptyline; 6, diphenhydramine; and 7, salbutamol in 20% acetonitrile, 20 mM ammonium carbonate, pH 10, λ ) 214 nm; (C) 20% acetonitrile, 20 mM borate, pH 10.1, λ ) 214 nm.

tion conditions. The NPs require high pH in order to ensure ionization, and so, all separations are conducted in high pH buffers. Separations under identical buffer conditions without the NPs are also presented, showing that in most cases the analytes are at least weakly cationic under the separation conditions and that the NPs have a significant effect on migration and separation selectivity. Figure 3A demonstrates that the NPs can be used to separate β-blockers under conditions where they are not resolved in the absence of the NPs. These and similar compounds can be at least partially separated at low pH where they are more fully ionized, but additives are often used to achieve full resolution.77,78 Polymeric surfactants have also been employed as additives for the simultaneous separation and chiral resolution of these compounds.50,54 Under the selected conditions, all four of these β-blockers are ionized to essentially the same extent, indicating that the observed separation selectivity is not due solely to ionic interactions with the NPs. In fact, the migration order in this separation compares well with the relative hydrophobicity of the compounds, as indicated by their octanol-water partition coefficients (PO/W).50 This is in contrast to early reports for NP PSPs, which found analyte-PSP interactions to be primarily ionic.30 Figure 3B,C demonstrates that the separation selectivity for another three selected pharmaceuticals is altered significantly by the addition of the NPs. Salbutamol has a net negative charge (77) Mazzarino, M.; de la Torre, X.; Mazzei, F.; Botre, F. J. Sep. Sci. 2009, 32, 3562–3570. (78) Zhou, S.; Wang, Y.; De Beer, T.; Baeyens, W. R. G.; Fei, G. T.; Dilinuer, M.; Ouyang, J. Electrophoresis 2008, 29, 2321–2329.

in the borate buffer system (Figure 3C) because it forms a complex with the borate anion. Nortriptyline gives poor efficiency in the carbonate buffer, presumably due to interactions with buffer components. The change in migration order and selectivity for these compounds is much greater than that reported earlier for a NP PSP of different chemistry,30 with which salbutamol showed weak or no interactions and ionic interactions were primarily responsible for the separation and selectivity. With the current NPs, nortriptyline shows the strongest interactions and salbutamol shows the weakest. Once again, the strength of the interactions increases with increasing PO/W,79,80 indicating that hydrophobic interactions are dominant. Salbutamol shows weak interactions with the NPs either as the weak cation or as the anionic borate complex. In some cases, the peaks for the pharmaceuticals in Figure 3 show significant tailing and relatively low plate numbers, although the peak tailing is emphasized by the expanded time scales. This broadening and tailing is not observed for all compounds or under all buffer conditions. The cause of this tailing may be related to nonlinear Langmuir type isotherms for some compounds,81,82 even at these relatively low concentrations. Reducing solute concentrations to 5 ppm, where signal-to-noise ratios are relatively low, did not improve the peak shapes. An alternative explanation for the band broadening and peak asymmetry could be slow mass transfer kinetics. It is not yet clear whether nonlinear isotherms or slow mass transfer and the resulting asymmetrical peak dispersion are a more general limitation of the NP PSP. Many of the performance characteristics of this NP PSP are advantageous and suggest that further study and development is warranted. The NPs form highly stable suspensions in aqueous and mixed aqueous-acetonitrile buffer systems. Suspensions showed no visible signs of instability over periods of several weeks. Single sets of buffer vials were used in the instrument without replenishment or agitation for several continuous hours of experiments with no measurable or systematic change in performance. The NP modified BGEs also showed no measurable increase in the separation current relative to the BGE alone under normal conditions. Only by dispersing a relatively high concentration of the NPs (0.4%) in a low concentration buffer (5 mM borate, pH 9.2) and applying a high field strength (625 V/cm) was a statistically significant (P < 0.001) increase in the current of 0.07 µA, or 0.4% of the total current, observed. Under typical operating conditions of e0.16% NP and 417 V/cm, the NPs would be expected to contribute a negligible 0.02 µA or less to the current. Finally, Figures 2 and 3 illustrate that, while the NPs do cause a statistically significant increase in the noise relative to the background electrolyte, this increase is not so severe as to prevent their use in this mode. Analysis of two minute regions of “typical” baseline from four separate separations run under identical conditions, two with and two without NPs, shows an average increase in the standard deviation of the signal of approximately 3-fold, from 0.021 to 0.061 mAU. The general nature of the noise is very similar, i.e., the NPs do not cause an increase in the (79) Brodin, A. Acta Pharm. Suec. 1974, 11, 141–148. (80) Hansch, C.; Leo, A.; Hoekman, D. H. Exploring QSAR; American Chemical Society: Washington, DC, 1995. (81) Smith, K. W.; Davis, J. M. Anal. Chem. 2002, 74, 5969–5981. (82) Williamson, Y.; Davis, J. M. Electrophoresis 2006, 27, 572–583.

