Anal. Chem. 1994, 66, 4127-4132
`
`Precolumn Reactions with Electrophoretic
`Analysis Integrated on a Microchip
`Stephen C. Jacobson, Roland Hergenroder, Alvin W. Moore, Jr., and J. Michael Ramsey*
`Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Building 4500S, MS 6142,
`Oak Ridge, Tennessee 37831-6142
`
`A glass microchip was constructed to perform chemical
`reactions and capillaiy electrophoresis sequentially. The
`channel manifold on the glass substrate was fabricated
`using standard photolithographic, etching, and deposition
`techniques. The microchip has a reaction chamber with
`a 1 nL reaction volume and a separation column with a
`separation length. Electrical control of the
`15.4 mm
`buffer, analyte, and reagent streams made possible the
`precise manipulation of the fluids within the channel
`manifold. The microchip was operated under a continu-
`ous reaction mode with gated injections to introduce the
`reaction product onto the separation column with high
`reproducibility (<1.8% rsd in peak area). The reaction
`and separation performances were evaluated by reacting
`amino acids with o-phthaldialdehyde to generate a fluo-
`rescent product which was detected by laser-induced
`fluorescence. Control of the reaction and separation
`conditions was sufficient to measure
`reaction kinetics and
`variation of detection limits with reaction time. Half-times
`of reaction of 5.1 and 6.2 s and detection limits of 0.55
`and 0.83 fmol were measured for arginine and glycine,
`respectively.
`
`The microfabrication of analytical instrumentation will poten-
`tially enable the laboratory to be transported to the samples rather
`than vice versa. Miniature chemical instrumentation can be based
`on conventional laboratory approaches to chemical measurement
`problems. Many laboratory-based procedures require sample
`manipulation prior to the actual measurement, and many analyses
`are fully automated to circumvent operator bias. Similarly, a
`portable instrument should have these sample preparatory steps
`incorporated into the design and function of the instrument.
`In
`instruments as opposed to
`addition, micromachined chemical
`single-analyte chemical sensors should be able to identify and
`quantify all members of a desired class of compounds using a
`single device. The approach taken is to micromachine a channel
`manifold in a monolithic device that includes all desired fluid
`manipulation for complete chemical analysis. Chemical separation
`techniques including capillary electrophoresis,1-6 free-flow elec-
`
`(1) Harrison, D. J.; Manz, A.; Fan, Z.; Liidi, H.; Widmer, H. M. Anal. Chem.
`1992, 64, 1926.
`(2) Manz, A; Harrison, J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A; Liidi,
`H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253.
`(3) Seiler, K; Harrison, D. J.; Manz, A Anal. Chem. 1993, 65, 1481.
`(4) Harrison, D. J.; Fluri, K; Seiler, K; Fan, Z.; Effenhauser, C. S.; Manz, A.
`Science 1993, 261, 895.
`(5) Jacobson, S. C.; Hergenroder, R; Koutny, L. B.; Warmack, R J.; Ramsey,
`J. M. Anal. Chem. 1994, 66, 1107.
`
`0003-2700/94/0366-4127$04.50/0 &copy; 1994 American Chemical Society
`
`trophoresis,1 2345*7 *open channel electrochromatography,8 and capillary
`electrophoresis with postcolumn derivatization9 have been dem-
`onstrated using microfabricated devices.
`To demonstrate a reactor/analyzer microchip, we chose the
`relatively simple reaction of amino acids with o-phthaldialdehyde
`(OPA), which yields a fluorescent product.10 For capillary elec-
`trophoresis, both pre- and postseparation labeling using OPA have
`been studied.11 The reaction is moderately fast (a half-time of
`s 12), but the
`reaction for alanine at room temperature of
`fluorescent product can be short lived («10 min for glycine13).
`This reaction is fast enough to perform postcolumn labeling when
`the analysis time is long compared to the reaction time, but as
`analysis times decrease, the postcolumn labeling becomes less
`attractive. Both detection limits and separation efficiency suffer
`due to “slow” product formation when high-speed analysis is
`performed.9 Analysis times on microchip electrophoresis devices
`can be significantly faster than the OPA reaction times; separation
`times for microchip electrophoresis as short as 150 ms have been
`reported.6 Precolumn derivatization becomes the desired method
`if the electrophoretic analysis can be performed before the product
`degrades and without hindering the quality of the separation.
