Monitoring Electroosmotic Flow by Periodic Photobleaching of a Dilute

Anal. Chem. 2000, 72, 4317-4321

Monitoring Electroosmotic Flow by Periodic Photobleaching of a Dilute, Neutral Fluorophore Kimberley F. Schrum,† Joseph M. Lancaster III, Stephen E. Johnston, and S. Douglass Gilman*

Department of Chemistry, University of Tennessee, Knoxville, Tennessee 37996-1600

Electroosmotic flow has been monitored in a capillary using a method based on periodic photobleaching of a neutral, fluorescent buffer additive. Rhodamine B was determined to be neutral between pH 6.0 and 10.8 and was added to the running buffer at a concentration of 400 nM. Rhodamine B was photobleached by opening a shutter under computer control for 250 ms every 5.00 s, to expose the dye to a laser beam and create a photobleached zone. The time was measured for the photobleached zone to migrate 6.13 mm to a downstream laserinduced fluorescence detector, to determine the rate of electroosmotic flow in the entire capillary. The flow rate was sampled every 5.00 s, and the precision of the flow measurements was 0.7% or better. Three fluorescent compounds were separated and detected by capillary electrophoresis with laser-induced fluorescence detection, while simultaneously monitoring the electroosmotic flow rate. Electroosmotic flow (EOF) plays a major role in capillary electrophoresis (CE) and in analytical methods in microfabricated devices based on electrophoresis. Typically, EOF is greater in magnitude than electrophoretic migration in fused-silica capillaries. Thus, cations and anions can be separated and detected simultaneously by CE.1-3 The relatively flat flow profile for EOF results in high-efficiency separations.1,4-10 In microfabricated analytical devices, EOF is useful for both electrophoretic separations and other applications requiring fluid transport.11-14 * Corresponding author: (fax) 865-974-3454; (e-mail) [email protected]. † Department of Chemistry, Whittier College, Whittier, CA, 90608. (1) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (2) Jorgenson, J. W.; Lukacs, K. D. Science (Washington, D.C.) 1983, 222, 266272. (3) Lukacs, K. D.; Jorgenson, J. W. J. High Resolut. Chromatogr. Chromatogr. Commun. 1985, 8, 407-411. (4) Rice, C. L.; Whitehead, R. J. Phys. Chem. 1965, 69, 4017-4024. (5) Taylor, J. A.; Yeung, E. S. Anal. Chem. 1993, 65, 2928-2932. (6) Tsuda, T.; Ikedo, M.; Jones, G.; Dadoo, R.; Zare, R. N. J. Chromatogr. 1993, 632, 201-207. (7) Tsuda, T.; Kitagawa, S.; Dadoo, R.; Zare, R. N. Bunseki Kagaku 1997, 46, 409-414. (8) Paul, P. H.; Garguilo, M. G.; Rakestraw, D. J. Anal. Chem. 1998, 70, 24592467. (9) Herr, A. E.; Molho, J. I.; Santiago, J. G.; Mungal, M. G.; Kenny, T. W.; Garguilo, M. G. Anal. Chem. 2000, 72, 1053-1057. (10) Tallarek, U.; Rapp, E.; Scheennen, T.; Bayer, E.; Van As, H. Anal. Chem. 2000, 72, 2292-2301. (11) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science (Washington, D.C.) 1993, 261, 895-897. 10.1021/ac0005114 CCC: $19.00 Published on Web 08/18/2000

