Anal. Chem. 1998, 70, 3585-3589
Capillary Electrophoretic Determination of Different Classes of Organic Ions by Potentiometric Detection with Coated-Wire Ion-Selective Electrodes Peter Schnierle,† Thomas Kappes, and Peter C. Hauser*
Department of Chemistry, The University of Basel, Spitalstrasse 51, 4056 Basel, Switzerland
Singly charged amines and sulfonic acids as cationic and anionic aliphatic model substances, respectively, were detected in capillary electrophoresis with all-solid-state ion-selective electrodes. The sensitivity for these compounds is a function of their lipophilicity. The signal detected is generally greater for molecules with a larger organic part. The utility of the method is further demonstrated by the detection of a group of aromatic compounds in the form of the anionic analgesics (S)-(+)-2-(4-isobutylphenyl)propionic acid, 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid, and o-acetylsalicylic acid. Further determined were the artificial sweeteners cyclamate, acesulfame-K, and saccharin. Detection limits for the different substances were between 2.5 × 10-6 and 1 × 10-5 M. Capillary electrophoresis (CE) has been developed in recent years for the analysis of inorganic and organic ions. Detection in this separation method is usually carried out via optical absorption or fluorescence methods. Both approaches are not ideal, however. Absorbance measurements suffer from the short optical path lengths available and the fact that not all analyte species absorb UV light. Many methods, therefore, rely on indirect absorbance measurements which work less well than direct absorbance measurements. Fluorescence is a more sensitive detection method, but intrinsically fluorescent analytes are even less common and derivatization usually has to be carried out in order to enable detection by fluorescence. Electrochemical detection means have also been reported but have not been as widely used as the spectroscopic methods. Altough very low detection limits can be achieved with amperometry, its application remains limited to electroactive analytes. Conductivity detection, in contrast, is a universal method, but, as in ion chromatography, its performance may be limited by the presence of a large background signal. An interesting alternative is potentiometric detection. Analyte species do not have to be electroactive, as, in principle, potentiometric ion-selective microelectrodes can respond to all ions with a charge of the correct sign. Additionally, potentiometry may offer wider dynamic ranges than spectrophotometry because of the logarithmic response of ion-selective electrodes. Potentiometric detection † Current address: Neuropharmacology Section, Neurocentre, The University of Freiburg, Breisacherstr. 64, 79106 Freiburg, Germany.
S0003-2700(98)00117-6 CCC: $15.00 Published on Web 07/30/1998
© 1998 American Chemical Society
is also fundamentally the simplest method of all, as the response is obtained directly in the electronic domain and no excitation signal has to be applied. The first potentiometric detectors used in CE were miniaturized liquid membrane ion-selective electrodes of a construction that is applied in physiological studies of individual cells. These electrodes consist of drawn out glass pipets, containing an internal reference electrode and electrolyte solution, with a tip of 1-3-µm diameter which is filled with a liquid organic membrane solution. The detection of inorganic cations1-3 and monovalent anions4-6 in CE with such electrodes has been reported. However, handling of these fragile probes is relatively difficult. Coated-wire electrodes are more robust because they consist of a PVC membrane in direct contact with a metallic conductor, and no internal reference is needed.7 This electrode type has a much higher mechanical resistance and a longer lifetime than the micropipet electrodes but gives comparable detection results in CE.8,9 Coated-wire electrodes have been applied in CE to determine aliphatic carboxylates,8 alkali and alkaline earth cations,9 and monovalent inorganic anions9. When using ion-selective electrodes as detectors in separation methods such as CE, their selectivity is an important criterion. Ideally, the sensor reponds equally well to all cations or anions, with the exception of any species employed in the running buffer. This is in contrast to the normal use of ion-selective electrodes, where a high selectivity is required for the single ion to be determined. Ion-selective electrodes with polymeric membranes usually contain an electrically neutral ionophore which possesses high selectivity for the analyte ion and lower but not negligible response to all other species. To use such an ISE for the detection of inorganic cations in capillary electrophoresis, an ionophore which displays a poor selectivity pattern in conventional terms (1) Haber, C.; Silvestri, I.; Ro¨o¨sli, S.; Simon, W. Chimia 1991, 45, 117-121. (2) Nann, A.; Simon, W. J. Chromatogr. 1993, 633, 207-211. (3) Nann, A.; Silvestri, I.; Simon, W. Anal. Chem. 1993, 65, 1662-1667. (4) Nann, A.; Pretsch, E. J. Chromatogr. A 1994, 676, 437-442. (5) Hauser, P. C.; Renner, N. D.; Hong, A. P. C. Anal. Chim. Acta 1994, 295, 181-186. (6) Hauser, P. C.; Hong, A. P. C.; Renner, N. D. J. Cap. Electrophor. 1995, 2, 209-212. (7) James, H.; Carmack, G.; Freiser, H. Anal. Chem. 1972, 44, 856-857. (8) De Backer, B. L.; Nagels, L. J. Anal. Chem. 1996, 68, 4441-4445. (9) Kappes, T.; Schnierle, P.; Hauser, P. C. Anal. Chim. Acta 1997, 350, 141147.
