Anal. Chem. 1995, 67, 4184-4189
`
`Microchip Separations of Neutral Species via
`Micellar Electrokinetic Capillary Chromatography
`Alvin W. Moore, Jr., Stephen C. Jacobson, and J. Michael Ramsey*
`Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, P. 0. Box 2008, Building 4500S, MS 6142,
`Oak Ridge, Tennessee 37831-6142
`
`Micellar electrokinetic capillary chromatography (MECC)
`of three neutral coumarin dyes was performed on glass
`microchips. Manifolds of channels for analyte injection
`and separation were machined into one surface of the
`glass substrates using standard photolithographic, etch-
`ing, and deposition techniques. Coverplates were then
`these channels to form capillaty
`directly bonded over
`networks, with fluid flow in these networks controlled by
`varying the applied high-voltage potentials at the outlets.
`The separation capillaty was 16.5 cm long for a serpentine
`channel chip and 1.3 cm long for a straight channel chip.
`Detection of analyte zones was accomplished by laser-
`induced fluorescence using the UV lines (~350 nm) of
`an argon ion laser. At low applied electric field strengths,
`MECC analyses with on-chip injections gave high repro-
`ducibilities in peak areas and migration times (<1% for
`two of the three coumarins) and near constant separation
`efficiencies throughout the analysis. At high fields (> 400
`V/cm), analysis times were shorter, but separation ef-
`ficiency decreased at later migration times. These peaks
`showed significant broadening, consistent with mass
`transfer effects.
`
`Many of today’s applications of analytical chemistry involve
`collection of samples from some remote site, transportation of
`these samples back to the laboratory, and analysis using benchtop
`instrumentation. These situations range from detection of pol-
`lutants in water supplies to process monitoring for quality control.
`In the typical analytical laboratory, a universal detection system
`such as a UV absorption detector and a separation technique such
`as liquid chromatography (LC) or capillary electrophoresis (CE)
`are used to quantify the analyte of interest. The separation
`method provides the selectivity between components and classes
`If the analysts’
`of components that the universal detector lacks.
`needs change, such that they must quantify other species, the
`selectivity can be easily modified by changing the separation
`conditions such as the pH of the mobile phase or buffer, the
`organic solvent content, or the elution gradient. The enhanced
`versatility of this method comes at the expense of requiring that
`samples be collected and transported to the lab.
`An attractive means of saving time and expense would be to
`take the instrument to the sample. Micromachining technology
`may allow the ultimate miniaturization of chemical measurement
`instrumentation. Liquid phase separation devices are particularly
`amenable to miniaturization because analytical separation perfor-
`mance often improves with decreasing size of the components
`involved. CE is a relatively new separation method which has
`4184 Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`used microcolumn capillaries from its inception. Because CE
`gives high-efficiency separations but requires no high-pressure
`pump or gas supply, it is especially well suited for miniaturization.
`Substantial progress has already been made in this area.1 23™4
`A problem with CE for general analysis is its inability to
`separate uncharged species. All neutral species in a particular
`sample will have zero electrophoretic mobility and thus the same
`migration time. Previously, our group has investigated open-
`channel electrochromatography (OCEC) on a microchip as a
`means of separating neutrals.5 In this previous work, sample
`components were separated by their partitioning interaction with
`a stationary phase coated on the channel walls. The mobile phase
`was driven not by a conventional pump but by electroosmotic flow.
`Micellar electrokinetic capillary chromatography (MECC) is a
`operational mode of CE which was developed by Terabe et al.6 7to
`address the separation of neutrals by CE. A surfactant such as
`is added to the CE buffer in
`sodium dodecyl sulfate (SDS)
`sufficient concentration to form micelles in the buffer.
`In a typical
`experimental arrangement, the micelles move much more slowly
`toward the cathode than does the surrounding buffer solution.
`The partitioning of solutes between the micelles and the sur-
`rounding buffer solution provides a separation mechanism similar
`to that of LC.
