Anal. Chem. 2006, 78, 5061-5067
Capillary Zone Electrophoresis of Some Extremely Weak Bases in Acetonitrile Simo P. Porras*
Laboratory of Analytical Chemistry, Department of Chemistry, University of Helsinki, Helsinki, Finland
In water capillary zone electrophoresis cannot be used to investigate basic compounds, which are so weak that their pKa,HB+ values are less than zero. In acetonitrile the basic strength of such compounds is increased by many orders of magnitude. Accordingly, several extremely weak bases are protonated at low pH in acetonitrile, thus, allowing their investigation by CZE. In this work the CZE separation of thioacetamide, acetamide, thiourea, benzamide, and 4-nitrobenzamide as well as that of N-methylformamide, N,N-dimethylformamide, formamide, and dimethyl sulfoxide is demonstrated in acetonitrile using 10 mmol/L perchloric acid as an electrolyte. The effect of BGE additives, like water and acetic acid, on the CZE performance is discussed. The problem of finding a suitable electroosmotic flow marker at low pH in acetonitrile is addressed, and nitromethane is shown to be a proper marker compound under such extreme conditions. This work demonstrates how organic solvents can enlarge the field of application of CZE. All transport and dispersive phenomena decisive in capillary zone electrophoresis (CZE) are affected by the solvent of the background electrolyte (BGE).1,2 Thus, a CZE separation may be affected considerably when a common water-based BGE is replaced with an organic solvent one. Nevertheless, it is not selfevident if the effect of organic solvent(s) on the CZE separation is an advantage or not.3 It is clear that organic solvents are the matter of choice for the separation medium, e.g., when stability problems arise in aqueous solutions. In most other cases, however, potential advantages of organic solvents need to be evaluated case by case. There is no doubt, however, that under carefully selected experimental conditions some organic solvents allow capillary electrophoretic separations not possible (or at least very difficult) in aqueous medium. Usually, the prerequisite is that the selected solvent should possess physicochemical properties rather different from those of water. An organic solvent, which fulfills such a criteria, is acetonitrile (AN) (for solvent properties, see ref 4). Even though AN has been often used in capillary electrophoresis (CE), it is rarely applied without cosolvent(s); most often it is mixed, e.g., with water or methanol. One reason for it not being * Corresponding author. Fax: +358 191 50253. E-mail: simo.porras(at)iki.fi. (1) Porras, S. P.; Riekkola, M.-L.; Kenndler, E. Electrophoresis 2003, 24, 14851498. (2) Porras, S. P.; Kenndler, E. J. Chromatogr. A 2004, 1037, 455-465. (3) Porras, S. P.; Kenndler, E. Electrophoresis 2005, 26, 3203-3220. (4) Marcus, Y. The Properties of Solvents; Wiley: Chichester, 1998. 10.1021/ac060243v CCC: $33.50 Published on Web 05/25/2006
© 2006 American Chemical Society
used in pure form is the low solubility of many common BGE chemicals in it. The rare CE investigations in pure AN are often devoted to some specific interaction or phenomenon not readily present in aqueous medium: complex formation between neutral analytes and cations,5-7 complex formation of neutral analytes and anions via heteroconjugation,8-12 heteroconjugation of anionic analytes with neutral additives,13,14 and separation of radical anions unstable in aqueous medium,15-17 for example. Another advantage of AN is that an electrochemical potential window is larger in AN when compared to water.18 This can be utilized, e.g., when electrochemical detection is used in CE.19-22 Clearly, special interaction phenomena in AN can be a very interesting way to enlarge the applicability of CE. However, sometimes such phenomena may also hamper the practical CE work in this solvent. A good example in this respect is the pH adjustment in AN. As is well-known, a requirement for reasonable CE work is the adjustment of pH of the BGE in a defined way. Organic solvent-based BGEs do not make any difference in this respect. The most convenient way to derive the pH of organic solvent-based BGE is a theoretical calculation. This can be done, e.g., with the Henderson-Hasselbalch equation (preferably accompanied with an activity correction). Unfortunately, calculation of the correct pH in AN is not an easy task due to other interactions, which operate parallel to ionic dissociation (e.g., ionpair formation and conjugation via hydrogen bonding; see ap(5) Miller, J. L.; Khaledi, M. G.; Shea, D. Anal. Chem. 1997, 69, 1223-1229. (6) Miller, J. L.; Khaledi, M. G.; Shea, D. J. Microcolumn Sep. 1998, 10, 681685. (7) Li, S.; Weber, S. G. J. Am. Chem. Soc. 2000, 122, 3787-3788. (8) Okada, T. Chem. Commun. 