Figure 4. (A) EKC-MS separation of alkyl phthalates and alkyl-phenyl ketones using 20% acetonitrile, 20 mM ammonium carbonate at pH 10 and 0.2% NP: 1, diethyl phthalate (135 + 223); 2, butyrophenone (149); 3, valerophenone (163); 4, dipropyl phthalate (191 + 251); 5, hexanophenone (177); and 6, dibutyl phthalate (205 + 279). Phthalates at 16 ppm, ketones at 40 ppm. Extracted ion electropherogram 135, 149, 163, 177, 191, 205, 223, 251, and 279 m/z. (B) Average background mass spectrum in the migration time range of 5-7 min. (C) Raw (upper) and background subtracted (lower) mass spectrum for dipropyl phthalate.

number or magnitude of noise spikes, as might be expected for severe light scattering. EKC-MS Detection. Figures 4-6 show the results for three representative EKC-MS separations. Figure 4 demonstrates that it is possible to separate and detect nonionic compounds by this approach, although low electrospray ionization efficiency of some compounds (e.g., the alkyl-phenyl ketones) makes their detection difficult. It is possible that the sensitivity for either or both series of compounds could be improved by specific optimization of ionization and MS conditions. The separation of dialkyl phthalates in Figure 4A shows significantly improved resolution and signalto-noise over similar NP EKC-MS separations reported earlier.29,55,56 The extracted ion electropherogram presented shows good signalto-noise, but all of the peaks except butyrophenone were detectable in the total ion electropherogram (120-300 m/z). An important performance criterion for additives and PSPs is the extent to which they affect the background spectrum. The mass spectrum in Figure 4B is the average spectrum over the period of baseline from 5 to 7 min and illustrates the background spectrum frequently observed in the presence of NP. The ions at 146 and 274 m/z are not observed in the absence of the NPs. Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4051

Figure 5. (A) EKC-MS separation of β-blockers using 10% acetonitrile, 20 mM ammonium carbonate at pH 10, and 0.2% NP: 1, atenolol; 2, metoprolol; 3, alprenolol; and 4, propranolol, each at 10 ppm. Total ion electropherogram 240-270 m/u. (B) Raw mass spectrum of alprenolol.

Figure 4C shows the mass spectrum for dipropyl phthalate before and after subtraction of the background spectrum. It is clear that this background spectrum is significant and can interfere with the mass spectral analysis of analyte peaks at these concentrations, although the relatively constant background allows the use of subtraction for the analysis of solutes at sufficient concentration. Figures 5 and 6 demonstrate the applicability of the NPs to EKC-MS analyses of more easily ionized basic pharmaceuticals. These separations take advantage of the separation selectivity offered by the NPs but continue to show some limitations in terms of separation efficiency and peak symmetry. Figure 5 shows the separation and detection of four β-blockers, and the mass spectrum for alprenolol without any background subtraction. Little or no background interference is observed in this case because of the relatively narrow mass range monitored (240-270 m/z). Figure 6 presents the extracted ion electropherogram for diphenhydramine, salbutamol, and nortriptyline, as well as the raw mass spectra for diphenhydramine and nortriptyline. All three peaks are clearly detectable in the total ion electropherogram (150-330 m/z). Although significant tailing and peak broadening are evident for nortriptyline, Figure 6C shows that it is still possible to obtain a good mass spectrum for the compound. Figure 6C also shows a reduced background level relative to that in Figure 6B. It is not yet clear why the background is more problematic in some separations than in others. The NPs also perform well in the EKC-MS mode, with no statistically significant increase in the noise and no evidence of fouling at the ESI interface over 2 weeks of continuous operation. The lack of fouling is probably attributable at least in part to the orthogonal spray design of the source, which results in the NPs being sprayed past the MS inlet and to waste. Methods developed using UV detection were easily transferred to the EKC-MS system. The background spectrum for NP suspensions (Figure 4B) often showed signals at 146 and 274 m/z. These are thought to be associated with impurities in the NP stock suspension most likely 4052