`In this paper, a simple channel manifold was fabricated on a
`glass microchip to perform on-line precolumn reactions coupled
`with electrophoretic analysis of the reaction products. Here, the
`reactor is operated continuously with small aliquots introduced
`periodically onto the separation column to be analyzed. The
`operation of the microchip consists of three elements: derivati-
`zation of the amino acids with OPA, injection of the sample onto
`the separation column, and separation/detection of the compo-
`nents of the reactor effluent. Each of these elements is evaluated.
`
`EXPERIMENTAL SECTION
`The microchips were fabricated using standard photolitho-
`graphic, wet chemical etching and bonding techniques.14 A
`
`(6) Jacobson, S. C.; Hergenroder, R; Koutny, L. B.; Ramsey, J. M. Anal. Chem.
`1994, 66, 1114.
`(7) Raymond, D. E.; Manz, A; Widmer, H. M. Anal. Chem. 1994, 66, 2858.
`(8) Jacobson, S. C.; Hergenroder, R; Koutny, L. B.; Ramsey, J. M. Anal. Chem.
`1994, 66, 2369.
`(9) Jacobson, S. C.; Koutny, L. B.; Hergenroder, R; Moore, A. W., Jr., Ramsey,
`J. M. Anal. Chem. 1994, 66, 3472.
`(10) Roth, M. Anal. Chem. 1971, 43, 880.
`(11) For example: Rose, D.J.; Jorgenson, J. W.J. Chromatogr. 1988,447,117.
`Pentoney, S. L, Jr.; Huang, X.; Burgi, D. S.; Zare, R N. Anal. Chem. 1988,
`60, 2625. Nickerson, B.; Jorgenson, J. W.J. Chromatogr. 1989, 480, 157.
`Albin, M.; Weinberger, R; Sapp, E.; Moring, S. Anal. Chem. 1991, 63,
`417.
`(12) Butchner, E. C.; Lowry, O. H. Anal. Biochem. 1976, 76, 502.
`(13) Lindroth, P.; Mopper, K Anal. Chem. 1979, 51, 1667.
`(14) For example: Ko, W. H.; Suminto, J. T. In Sensors; Gopel, W., Hasse, J.,
`Zemmel, J. N., Eds.; VCH: Weinhein, Germany, 1989; Vol. 1, pp 107-168.
`Analytical Chemistry, Vol. 66, No. 23, December 1, 1994 4127
`
`Downloaded via UNIV OF VIRGINIA on November 6, 2018 at 17:53:41 (UTC).
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`Agilent Exhibit 1283
`Page 1 of 6
`
`

`

`Figure 1. Schematic of the microchip with integrated precolumn
`reactor. The reaction chamber is 2 mm long, and the separation
`column is 15.4 mm long.
`
`photomask was fabricated by sputtering chrome (50 nm) onto a
`glass slide and ablating the microchip design (Figure 1) into the
`chrome film using a CAD/CAM excimer laser machining system
`(ArF, 193 nm; Resonetics, Inc.). The column design was then
`transferred onto the substrates using a positive photoresist
`(Shipley 1811). The channels were etched into the substrate in
`a dilute, stirred HF/NH4F bath. To form the closed network of
`channels, a cover plate was bonded to the substrate over
`the
`etched channels by use of a direct bonding technique.5 Cylindrical
`glass reservoirs were affixed on the substrate using epoxy.
`Platinum electrodes provided electrical contact from the power
`supply (Spellman CZE1000R) to the solutions in the reservoirs.
`The reaction chamber was designed to be wider than the
`separation column to give lower electric field strengths in the
`reaction chamber and thus longer residence times for the
`reagents. Figure 2a shows a scanning electron microscope (SEM)
`image of the injection cross and the lower portion of the reaction
`chamber prior to bonding of the cover plate. The image has the
`same orientation as the microchip schematic in Figure 1. The
`reaction chamber is 96 /im wide at half-depth and 6.2 /<m deep,
`and the separation column is 31 /4m wide at half-depth and 6.2
`«m deep. Figure 2b shows an SEM image of the cross section of
`the separation channel. The microchip was broken perpendicular
`to the separation column ~1 mm downstream from the injection
`cross. The isotropic etch of the glass substrate, i.e., uniform etch
`in all directions, is evident in the trapezoidal geometry of the cross
`section.