1996, 1779-1780, 2495. (9) Okada, T. J. Chromatogr. A 1997, 771, 275-284. (10) Miller, J. L.; Shea, D.; Khaledi, M. G. J. Chromatogr. A 2000, 888, 251266. (11) Porras, S. P.; Kuldvee, R.; Palonen, S.; Riekkola, M.-L. J. Chromatogr. A 2003, 990, 35-44. (12) Kuldvee, R.; Vaher, M.; Koel, M.; Kaljurand, M. Electrophoresis 2003, 24, 1627-1634. (13) Esaka, Y.; Yoshimura, K.; Goto, M.; Kano, K. J. Chromatogr. A 1998, 822, 107-115. (14) Esaka, Y.; Inagaki, S.; Uchida, D.; Goto, M.; Kano, K. J. Chromatogr. A 2001, 905, 291-297. (15) Esaka, Y.; Okumura, N.; Uno, B.; Goto, M. Anal. Sci. 2001, 17, 99-102. (16) Esaka, Y.; Okumura, N.; Uno, B.; Goto, M. J. Chromatogr. A 2002, 979, 91-96. (17) Esaka, Y.; Okumura, N.; Uno, B.; Goto, M. Electrophoresis 2003, 24, 16351640. (18) Izutsu, K. Electrochemistry in Nonaqueous Solutions; Wiley-VCH: Weinheim, 2002. (19) Salimi-Moosavi, H.; Cassidy, R. M. Anal. Chem. 1995, 67, 1067-1073. (20) Matysik, F.-M. J. Chromatogr. A 1998, 802, 349-354. (21) Matysik, F.-M. Electrochim. Acta 1998, 43, 3475-3482. (22) Matysik, F.-M. Electroanalysis 1999, 11, 1017-1021.
Analytical Chemistry, Vol. 78, No. 14, July 15, 2006 5061
propriate textbooks18,23 for details). The problem of calculation of pH in AN with the Henderson-Hasselbalch equation was demonstrated in recent CZE investigation.24 It was shown that the mobility vs pH curves of some cationic drugs did not follow a well-known sigmoid shape even though the ionic strength of the BGE was kept constant at each pH (which is essential for a reliable interpretation of the data). The proposed reason for the reported deviations was homoconjugation of buffer acids and their conjugate bases, which hampered the pH calculated from the Henderson-Hasselbalch equation. Mobility vs pH behavior of cationic drugs in AN was investigated also in an earlier contribution, and pKa values of the drugs were derived from the data.25 Also, in this paper some data points were reported, which did not fit with the theoretical mobility vs pH curve (note that these data were ignored from further use). It was shown that heteroconjugation of the analytes did not play a significant role in mobility deviations, and ion-pair formation was suggested as a potential reason for the deviations. A more plausible explanation would be, however, that homoconjugation of the BGE components (dichloroacetic acid-dichloroacetate) ruined the calculation of pH with the Henderson-Hasselbalch equation, and this is the main reason for the observed deviations. This phenomenon has been discussed in detail by Subirats et al.24 and need not be repeated herein. Accordingly, the calculated pKa data presented in the given papers25,24 can be only considered as very approximate. As is evident, the calculation of correct pH in AN can be cumbersome for a CE operator. Nevertheless, acid-base chemistry in AN may offer some other advantages. The acid strength of analytes is known to vary substantially within solvents (for details, see refs 23 and 18). AN is an especially good example in this respect due to large shifts in pKa values when compared to water. As pointed out in previous work,3 it is usually not the shift in pKa, which is of importance in CZE, but the relative difference in acid strength. However, there is at least one case when the shift in pKa can be advantageous, namely when CZE of extremely weak bases is investigated. Here it should be recalled that it is normal practice to speak about pKa of a conjugated acid of a base (BH+; sometimes named as cation acid23) instead of pKb of a base (B).26,27 This practice is followed throughout the present work. There are some bases, which are too weak to be protonated under normal pH range in water (pKa,HB+ < 0). Thus, their zone electrophoretic separation is infeasible in water-based BGEs. On the other hand, in AN the basic strength of the bases increases. This means that their CZE separation becomes possible, i.e., these bases carry a positive charge within a practically reachable pH range. It is the aim of the present investigation to show that under properly selected experimental conditions the CZE separation of extremely weak bases is possible and to discuss the matters that need to be taken into account while working under such conditions. (23) Kolthoff, I. M.; Chantooni, M. K., Jr. In Treatise on Analytical Chemistry Part 1; Kolthoff, I. M., Elving, P. J., Eds.; Wiley: New York, 1979; Chapter 19. (24) Subirats, X.; Reinstadler, S.; Porras, S. P.; Raggi, M. A.; Kenndler, E. Electrophoresis 2005, 26, 3315-3324. (25) Porras, S. P.; Riekkola, M.-L.; Kenndler, E. Chromatographia 2001, 53, 290294. (26) Perrin, D. D. Dissociation Constants of Organic Bases in Aqueous Solution; Butterworths: London, 1965. (27) Perrin, D. D. Dissociation Constants of Organic Bases in Aqueous Solution: Supplement 1972; Butterworths: London, 1972.