Analytical Chemistry, Vol. 70, No. 17, September 1, 1998 3585
has to be chosen. Another possibility is the use of membranes which contain lipophilic charged ion-exchanger sites. Such electrodes usually display no specific interaction with the ion, and the following selectivity series are obtained for anions, R- > ClO4> SCN- > I- > NO3- > Br- > Cl- > HCO3- > F- > AcO- > X2-, and cations, R+ > Cs+ > Rb+ > K+ > Na+ > Li+ > Ca2+ > Mg2+. R- and R+ denote singly charged organic anions and cations. For the anions, the sequence is known as the Hofmeister series, and such electrodes can, for example, be used for the determination of nitrate. The results are in good agreement with the free enthalpy of hydration of these analyte ions and thus with the lipophilicity of these substances. Therefore, the selectivity of an ion-selective electrode is higher the more lipophilic the ion is. For organic ions, it has been shown that the number of carbon atoms, degree of nitrogen substitution, branching of the hydrocarbon chains, and the number and type of hydrophilic substituents of an ion affect the selectivity displayed by such ion-selective electrodes.10 With macroscale ion-selective electrodes for N-based substances, an improved selectivity with the extent of nitrogen substitution was found, quaternary amines giving the highest response.11 This means that, also for organic ions, the response of such an electrode is higher the more lipophilic the organic ion is. Ion-selective electrodes are used widely, mainly for the determination of certain inorganic ions. However, many reports have also been published on ion-selective electrodes for pharmaceutical analysis. Alkaloids,12 antibiotics,13 and vitamins14 are some of the substances that have been determined. For the analysis of organic ions in CE, the application of ion-selective electrodes based on charged ion-exchangers as potentiometric detectors is also very promising because of the high sensitivity the electrodes show for organic ions compared to inorganic ones. This should be of interest because the main applications of CE are in the determination of organic ions. However, with the exception of dopamine, histamine, imidazole,2,3 and aliphatic carboxylic acids,8 potentiometric detection in CE so far has been limited to inorganic ions. In the present work, the use of all-solid-state coated-wire ionselective electrodes as potentiometric detectors in the CE determination of different monovalent organic ions is described. PVCbased membranes were used which contained ion-exchangers that do not interact specifically with the analytes. First, three different alkylamines and three different sulfonic acids were investigated. These can be regarded as model substances for proving the properties of ion-selective electrodes when used as potentiometric CE detectors. The applicability and usefulness of this detection method for the analysis of organic ions in real samples was then further investigated for different organic analytes. It was found that potentiometric detection can be used for the determination of the analgesics (S)-(+)-2-(4-isobutylphenyl)propionic acid, 2-[(2,6dichlorophenyl)amino]benzeneacetic acid, and o-acetylsalicylic acid in CE. Also determinated were the three nonsugar sweeteners cyclamate, acesulfame-K, and saccharin, and a popular low(10) Cunningham, L.; Freiser, H. Anal. Chim. Acta 1981, 132, 43-50. (11) Martin, C. R.; Freiser, H. Anal. Chem. 1980, 52, 562-564. (12) Eppelsheim, C.; Aubeck, R.; Hampp, N.; Brauchle, C. Analyst 1991, 116, 1001-1003. (13) Yao, S. Z.; Shiao, J.; Nie, L. H. Talanta 1989, 36, 1249-1252. (14) Zhang, Z. H.; Imato, T.; Asano, Y.; Sonoda, T.; Kobayashi, H.; Ishibashi, N. Anal. Chem. 1990, 62, 1644-1648.