`Column performance and separations were monitored on-
`microchip using a single-point detection scheme via laser-induced
`fluorescence (LIF). An argon ion laser
`(351.1 nm, 10 mW;
`Coherent Innova 90) was used for excitation and focused to a spot
`onto the microchip using a lens (100 mm focal
`length). The
`fluorescence signal was collected using a 20x objective lens,
`followed by spatial filtering (0.6 mm diameter pinhole) and spectral
`filtering (440 nm bandpass, 10 nm bandwidth), and measured
`using a photomultiplier tube (PMT; Oriel 77340). The schematic
`
`4128 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
`
`(a, top) Scanning electron microscope (SEM) image of
`Figure 2.
`is the lower portion of the
`injection cross. The broader channel
`is 6.2 /<m deep and 31
`reaction chamber. The separation channel
`//m wide, and the reaction chamber is 6.2 nm deep and 96 /<m wide.
`The SEM is in the same orientation as Figure 1. (b, bottom) SEM
`section of separation column after bonding of
`image of cross
`coverplate to the substrate.
`
`in Figure 3a shows the data acquisition/voltage switching ap-
`paratus, which is computer controlled using programs written in-
`house in Labview 3.0 (National Instruments). The compounds
`used for the experiments were arginine (0.48 mM), glycine (0.58
`mM), and o-phthaldialdehyde (5.1 mM; Sigma Chemical Co.). The
`buffer in all of the reservoirs was 20 mM sodium tetraborate with
`2% (v/v) methanol and 0.5% (v/v) 2-mercaptoethanol. 2-Mercap-
`is added to the buffer as a reducing agent for the
`toethanol
`derivatization reaction.10
`Figure 3b shows the ratio of the potentials at each of the
`reservoirs for a given potential applied to the system. This
`configuration allowed the lowest potential drop across the reaction
`chamber (25 V/cm for 1.0 kV applied to the microchip) and the
`highest across the separation column (300 V/cm for 1.0 kV applied
`to the microchip) without significant bleeding of the product into
`the separation column. The voltage divider used to establish the
`potentials applied to each of the reservoirs had a total resistance
`of 100 with 10 MQ divisions. The sample and reagent are
`electroosmotically pumped into the reaction chamber with a
`the solutions from the
`volumetric ratio of 1:1.06. Therefore,
`analyte and reagent reservoirs are diluted by a factor of %2. Buffer
`
`Agilent Exhibit 1283
`Page 2 of 6
`
`

`

`Figure 3.
`(a, top) Diagram of the high-voltage switching apparatus
`and detection/data acquisition system, (b, bottom) Block diagram of
`flow pattern for precolumn reactor microchip with the relative potentials
`applied at each reservoir (arrows depict direction of flow).
`
`was simultaneously pumped by electroosmosis from the buffer
`reservoir toward the waste and analyte waste reservoirs. This
`buffer stream prevents the newly formed product from bleeding
`into the separation column. A gated injection scheme, described
`previously,9 is used to inject effluent from the reaction chamber
`onto the separation column. The potential at the buffer reservoir
`is simply floated (opening of the high-voltage switch in Figure
`3a) for a brief period of time (0.1—1.0 s), and sample migrates
`into the separation column as in an electrokinetic injection.15 To
`break off the injection plug, the potential at the buffer reservoir
`is reapplied (closing of the high-voltage switch in Figure 3a). The
`length of the injection plug is a function of both the time of the
`injection and the electric field strength. With this configuration
`of applied potentials, the reaction of the amino acids with the OPA
`continuously generates fresh product to be analyzed.
`
`RESULTS AND DISCUSSION
`Fluid manipulation, including sample injections, is performed
`by simply controlling the potentials applied to the reservoirs; i.e.,
`no valves or pumps are required. This allows reagents to be
`mixed accurately and samples to be injected reproducibly onto
`the separation column. Using the gated injection scheme de-
`scribed above, Figure 4 shows four injections of glycine derivatized
`with OPA for increasing injection times of 0.2, 0.4, 0.6, and 0.8 s
`with an injection field strength of 0.6 kV/cm. The injection field
`strength is the electric field strength in the separation column
`during the injection; i.e., the high-voltage switch in Figure 3a is
`open. Similar profiles were obtained for arginine. The injections
`were observed in the separation column 0.05 mm downstream
`from the injection cross using the single-point detection scheme.