5062
Analytical Chemistry, Vol. 78, No. 14, July 15, 2006
EXPERIMENTAL SECTION Reagents. Acetic acid (HAc), 4-nitrobenzamide, and nitromethane (NM) were from Aldrich (Steinheim, Germany), acetonitrile (AN; HPLC grade, far-UV) and dimethyl sulfoxide (DMSO) were from Lab-Scan (Dublin, Ireland), acetamide, crystal violet, 70-72% perchloric acid (HClO4), potassium hydrogen phthalate, and thioacetamide were from E. Merck (Darmstadt, Germany), and acetic anhydride, Coulomat AD Karl Fischer (KF) reagent, N,N-dimethylformamide (DMF), and formamide (FA) were from Riedel-de Hae¨n (Seelze, Germany). Other reagents were benzamide (Eastman Kodak Company, Rochester, NY), N-methylformamide (NMF; Fluka, Buchs, Switzerland), and thiourea (BDH, Poole, UK). All above-mentioned chemicals were used as received except potassium hydrogen phthalate, which was dried in an oven prior to use. Distilled water was further purified with a Milli-Q apparatus (Millipore, Molsheim, France). CZE and Related Parameters. A commercial capillary electrophoresis instrument equipped with a UV detector (HewlettPackard, Waldbronn, Germany) was used. Detection was carried out at 195 and 220 nm. Data were collected with ChemStation software (Agilent Technologies, Waldbronn, Germany), fitted with laboratory written program epeaks (for details, see ref 28), and depicted with Origin (OriginLab, Northampton, MA). Untreated fused-silica capillaries (i.d./o.d. 50/375 µm; total length 78.5 cm, effective length 70.0 or 8.5 cm) were purchased from Composite Metal Services (Ilkley, U.K.). Sample introduction was performed hydrodynamically ((50 mbar for 1 s). A sample was typically injected at both ends of the capillary using the following sequence: (i) 50 mbar for 2 s (normal-end) and (ii) -50 mbar for 1 s (short-end). The capillary cassette was thermostated at 25.0 °C with forced air-cooling. Note that the whole capillary length was not under temperature control, but a significant part (∼23%) was outside thermostation. The separation voltage was 15-30 kV (either positive or negative polarity), and the maximum resulting current was ∼22 µA (typically around 12 µA). Before experiments the capillary was rinsed with 0.1 mol/L NaOH (aqueous), water, and the BGE, respectively. This procedure guaranteed sufficient reproducibility of the electroosmotic flow (EOF). Procedures. Stock solutions of analytes were prepared in AN at a concentration of 100 mmol/L each. They were diluted with AN to the final concentration (0.2-10 mmol/L). One mmol/L NM in AN served as an electroosmotic flow marker. Note that AN is a hygroscopic solvent. The water uptake from the laboratory air depends on the humidity, which may vary from day to day. Thus, special care was taken to minimize the time in which solvent and BGE vessels were open. However, the water uptake could not be fully avoided, and, thus, the measured water content of the solvent and other solutions prepared in AN is given. Perchloric acid BGEs in AN were diluted daily from 2.0 mol/L HClO4 stock solution. The stock solution was prepared in HAc as follows. First, commercial 70-72% HClO4 was considered to be 70% of HClO4, and the rest (30%) was assumed to be water. The required amount of HClO4 was added to a small volume of acetic acid. Then, it was calculated how much water (in mol/L) was introduced in HAc by addition of HClO4, and exactly the same molarity of acetic anhydride was added. Acetic anhydride tends (28) Palonen, S. Thesis, Helsinki University Printing House; Helsinki, 2005. http://ethesis.helsinki.fi/
to heat up the solution, and, thus, it was added carefully one drop at a time keeping the HClO4 solution in an ice-bath. After addition of anhydride, the flask was filled to the mark by HAc. Concentration of the 2.0 mol/L HClO4 solution was confirmed by titration with potassium hydrogen phthalate solution (prepared in HAc) where crystal violet served as an indicator (one drop of 1 g/100 mL crystal violet in HAc). The result showed that the commercial perchloric acid was indeed 70%. Note that perchloric acid is a hazardous chemical and should be handled with extreme care. HClO4 solutions are to be prepared under a hood using appropriate protections. Mixed AN-water BGEs were prepared from the 2.0 mol/L stock solution with an appropriate amount of water added. Fresh BGE was used in every CZE run throughout the work. The water content of AN solutions (those without added water) was measured by a KF coulometer (Metrohm, Herisau, Switzerland), and the result was given in % w/w. For AN the water content was ∼0.01-0.025%, for freshly prepared BGEs it was
