Anal. Chem. 1994, 66, 1114-1118
`
`High-Speed Separations on a Microchip
`Stephen C. Jacobson, Roland Hergenroder, Lance B. Koutny, and J. Michael Ramsey*
`Chemical and Analytical Sciences Division, Oak Ridge National Laboratory,
`Oak Ridge, Tennessee 37831-6142
`
`Fast, efficient separations are sought in liquid-phase analyses
`which incorporate a nondiscriminatory sample injection scheme
`and can implement a variety of detection modes. A glass
`microchip device for free solution electrophoresis was fabricated
`using standard photolithographic procedures and chemical wet
`etching. Separations were performed at several separation
`lengths from the injector to the detector with electric field
`strengths from 0.06 to 1.5 kV/cm. For a separation length of
`0.9 mm, electrophoretic separations with baseline resolution
`are achieved in less than 150 ms with an electric field strength
`of 1.5 kV/cm and an efficiency of 1820 plates/s. For a
`separation length of 11.1 mm, a minimum plate height of 0.7
`jtm and a maximum number of plates per second of 18 600
`were achieved.
`
`Microfabricated chemical instruments show great promise
`for the laboratory and as advanced chemical sensors. Micro-
`fabrications of chemical separation techniques have received
`noticeable attention over
`the past several years and have
`included the techniques of gas chromatography,1 liquid
`3456and capillary electrophoresis.3-7 Micro-
`chromatography,1 2
`electronic devices have been able to achieve even faster response
`times in part due to miniaturization. Similar benefits may
`also accrue from miniaturization of some chemical measure-
`ment techniques. The response times of chemical measure-
`ments can often be an important issue, in particular in the
`case of chemical sensing.
`With microfabrication, the performance of many liquid
`separation techniques improves, especially capillary electro-
`phoresis.8-10 For capillary electrophoresis, smaller column
`dimensions enable the power generated to be dissipated more
`efficiently, and as a direct result, separation devices can be
`operated at higher electric field strengths. The efficiency of
`the separations, therefore, improves in two areas. First, Joule
`heating, which leads to thermal gradients within the channel
`and ultimately contributes to the dispersion of the analyte
`
`(1) Terry, S. C.; Jerman, J. H.; Angell, J. B. IEEE Trans. Electron Devices 1979,
`26, 1880.
`(2) Manz, A.; Miyahara, Y.; Miura, J.; Watanabe, Y.; Miyagi, H.; Sato, K. Sens.
`Actuators 1990, Bl, 249.
`(3) Manz, A.; Harrison, J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Ludi,
`H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253.
`(4) Harrison, D. J.; Manz, A.; Fan, Z.; Ludi, H.; Widmer, H. M. Anal. Chem.
`1992, 64, 1926.
`(5) Seiler, K.; Harrison, D. J.; Manz, A. Anal. Chem. 1993, 65, 1481.
`(6) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A.
`Science 1993, 261, 895.
`(7) Jacobson, S. C.; Hergenroder, R.; Koutny, L. B.; Warmack, R. J.; Ramsey,
`J. M. Anal. Chem., preceding paper in this issue.
`(8) Mikkers, F. E. P.; Everaerts, F. M.; Verheggen, Th. P. E. M. J. Chromatogr.
`1979, 169, 11.
`(9) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298.
`(10) Jorgenson, J. W.; Lukacs, K. D. J. Chromatogr. 1981, 218, 209.
`1114 Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
`
`band, is minimized.11 The second and primary advantage is
`that higher field strengths enable separations to be performed
`faster, thereby minimizing the contribution of axial diffusion
`to the dispersion of the analyte band.9 10*To take advantage of
`these two properties, on-column optical gating12 was used to
`injection plugs for high-speed separations.
`define narrow
`However, this application is limited because the analyte must
`be tagged with a fluorophore that can be photobleached to
`implement such a scheme.
