Anal. Chem. 2001, 73, 2536-2540
Development of Rotating Electrochemical Detectors for Capillary Electrophoresis Abdelkader Hilmi and John H. T. Luong*
Biotechnology Research Institute, National Research Council Canada, Montreal, Quebec, Canada H4P 2R2
A rotating disk electrode (RDE) has been evaluated and optimized for the detection of electroactive species separated by capillary electrophoresis (CE). With catechol as a working model, the limit of detection was estimated to be 0.3 µM, i.e., ∼2.5-fold better than that of the stationary disk electrode (0.7 µM). Separation efficiency was significantly improved as exemplified by an increase of theoretical plates from 26 000 plates/m at 0 rpm to 67 000 plates/m at 500 rpm. Of particular importance was the capability of RDE to alleviate electrode passivation and electrical interference associated with high separation potential fields. Therefore, rotation amperometry was especially useful for analytes such as phenolic compounds that tended to rapidly foul the electrode surface. The RDE/ CE system was capable of separation and determination of pentachlorophenol in contaminated soils, and the result obtained agreed well with conventional liquid chromatography, an EPA recommended procedure. Electrochemical detection (EC) and capillary electrophoresis (CE), known as CEEC, have been combined as an extremely sensitive and powerful technique for analysis of many important electroactive species in the low-nanomolar-to-subnanomolar range in biological and environmental media.1-5 Indirect electrochemical detection, involving use of an electroactive substance as a major constituent of the separation buffer,6 also shows great promise for analysis of electrochemically inactive species without prior sample derivatization or electrode modification. Numerous studies have demonstrated several advantages of EC such as low sample volume, simplicity, miniaturization, and cost-effectiveness.7-10 EC also offers a great deal of flexibility since this detection technique is tunably selective, through proper selection of the applied potential. To date, constant-voltage amperometry with stationary electrodes has been widely used for detection of many biological * Corresponding author: (phone) 514-496-6175; (fax) 514-496-6265; (e-mail)
[email protected]. (1) Olefirowicz, T. M.; Ewing, A. G. Anal. Chem. 1990, 62, 1872-1876. (2) Lu, W.; Cassidy, R. M. Anal. Chem. 1993, 65, 2878-2881. (3) Huang, X.; Zare, R. N.; Sloss, S.; Ewing, A. G. Anal. Chem. 1991, 63, 189192. (4) Hilmi, A.; Luong, J. H. T.; Nguyen, A. L. Anal. Chem. 1999, 71, 873-878. (5) Sloss, S.; Ewing, A. G. Anal. Chem. 1993, 65, 577-581. (6) Olefirowicz, T.; Ewing, A. G. J. Chromatogr. 1990, 499, 713-719. (7) Zhong, M.; Lunte, S. M. Anal. Chem. 1999, 71, 251-255. (8) Zhong, M.; Zhou, J.; Lunte, S. M.; Zhao, G.; Giolando, D. M.; Kirchhoff, J. R. Anal. Chem. 1996, 68, 203-207. (9) Pentoney, S. L.; Huang, D. S.; Burgi, D. S.; Zare, R. N. Anal. Chem. 1988, 242, 2625-2629. (10) Voegel, P. D.; Baldwin, R. P. Electrophoresis 1997, 18, 2267-2278.