`The peak fronts coincide as expected, and the durations of the
`injection profiles correspond to the injection times.
`
`(15) Jorgenson, J. W.; Lukacs, K D. Anal. Chem. 1981, 53, 1298.
`
`time [s]
`Figure 4. Series of injection profiles for injection times of (a) 0.2,
`(b) 0.4, (c) 0.6, and (d) 0.8 s with £j„j = 0.6 kV/cm and Ubs = 0.05
`
`20
`
`15
`
`T3
`
`5
`
`2 % -> 
`0
`
`0.0
`
`0.2
`
`1.0
`
`0.6
`0.8
`0.4
`injection time [s]
`Figure 5. Reproducibility of injections as percent relative standard
`deviation (% rsd) of the peak area with injection time for arginine ( )
`and glycine (•) at £jnj = 1.2 kV/cm and for arginine (a.) and glycine
`( ) at £jnj = 0.6 kV/cm with Ubs = 0.05 mm. Dashed line represents
`2% rsd.
`
`A significant shortcoming of many capillary electrophoresis
`experiments has been the poor reproducibility of the injections.
`Here, because the microchip injection process is computer
`controlled, and the injection process involves the opening of a
`single high-voltage switch, the injections can be accurately timed
`events. Figure 5 shows the reproducibility of the amount injected
`(percent relative standard deviation, % rsd, for the integrated areas
`for both arginine and glycine at
`injection field
`of the peaks)
`strengths of 0.6 and 1.2 kV/cm and injection times ranging from
`0.1 to 1.0 s. For injection times of >0.3 s, the percent relative
`standard deviation is below 1.8%. This is comparable to reported
`values for commercial, automated capillary electrophoresis instru-
`ments.16 However, injections made on the microchip are «100
`times smaller in volume, e.g., 100 pL on the microchip versus 10
`nL on a commercial instrument. Part of this fluctuation is due to
`the stability of the laser, which is «0.6%. For injection times of
`
`(16) For example, Hewlett-Packard reports 1-2% rsd for peak area measure-
`ments.
`
`Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
`
`4129
`
`Agilent Exhibit 1283
`Page 3 of 6
`
`

`

`Figure 6. Variation of injection widths with injection time for arginine
`( ) and glycine (•) at Enj = 1.2 kV/cm and for arginine (a) and glycine
`( ) at Enj = 0.6 kV/cm with Lobs = 0.05 mm. Lines represent best fit
`for injection times of >0.2 s for arginine and >0.3 s for glycine.
`
`>0.3 s, the error appears to be independent of the compound
`injected and the injection field strength.
`In Figure 6, the injection plug lengths at half-height are plotted
`versus the injection time. For injection times of >0.2 s for arginine
`and >0.3 s for glycine at the two selected injection field strengths,
`the lengths of the injected plugs are a linear function of the
`injection times and injection field strengths. Because the injection
`is electrophoretically biased, and arginine has a higher electro-
`phoretic mobility than glycine, injection widths are linear at shorter
`injection times for arginine. The injection widths are not propor-
`to the injection times for shorter injection times. The
`tional
`region has a finite width, and once filled with
`injection cross
`analyte, the area must be swept clean after the injection. Lines
`with a y-intercept equal to zero are fitted to the data for injection
`times of >0.2 s for arginine and >0.3 s for glycine. The slopes of
`the lines are 3.13 and 1.51 mm/s for arginine with injection field
`strengths of 1.2 and 0.6 kV/cm, respectively, and 1.44 and 0.692
`mm/s for glycine with injection field strengths of 1.2 and 0.6 kV/
`cm, respectively. For both arginine and glycine, the slopes for
`the injection field strength of 1.2 kV/cm are 2 times greater than
`at 0.6 kV/cm. For the experiments described below, the injector
`was operated in the reproducible (Figure 5) and linear ranges
`(Figure 6), and the injection times were scaled to give comparable
`peak widths and, thus, injection volumes for separations performed
`at different separation field strengths. The electrophoretic bias
`in the injections requires the injection time to be long enough to
`inject accurately the compound with the lowest electrophoretic
`mobility, e.g., glycine. This, in turn, means compounds with a
`higher electrophoretic mobility, e.g., arginine, are injected in
`relative excess.