`Planar glass microchips have been constructed for elec-
`trophoretic separations using a cross-column geometry.3-6 The
`cross-column geometry is not an elaborate design, but it
`effectively illustrates the power of a microchip separation
`device. For these fast separations, we have fabricated a glass
`microchip which has a cross-column geometry and utilizes a
`nondiscriminatory injection procedure and a fluorescence
`detection scheme. To evaluate the microchip, a series of
`separations were performed at several separation lengths and
`over a range of field strengths employing a binary analyte
`mixture.
`
`EXPERIMENTAL SECTION
`The channels on the microchip substrate were fabricated
`using standard photolithographic techniques and chemical
`wet etching. The micromachined substrate and the cover plate
`were joined using a direct bonding technique as described
`previously.7 8Figure 1 shows a schematic of the column design
`implemented for the experiments. The cross-column geometry
`is the simplest structure that allows injection and separation
`to occur quickly and efficiently. The enclosed column lengths
`are 3.8 mm from the buffer reservoir to the injection cross,
`7.0 mm from the analyte reservoir to the injection cross, 9.8
`mm from the analyte waste reservoir to the injection cross,
`and 17.8 mm from the waste reservoir to the injection cross.
`The dimensions of the channel are 15 pm deep, 94 pm wide
`at the top, and 64 pm wide at the bottom. The dimensions
`were obtained using a profilometer (Alpha-Step 200, Tencor
`Instruments).
`The separations were monitored on-microchip via fluo-
`rescence using an argon ion laser (514.5 nm, Coherent Innova
`90) for excitation and a photomultiplier tube (PMT; Oriel
`77340) to collect the fluorescence signal. The PMT with
`collection optics was situated below the microchip with the
`optical axis perpendicular to the microchip surface. The laser
`was operated at ~20 mW, and the beam impinged upon the
`microchip at a 45° angle from the microchip surface and a
`45° angle from the separation channel. The laser beam and
`
`(11) Hjerten, S. Chromatogr. Rev. 1967, 9, 122.
`(12) Monnig, C. A.; Jorgenson, J. W. Anal. Chem. 1991, 63, 802.
`
`0003-2700/94/0366-1114$04.50/0
`© 1994 American Chemical Society
`
`Downloaded via UNIV OF VIRGINIA on November 6, 2018 at 17:52:07 (UTC).
`
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`
`Agilent Exhibit 1280
`Page 1 of 5
`
`

`

`Figure 2. CCD Image ot
`(shaded area).
`
`'sample loading" mode for rhodamine B
`
`stability of the injection plug.
`Ideally, when the analyte plug
`is injected into the separation column, only the analyte in the
`injection cross would migrate into the separation column. From
`Figure 2, the volume of the injection plug in the injection
`cross is — 120 pL with a plug length of 130 /am. A portion
`of the analyte in the analyte channel and the analyte waste
`channel is drawn into the separation column. Following the
`switch to the separation mode, the volume of the injection
`plug is —250 ± 24 pL with a plug length of 208 ± 19 pm.
`These dimensions are estimated from a series of CCD images
`taken immediately after the switch is made to the separation
`mode.
`One particular advantage to the planar microchip design
`is that with laser-induced fluorescence the point of detection
`can be placed anywhere along the separation column. The
`electropherograms are detected at separation lengths of 0.9,
`I. 6, and 11.1 mm from the injection cross.
`The 1.6- and
`II. 1 -mm separation lengths were used over a range of electric
`field strengths from 0.06 to 1.5 kV/cm, and the separations
`this range. At an electric field
`had baseline resolution over
`strength of 1.5 kV/cm,
`rhodamine B and
`the analytes,
`fluorescein, are resolved in less than 150 ms for the 0.9-mm
`separation length (Figure 3a), in less than 260 ms for the
`1.6-mm separation length (Figure 3b), and in less than 1.6 s
`for the 11.1-mm separation length (Figure 3c). Due to the
`trapezoidal geometry of the channels, the upper corners make
`it difficult to cut
`the sample plug away exactly when the
`potentials are switched from the sample loading mode to the
`separation mode. Thus, the injection plug has a slight tail
`associated with it, and we believe that this effect accounts for
`the tailing observed in some of the separated peaks.