© 2000 American Chemical Society

Two related problems associated with EOF are its susceptibility to changes in the surface-solution interface of a capillary or microfabricated channel and the difficulty in monitoring EOF rates (typically on the order of nanoliters per minute) over the course of an experiment. Electroosmotic flow is generated at the surfacesolution interface in a capillary or microfabricated channel, and the magnitude of EOF will change due to changes in the chemical composition of the surface, changes in the pH and buffer composition, and changes in temperature.3,4,12,13,15 Unfortunately, the injection of a sample zone and separation of its components will alter EOF if any sample components adsorb to the capillary surface and alter the ζ potential.2,16-18 Because it is difficult to monitor extremely low-volume flow rates typical of capillaries and microfabricated devices, it is hard to account for these changes in EOF and the effects they will have on a separation or other analytical operations involving EOF. One result of the alteration of EOF due to sample interactions with the flow-generating surface is poor precision for CE.19 Clearly it is desirable to continuously monitor EOF during the course of an entire experiment. The earliest reported method for measuring EOF in CE is commonly referred to as the neutral marker method.3,20 A neutral compound is injected with the sample zone, and its migration time to the detector is used to calculate the EOF rate. It is important to realize that this experiment provides an average EOF rate. Furthermore, this average EOF rate only covers the time period from the beginning of the separation until the neutral marker migrates past the detector. Depending on the experimental conditions, anionic or cationic analytes will elute past the detector after the neutral marker. Changes in the EOF after the neutral marker has passed the detector will not be included in the average EOF measured using this method. A closely related technique, the current-monitoring method, can be used to measure average EOF.21 For this method, the solution reservoir at the injection end of the capillary has a (12) Effenhauser, C. S.; Bruin, G. J. M.; Paulus, A. Electrophoresis 1997, 18, 2203-2213. (13) Fletcher, P. D. I.; Haswell, S. J.; Paunov, V. N. Analyst (Cambridge, U.K.) 1999, 124, 1273-1282. (14) Figeys, D.; Pinto, D. Anal. Chem. 2000, 72, 330A-335A. (15) Knox, J. H.; McCormack, K. A. Chromatographia 1994, 38, 207-214. (16) Towns, J. K.; Regnier, F. E. Anal. Chem. 1992, 64, 2473-2478. (17) Lee, T. T.; Dadoo, R.; Zare, R. N. Anal. Chem. 1994, 66, 2694-2700. (18) Locascio, L. E.; Perso, C. E.; Lee, C. S. J. Chromatogr., A 1999, 857, 275284. (19) Schaeper, J. P.; Sepaniak, M. J. Electrophoresis 2000, 21, 1421-1429. (20) Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T. Anal. Chem. 1984, 56, 111-113. (21) Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1988, 60, 1837-1838.