`In this work, MECC was performed in a capillary etched into
`the surface of a glass chip. All of the fluidic manipulation
`necessary for sample injection and analysis was done through a
`manifold of channels, with flow controlled by the voltage applied
`the end of each channel. Laser-induced
`to the reservoir at
`fluorescence detection was used to monitor the separations. Some
`results of this implementation of MECC will be shown, and some
`of its advantages and disadvantages for chip-based microinstru-
`mentation will be discussed.
`
`OVERVIEW
`An in-depth treatment of the theory of MECC has been given
`by others6™8 and is beyond the scope of this paper. Only a brief
`
`(1) Harrison, D. J.; Manz, A; Fan, Z.; Liidi, H.; Widmer,  . M. Anal. Chem.
`1992, 64, 1926-1932.
`(2) Seiler, K; Harrison, D. J.; Manz, A Anal. Chem. 1993, 65, 1481-1488.
`(3) Jacobson, S. C.; Hergenrtider, R.: Koutny, L. B.; Warmack, R J.; Ramsey, J.
`M. Anal. Chem. 1994, 66, 1107-1113.
`(4) Jacobson, S. C.; Hergenrtider, R; Koutny, L. B.; Ramsey, J. M. Anal. Chem.
`1994, 66, 1114-1118.
`(5) Jacobson, S. C.; Hergenrtider, R; Koutny, L. B.; Ramsey, J. M. Anal. Chem.
`1994, 66, 2369-2373.
`(6) Terabe, S.; Otsuka, K; Ando, T. Anal. Chem. 1985, 57, 834-841.
`(7) Sepaniak, M. J.; Powell, A C.; Swaile, D. F.; Cole, R 0. In Capillary
`Electrophoresis: Theory and Practice·, Grossman, P. D„ Colburn, J. C., Eds.;
`Academic Press, Inc.: San Diego, CA 1992; pp 159—89.
`(8) Sepaniak, M. J.; Cole, R. 0. Anal. Chem. 1987, 59, 472-476.
`
`0003-2700/95/0367-4184$9.00/0 &copy; 1995 American Chemical Society
`
`Downloaded via UNIV OF VIRGINIA on November 6, 2018 at 18:16:15 (UTC).
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`Agilent Exhibit 1277
`Page 1 of 6
`
`

`

`review will be given here. For neutral solutes, the separation
`mechanism is essentially chromatographic and can therefore be
`described with modified chromatographic relationships. The
`capacity factor, k', for a given solute and experimental conditions
`is the ratio of the total moles of solute in the stationary phase
`(here, the pseudostationary phase of the micelles) to those in the
`is modified to account for the
`mobile phase. For MECC, k'
`movement of the micelles,
`
`k' =
`
`^o(l ""
`
`*rAji)
`
`= TO
`
`(i)
`
`where ír is the solute retention time, k is the “void time” (retention
`time for a solute moving at the rate of the electroosmotic flow),
`tm is the micelle retention time (i.e„ retention time of a completely
`retained solute), Kis the partition coefficient, 7S is the volume of
`the micellar phase, and 7m is the volume of the mobile phase. If
`the micelles were indeed stationary, tm would become infinite and
`the equation would reduce to its conventional form for chroma-
`tography.
`Because to is the retention time for a solute moving with the
`electroosmotic flow, and tm is that of a solute completely retained
`in the micelles, neutral solutes must elute between to and tm.
`Resolution can be improved by increasing the difference between
`to and tm, thus increasing the “window” of time over which sample
`components can elute. Resolution can also be improved by
`changing the k' values of the solutes, which affects both the
`retention and the selectivity of the separation. This can be done
`in many of the same ways it is done in liquid chromatography,
`such as by adjusting the temperature, buffer concentration, or pH,
`or by the addition of organic modifiers. Organic solvents such
`as methanol and acetonitrile can have multiple effects in that they
`may modify the electroosmotic flow in the system or alter the
`hydrophobic interactions between solute and micelle, but they
`also affect micelle structural interactions and so alter the partition-
`ing kinetics. Changes in the organic solvent content of the buffer
`also modify the electroosmotic flow in the system.