`Applying these separations to real samples depends largely
`on the capacity of the separation and the types of samples.
`The peak capacity with a resolution of 1.0 can be calculated;1}
`I + (TV05/ 4) lnd^/V^)
`
`nc =
`
`(1)
`
`where N is the number of theoretical plates and Kma, and Fmj„
`are the largest and smallest volumes in which zones can be
`eluted and detected. An experimentally viable value for the
`ratio of Vmax to Kmjn for capillary electrophoresis is 5. For
`Figure 3a-c, the number of plates and the peak capacities for
`rhodamine B and fluorescein at the three separation lengths
`are listed in Table 1. Although the peak capacities of 5.5 for
`
`(13) Giddings, J. C. Anal. Chem. 1967, 39, 1027.
`
`Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
`
`1115
`
`Figure 1. Schematic of the microchip with a cross-column geometry.
`The large circle represents the cover plate that forms the closed
`channels.
`
`PMT observation axis were separated by a 135° angle.
`Labview (National Instruments) software was used to monitor
`the ADC board, which was connected to the output of the
`PMT. A charge-coupled device (CCD; Princeton Instruments,
`Inc. TE/CCD-512TKM) camera was used to determine the
`sample plug length. The CCD camera was mounted on a
`stereomicroscope (Nikon SMZ-U), and the microchip was
`illuminated using the argon ion laser operating at 1.5 W with
`the beam expanded to ~20 mm.
`The analytes used for the
`rhodamine B and disodium fluorescein
`experiments were
`(Exciton Chemical Co., Inc.) at 20 #<M. A sodium tetraborate
`buffer (4 mM, pH 9.2) was the buffer in all experiments.
`The microchip was operated under a “pinched sample
`loading” mode and a “separation" mode as described previ-
`ously.7 During the separation mode,
`the analyte and the
`analyte waste reservoirs are maintained at 43% of the potential
`applied to the buffer reservoir.
`
`RESULTS AND DISCUSSION
`The sample loading scheme does not discriminate against
`an analyte because of its net charge as do the electromigration9
`and the photolytic12 injections used for capillary electro-
`phoresis. By loading the sample using a frontal electrophero-
`gram, the injection plug is stoichiometrically representative
`of the solution being analyzed.
`In Figure 2, a CCD image
`displays the flow pattern of the analyte (shaded area) and the
`buffer (white area) through the region of the injection cross.
`By pinching the flow of the analyte, the volume of the analyte
`time.7 The slight asymmetry of the plug
`plug is stable over
`shape is due to the different electric field strengths in the
`buffer channel (470 V/cm) and the separation channel (100
`V/cm) for 1.0 kV applied to the analyte, buffer, and waste
`reservoirs and with the analyte waste reservoir grounded.
`However, the different field strengths do not
`influence the
`
`Agilent Exhibit 1280
`Page 2 of 5
`
`

`

`Table 1. Plata Numbers, Peak Capacities, and Plata* par
`Second for the Electric Field Strength of 1.5 kV/cm*
`plate
`separation
`peak
`no., N
`length, mm
`capacity, nc
`5.5
`120
`0.9
`130
`450
`450
`15100
`11900
`
`compound
`rhodamine B
`fluorescein
`rhodamine B
`fluorescein
`rhodamine B
`fluorescein
`
`1.6
`
`11.1
`
`plates/s
`1820
`1070
`3820
`2080
`18600
`7900
`
`9.5
`
`47.7
`
`0 For electropherograms, see Figure 3.