Analytical Chemistry, Vol. 72, No. 18, September 15, 2000 4317

different ionic strength than the solution filling the capillary. Consequently, the electrophoretic current changes as the solution at the injection end of the capillary fills the capillary and changes the total conductivity in the capillary. Average EOF is measured by determining the time it takes for the current to stop changing when the entire capillary has been filled with the solution from the injection reservoir. Again, an average EOF rate is obtained and EOF changes occurring after the time that a neutral marker would elute will not be detected. The current-monitoring method has been used to observe changes in EOF during an experiment by observing changes in the slope of a plot of current versus time; however, when quantitative values for EOF were required, the average EOF was measured as described above.22 Reported precision for average EOF rates measured using this method in capillaries and microfabricated devices typically are between 5 and 15%.18, 21-24 Conductivity measurements have been used to measure average EOF and to continuously monitor EOF in capillaries.21,25 Zare and co-workers measured average EOF using a conductivity detector at the detection end of a CE capillary.21 This approach was used to validate the current monitoring method, and the conductivity change was detected resulting from the elution of the solution from the injection reservoir at the end of the capillary. Everaerts and co-workers used conductivity detection to monitor changes in EOF over the course of an experiment with a time resolution of 20 s.25 No precision was reported for these EOF monitoring experiments. The average EOF rate and the EOF rate over the course of an experiment have both been measured by weighing the effluent from a capillary with an analytical balance.26-28 For average EOF determinations, a precision of 1.3% was reported.28 The time resolution for EOF monitoring based on effluent weighing was 5 min.26 The precision for these EOF monitoring experiments was not reported. Zare and co-workers developed a method for continuous monitoring of EOF with a time response of ∼1 s and precision as low as 1%.17 This method was based on the measurement of the dilution of a fluorophore solution by the effluent from the CE capillary. The CE effluent and the fluorophore solution were mixed on-line using a concentric capillary design for postcolumn reagent addition. The fluorescence measured in the postcolumn capillary was inversely related to the EOF rate in the CE capillary because the fluorophore was diluted by the EOF. Several research groups have imaged EOF using fluorescence and NMR.5-10,29 Most of these papers are fundamental studies of the axial flow profiles for EOF and pressure-driven flow in capillaries.5-10 Fluorescence imaging is also used to study fluid flow in microfabricated devices,11,12 although routine monitoring of EOF rates using this approach does not appear to be practical. (22) Lee, C. S.; Blanchard, W. C.; Wu, C.-T. Anal. Chem. 1990, 62, 1550-1552. (23) Lee, C. S.; McManigill, D.; Wu, C.-T.; Patel, B. Anal. Chem. 1991, 63, 15191523. (24) Ocvirk, G.; Munroe, M.; Tang, T.; Oleschuk, R.; Westra, K.; Harrison, D. J. Electrophoresis 2000, 21, 107-115. (25) Wanders, B. J.; van de Goor, T. A. A. M.; Everaerts, F. M. J. Chromatogr., A 1993, 652, 291-294. (26) van de Goor, A. A. A. M.; Wanders, B. J.; Everaerts, F. M. J. Chromatogr. 1989, 470, 95-104. (27) Altria, K. D.; Simpson, C. F. Anal. Proc. 1986, 23, 453-454. (28) Altria, K. D.; Simpson, C. F. Chromatographia 1987, 24, 527-532. (29) Priesler, J.; Yeung, E. S. Anal. Chem. 1996, 68, 2885-2889.

4318 Analytical Chemistry, Vol. 72, No. 18, September 15, 2000

Figure 1. Schematic of the instrument used for detection of EOF by periodic photobleaching. BS, beam splitter; M, mirror; L, planoconvex lens; C, capillary; F1, bleaching zone in the capillary; F2, LIF detection zone in the capillary; MO, microscope objective; DM, dichroic mirror; F, filters; PH, pinhole; PMT, photomultiplier tube. Positions F1 and F2 are the same as illustrated in Figure 2.

Priesler and Yeung monitored EOF by fluorescence imaging of a zone of a neutral dye over a 23.2-cm region in a capillary.29 The EOF was imaged every 1-5 min. Tallarek et al. used NMR to image EOF in a 250-µm i.d. capillary, and the EOF rate was measured as a function of applied potential.10 The goal of the research presented in this paper is to develop a method for continuous and direct monitoring of EOF rates that can be used for routine monitoring of EOF in capillaries. Our approach to EOF monitoring is based on the periodic photobleaching of a dilute, neutral fluorophore and is similar to methods developed to measure plasma flow rates in capillaries in animals.30-32 Photobleaching has also been used for sample gating and highspeed CE.33 Ultraviolet laser pulses have been used to release a caged fluorescent dye in a spatially defined zone in a capillary for imaging studies of EOF profiles.8,9 Our method in its present form samples EOF every 5-6 s with a precision of 0.7% or better. No postcolumn detection or reagent addition is necessary for this method. Instead, the dilute, neutral fluorophore is simply added to all the buffers used in the experiments, and EOF is monitored throughout the experiment. EXPERIMENTAL SECTION Chemicals. Laser-grade rhodamine B and fluorescein isothiocyanate were obtained from Acros (Pittsburgh, PA), and mesityl oxide was obtained from Aldrich (Milwaukee, WI). Fluorescein was purchased from Sigma (St. Louis, MO), and boric acid was purchased from J. T. Baker Chemical (Phillipsburg, NJ). All solutions were prepared in doubly distilled water. Instrumentation. A schematic of the instrument constructed for these experiments is shown in Figure 1. All components indicated in the schematic, with the exception of the laser, were housed in a light-tight, black, Plexiglas box. The beam from the 457.9-nm line of an argon ion laser (Coherent Innova 90C-5; Santa Clara, CA) was split with a broadband cubic beam splitter. The bleaching beam (52 mW) was directed by a mirror through a computer-controlled, electronic shutter (Uniblitz 310 B; Vincent (30) Wieringa, P. A.; Van Putten, M. J. A. M.; Duling, B. R. Microvasc. Res. 1993, 46, 263-282. (31) Wieringa, P. A.; Damon, D. N.; Duling, B. R. Med. Biol. Eng. Comput. 1995, 33, 563-570. (32) Berk, D. A.; Swartz, M. A.; Leu, A. J.; Jain, R. K. Am. J. Physiol.: Heart Circ. Physiol. 1996, 270, H330-H337. (33) Monnig, C. A.; Jorgenson, J. W. Anal. Chem. 1991, 63, 802-807.