`
`EXPERIMENTAL SECTION
`The microchips were constructed using standard photolitho-
`graphic, wet chemical etching and bonding techniques described
`previously.9 An ordinary soda lime glass microscope slide was
`used as a substrate, into which was etched a network of open
`channels (see Figure 1). The serpentine channel geometry
`(Figure la) allowed use of a longer separation capillary within a
`small area, while the straight channel geometry (Figure lb)
`allowed experiments with high applied electric field strengths.
`closed with thin coverplates,
`These channel networks were
`to form capillary networks.
`directly bonded to the substrate,
`Cylindrical glass reservoirs were then bonded with epoxy to the
`capillary outlets. The lengths of the capillary channels may vary
`with the placement of the coverplate. Figure 1 gives the channel
`lengths for the two chips used in this work. Use of the wet
`chemical etch on a glass substrate results in an isotropic etch.
`That is, the glass etches at the same rate in all directions, and
`the resulting channel geometry is trapezoidal.5 The channel cross
`the same for both chips used in this
`section dimensions were
`
`(9) Jacobson, S. C.; Koutny, L. B.; Hergenroder, R.; Moore, A W„ Jr.; Ramsey,
`J. M. Anal. Chem. 1994, 66, 3472-3476.
`
`Figure 1. Schematic diagrams of microchip geometry. The large
`outer boxes show the outlines of the glass substrates.
`Individual
`capillary lengths are measured from the point of Injection to the point
`where the capillary passes beyond the cover plate Into the respective
`reservoir, (a) The serpentine channel geometry used for the majority
`lengths as follows: analyte channel, 9.2
`of the work, with channel
`mm; buffer channel, 6.6 mm; analyte waste channel, 7.7 mm; waste
`(b) The straight channel geometry
`(separation) channel, 171 mm.
`used for the high-field experiments, with channel lengths as follows:
`analyte channel, 8.0 mm; buffer channel, 4.6 mm; analyte waste
`channel, 9.8 mm; waste (separation) channel, 17.2 mm.
`
`work. The channels were 10 µ   deep, 60 µ   wide at the bottom,
`and 80 µ   wide at the top. These dimensions were measured
`with a profilometer (Alpha-Step 200, Tencor Instruments, Moun-
`tain View, CA) after bonding the coverplate but before attaching
`the reservoirs.
`Separation of the analyte zones was monitored via laser-induced
`fluorescence using an argon ion laser (351.1-363.8 nm, all lines;
`~50 mW; Coherent Innova 90, 5 W; Palo Alto, CA) for excitation.
`The UV beam from the laser was prefiltered with a Coming 7-37
`colored glass filter (blue, narrow bandpass; transmittance, 30% at
`360 nm) to reduce plasma emissions and then focused at the
`desired point on the capillary with a planoconvex fused silica lens
`(focal length, 100 mm; Newport Corp., Irvine, CA). The laser
`impinged on the chip at an angle of 45° with the surface. The
`fluorescence signal was collected from below the chip by a 21 x
`microscope objective (Bausch & Lomb Opt. Co., Rochester, NY),
`filtered with a Coming 3-73 colored glass filter (yellow, sharp cut-
`on at 426 nm), and detected with a photomultiplier tube (PMT;
`Oriel 77340, Stratford, CT). The PMT current was amplified and
`converted to a proportional voltage with a Keithley 617 electrom-
`eter (Keithley Instruments Inc., Cleveland, OH). The analog
`voltage output of the electrometer was measured with a multi-
`function interface board (AT-MIO-16X, National
`Instruments,
`Austin, TX) controlled by software written in-house using Lab-
`VIEW 3.0 for Windows (National Instruments, Austin, TX) on a
`PC compatible computer. Separation efficiencies were obtained
`by calculation of peak statistical moments, also in LabVIEW.