`
`separations the total plate height can be expressed as
` ^total = H'mj + #det + ^diff
`
`(2)
`
`where Hinj, H&tt, and Hm are the contributions to the plate
`height from the injection plug, the detector path length, and
`the axial diffusion, respectively. The contributions from the
`injection plug length and the detector path length are time
`independent, and the contribution to the plate height from
`axial diffusion is time dependent. Effects such as Joule heating
`were not considered because the power dissipation was well
`below 1 W/m for all experiments.12
`The contributions to the plate height from the injection
`plug length and the detector length are15
`H-mj = jy (12L)
`#det = PdJWL)
`
`(3)
`
`(4)
`
`where /j„j and ket are the lengths of the injection plug and of
`the detector path, respectively, and L is the separation length.
`The lengths of the injection plug and the detector path are
`If
`for all experiments.
`constant
`these 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
`is16
`
`J^diff= 2Dm/u
`
`(5)
`
`where u is the linear velocity of the analyte and Dm is the
`diffusion coefficient of
`the analyte in the buffer. The
`contribution from axial diffusion to the plate height is reduced
`by large linear velocities for the analyte.
`The total plate height can be rewritten17
`= A + B/u
`
`(6)
`
`where A is the sum of the contributions from the injection
`plug length and the detector path length and B is equal to
`2Dm.
`
`(15) Sternberg, J. C. Adv. Chromatogr. 1966, 2, 205.
`(16) Giddings, J. C. Dynamics of Chromatography, Part I: Principles and Theory,
`Marcel Dekker: New York, 1965; Chapter 2.
`(17) van Deemter, J. J.; Zuiderweg, F. J.; Klinkenberg, A. Chem. Eng. Sci. 19S6,
`5, 271.
`
`0
`
`:
`
`.
`
`300
`
`400
`
`100
`
`200
`time [ms]
`.............
`
`rhodamine B
`
`fluorescein 7
`
` 
`
`I
`
`j
`I l
`
`L.
`
`1500
`
`2000
`
`0
`
`500
`
`1000
`time [ms]
`Figure 3. Electropherograms of rhodamine B and fluorescein with a
`separation field strength of 1.5 kV/cm and a separation length of (a,
`top) 0.9, (b, middle) 1.6, and (c, bottom) 11.1 mm.
`the 0.9-mm separation length and 9.5 for the 1.6-mm
`in less
`separation length are not high, the separations occur
`than 150 and 260 ms, respectively. This is appealing when
`only a few species require analysis or when high-speed capillary
`electrophoresis is used as the second dimension to a two-
`dimensional separation.14 If the time duration of the first
`dimension is on the order of tens of seconds to minutes, having
`such a rapid second step enables information to be extracted
`without compromising the resolution obtained from the first
`step.
`In order to improve the peak capacity of the separation,
`the band dispersion needs to be minimized. For electrophoretic
`
`(14) Lemmo, A. V.; Jorgenson, J. W. Anal. Chem. 1993, 65, 1576.
`
`1116 Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
`
`Agilent Exhibit 1280
`Page 3 of 5
`
`

`

`the contribution of the axial dispersion to the total plate height
`is negligible relative to the contributions from the injection
`and detector. The average values obtained from the data in
`Figure 4 (Table 2, term A) compare well with theory for the
`1.6-mm separation length, 3.46 versus 3.42 Mm, but are 50%
`higher than predicted for the 11.1-mm separation length, 0.76
`versus 0.50 Mm. How the column dynamics change down-
`stream from the 1.6-mm observation point is not understood.
`Also, the 11.1-mm separation length is 6.9 times longer than
`the 1.6-mm separation length, and the average time-inde-
`pendent contributions to the plate height should differ by 6.9
`times, but a factor of 4.6 is observed between 3.46 and 0.76
`Mm (see Table 2). Consequently, the separations at the 11.1 -
`mm separation length are not as efficient as expected. The
`efficiencies of the separations at the 0.9- and 1,6-mm separation
`lengths are limited by the lengths of the injection plug and the
`Improvements can be made on both counts by
`detector path.