Associates; Rochester, NY). The beam was then focused onto the capillary at position F1 with a 38-mm focal length fused-silica plano convex lens. Position F1 was actually located vertically above the detection position at F2. The detection portion of the capillary was mounted in a homebuilt holder and secured with Duco cement (Devcon; Danvers, MA). The detection beam was reduced to 28 mW with a neutral density filter (not shown). A dichroic mirror (475DRLP; Omega Optical; Brattleboro, VT) directed the beam toward the capillary. The detection beam was focused onto the capillary through a 20× microscope objective (0.4 NA) at position F2. The fluorescence was back-collected, and residual scattering from the capillary was reduced by two glass cut-on filters. Scatter was further reduced by passing the fluorescence through a pinhole (1.5 mm). The fluorescence was detected using a PMT (Hamamatsu HC120; Bridgewater, NJ), which was powered with a home-built power supply. A small amount of the bleaching beam that passed through the capillary at position F1 was redirected by two mirrors through the pinhole as indicated in Figure 1. This light provided a small signal at the PMT that indicated when the shutter was opened. An instrument control program was developed using LabVIEW (National Instruments, Austin, TX). The program controlled the shutter timing and collected data from the PMT using a National Instruments Lab-PC-1200 data acquisition board. The signal from the PMT was filtered using a 50 Hz low-pass filter before the data acquisition board. The data files were analyzed in Microsoft Excel. Capillary Electrophoresis. The CE system was constructed in-house and was enclosed in a Plexiglas box for operator safety. The electrophoretic potential was applied using a Spellman CZE1000R power supply (Hauppague, NY), and platinum wires were used as electrodes in the buffer reservoirs at the injection and detection ends of the capillary. The fused-silica capillaries used were 50-µm i.d./220-µm o.d. (SGE, Austin, TX), and detection windows were made by burning away ∼1 cm of the polyimide coating. The capillary used for EOF measurements was 70.0 cm in total length and 48.5 cm from the injection end to the laserinduced fluorescence (LIF) detector. Injections were made electrokinetically. All electrophoretic buffers used were 50 mM borate at pH 9.5, and buffers for EOF monitoring also contained 400 nM rhodamine B. Rhodamine B Neutrality Studies. For experiments to test the pH range over which rhodamine B was neutral, UV-vis absorbance detection at 200 nm was used. The detector was a Linear UVIS 204 Detector (Linear Instruments, Reno, NV) equipped with an on-column capillary cell. The UV-vis data were collected with a strip chart recorder. Mesityl oxide and methanol were used as neutral markers and were injected electrokinetically for 3.0 s at 15.0 kV. Separations were carried out at 15.0 kV in 50 mM borate solutions adjusted to the desired pH using either NaOH or HCl. The capillary used for these experiments was 70.0 cm in total length, and 51.0 cm from the injection end to the absorbance detector. RESULTS AND DISCUSSION Monitoring Electroosmotic Flow. A schematic of the instrument constructed for this work is presented in Figure 1, and our technique for measurement of EOF is illustrated in Figure 2. The capillary and both the injection and detection buffer reservoirs are filled with a solution containing a dilute, neutral fluorophore.