`Platinum electrodes in each reservoir provided electrical
`contact between the buffer solutions and the CE high-voltage
`power supply (CZE1000R, Spellman Inc., Plamview, NY). A
`
`Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`4185
`
`Agilent Exhibit 1277
`Page 2 of 6
`
`

`

`voltage divider/relay apparatus described earlier10 was used to
`set the relative voltages applied to each reservoir and to switch
`between run and inject modes under computer control. For both
`chip geometries, the separation voltage was applied over the entire
`length of the capillary channel for all analyses. However, for
`individual groups of analyses, the length of the separation channel
`used was set by the location of the point of detection.
`The reservoirs at the end of each capillary channel had a
`volume of only 200 µ . To avoid problems with changes in buffer
`concentration due to evaporation, the reservoirs were sealed with
`thin rubber septa held in place with parafilm. The platinum
`electrodes could be inserted through the septa if the septa was
`initially pierced with a syringe needle. The reservoirs could be
`flushed and filled a in similar manner with a syringe of buffer or
`In use, the seal formed around the electrodes was almost
`sample.
`gas-tight. The resolution between peaks could be maintained from
`run to run for several hours before significant evaporation effects
`were noticed. Only the data in Figure 3, as mentioned below,
`were acquired with open buffer reservoirs.
`The analytes used in these experiments were the neutral dyes
`coumarin 440 (C440), coumarin 450 (C450), and coumarin 460
`Individual stock solutions of each dye were
`(C460, Exciten Inc.).
`prepared in methanol and then diluted in the analysis buffer before
`use. The concentration of each dye was ~50 µ  unless indicated
`otherwise. The substrate is glass rather than quartz, so there is
`substantial background fluorescence where the UV laser strikes
`the chip. No attempt was made to measure limits of detection
`with the present system. The MECC buffer was composed of 10
`mM sodium borate (pH 9.1), 50 mM SDS, and 10% (v/v) methanol.
`The methanol aids in solubilizing the coumarin dyes in the
`aqueous buffer system and also affects the partitioning of some
`of the dyes into the micelles. Due care must be used in working
`with coumarin dyes as the chemical, physical, and toxicological
`properties of these dyes have not been fully investigated.* 11
`Sample Injection. The microchips were operated in a
`“pinched injection” mode described previously.3 The voltages
`applied to the reservoirs are set to either an “inject” (sample
`In the inject mode,
`loading) or a “run” (separation) configuration.
`a frontal chromatogram of the solution in the analyte reservoir is
`pumped electroosmotically through the intersection and into the
`analyte waste reservoir. Voltages applied to the buffer and waste
`reservoirs also cause weak flows into the intersection from the
`sides and then into the analyte waste reservoir. These flows serve
`to confine the stream from the analyte reservoir to give a well-
`defined plug of sample in the intersection. The chip remains in
`this mode until the slowest moving component of the sample has
`passed through the intersection. At this point, the sample plug
`in the intersection is representative of the analyte solution, with
`no electrokinetic bias.
`An injection is made by switching the chip to the run mode,
`which changes the voltages applied to the reservoirs such that
`buffer now flows from the buffer reservoir through the intersec-
`tion, into the separation column, and eventually into the waste
`reservoir. The plug of sample which was previously in the
`intersection is swept onto the separation column. Proportionately
`lower voltages are applied to the analyte and analyte waste
`reservoirs to cause a weak flow of buffer from the buffer reservoir
`
`(10) Jacobson, S. C.; Hergenroder, R; Moore, A. W„ Jr.; Ramsey, J. M. Anal.
`Chem. 1994, 66, 4127-4132.
`(11) Exciton, Inc., Dayton, OH, 1991.
`
`4186 Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`Figure 2. Microchip MECC analysis of a mixture of coumarin dyes.
`The small peak at 4.6 min is an unidentified impurity in one of the
`dyes. Concentration of each dye was ~50 µ . Other analysis
`conditions are given in the text.
`
`into these channels. These flows ensure that the sample plug is
`cleanly “broken off” from the analyte stream and that no excess
`analyte leaks into the separation channel during the analysis. This
`sample loading method is time-independent (after the initial time
`necessary to pump all components through the intersection),
`nonbiased, and reproducible.