2536 Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
and environmental molecules separated by capillary electrophoresis or CE chips. The end-column amperometric detection scheme has increased in popularity since the electrochemical sensor can be simply positioned close to the outlet of the separation capillary without any conductive junction.11-14 In this arrangement, most of the separation potential field will drop at the end of the capillary before reaching the detecting electrode. Significant advances have also been made in the design of end-column detection in addition to the fostered use of small capillaries (5-25 µm, i.d.) to facilitate alignment as well as to circumvent electrical interference associated with separation potential fields.15,16 In principle, forced convection electrodes, and particularly the rotating disk electrode (RDE) and vibrating electrode, should outperform the stationary electrode because the controlled hydrodynamics of RDE provides reproducible and experimentally controlled transport of substrate to, and product from, the electrode surface.17,18 RDE is rather simple to construct and consists of a disk of the electrode material imbedded in a rod of an insulating material. RDEs are also commercially available with platinum, gold, or glassy carbon as the electrode material and their hydrodynamic/convective-diffusion equations have been well defined.19 In this paper, the applicability of end-column rotating electrochemical detection in capillary electrophoresis has been demonstrated to minimize electrode passivation as well as to improve detection sensitivity and separation efficiency. The characterization, optimization, and advantages of RDE are reported with catechol and pentachlorophenol (PCP) as model systems. To our knowledge, rotating amperometry has not been coupled with CE as an attractive technique for routine analysis of electroactive species although CEEC has attracted increasing attention over the last two decades. (11) Hilmi, A.; Luong, J. H. T.; Nguyen, A. L. J. Chromatogr., A 1997, 761, 259268. (12) Fermier, A. M.; Gostkowski, M. L.; Colon, L. A. Anal. Chem. 1996, 68, 1661-1664. (13) Salimi-Moosavi, H.; Cassidy, R. M. Anal. Chem. 1995, 67, 1067-1073. (14) Wallenborg, S. R.; Nyholm, L.; Lunte, C. E. Anal. Chem. 1999, 71, 544549. (15) Hu, S.; Wang, Z. L.; Li, P. B.; Cheng, J. K. Anal. Chem. 1997, 69, 264-267. (16) Lu, W.; Cassidy, R. M.; Baranski, A. S. J. Chromatogr. 1993, 640, 433440. (17) Von Karman, T. Angew Math. Methods 1921, 1, 233-252. (18) Sedahmed, G. H.; El-Abd, M. Z.; Zatout, A. A.; El-Taweel, Y. A.; Zaki, M. M. J. Electrochem. Soc. 1994, 141, 437-440. (19) Cochran, W. G. Proc. Cambridge Philos. Soc. Math. Phys. Sci. 1934, 30, 365-375. 10.1021/ac001192j CCC: $20.00 Published 2001 Am. Chem. Soc. Published on Web 03/09/2001
Figure 1. Amperometric detection cell equipped with a rotating disk electrode (RDE).
EXPERIMENTAL SECTION Materials. Except for 6-amino-β-cyclodextrin (6-ACD), a product of Advanced Separation Technologies (Whippany, NJ), all chemical products used in this study were purchased from Aldrich (Milwaukee, WI). Metal wires (platinum and silver) were purchased from Alfa Aesar (Ward Hill, MA), whereas electrodes used for cyclic voltammetry studies were purchased from Bioanalytical Systems (BAS, West Lafayette, IN). The contaminated soil was obtained from Resource Technology Corporation (Laramie, WY). Sample Preparations. Stock solutions of PCP, phenol, and 2,3-dimethylphenol (2,3-DMPh) were prepared in methanol (HPLC grade), whereas catechol was prepared in 0.1 M perchloric acid solution. Prior to CE analysis, all samples were diluted in the buffer and passed through a 0.45-µm filter (Gelman Sciences, Montreal, QC, Canada). To avoid any decomposition, all stock solutions were protected from light and kept at 4 °C until needed. Soil Extraction. A soil sample (0.1 g) was placed in a 15-mL vial that contained 5 mL of an extracting medium (buffer/methanol containing 25 mM 6-ACD). Magnetic agitation was effected for the desired time, and then the slurry was centrifuged and filtered with a 0.45-µm Gelman filter to obtain the sample for electrophoresis. End-Column Rotating Amperometric Detection and Capillary Electrophoresis Arrangement. Polyimide-coated fusedsilica capillaries of 20-µm internal diameter (i.d.), 42 cm long for catechol and 60 cm for phenolic compounds, were purchased from Polymicro Technologies (Phoenix, AZ). The separation potential was effected by a high-voltage power supply with an adjustable voltage range between 0 and 50 kV (model EH, Glassman, Whitehouse Station, NJ). The end of the capillary was polished by an abrasive paper aided by an alumina slurry (CF-1050, BAS) leading to a flat and smooth surface, and the