Figure 6. EKC MS of 1, diphenhydramine (222 + 240); 2, salbutamol (167 + 256); and 3, nortriptyline (264) in 20% acetonitrile, 20 mM ammonium carbonate at pH 10, and 0.2% NP. Each solute at 16 ppm. Extracted ion electropherogram 167, 222, 240, 256, and 264 m/u. Raw mass spectra of (B) diphenhydramine and (C) nortriptyline.

from one or more as yet unidentified reagents used in the synthesis. These relatively low molecular weight materials might be eliminated or reduced by dialysis purification. The background spectrum did not interfere with detection or the mass spectra of the selected analytes but would be problematic for the detection and analysis of analytes with signals at those m/z values. A significant concern with EKC-MS is that the added PSP will suppress ionization of analytes, thereby reducing the sensitivity of the analysis.33-35,37 Although not as severe as with conventional surfactants, some loss in signal has also been reported for higher concentrations of polymeric PSPs50,83 and for NP PSPs.30 The peak areas for selected analytes with and without the NP PSP are presented in Figure 7. In each case, the area represents the average of seven replicate injections of a single sample dissolved in background electrolyte without NP. Salbutamol, diphenhydramine, and nortriptyline are fully resolved with or without the NP and were injected as a mixture. Alprenolol is partially resolved from the other three β-blockers in the absence of NP and was analyzed as part of the mixture of four β-blockers in Figure 5. Individual sample solutions of diethyl phthalate and dipropyl phthalate were injected because these compounds are not separated in the absence of the NP, and comigration could result in ionization suppression. Salbutamol (P ) 0.03) and nortriptyline (P < 0.01) did show significant reductions in signal in the presence of 0.2 wt %/vol NPs at greater than 95% confidence, with a 19% and 37% reduction in signal, respectively. However, no significant (83) Shamsi, S. A. Anal. Chem. 2001, 73, 5103–5108.

Table 2. Quantitative Figures of Merit for Sweeping-EKC-MS Experiments analyte

enhancement factor

detection limit (ppb)

RSD (%)

dipropyl phthalate dibutyl phthalate

30 90

16a 10a

4.7b 7.3b

a Detection limits are the concentration calculated to give a peak height of two times the typical peak to peak noise in the extracted ion chromatogram. b Average RSD in peak height for concentrations ranging from 20 to 200 ppb, 5 levels, n ) 3-5 per level.

Figure 7. Area counts for six analytes (Sal ) salbutamol, Diphen ) diphenhydramine, Nor ) nortriptyline, Alp ) alprenolol, diEtph ) diethylphthalate, and diPrph ) dipropylphthalate) run under the conditions and at the concentrations indicated in Figure 6, with and without NP. Error bars are 95% CI, n ) 7.

Figure 8. Total ion chromatogram (175-285 m/u) results for sweeping EKC-MS of dialkyl phthalates. (A) Conventional 5 s 5 kPa injection of 1 ppm: 1, diethyl; 2, dipropyl; and 3, dibutyl phthalates dissolved in separation buffer. (B) Sweeping 30 s 90 kPa injection (27 cm, 28% of capillary) of 0.05 ppm (2) dipropyl and (3) dibutyl phthalates in 15 mM NH4HCO3 buffer at pH 9.7. Separation buffer for both separations is 10 mM NH4HCO3 (pH 9.7), 20% acetonitrile, and 0.4% NP.