`
`RESULTS AND DISCUSSION
`The results of an MECC analysis of a mixture of C440, C450,
`and C460 are shown in Figure 2. The peaks were identified by
`individual analyses of each dye. The migration time stability of
`the first peak, C440, with changing methanol concentration (see
`below) was a strong indicator that this dye did not partition into
`the micelles to a significant extent. Therefore, it was considered
`an electroosmotic flow marker with migration time to. The last
`peak, C460, was assumed to be a marker for the micellar migration
`time, tm. Using these values of to and tm from the data in Figure
`is 0.43. This agrees well
`2, the calculated elution range, to/tm,
`with a literature value of to/tm = 0.4 for a similar buffer system8
`and supports our assumption. This analysis was done in the
`serpentine chip with a separation length of 21.3 mm and an applied
`electric field strength of 47 V/cm.
`Buffer Evaporation. Figure 3 shows several successive
`analyses of the same three component sample over a period of
`30 mb. These analyses were done with an applied field of 96
`V/cm and a separation length of 17 mm. Notice that from parts
`a to c of Figure 3, the C440 peak and the C450/C460 peak pair
`tend to move closer together (the elution range decreases), and
`the resolution between the C450 and C460 peaks gradually
`decreases. The high voltage was switched off for 5 min between
`parts b and c of Figure 3, so that if gradual heating of the MECC
`buffer m the chip were responsible for the lost resolution, it should
`have been partially restored in Figure 3c. Obviously, this was
`Immediately before the analysis shown in Figure
`not the case.
`3d, the buffer and sample solutions were replaced. Notice that
`the resolution between the last two peaks has been restored,
`indicating some change in the buffer solutions during the course
`of the analyses shown in Figure 3a-c.
`For the data in Figure 3, the reservoirs at the end of each
`capillary channel were open to the surrounding air. When the
`
`Agilent Exhibit 1277
`Page 3 of 6
`
`

`

`Table 1. Peak Parameters*
`
`migration
`time (s)
`
`209.1
`1.94
`0.93
`
`384.6
`3.27
`0.85
`
`438.1
`5.76
`1.31
`
`peak
`area
`
`C440
`2.82
`0.03
`0.91
`C450
`4.92
`0.05
`0.92
`C460
`7.62
`0.17
`2.21
`
`theoretical
`plates (A)
`
`HETP
`(µ  )
`
`3523
`60
`1.71
`
`3855
`80
`2.09
`
`3705
`101
`2.73
`
`6.05
`0.10
`1.71
`
`5.53
`0.12
`2.09
`
`5.75
`0.16
`2.78
`
`av (  = 5)
`SD
`%RSD
`
`av (  = 5)
`SD
`%RSD
`
`av {  = 5)
`SD
`%RSD
`
`“ Data for five replicate analyses with applied electric field of 47
`V/cm, separation length of 21.3 mm.
`
`Table 2. Ratio Peak Parameters*
`peak area ratio
`migration time ratio
`C460/
`C440/
`C440/
`C460/
`C450
`C450
`C450
`C450
`
`capacity
`factor (kr),
`C45Ó
`
`av (  = 5)
`SD
`%RSD
`
`0.544
`0.01
`1.57
`
`1.14
`0.01
`0.54
`
`0.574
`0.01
`1.11
`
`1.55
`0.02
`1.51
`
`6.89
`0.05
`0.78
`
`“ Ratio values and capacity factor for C450 calculated from the data
`used in Table 1.
`
`generally even higher. Here, for the C440 and C450 peaks, peak
`the %RSD values for
`area %RSD values are <1%. For C460,
`migration times and peak areas are 1.3 and 2.2%, respectively. Such
`low values are presumably due to the on-chip injection scheme;
`all fluid handling for sample injection and analysis is done on-
`chip with no moving parts (other than the buffer itself). Thus,
`of variance associated with benchtop
`many of
`the sources
`instruments, such as those resulting from moving the separation
`capillary between vials, are minimized.