`Figure 7 shows the overlay of three electrophoretic separations
`of arginine and glycine after on-microchip precolumn derivatization
`with OPA with a separation field strength of 1.8 kV/cm and a
`separation length of 10 mm. The separation field strength is the
`electric field strength in the separation column during the
`separation; i.e., the high voltage switch in Figure 3a is closed.
`The field strength in the reactor is 150 V/cm. The reaction times
`
`4130 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
`
`time [s]
`Figure 7. Overlay of
`three electropherograms of arginine and
`glycine using precolumn derivatization with o-phthaldialdehyde with
`Esep = 1.8 kV/cm and Lsep = 10 mm.
`
`for the analytes are inversely related to their mobilities; e.g., for
`arginine the reaction time is 4.1 s and for glycine the reaction
`time is 8.9 s. The volumes of the injected plugs were 150 and 71
`pL for arginine and glycine, respectively, which correspond to 35
`and 20 fmol of the amino acids injected onto the separation
`column. The gated injector allows rapid sequential injections to
`In this particular case, an analysis could be performed
`be made.
`every 4 s. The observed electrophoretic mobilities for the
`compounds are determined by a linear fit to the variation of the
`linear velocity with the separation field strength. The slopes were
`29.1 and 13.3 mm2/ (kV-s) for arginine and glycine, respectively.
`No evidence of Joule heating was observed, as indicated by the
`field strength data. A linear fit
`linearity of the velocity versus
`produced correlation coefficients of 0.999 for arginine and 0.996
`for glycine for separation field strengths from 0.2 to 2.0 kV/cm.
`With increasing potentials applied to the microchip, the field
`strengths in the reaction chamber and separation column increase.
`This leads to shorter residence times of the reactants in the
`reaction chamber and faster analysis times for the products. By
`the reaction
`varying the potentials applied to the microchip,
`kinetics can be studied. The variation in amount of product
`generated with reaction time is plotted in Figure 8. The response
`is the integrated area of the peak corrected for the residence time
`in the detector observation window and photobleaching of the
`product (discussed below). The offset between the data for the
`arginine and the glycine in Figure 8 is due primarily to the
`difference in the amounts injected, i.e., different electrophoretic
`mobilities, for the amino acids. A 10-fold excess of OPA was used
`to obtain pseudo-first-order reaction conditions. The slopes of the
`lines fitted to the data correspond to the rates of the derivatization
`reaction. The slopes are 0.13 s_1 for arginine and 0.11 s_1 for
`glycine, corresponding to half-times of reaction of 5.1 and 6.2 s,
`respectively. These half-times of reaction are comparable to the
`4 s previously reported for alanine.12 We have found no previously
`reported data for arginine or glycine.
`The reaction time varies with the potential applied to the
`microchip. For lower field strengths in the reaction chamber, the
`reaction time is longer and should yield more product and larger
`In Figure 9, the mass detection limits (signal-to-noise
`signals.
`ratio (S/N) equal to 2) are estimated for a series of reaction times
`for arginine and glycine. For longer reaction times, the mass
`detection limits improve as expected. Detection limits of 0.55 fmol
`
`Agilent Exhibit 1283
`Page 4 of 6
`
`

`

`0.0
`
`S = kIoFNe~At/r
`
`(1)
`
`where k is a geometrical factor, I is the intensity of the laser (~400
`W/cm2), oF is the fluorescence cross section, N is the number of
`molecules in the detection volume, Af is the detector residence
`time (0.01-0.1 s for arginine and 0.02-0.25 s for glycine), and r
`is the photochemical lifetime, k, /, oF, and N are constant for the
`experiments. The residence time of the product in the detection
`volume is known, and thus, the photochemical lifetime, r, of the
`reaction product can be estimated by use of a linear fit of the
`variation of In S with At. The photochemical lifetimes are 51 ms
`for arginine and 58 ms for glycine under these experimental
`conditions. The products are known to have poor stability,13 and
`not surprisingly, the photochemical lifetimes are short compared
`to the longer observation times. Accurate quantitative measure-
`ments must take account of the photobleaching effect.