reduction in signal was observed for the phthalates, and a statistically significant (P < 0.01) increase in signal was observed for alprenolol. The increase in signal for alprenolol in the presence of NP may be related to the fact that it is not fully resolved from the other β-blockers in the absence of NP (Rs ) 0.6), and thus, ionization may be partly suppressed by the other β-blockers. Sweeping-EKC-MS. The results for the sweeping EKC-MS experiments for dialkyl phthalates are presented in Figure 8 and Table 2. Sweeping with the NPs does provide significant enhancement in sensitivity, approaching an enhancement factor of nearly 100 in the best case. The combination of this with sensitive MS detection allows low detection limits, particularly for hydrophobic compounds with the highest enhancement factors using extracted ion electropherograms. The focusing and enhancement factors are, as predicted for sweeping, better for the more highly retained species. Sweeping was unsuccessful for diethyl phthalate and most effective for dibutyl phthalate. The reduced migration times and retention for the phthalates with sweeping injections result from

the separations having been carried out over a shorter effective length of capillary. Reproducibility of the peak heights for the sweeping experiments with 20-100 ppb solute concentrations is 5-10% RSD, which is only fair but compares well with typical relative standard deviations in peak heights (10-15%) achieved with this system for conventional injections of 1 ppm solutions. Five level calibration curves for dipropyl and dibutyl pththalates were linear in the range of 20-200 ppb with sweeping injection, giving r2 values of 0.995 and 0.997, respectively. CONCLUSIONS A new chemistry is introduced for the synthesis of nanoparticles that are suitable for use as PSPs in EKC with UV and MS detection. The NPs are relatively monodisperse and are smaller in diameter (63 nm) than those previously reported. Importantly, they can be used as PSPs without complicated post synthetic purification or modification. The synthetic approach can be easily modified to utilize different monomer chemistries or to alter the size of the particles. Thus, this approach could potentially offer NP PSPs with altered separation selectivity and improved performance. As a PSP, the NPs provide efficient and selective separations of a variety of compounds, although some more highly retained cationic solutes do show significant tailing and peak broadening. These NPs form highly stable suspensions in aqueous and modified aqueous buffers. They provide very strong hydrophobic interactions, and the separation selectivity for partially ionized basic compounds appears to be based primarily on hydrophobic interactions. The NP PSP causes a negligible increase in the separation current, potentially making it suitable for high-speed separations and compatible with various field-amplified sample preconcentration methods. The NPs do cause a small but significant increase in the baseline noise with UV detection, but the noise does not prevent their use in this mode. The NPs are compatible with MS detection in that they do not cause fouling of the interface and in most cases do not cause a significant reduction in analyte signal. The compatibility with both UV and MS detection allows methods to be developed with UV detection and transferred directly to the MS system. There is often a significant background MS spectrum associated with the NPs. This background spectrum does not interfere with detection or MS analysis of analytes, so long as they have m/z different from the background and are at sufficient concentration to permit background subtraction. It is expected that purification of the NPs, possibly through dialysis, would eliminate this problem. The NP PSP is also suitable for sweeping EKC, providing a substantial enhancement in sensitivity particularly for highly Analytical Chemistry, Vol. 82, No. 10, May 15, 2010

4053

retained analytes. Using the combination of sensitive MS detection and sweeping injection allows estimated detection limits in the 10-20 ppb range. This approach could be used for online preconcentration of a variety of analytes before MS detection, although it is clearly most effective for more hydrophobic compounds. The demonstrated qualities of these NPs in EKC separations in both the UV and MS modes and for online preconcentration through sweeping indicate that this approach has significant promise for the continuing development of high performance PSPs. Further study is necessary to elucidate the interaction mechanisms responsible for the separation selectivity, to understand and eliminate the observed peak asymmetry and broadening, and to identify or eliminate the source of the background using MS detection. The synthetic approach used here allows independent control over the lengths and chemistries of the hydrophilic and hydrophobic blocks and the size of the NPs. It

4054


would be useful to use this synthetic flexibility and control to perform systematic studies into the factors that affect performance and selectivity, leading to NPs with improved performance characteristics and/or altered separation selectivity. ACKNOWLEDGMENT This research was supported under Australian Research Council’s Discovery (DP0666121) and Linkage, Infrastructure, Equipment, and Facilities (LE0668471) Funding Schemes. P.R.H. is the recipient of an ARC Federation Fellowship. We acknowledge Prof. Robert Gilbert and Dr. Joost Leswin for assistance in the synthesis of the nanoparticles used in this work.

Received for review December 21, 2009. Accepted April 6, 2010. AC902922U

Electrokinetic Chromatography and Mass Spectrometric Detection

Recommend Documents