`To better correct for environmental variances, peak migration
`time and peak area ratios may be calculated relative to an internal
`standard peak. These ratios help to account for differences in
`flow rate and injection volume from run to run. The %RSD for
`ratio data is usually <2%.7 In Table 2, we have calculated ratio
`values for C440 and C460 considering C450 an internal standard,
`based on the same raw data used to construct Table 1. As
`expected, our %RSD values for these ratios are also <2%.
`Alternately, some have calculated capacity factors, k', for the
`analytes of interest. Because the partitioning into the micelles is
`similar to the partitioning found in liquid chromatography, the k'
`values are distinctive for a given compound under given experi-
`mental conditions.
`If we consider the migration time of C440 to be that of the
`electroosmotic flow, and that of C460 to be that of the micelles,
`then a k' value for C450 can be calculated for each analysis. The
`%RSD for the k' value of C450 is also below 1%, which compares
`well to a value of 0.5% found in the literature.13
`For a given applied electric field (47 V/cm, as above), and
`consequently, a constant plate height, the number of theoretical
`plates should increase as the length of the separation capillary
`injection to point of detection)
`(the distance from point of
`increases. Figure 4 is a plot of efficiency data gathered from
`
`Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`4.187
`
`Figure 3. Effects of methanol evaporation in microchip MECC. The
`times indicated in the figure are times between the ends of the
`analyses, immediately after acquisition when the data were saved.
`Peaks are identified as in Figure 2.
`
`buffers were entirely aqueous, buffer evaporation was not a
`significant problem. However, the MECC buffer contained 10%
`(v/v) methanol, so methanol evaporation might be significant. A
`decrease in the methanol concentration would cause the C450 to
`prefer the free buffer solution less and partition to a greater extent
`into the micelles. That is, the partition coefficient of the C450
`would increase, and the selectivity between C450 and C460 would
`decrease, lowering the resolution between them. Also, addition
`of methanol is known to extend the elution range in MECC,7 so
`the decrease seen in Figure 3a-c agrees well with a decrease in
`methanol concentration due to evaporation. Weinberger and
`Lurie12 have suggested that similar evaporation problems with
`MECC buffers containing organic modifiers contribute to poor
`run-to-run reproduciblity.
`Reproducibility, The results for five replicate analyses of the
`three coumarin dyes mix as in Figure 2 are shown in Table 1.
`The migration times, peak areas, and peak variances were
`obtained from the best-fit values for Gaussian curves fitted to the
`experimental data. The percent
`relative standard deviation
`(%RSD) of the migration times and peak areas are slightly lower
`than expected for MECC. Some authors have reported %RSD of
`less,712 with some as high as 10% over
`migration times of 2% or
`the course of a day,13 while %RSD values for peak area are
`
`(12) Weinberger, R.: Lurie, I. Anal. Chern. 1991, 63, 823—827.
`(13) Northrop, D. M.; Matire, D. E.; MacCrehan, W. A Anal. Chem. 1991, 63,
`1038-1042.
`
`Agilent Exhibit 1277
`Page 4 of 6
`
`

`

`14-
`
`Capillary Length (mm)
`Figure 4. Plot of peak efficiency versus
`length of separation
`capillary. Each point represents the average of three analyses, with
`error bars above and below the average at 1 SD. See text for further
`description.
`
`triplicate analyses at four different capillary lengths. The dotted
`lines are the best-fit lines for each of the components (C440, C450,
`and C460). As expected, at each capillary length, the measured
`peak efficiencies for each of the three compounds are similar. The
`linear increase in efficiency as the separation length increases
`demonstrates the homogeneity of the capillary walls over
`the
`length of the separation.
`MECC at High Applied Field Strengths. The straight
`channel chip (Figure lb) was used to study the effects of applied
`electric field strength on the MECC separations. Because the
`separation capillary in the straight chip is much shorter, signifi-
`cantly higher applied fields can be obtained for the same total
`In our present experimental arrangement,
`applied voltages.