`To improve the peak capacity of the separation,
`the band
`dispersion needs to be minimized. For electrophoretic separations
`the total plate height can be expressed as
`^total = ^inj + #det + #diff
`
`(2)
`
`where Hmi, Ha&, and Hm are the contributions to the plate height
`from the injection plug length, the detector observation length,
`and axial diffusion, respectively. The contributions from the
`injection plug length and the detector observation length are time
`independent, and the contribution to the plate height from axial
`diffusion is time dependent. Effects such as Joule heating are
`not considered because the variation of the linear velocity with
`the separation field strength is linear, as mentioned above.
`The contributions to the plate height from the injection plug
`length and the detector observation length are17
`
`Hh =
`
`(jnj2/ (16Lsep)
`
`tfdet =
`
`/det2/(16LSep)
`
`(3)
`
`(4)
`
`where lm) and ldet are the lengths of the injection plug and detector
`observation, respectively, and Lsep is the separation length. The
`contributions from the injection plug length (see injection profiles
`in Figure 4) and detector observation length are written as
`Gaussian functions. The lengths of the injection plug and the
`If these
`detector observation are constant for all experiments.
`time-independent contributions predominate in their contribution
`to the plate height, then the total plate height decreases as the
`separation length increases.
`The contribution to the plate height from axial diffusion is18
`Hm = 2DJ{uE)
`
`(5)
`
`where pi is the effective electrophoretic mobility of the analyte, E
`is the electric field strength, and Dm is the diffusion coefficient of
`the analyte in the buffer. The contribution from axial diffusion
`to the plate height is reduced by increasing the separation field
`strength and, consequently, reducing the analysis time.
`
`(17) Sternberg, J. C. Adv. Chromatogr. 1966, 2, 205.
`(18) Giddings, J. C. Dynamics of Chromatography, Part I: Principles and Theory,
`Marcel Dekker: New York, 1965; Chapter 2.
`
`Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
`
`4131
`
`reaction time [s]
`Figure 8. Variation of product formation (In [response]) with reaction
`time for arginine (•) and glycine ( ) derivatized with o-phthaldialde-
`hyde. Lines represent linear fits. Error bars are ±o for three runs.
`
`Figure 9. Variation of detection limits (S/N = 2) with reaction time
`for arginine (•) and glycine ( ) derivatized with o-phthaldialdehyde.
`Error bars are ±o for three runs.
`
`for arginine and 0.83 fmol for glycine are reached for reaction
`times of 36 and 80 s, respectively. The difference in detection
`limits between arginine and glycine is assumed to be differences
`in relative fluorescence and in separation efficiency. The source
`of the limiting background for this experiment is fluorescence
`from the glass substrate. Detection limits could be improved by
`using fused silica substrates, which would minimize the back-
`ground fluorescence.
`Photobleaching can affect quantitative results in fluorescence
`measurements. The degree of photobleaching was estimated by
`determining the photochemical lifetimes of the fluorescent prod-
`ucts. The fluorescence signal was measured as a function of the
`residence time in the detector observation window by running a
`frontal electropherogram of the individual products. The fluo-
`rescence signal, S, for the continuous stream measurements is
`
`Agilent Exhibit 1283
`Page 5 of 6
`
`

`

`especially at the channel walls. As seen in the SEM image in
`Figure 2a, the separation channel walls have a noticeable rough-
`ness. These etch defects derive primarily from the imperfections
`in the photomask. The photomask that was generated using UV
`ablation of a thin chrome film had an edge smoothness of ± 1
`wide line. This roughness in the mask is then
`fxm on a 25
`transferred to the photoresist during exposure and, subsequently,
`In Figure 10, the plate height increases
`etched into the substrate.
`If related to the surface
`with increasing electric field strength.
`roughness, the magnitude of such a contribution is difficult to
`quantitate since the exact nature of the channel surface is not
`known. Prior work at comparable separation field strengths
`yielded more favorable results and agrees with eq 2.6 However,
`in that work, an e-beam written photomask, which had edge
`smoothness more than 1 order of magnitude better, was used for
`the microchip fabrication. The threshold of roughness that can
`be tolerated for the photomask and thus the channel wall is under
`further investigation.