3586 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
calorie soft drink was analyzed. To our knowledge, no CE method for the determination of these sweeteners has been described previously. EXPERIMENTAL SECTION Apparatus. The CE system consisted of a high-voltage power supply (Spellman CZE 1000R/220, Spellman, Hauppauge, NY) and an uncoated fused silica capillary (Scientific Glass Engineering, Melbourne, Australia) of 25-µm i.d. and 98-cm length. The injection end of the capillary was dipped into a container with running electrolyte together with a platinum electrode connected to the high-voltage output. For safety reasons, the whole injection unit was placed in a Perspex box fitted with a switch that interrupted the power supply when the box was opened. The detection unit was located under a microscope in a shallow Perspex vessel. The capillary end was immersed at a low angle, and a platinum wire was placed as counter electrode at a distance of approximately 500 µm. This electrode also served as reference electrode. The detector electrode was placed 25 µm from the capillary end. This was facilitated by the use of a micromanipulator fixed to the microscope stage and could be reproduced to better than 5 µm. More details regarding the alignment can be found in ref 9. An electrometer operational amplifier (OPA12lKP from Burr-Brown, Tucson, AZ) in the voltage follower configuration served to lower the impedance of the measured voltage. This amplifier was mounted in close proximity and shielded in a grounded metal box. The microscope, electrode assembly, and amplifier were contained in a Faraday cage. The electropherograms were recorded on a MacLab/4s data acquisition system (ADInstruments Pty Ltd., Castle Hill, Australia) and an Apple Power Macintosh personal computer. For the anions, the signal polarity was inverted in order to obtain positive-going peaks. Coated-Wire Ion-Selective Electrodes. For the cation- and anion-selective electrodes, only ion-exchangers with no specific interactions with the analytes were used. The mixtures for the PVC-based sensor membranes consisted, for the cation-selective electrode, of 3.4% potassium tetrakis(4-chlorophenyl)borate, 64.4% o-nitrophenyl octyl ether (avoid skin and eye contact), and 32.2% PVC and, for the anion detector, of 6.2% tridodecylmethylammonium chloride, 62.5% o-nitrophenyl octyl ether, and 31.3% PVC. These components were dissolved in tetrahydrofuran. All membrane components were purchased from Fluka (Buchs, Switzerland). The electrodes themselves were constructed by inserting a 90-µm platinum wire (California Fine Wire Co., Grover Beach, CA) into a piece of fused silica capillary (Scientific Glass Engineering, Melbourne, Australia) of 100-µm i.d., 300-µm o.d., and 5-cm length. At the rear, the wire protruded ∼1 cm for making the electrical connection and was sealed to the capillary with a drop of epoxy glue. At the front, the wire was cut off flush with the capillary end and polished with a 5-µm-grid polishing sheet in order to obtain a clean and even disk. The surface was coated with the membrane phase by dipping five times in the membrane solution until a suitable layer was formed. In this way, disk-shaped membrane surfaces with 300-µm overall diameter were obtained. Reagents and Methods. Running electrolytes were prepared from the analytical grade reagents tartaric acid and sodium tetraborate decahydrate, purchased from Merck (Darmstadt, Germany). Ammonium chloride, hydrochlorides of methylamine, dimethylamine, and trimethylamine, and sodium salts of 1-pro-