`applied voltages of >4 kV begin to give arcing between the buffer
`reservoirs, so studies of higher applied field strengths are best
`done with shorter separation capillaries.
`A single analyte, C450, was used for the range of applied fields
`studied. C450 was chosen because it is partially retained by the
`micelles, and thus its behavior should be indicative of free solution
`and micellar effects.
`In Figure 5a, average analyte velocity is
`plotted versus applied electric field strength, with a best-fit line
`through the four points at the lowest field strengths. Theoreti-
`cally, the analyte velocity should increase linearly with applied
`field. The curve upward seen in the actual data indicates Joule
`heating effects at the higher applied fields. This is not unexpected,
`considering the high ionic strength of the MECC buffer. Analyses
`also done at 1340 V/cm, but the C450 migration time
`were
`continually drifted to earlier times. The precision is greatest for
`the points at low field. This is due both to the lack of heating
`its point of
`thermal
`effects, so that
`the capillary is nearer
`equilibrium, and to the longer migration times, any variations in
`which will be a smaller percentage of the average value.
`In Figure 5b, the average plate height for these analyses is
`plotted against the average analyte velocity. Error bars are shown
`at the 1 standard deviation (SD) points for both dimensions
`because both are calculated from the experimental data. The data
`give a Van Deemter-like plot, similar to that found in LC. The
`plate height at 0.9 mm/s is high relative to the trend of the other
`points and is probably the result of a systematic error
`in the
`analyses at that point. With this point omitted, a curve fit of the
`Van Deemter equation to the experimental data gave an Is- value
`of 0.9990.
`The general form of the Van Deemter equation is
`
`4188 Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`Figure 5.
`(a) Plot of analyte velocity versus applied electric field
`strength. Each point represents the average of three analyses, with
`error bars above and below the average at 1 SD. The fitted data are
`calculated from a linear regression fit of the four data points at the
`lowest applied fields. See text for further description, (b) Plot of plate
`height versus analyte velocity. Each point represents the average of
`three analyses, with error bars In both dimensions at plus and minus
`1 SD. The point at 0.9 mm/s Is omitted from the data fit to the Van
`Deemter equation. The fit equation is of the form  
`= A + B/u + Cu,
`where HK = A, H\ = B/u, and (Hmc + Hep) = Cu, to give hh0, = Hec
`+ H\ + Hmc + Hep. From the curve fit, A = 2.32 µ \, B = 0.141 («m2V
`s, and C = 0.651 s. For this fit,  2 = 0.0063 and r2 = 0.9990. See
`text for further description.
`H = A + B/u + Cu
`
`(2)
`
`where H is the theoretical plate height, u is the velocity of the
`mobile phase, A represents band broadening effects independent
`of u (such as eddy diffusion in LC), B represents band broadening
`effects inversely dependent on u (such as longitudinal diffusion),
`and C represents band broadening effects directly dependent on
`u (such as the mass transfer term in LC).
`For MECC, the major causes of on-column band broadening
`have been studied by Terabe et al.14 and Sepaniak and Cole.8 The
`mobile phase velocity represented as u above is given by the
`electroosmotic flow velocity veo in MECC. The A term can include
`all constant additions to the plate height, such as extracolumn
`effects of the injector and detector, because with free solution CE
`or MECC, there is no eddy diffusion term. This is written as HK
`and is independent of the electroosmotic flow velocity. The B
`term is a longitudinal diffusion term as in LC which is inversely
`proportional to the electroosmotic flow velocity, written as H\. The
`C term for MECC includes two terms, Hmc and Hev. The sorption/
`desorption kinetics term, Hmc, is the plate height due to the rate
`at which solutes move into and out of the micelles.
`In most cases,
`
`(14) Terabe, S.; Otsuka, K; Ando, T. Anal. Chem. 1989, 61, 251-260.