resulting capillary was washed extensively with deionized water and sonicated for 10 min to remove any alumina particles that could adhere inside the capillary wall. The capillary was firmly inserted into the cathodic reservoir through a septum opening at the bottom of the reservoir, with the capillary outlet evenly positioned to the flat surface of the septum (Figure 1). Two other septa were installed on the side of the reservoir to allow insertion of the two electrodes, whereas a platinum wire was inserted into this reservoir through a septum to serve as a cathode for electrophoresis. The Ag/AgCl reference electrode was prepared by electrochemical oxidation of a silver wire (0.5-mm diameter) in 0.1 M HCl, while a platinum wire (0.5-mm diameter) served as the counter electrode. A rotating disk electrode system RDE-1 from BAS equipped with a glassy carbon RDE (MF-2066) was used as the detecting
electrode, which was centered directly at the capillary outlet. A microscopic inspection showed that the space between the rotating electrode and the outlet of the capillary was ∼60 µm. With such a distance and as the capillary outlet was firmly inserted into a rigid septum with the outlet evenly positioned to the surface of the septum, the vibration of the capillary was effectively eliminated. Due to the large diameter of the detecting electrode (3 mm) compared to the internal diameter of the capillary (20 µm), good electrochemical efficiency was achieved without precise microalignment. During electrophoresis, a CV-1B voltammograph (BAS) was used to apply +700 mV to the detecting electrode. Cyclic voltammetry experiments used a potentiostat/galvanostat (model 263 A, EG&G, Princeton Applied Research, Princeton, NJ) to operate the three-electrode amperometric detection system. These experiments allowed us to characterize each compound and select the detecting potential. The time response data were digitized and treated by an A/D board (DP 500-AD) supplied with an interface box (Labtronics, Guelph, ON, Canada) that was installed on a PC 486 computer. The data were stored in ASCII files and converted to PRN files for treatment by a graphic program. No software filtration of the signal was used. Electrophoretic Procedure. All analyses were effected by a system composed of a high-voltage power supply with an adjustable voltage range between 0 and 50 kV and a rotating electrochemical detector (Figure 1). Prior to use, the capillary was flushed with methanol for 3 min to remove any organic contamination and rinsed with deionized water for another 3 min. The capillary was then filled with 0.1 M NaOH for 30 min, followed by a rinsing sequence with 0.1 M HClO4 and deionized water for 5 and 3 min, respectively. Depending on the target analyte, the capillary was filled with one of the following running buffers: (i) 20 mM 4-morpholineethanesulfonic acid (MES), pH 5.6; (ii) 10 mM borate and 6 mM phosphate, pH 7.5 containing 10 mM sodium dodecyl sulfate (SDS); or (iii) a mixture of 25 mM phosphate and 25 mM borate, pH 8.7. Initially, the capillary was preconditioned by applying low potential (+5 kV) for 20 min to the anodic reservoir with the cathodic reservoir grounded. An initial stable baseline was then established by applying +10 kV to the anodic reservoir. The sample was then injected by applying +5 kV for 4 s in the case of catechol and +8 kV for 5 s in the case of PCP. PCP is a neutral compound; its injection was effected by adding some positively charged 6-amino-β-cyclodextrin (25 mM) to the diluted sample to form an inclusion complex with PCP. Unless otherwise stated, the solutions were not deaerated, and all separations were performed at room temperature (21-23 °C). The electropherogram peaks were identified by spiking with individual standard compounds. For each compound, a series of runs at different concentrations was performed to obtain a peak height vs concentration plot, used for estimation of each compound in unknown samples. Efficiency, defined as the number of theoretical plates (N) for each peak in the electropherogram, was calculated as 5.54 (tr/w1/2)2, where tr and w1/2 represent the migration time of the analyte and the full-peak width at halfmaximum, respectively.20 Safety Considerations. PCP, phenol, and 2,3-dimethylphenol are harmful and cause headache and weakness, and the vapor (20) Jorgenson, J. W.; Lukas, K. D. Anal. Chem. 1981, 53, 1298-1302.
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
2537
Figure 3. Electrophoretic detection of catechol (0.4 mM) with rotating electrode (solid line) at 500 rpm and with stationary electrode (dotted line). The buffer consisted of 20 mM MES, pH 5.6, sample injection at +5 kV for 4 s, separation potential +8 kV, and detection potential +0.7 V vs Ag/AgCl.