panesulfonic acid, 1-heptanesulfonic acid, 1-undecanesulfonic acid, o-acetylsalicylic acid, and (S)-(+)-2-(4-isobutylphenyl)propionic acid were obtained from Fluka (Buchs, Switzerland). 2-[(2,6Dichlorophenyl)amino]benzeneacetic acid was from Sigma (Buchs, Switzerland). Acesulfame-K was from Riedel-de Hae¨n (Seelze, Germany), cyclamic acid was purchased from Sigma, and the sodium salt of saccharin was from Fluka. All substances were of the highest purity available. Voltaren (Ciba-Geigy, Basel, Switzerland) for 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid, ibuprofen (Heumann-Pharma, Nu¨rnberg, Germany) for (S)-(+)-2-(4isobutylphenyl)propionic acid, and aspirin (Bayer, Leverkusen, Germany) for o-acetylsalicylic acid were used as pharmaceutical preparations of the analgesics and purchased from pharmacy stores. Popular soft drinks were obtained from local food stores. Sample solutions for the drinks were made by diluting 5 mL of soft drink with 3 mL of 1.5 mM tetraborate buffer. Water was purified by a Milli-Q (Millipore, Bedford, MA) purifying system. For cation determinations, 2.5 mM tartaric acid and, for anion analysis, 0.5 mM (pharmaceuticals) and 1.5 mM (all other anions) tetraborate buffer were used without any pH adjustment. Tartaric acid and borate were chosen because of their very low interactions with the ion-exchangers used in the PVC membranes of the ionselective electrodes. Stock solutions of the analytes with different concentrations were prepared by dissolving known amounts of the pure analytes in running electrolyte. Analyte sample solutions or analyte sample solution mixtures were prepared by dilution of the stock solutions. Sample injection was carried out electrokinetically at + 5.0 kV for 7 s for all determinations. The sample solution and running electrolyte always had the same concentration and approximately same ionic strength to avoid the electrostacking effect. The electrodes were conditioned with running electrolyte under separation voltage until a stable signal was obtained (about 1 h). All solutions were filtered through 0.2-µm Nylon filters. The capillary was regenerated periodically with 0.5 M NaOH. RESULTS AND DISCUSSION Amines. The electropherogram for a mixture of ammonium ions, methylamine, dimethylamine, and trimethylamine is shown in Figure 1. Under the conditions employed, they were all in the protonated, positively charged form and were detected with a cation-selective electrode. The analytes are separated baselineresolved. First, the ammonium ions escaped from the capillary, followed by methylamine, dimethylamine, and trimethylamine. This sequence is expected, as the mobility of the ions is a function of their size. An interesting feature of the electropherogram is the variation in relative peak heights for the analytes. These differed although the concentrations (2.5 × 10-4 M) were identical. The amine with the largest organic substituent content (the highest number of methyl groups) displayed the highest response of the electrode, followed by amines with fewer organic substituents (methyl groups). Therefore, the selectivity of the ionselective electrode increases with higher organic substituent contents of the amines, which is in agreement with the pattern observed with macroscale electrodes. Sulfonic Acids. The anions of the three sulfonic acids 1-propanesulfonic acid, 1-heptanesulfonic acid, and 1-undecanesulfonic acid displayed the same behavior as the amines, as evidenced in the electropherogram of Figure 2. 1-Undecane-
Figure 1. Electropherogram of a mixture of ammonium (1), methylamine (2), dimethylamine (3), and trimethylamine (4) using a coated-wire ISE based on potassium tetrakis(4-chlorophenyl)borate. Concentrations: 2.5 × 10-4 M each in 2.5 mM tartaric acid. Sample injection: electrokinetically, 7 s at +5.0 kV. Separation potential: +30 kV. Capillary: 25-µm i.d., 98-cm length, fused silica, uncoated. Electrolyte: 2.5 mM tartaric acid.