`
`Agilent Exhibit 1277
`Page 5 of 6
`
`

`

`microchip which employed a serpentine column geometry.5 For
`the same three dyes were analyzed here on a
`comparison,
`microchip of the serpentine configuration as well as on a straight
`channel chip. As would be expected, the elution order was the
`same as that found with OCEC.
`In Figure 2, the three dyes are
`baseline resolved in 8 min. The longer analysis time results from
`the lower applied field strength for the long serpentine channel.
`In Figure 6, with the straight channel and resultant higher applied
`fields, the analysis time is shorter than for OCEC. The MECC
`resolution is better than that of the OCEC, and the peak shape is
`significantly improved. At low applied fields,
`the separation
`the course of the analysis is relatively constant
`efficiency over
`for MECC. As mentioned above, at high fields, peaks for later-
`eluting sample components show more broadening due to the
`higher capacity factors of those components. Even with these
`high-field effects, however, the efficiency of the later peaks does
`not decrease as rapidly as in the OCEC analysis.
`The MECC has the advantage that
`the stationary phase
`(pseudostationary phase) is continuously replaced as fresh buffer
`is brought into the chip.
`In addition, the capacity factor of solutes
`can be varied by changing the micelle concentration.7 This
`In OCEC, the
`effectively changes the “stationary phase loading”.
`stationary phase is bonded to the capillary surface before use, so
`the stationary phase loading is fixed. Also, as with other bonded
`LC columns, the performance of the bonded stationary phase in
`time with repeated use.
`the microchip degrades over
`As a microchip technique, MECC gives better than average
`reproducibility for neutral species and is relatively simple to
`implement. The same chip used for CE can be used for MECC
`with only a change in buffer. Analysis times can be decreased
`by using high applied fields, while further addition of organic
`modifiers such as methanol or acetonitrile would increase the rate
`of solute exchange kinetics and so sharpen later peaks. Use of
`higher micelle concentrations would limit the contribution of
`micelle polydispersity and thus would also help sharpen later
`peaks. Higher micelle concentrations would also give higher
`electrophoretic currents, and thus more problems with Joule
`heating of the buffer, but these could be circumvented by the
`use of narrower
`channels in the chip. Finally, evaporation
`problems would be eliminated in an actual portable instrument
`by using sealed reservoirs.
`
`ACKNOWLEDGMENT
`This research was sponsored by U.S. Department of Energy
`(DOE), Office of Research and Development Oak Ridge National
`Laboratory is managed by Lockheed-Martin Energy Systems, Inc.,
`for the U.S. Department of Energy under Contract DE-AC05-
`840R21400. Also, this research was sponsored in part by an
`appointment for AW.M. to the ORNL Postdoctoral Research
`Associates Program. These postdoctoral programs are adminis-
`tered by the Oak Ridge Institute for Science and Education and
`ORNL.
`
`Received for review June 21, 1995. Accepted August 18,
`1995.®
`AC950629Y
`
`® Abstract published in Advance ACS Abstracts, October 1, 1995.
`
`Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
`
`4189
`
`Figure 6. High-field MECC analysis of a mixture of coumarin dyes.
`Analysis conditions are given in the text. As described in the text,
`the analysis time is much shorter, but the separation efficiency is no
`longer constant over
`the course of the analysis. The number of
`theoretical plates calculated for each peak was as follows: C440, N
`= 8800; C450, N = 3900; C460, N = 3200.
`
`fíne is small, but it can be significant if the sorption/desorption
`kinetics are very slow, such as with ionic interactions between
`the solute and the micelle. The electrophoretic dispersion term,
`fíep, is the plate height due to the variance in micelle size (micelle
`polydispersity). The micelle solution gives a distribution of micelle
`sizes with a standard deviation of as much as 20% from the
`average. Both fímc and fíep increase with increasing veo. Thus
`fíec + fí, + fímc + fíep
`
`fíTot =
`
`(3)
`
`where HTot is the total plate height.
`From the fit equation in Figure 5b, fíec is the major fraction of
`the total plate height This indicates a large constant contribution
`to the plate height, such as from extracolumn effects. A detailed
`examination of these effects was