Figure 2. Illustration of the method for EOF monitoring. (A) EOF is measured by determining the time required for a photobleached zone of a neutral fluorophore to migrate through the capillary from position F1 (bleaching beam focal point) to position F2 (detection beam focal point) over a distance dF1-F2. (B) Before the shutter blocking the bleaching beam has opened; the capillary contains a constant concentration of the neutral fluorophore (CNF). (C) The shutter opens briefly, and a bleached zone of fluorophore is created at F1 (reduced CNF). When the shutter opens, a positive peak is observed (see Figure 3) due to light from the bleaching beam that is redirected to the detector at F2 by two mirrors (Figure 1). (D) After the shutter has closed, the photobleached zone flows to F2, and a negative peak is observed at the LIF detector.

The photobleached zone is created at position F1 by laser irradiation, and this zone is detected downstream at position F2 by LIF (Figure 2A). Just before the flow measurement is started, as indicated in Figure 2B, the shutter blocking the bleaching laser beam is closed and the concentration of the neutral fluorophore is constant in the capillary. Then the shutter is opened briefly, and a photobleached zone is created at position F1 (Figure 2C). The photobleached zone migrates to position F2, due to EOF, where it is detected by an LIF detector at F2 (Figure 2D). After the photobleached zone has migrated past the detector at F2, the shutter is briefly opened again and the flow measurement is repeated. The time required for a photobleached zone of a neutral fluorophore to migrate from position F1 in the capillary downstream to position F2 is measured (Figures 1 and 2A). The linear flow rate is simply

flow rate ) dF1-F2/∆tF1-F2

(1)

where dF1-F2 is the distance between positions F1 and F2 on the capillary and ∆tF1-F2 is the time required for the photobleached zone to travel from position F1 to position F2. Figure 3 presents initial data for EOF monitoring using this approach. Before the shutter is opened, a constant background fluorescence is observed at the LIF detector at position F2 due to rhodamine B (400 nM) in the separation buffer (Figure 3A, 2B). The shutter is then opened for 250 ms, creating a photobleached zone. At the same time, a positive peak is observed at the LIF detector at F2 (Figure 3A, 2C). This positive peak is generated by redirecting some of the light from the bleaching beam (F1) to the PMT, using mirrors as indicated in Figure 1. This positive Analytical Chemistry, Vol. 72, No. 18, September 15, 2000

4319

Table 1. Change of Electroosmotic Flow Rate with Applied Potentiala applied potential (kV)

EOF (cm/s)

RSD (%)

15.0 17.5 20.0 22.5 25.0 27.5 30.0

0.1186 0.1436 0.1732 0.2092 0.2502 0.2976 0.3543

0.2 0.2 0.6 0.7 0.4 0.5 0.6

a Experimental conditions were as described in Figure 3. RSD values are based on 8-10 measurements.

of the capillary to the detector at F2. At the same time, the EOF is monitored at 5.00-s intervals, as described above. The mean time for the photobleached zone to migrate from F1 to F2 is calculated using all of the EOF monitoring data during the time that the concentrated rhodamine B zone migrated to the detector. The distance between F1 and F2 is simply

dF1-F2 ) mean photobleached zone migration time × injection end to F2 distance (2) neutral marker migration time

Figure 3. EOF monitoring data. (A) The labeled positions, (2B, 2C, and 2D) correspond to the illustrations in Figure 2. The distance between F1 and F2, as shown in Figure 2, is 6.13 mm. The shutter was opened for 250 ms every 6.00 s. All other conditions were as described in the Experimental Section. At ∼121 s, the applied potential was increased from 17.5 to 20.0 kV. (B) EOF monitoring data at an applied potential of 30.0 kV. (C) A series of EOF monitoring peaks at an applied potential of 15.0 kV.