Figure 2. Electropherogram of a mixture of 1-undecanesulfonic acid (1), 1-heptanesulfonic acid (2), and 1-propanesulfonic acid (3) using a coated-wire ISE based on tridodecylmethylammonium chloride. Concentrations: 2.5 × 10-4 M each in 1.5 mM tetraborate buffer. Sample injection: electrokinetically, 7 s at +5.0 kV. Separation potential: +30 kV. Capillary: 25-µm i.d., 98-cm length, fused silica, uncoated. Electrolyte: 1.5 mM tetraborate buffer.
sulfonic acid displayed the highest peak, followed by 1-heptanesulfonic acid and 1-propanesulfonic acid, with a peak height of approximately one-third of that for 1-undecanesulfonic acid for identical concentrations of again 2.5 × 10-4 M. The selectivity thus follows the length of the organic residue. Again, this pattern is expected from the experience with normal macroelectrodes. Note that the separation of the sulfonic acid anions was carried out with the detection end at cathodic polarity. This implies that the electroosmotic flow in the capillary is stronger than the electrophoretic migration of the anions in the opposite direction. With this method, the analyte with the lowest electrophoretic mobility elutes first from the capillary and the fastest-migrating analyte last. Therefore, the sulfonic acid with the largest organic substituent and highest response with the electrode is detected first. Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
3587
Figure 3. Electropherograms (A) of 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid (1) (6.3 × 10-4 M), (B) of (S)-(+)-2-(4-isobutylphenyl)propionic acid (2) (9.7 M × 10-4 M), (C) of o-acetylsalicylic acid (3) (1.1 × 10-3 M) and impurity (4), and (D) of (1), (3), and (4) in 0.5 mM tetraborate buffer using a coated-wire ISE based on tridodecylmethylammonium chloride. Sample injection: electrokinetically, 7 s at +5.0 kV. Separation potential: +30 kV. Capillary: 25µm i.d., 98-cm length, fused silica, uncoated. Electrolyte: 0.5 mM tetraborate buffer.
The results for the two groups of model substances showed that it is, indeed, possible to detect organic cations and anions with ion-selective electrodes based on charged ion-exchangers as ionophores. The fact that larger peaks were obtained for the organic amines compared to those for the ammonium ion confirms that the selectivity of such electrodes is, indeed, favorable for the detection of organic ions. The method was then tested for two further groups of organic analytes. Analgesics. The three analgesics 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid, (S)-(+)-2-(4-isobutylphenyl)propionic acid, and o-acetylsalicylic acid were detected. The electropherograms for these compounds are given in Figure 3. It is evident from the first three traces of Figure 3 that all three compounds can be detected. This demonstrates that not only the detection of aliphatics but also the detection of aromatic compounds is possible. The electropherogram for o-acetylsalicylic acid showed a second peak due to an impurity which may be salicylic acid. The analysis times are close, and it is not possible to separate a mixure of all three compounds. However, a mixture of 2-[(2,6dichlorophenyl)amino]benzeneacetic acid and o-acetylsalicylic acid could be resolved, as shown by the last trace of the figure, although complete baseline resolution was not achieved under the conditions employed. The peaks for these substances are broader than the ones obtained for the simple aliphatic molecules. This may be an artifact of the detection method, but it is not 3588 Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
Figure 4. Electropherograms (A) of cyclamate (1) (1.1 × 10-3 M), (B) of acesulfame-K (2) (9.9 × 10-4 M), (C) of saccharin (3) (1.1 × 10-3 M), (D) of soft drink I, and (E) of soft drink II ((4) ) benzoic acid) using a coated-wire ISE based on tridodecylmethylammonium chloride. Sample injection: electrokinetically, 7 s at +5.0 kV. Separation potential: +30 kV. Capillary: 25-µm i.d., 98-cm length, fused silica, uncoated. Electrolyte: 1.5 mM tetraborate buffer.