Figure 2. Effect of rotation frequency on the amperometric signal. (A) Current response of the rotating amperometric signal to catechol (3 mM in the cathodic reservoir) as a function of rotation frequency. (B) Influence of the rotation rate on the electrophoretic detection of catechol (0.4 mM). The buffer consisted of 20 mM MES, pH 5.6, sample injection at +5 kV for 4 s, separation potential +8 kV, and detection potential + 0.7 V vs Ag/AgCl.
of such compounds is very dangerous. Stock solutions of the phenolic compounds must be prepared and handled in a ventilated hood. Skin contact must be avoided since PCP and 2,3-DMPh may be absorbed through the skin. Special care must be taken to dispose of waste solutions. During the course of experiments, to avoid electrical shock, the high-voltage power supply must be handled with extreme care. RESULTS AND DISCUSSION Effect of Rotation Rate. The rotating disk electrode acts as a pump dragging fresh solution up from the bulk of the solution toward the electrode surface and then spinning it around and flinging it out sideways in a radial direction because of centrifugal force. The action of the rotating electrode creates a stationary boundary, known as the diffusion layer, at the electrode surface that rotates with the electrode. Outside this stationary layer the solution is well stirred. It has been well known that the thickness of the diffusion layer is dependent on the rotation speed and so can be varied experimentally. According to the Koutecky-Levich equation,21 the current of the rotating electrode is proportional to the square root of the rotation speed; i.e., the faster the electrode is rotated the thinner the stagnant layer and thus the more efficient the transport of substrate from the bulk solution to the electrode surface. Figure 2A shows the current obtained plotted against the rotation speed for analysis of 3 mM catechol (bulk concentration) in the outlet reservoir. As can be seen, with increasing rotation (21) Koutecky, J.; Levich, V. G. Zh. Fiz. Khim. 1958, 32, 1565-1575.
2538 Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
speed, the current eventually reached a plateau at 8000 rpm as expected. To characterize the CEEC system, a preliminary experiment was first performed by the electrophoretic separation of catechol. The polished end capillary was positioned ∼60 µm from the detecting electrode with a much larger surface area (3 mm). The analyte, emerging from the outlet of the separation capillary, would be confined in the diffusion layer which rotated with the electrode; i.e., the analyte is detected as it exits the capillary. As the diffusion layer thickness is dependent on the frequency of the rotating electrode, the amperometric signal of RDE will be governed by the rotation speed up to a certain frequency as anticipated from the Koutecky-Levich equation.21 The experimental data confirmed that the response of the electrochemical sensor increased with the rotation speed and reached a plateau (2.5 µA) at ∼500 rpm. Beyond this frequency, the rotating electrode current was almost independent of the rotation rate as expected from the Koutecky-Levich equation (Figure 2B). As a result, this rotation frequency was used to pursue CE with rotating amperometric detection. For comparison, the electropherogram obtained for the electrophoretic separation of catechol using the stationary disk electrode was compared to that of the rotating disk electrode (Figure 3). As shown in this figure, an important benefit of rotation during the electrophoretic separation was noted by an increase of the amperometric signal and the peak skewness (sharpness). The current response of RDE was almost 2-fold higher than that obtained with the stationary electrode, and the emerged peak became very sharp with no tailing (see Figure 3, inset). The number of theoretical plates for catechol was estimated to be 67 000 plates/m at 500 rpm compared to 26 000 plates/m at 0 rpm. Under rotation, there was only a little increase in the signal baseline and the noise level, indicating a successful isolation of the detector from the high separation potential applied. It should be noted that any failure of such isolation leads to high noise levels, which could be several orders of magnitude higher than the amperometric signal. It should be noted that the separation capillary was firmly inserted into a septum and the space between the rotating electrode and the separation capillary was ∼60 µm as observed by a microscope. No abrasion of the rotating electrode was noted throughout the course of investigation.
Figure 4. Effect of the separation potential on the detection of 0.4 mM catechol. The buffer consisted of 20 mM MES, pH 5.6, sample injection at +5 kV for 4 s, and detection potential + 0.7 V vs Ag/ AgCl. (A) Rotation frequency: 500 rpm, curve a, +8 kV, curve b +11 kV, and curve c +14 kV. (B) Rotation frequency: 0 rpm, curve a, +8 kV, curve b +11 kV, and curve c +14 kV.