peak serves as a time stamp, indicating exactly when the shutter was opened and closed to create the photobleached zone. After 4.27 s, the photobleached zone migrates past the LIF detector at F2 and is detected as a negative peak (Figure 3A, peak 2D). It is desirable to select a neutral fluorophore for EOF monitoring so that electrophoretic migration of the fluorophore can be neglected. Separate experiments with UV-vis absorbance detection show that rhodamine B is neutral (zwitterionic) from a pH of 6.0 to 10.8 based on coelution with a neutral marker. Rhodamine B is also highly fluorescent and, therefore, can be added to the electrophoresis solutions at low concentration for EOF monitoring. To make absolute measurements of flow magnitude, it is necessary to accurately measure the distance between positions F1 and F2 on the capillary; however, dF1-F2 is not easy to measure directly. Instead, this is accomplished by injecting a neutral marker, in this case a more concentrated plug of rhodamine B (1.6 × 10-6 M, 1-s injection). The separation buffer still contains 400 nM rhodamine B. The total migration time of the concentrated zone is determined, and the length of the capillary from the injection end to the detector at F2 is measured. The migration time of the neutral marker provides an average EOF rate during the time that the neutral marker migrated from the injection end 4320

Analytical Chemistry, Vol. 72, No. 18, September 15, 2000

For the experiments shown here, dF1-F2 was 6.13 mm. EOF versus Applied Potential. Figure 3 shows data for EOF monitoring experiments at several applied potentials, obtained using our technique. In Figure 3A, a potential of 17.5 kV (250 V/cm) was applied, and the measured linear flow rate was 0.144 cm/s. This corresponds to a volume flow rate of 168 nL/min for the 50-µm i.d. capillary used in these experiments. At ∼121 s (Figure 3A), the applied potential was increased to 20.0 kV (286 V/cm) and the distance between the next pair of positive and negative peaks decreased from 4.27 to 3.54 s, indicating an increase in flow rate to 0.173 cm/s (203 nL/min). Figure 3B shows EOF measurements at an applied potential of 30.0 kV (429 V/cm) and a measured flow rate of 0.354 cm/s (415 nL/min). The positive and negative peaks are now only 1.73 s apart. Figure 3C shows data from EOF monitoring at a constant potential (15.0 kV, 214 V/cm) over a longer period of time (50 s). The measured flow rate was 0.119 cm/s (139 nL/min). The shutter was opened every 6.00 s for 250 ms for the experiments shown in Figure 3. Table 1 presents the measured EOF rate at applied potentials from 15.0 to 30.0 kV. As expected, the EOF rate increases with applied potential. Ideally, the EOF rate should increase linearly with applied potential, but a nonlinear response is observed due to increased temperature in the capillary caused by Joule heating at higher applied potentials. The nonlinear dependence of EOF and electrophoretic current due to Joule heating has been observed and examined for CE by several groups.1,10,15,24-27,33 Time Resolution, Dynamic Range, Precision, and Accuracy. The EOF rate was measured every 6.00 s for the data presented in Figure 3 and Table 1. What is actually measured is the average flow rate over the time required for the photobleached zone to migrate from position F1 to F2 in the capillary. This time ranges from 5.17 s at 15.0 kV to 1.73 s at 30.0 kV. The dynamic

Figure 4. Separation of three fluorophores with simultaneous monitoring of EOF. Both analytes and EOF were detected at position F2 as illustrated in Figures 1 and 2. A 1.0-s, 20.0 kV electrokinetic injection was made of a solution containing 1.2 × 10-6 M rhodamine B, 2.5 × 10-7 M fluorescein isothiocyanate, and 5.0 × 10-9 M fluorescein. The separation was carried out at an applied potential of 20.0 kV. The separation buffer contained 400 nM rhodamine B and 50.0 mM borate at pH 9.5. The distance between F1 and F2 (Figures 1 and 2) was 6.13 mm. The shutter was opened for 250 ms every 5.00 s.