possible to clearly discern detector response from band broadening. Pharmaceutical preparations of the three compounds were also analyzed in the form of voltaren tablets (for 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid), ibuprofen tablets (for (S)-(+)-2(4-isobutylphenyl)propionic acid), and aspirin tablets (o-acetylsalicylic acid). These were dissolved and diluted in the running electrolyte. The electropherograms obtained were virtually identical to the ones given in Figure 3. Therefore, no other major anions that interacted with the ion-selective electrode seemed to be contained in the pharmaceutical preparations. This method may be applicable in production control, clinical analysis, or biomedical research. Sweeteners. It was also possible to detect and separate the three commonly used sweeteners cyclamate, acesulfame-K, and saccharin as anions, as evidenced by the electropherograms of Figure 4. The first three traces of the figure (A-C) were obtained for solutions of the pure substances. Electropherograms D and E were obtained for an identical brand of a popular low-calorie soft drink. The difference is the origin: one bottle was purchased in Switzerland, the other in France. By comparison with the standards, it is found that the first sample contains saccharin and cyclamate, whereas the second sample contains only cyclamate. This is in agreement with the declarations by the beverage
producer on the label of the soft drink bottles. Also detected were other, smaller peaks, one of which could be identified as being the preservative benzoic acid (peak 4) contained in one of the samples. We did not perform quantitative determination of the two sweeteners in the soft drink because no information was available from the producer about the amount of sweeteners used. But, in principle, quantification of the analytes should be possible. Until now, at least to our knowledge, no CE method is available for the analysis of these substances. For the determination of single sweeteners, HPLC methods with UV and fluorescent detection or ion chromatographic methods have been published. But these methods are often complicated, and cyclamate in particular requires sample pretreatment (oxidation) and derivatization.15-17 Calibration Curves. Quantitation by using this method is, indeed, possible, as indicated by the calibration curves of Figure 5. One compound was chosen as a representative for each of the four groups of species investigated. As can be seen from the figure, very similar working curves were obtained. The detection limits are good. The values based on a signal-to-noise ratio of 3 were 2.5 × 10-6 M for dimethylamine, 4.1 × 10-6 M for 2-[(2,6dichlorophenyl)amino]benzeneacetic acid, 1.0 × 10-5 M for heptanesulfonic acid, and 1.1 × 10-5 M for cyclamate. Note that these values were obtained for electrokinetic injection under conditions which do not induce electrostacking. The limits of detection are comparable with or better than those obtained with potentiometric detection for inorganic ions.9 Dynamic ranges up to 10-3 M were found possible; higher concentrations were problematic due to increased peak broadening. An interesting feature of Figure 5 is the fact that the calibration curves for the three anions are very similar, while the single-cation response (dimethylamine) is distinct. It is thought that this probably is a feature of the electrodes (perhaps related to the selectivity to the background electrolyte) and is not due to the individual analytes. CONCLUSIONS It was found that different classes of organic ions could be readily detected in capillary electrophoresis with microscale (15) Casals, I.; Reixach, M.; Amat, J.; Fuentes, M.; Serra-Majem, L. J. Chromatogr. A 1996, 750, 397-402. (16) Ruter, J.; Raczek, D. I. Z. Lebensm. Unters. Forsch. 1992, 194, 520-523. (17) Biemer, T. A. J. Chromatogr. 1989, 463, 463-468.
Figure 5. Calibration curves: O, dimethylamine; ×, 2-[(2,6-dichlorophenyl)amino]benzeneacetic acid; 0, 1-heptanesulfonic acid; +, cyclamate. Sample injection: electrokinetically, 7 s at +5.0 kV. Solutions were prepared in the running buffer. Separation potential: +30 kV. Capillary: 25-µm i.d., 98-cm length, fused silica, uncoated. Electrolyte: tetraborate buffer or tartaric acid.
coated-wire ion-selective electrodes. Their performance in terms of reproducibility, stability, and lifetime was found to be similar to that in the detection of inorganic ions.9 Detection limits were generally found to be comparable with or better than those for optical absorption methods. However, this method is inherently much simpler than optical methods of detection and may be applied to species which do not show UV absorption. It is, therefore, considered a promising and versatile alternative to established detection techniques. ACKNOWLEDGMENT The authors thank the Swiss National Science Foundation for providing financial support (Grant No. 21-45282.95). Received for review February 2, 1998. Accepted June 30, 1998. AC980117U
Analytical Chemistry, Vol. 70, No. 17, September 1, 1998
3589