Effect of the Separation Potential. A series of experiments was then conducted to evaluate the effect of the separation potential on the performance of the rotating electrode. At 500 rpm, increasing the separation potential from +8 to +11 kV and then +14 kV resulted in a decrease of the migration time from 6.5 to 5 min and then to 3.3 min. The current responses of RDE increased appreciably with increasing separation potentials without any significant change in the baseline slope and the noise level (Figure 4A). Such a result indicated a negligible electrical interference between the applied potential of RDE and the separation potential under such operating conditions. To date, all studies related to the conjunction of CE with an external and stationary electrode without any decoupler have showed serious electrical interferences caused by the separation potential.11 In general, increasing the separation potential results in decreased amperometric signal (stationary electrode) and a sloping baseline. The number of theoretical plates for catechol using RDE was estimated to be 52 600 plates/m at +8 kV and 57 500 plates/m at 14 +kV. In the case of the stationary electrode, although the migration time of the catechol peak decreased by increasing the separation potential as expected, there was no significant increase in the current responses of the detecting electrode. Furthermore, the baseline current increased and reached 0.4 µA at +14 kV (curve c in Figure 4B). This was probably due to the difference between the potential in the buffer above the detecting electrode and the potential set at the working electrode by the potentiostat. In most cases, there was a small dip in the front of the main peak, particularly at high separation potentials, an unknown phenomenon that remains to be investigated.
During all electrophoretic measurements, the activity of the rotating electrode toward catechol was stable and the peak height variation for 10 repeated injections was ∼4% compared to 7% for the stationary electrode. To regenerate the surface, a simple electrochemical treatment was performed by recycling the potential of the electrode between -0.5 and +1 V at 500 mV/s for 30 s. The detection limit of catechol was estimated on the basis of the signal-to-noise characteristics (S/N ) 3). The limit of detection obtained with a rotating detector was estimated to be 0.3 µM, 2-fold lower than the value obtained with the stationary electrode (0.7 µM). Electrophoretic Separation and Detection of Catechol from a Mixture. An experiment was conducted to investigate the effect of the rotation on the electrochemical activity of RDE during the electrokinetic separation of a mixture containing catechol, phenol, and 2,3-DMPh. These latter two phenolic compounds are prone to polymerization, which rapidly fouls electrodes; therefore, they were selected to characterize the passivation of RDE versus the stationary electrode. A separation buffer consisting of 10 mM phosphate, 6 mM borate, pH 7.5, containing 10 mM SDS was used to perform all analyses. With respect to RDE, increasing the separation potential from +14 to +20 kV exhibited a pronounced effect on the migration time. The baseline currents remained stable, and there were practically no changes in the noise level and the baseline slope (Figure 5A). The electropherogram, however, revealed that, unlike the sharp peak for catechol, tailing was present in the peaks corresponding to phenol and 2,3-DMPh. Such a result indicated that adsorption to the negatively charged capillary walls was responsible. If the electrode surface was not cleaned prior to the next sampling stage, the current responses of RDE for catechol decreased with the number of injections, about 3-4%/analysis. As shown in Figure 5B, the peak tailing became obvious and the current response decreased about 30-40% after 10 repeated analyses due to the formation of strongly adsorbed species on the electrode surface. Notice also that there was a noticeable variation in the migration time when the capillary was not treated between runs, likely due to the adsorption of phenol and 2,3-DMPh on the internal wall of the capillary to affect the electroosmotic mobility. In the case of the stationary electrode, the amperometric signal for catechol decreased up to 20-30%/run if the electrode and the capillary were not cleaned and conditioned between runs. After the third injection, no peaks were observed in the electropherogram, indicating a total electrode fouling (Figure 5C). Comparison of these results attested that the rotating electrode effectively circumvented or at least minimized a fast and total passivation of the electrode. Such results were not totally unexpected since the rotating electrode acted as a pump constantly drawing fresh solution up from the bulk liquid and then flinging the solution outward from the center. The fluid at the disk surface was replenished by a flow normal to the surface. As a consequence, the contact time between the electrode and the analyte and/or other intermediate products formed during electrooxidation was much shorter than that of the stationary electrode. Regardless of the operation mode, to obtain a good reproducibility for the migration time (3%), after each analysis, the capillary had to be flushed with methanol for 2 min and followed by a rinsing sequence of 0.1 M NaOH for 5 min, water for 2 min, and Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
2539
Figure 6. Electropherogram of an extract prepared from a PCPcontaminated soil at 0 (+++) and 500 rpm (- - -), 10 mM PCP standard (s). Separation buffer: borate/phosphate, 25 mM each, pH 8.7, sample injection at 8 kV for 5 s, detection potential +0.7 V vs Ag/AgCl, separation potential +11 kV.