`
`CONCLUSION
`Chemical reactions followed by analysis of the products have
`been demonstrated on a single microchip. Devices can be tailored
`to accommodate any such reaction to be performed in a miniatur-
`ized environment. The reactor volumes and separation lengths
`can be optimized accordingly. Electroosmotically driven flow
`enables buffer, analyte, and reagent streams to be controlled
`precisely in the channel manifold of the microchip without the
`use of valves or pumps. The gated injector allows continuous
`loading of sample and rapid sequential injections to be made. Low
`dead volume connections between channels can be easily fabri-
`cated allowing pre- and postseparation reactions to be incorporated
`on-microchip. Having an electrode placed at the injection cross
`would allow independent control of the potentials in the reaction
`chamber and separation column. To have a long reaction time
`for the chemical reaction,
`the field strength in the reaction
`chamber should be minimized, but for capillary electrophoresis,
`the resolution between compounds increases with increasing
`electric field strength. With the current design, a compromise
`must be reached with the potentials applied to the microchip so
`that the OPA has sufficient time to react with the amino acids
`without seriously impeding the quality of the separation.
`
`ACKNOWLEDGMENT
`This research was sponsored by the U.S. Department of
`Energy (DOE), Office of Research and Development. Oak Ridge
`National Laboratory is managed by Martin Marietta Energy
`Systems, Inc. for the U.S. Department of Energy under contract
`DE-AC05-840R21400. Also, this research was sponsored in part
`for S.C.J.
`to the Alexander Hollaender
`by an appointment
`Distinguished Postdoctoral Fellowship Program sponsored by the
`U.S. DOE and for A.W.M. and R.H. to the ORNL Postdoctoral
`Research Associates Program. These postdoctoral programs are
`administered by the Oak Ridge Institute for Science and Education
`and ORNL.
`
`Received for review August 1, 1994. Accepted September
`26, 1994.®
`
`Abstract published in Advance ACS Abstracts, October 15, 1994.
`
`separation field strength [kV/cm]
`Figure 10. Variation of plate height with the separation field strength
`for arginine (•) and glycine ( ) derivatized with o-phthaldialdehyde
`with Lsep =
`10 mm. The contribution to the plate height from the
`injection and detection is calculated from eqs 3 and 4 for arginine
`(a) and glycine ( ). Error bars are ±o for three runs.
`
`In Figure 10, the total plate height versus
`the electric field
`strength is plotted along with the calculated contributions from
`the injection plug length and the detector observation length. The
`time-independent contributions differ for arginine and glycine
`because the injections have an electrophoretic bias, and conse-
`quently, compounds with a greater electrophoretic mobility will
`have a larger contribution to the plate height. The injection plug
`lengths at half-height were 0.80 and 0.37 mm for arginine and
`glycine, respectively, which correspond to
`equal to 11.6 and
`2.5 /<m. These contributions are substantial, but in order to run
`the injector in the linear, reproducible regime for all separation
`field strengths, longer injection times were required. The detector
`observation length, i.e., the laser spot size, was «50 /xm, which
`In the range of separation
`corresponds to Hi# equal to 21 nm.
`field strengths used, the contribution from axial dispersion is small
`relative to the other contributions to the total plate height and is
`not discussed further. Unfortunately, as the field strength
`the total plate height increases. This behavior is
`increases,
`reminiscent of liquid chromatography, where mass transfer in the
`mobile and/or stationary phases dominates at high mobile phase
`velocities. The contribution from mass transfer to the plate height
`is a linear function of the mobile phase velocity:
`Hmt = Cu
`
`(6)
`
`where
`
`(7)
`u=
`fxE
`u is the linear velocity of the mobile phase and C is the coefficient
`of mass transfer. Typically, in capillary electrophoresis, contribu-
`tions to the plate height from mass transfer are neglected because
`there should be no interaction of the analyte with the walls of the
`capillary. However, in Figure 10 the magnitude of this contribu-
`tion is surprisingly large especially for small molecules which tend
`to have little or no interaction with the walls.
`An alternative explanation might be that the source of band
`broadening is due to inhomogeneity of the channel surface,
`
`4132 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994
`
`Agilent Exhibit 1283
`Page 6 of 6
`
`