range for these EOF measurements is limited by the time that the shutter is opened (250 ms), the width of the photobleached zones at F2, and the frequency that the shutter is opened (6.00 s)-1. Under these conditions (dF1-F2 ) 6.13 mm), the fastest EOF rate that could be measured is ∼2.4 cm/s, before the positive and negative peaks begin to overlap. The slowest EOF rate that could be measured is ∼0.10 cm/s, and it is limited by the frequency that the shutter is opened (6.00 s)-1. Our method provides extremely precise values for EOF. The standard deviation of 8-10 flow measurements at each applied potential is reported in Table 1 and ranges from 0.2 to 0.7%. The precision of these EOF measurements is superior to that typically obtained using the current monitoring method for average EOF measurements.18,21-24 Our precision values are similar to those obtained for average EOF measurements using neutral markers and effluent weighing and for EOF monitoring by postcolumn addition of a fluorophore.3,17,20,28 The accuracy of our method was assessed by comparing EOF measurements made using our technique to measurements carried out using the neutral marker method. Our method for EOF monitoring was calibrated using a neutral marker at 390 V/cm as described earlier in this manuscript. The applied potential was then adjusted between 234 V/cm and 430 V/cm, and EOF was determined at each potential using both techniques (excluding 390 V/cm, which was used for calibration). As expected, a plot of the average EOF measured using the neutral marker method versus the average of measurements made using our technique was linear (R2 ) 0.9996). The average relative error of EOF rates measured using the two methods is 0.35%. Separations. To demonstrate the feasibility of using this technique to measure EOF in real time during a separation, a mixture of three fluorophores was injected and separated while

the flow was monitored with rhodamine B in the running buffer. The electropherogram for this experiment is presented in Figure 4. The separation buffer contained 400 nM rhodamine B, and the shutter was opened for 250 ms every 5.00 s for EOF monitoring. The sample contained 1.2 × 10-6 M rhodamine B, 2.5 × 10-7 M fluorescein isothiocyanate, and 5.0 × 10-9 M fluorescein, and it was injected electrokinetically for 1.0 s at the separation potential (20 kV). The analyte peaks can be readily distinguished from the series of small EOF monitoring peaks. The most efficient peak in the separation is that for fluorescein at 515 s (N ) 360 000). Even for this peak, the full width at half-maximum of 2.01 s is 7.7 times wider than the EOF monitoring peaks at F2 (260 ms). The mean EOF rate was 0.1747 ( 0.0004 cm/s from 0 to 545 s. The flow rate varied only 2.3% (0.1722-0.1761 cm/s) over the entire experiment, indicating that the injection of this sample had very little effect on the capillary surface and EOF. CONCLUSIONS This work demonstrates a new and practical method for monitoring extremely low-volume flow rates typically encountered in microscale separation structures. We have demonstrated that this technique can measure small changes in flow over the course of an entire experiment with excellent precision and time resolution using the prototype device. The development of a practical method for measuring such flow rates is critical for the continued improvement and development of increasingly sophisticated microscale analytical devices that depend on EOF and pressuredriven flow. Although this prototype instrument is moderately complicated in design and construction, its operation is simple. It does not require any on-column fluid junctions or sophisticated microscopic imaging techniques. A neutral fluorophore is selected and added to the separation buffer at low concentration before it is introduced into the capillary. The user initiates the computer program that controls the shutter operation and EOF monitoring data collection. The data contains a record of the EOF throughout the experiment. Our approach using a 180° detection geometry should be equally applicable to EOF monitoring in capillaries and planar microfabricated devices. ACKNOWLEDGMENT The authors would like to recognize Angela Whisnant and Jason Pittman for their assistance with programming and instrument construction. The authors thank David Schrum for suggestions regarding dye selection and Robert Compton for providing some of the optics used for this project. The authors also thank Mark Hayes for helpful discussions. This research was supported by startup funds from the University of Tennessee. K.F.S. was supported by the NSF Macro-ROA Grant at UT, and S.E.J. was supported by an NSF REU grant at UT. Received for review May 3, 2000. Accepted July 3, 2000. AC0005114

Analytical Chemistry, Vol. 72, No. 18, September 15, 2000

4321