Figure 5. Performance of the rotating disk electrode during the electrophoretic separation of 0.4 mM catechol, 0.4 mM phenol, and 0.4 mM 2,3-dimethylphenol. Separation buffer consisted of a mixture of 10 mM phosphate, 6 mM borate, pH 7.5, containing 10 mM SDS, sample injection at + 8 kV for 5 s, detection potential + 0.7 V vs Ag/AgCl. (A) Rotation rate 500 rpm, (s) separation potential +14 kV, (- - -) separation potential +20 kV. (B) Rotation rate 500 rpm, (s) first injection and (- - -) tenth injection, separation potential +17 kV. (C) Rotation rate 0 rpm, (s) first injection, (- - -) second injection, and (+++) third injection, separation potential +17 kV.
the separation buffer for 3 min. As discussed earlier, a simple electrochemical treatment (recycling the potential between -0.5 and +1 V at 500 mV/s for 30 s) was effective in regenerating the electrode surface. Analysis of Pentachlorophenol in Contaminated Soil. Figure 6 shows the electropherogram of the soil extract prepared from a standard soil sample (U.S. EPA certified, CRM 103-100, Resource Technology Corp., Laramie, WY). Several portions (0.1 g) were extracted with 5 mL of a 50/50 volume mixture of methanol and buffer (mixture of 25 mM phosphate, 25 mM borate, pH 8.7) containing 25 mM 6-amino-β-cyclodextrin. Spiking the extract with a PCP solution augmented the peak at 6.6 min, thus confirming its identification. Besides PCP, such certified samples were known to contain several compounds including polyaromatic hydrocarbons (PAHs) which might be extracted in the process together with PCP. Therefore, it was not surprising to notice the
2540 Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
presence of some unidentified peaks in the electropherogram (Figure 6). Notice also that the PCP peak obtained from soil samples appeared at the same retention time as the PCP standard sample, which indicated that the composition did not affect the mechanism of separation. Current responses of the rotating electrode were higher corresponding to sharper peaks in comparison to those obtained with the stationary electrode. The baseline currents increased during the course of measurement, possibly due to the surface adsorption and/or polymerization of endogenous electroactive species present in the soil samples. Electrochemical treatment between two consecutive runs was not expected to remove them completely from the electrode surface. The detection limit obtained with RDE was estimated to be 20 µM, i.e., 2-fold lower than the value obtained with the stationary electrode (45 µM). The concentration of PCP (analyzed for three different soil samples and performed in triplicate) was determined to be 1370 ( 180 mg of PCP/kg of soil, which agreed well with the result obtained by U.S. EPA liquid chromatography (1425 ( 320 mg of PCP/kg of soil).22 Conclusion. Coupling the rotating disk electrode with CE leads to reusable analytical systems which provide high sensitivity, about 2-3-fold over the stationary electrode, better separation efficiency (2-3-fold), reduced electrode fouling, and suppressed interference at the detecting electrode. The rotating amperometric detector/CE system successfully measured pentachlorophenol in extracts prepared from contaminated soils, and the results were in good agreement with those obtained with the U.S. EPA liquid chromatography procedure. Received for review October 6, 2000. Accepted February 1, 2001. AC001192J (22) Hilmi, A.; Luong, J. H. T.; Nguyen, A. L. Environ. Sci. Technol. 1997, 31, 1794-1800.