`channels and by using smaller laser spot sizes.
`using narrower
`The time-dependent contribution of axial diffusion to the
`plate height is the dominant contribution at linear velocities
`below 3 mm/s (see Figure 4). From the B coefficient (see
`Table 2 and eq 6), the diffusion coefficients for each analyte
`can be calculated at
`the two separation lengths. The
`for
`experimentally determined diffusion coefficients, Dm,
`rhodamine B are 1.3 X 10-5 and 8.9 X 10-6 cm2/s for the 1.6-
`and 11.1-mm separation lengths, respectively, and for fluo-
`rescein are 1.2 X 10~5 and 7.8 X 10-6 cm2/s for the 1.6- and
`1.1-mm separation lengths, respectively. The values for each
`analyte at the two separation lengths differ by as much as
`50%, and this error can be attributed to an insufficient number
`of points in the lower liner velocity range. However, the values
`for two analytes at the same separation length compare well,
`1.3 X 10-5 cm2/s for rhodamine B and 1.2 X 10~5 cm2/s
`fluorescein at the 1.6-mm separation length and 8.9 X KH
`cm2/s for rhodamine B and 7.8 X 10-6 cm2/s for fluorescein
`at the 11.1 -mm separation length. The diffusion coefficients
`obtained at the 11.1 -mm separation length are consistent with
`diffusion coefficients for molecules of similar mass.
`An important measure of the utility of a separation system
`is the number of plates generated per unit time:18
`N/t = L/(Ht)
`
`(7)
`
`Combining eqs 6 and 7 and substituting u = pE where p is
`the effective electrophoretic mobility of the analyte and E is
`the electric field strength, the plates per unit time can be
`expressed as a function of the electric field strength:
`N/t = (pEf/iApE + B)
`
`(8)
`
`At low electric field strengths, when axial diffusion is the
`dominant form of band dispersion, the term ApE is small
`relative to B, and consequently, the number of plates per second
`increases with the square of the electric field strength. As the
`electric field strength increases, the plate height approaches
`a constant value, and the plates per unit time increase linearly
`with the electric field strength because B is small relative to
`ApE.
`
`(18) Giddings, J. C. Sep. Sci. 1969, 4, 181.
`
`Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
`
`1117
`
`Figure 4. Variation of the plate height as a function of the linear
`velocity for rhodamine B at separation lengths of 1.6 (circle) and 11.1
`mm (square) and for fluorescein at separation lengths of 1.6 (diamond)
`and 11.1 mm (triangle). Lines are eq 6 fitted to the experimental data
`with the resulting coefficients listed In Table 2.
`
`Table 2. Plate Height Coefficients Determined Using Equation 6
`Fitted to the Data in Figure 4
`separation
`length, mm
`1.6
`
`A, mm
`0.003 46
`0.003 46
`0.000 716
`0.000 802
`
`B, mm2/s
`0.002 59
`0.002 47
`0.001 79
`0.001 56
`
`compound
`rhodamine B
`fluorescein
`rhodamine B
`fluorescein
`
`11.1
`
`In Figure 4, the plate height is plotted versus
`the linear
`velocity of the analyte. As expected, the plate height decreases
`with increasing linear velocity until a constant value is obtained.
`Equation 6 is fitted to the experimental data, and the resulting
`coefficients are listed in Table 2. The time-independent
`contributions to the plate height, A, reach average constant
`values of 3.46 and 0.76 *im for the 1.6- and 11.1-mm separation
`lengths, respectively. The longer separation length of 11.1
`mm is expected to have a lower plate height contribution from
`the time-independent events because the only variable is the
`separation length (see eqs 3 and 4). Also, no adsorption of
`the analytes to the column wall
`is observed. Otherwise, a
`linear increase in the plate height with increasing linear
`velocities would be seen due to a resistance to mass transfer.
`For these experiments the injection plug length, /inj, was
`208 Mm and the detector path length, i.e., laser spot size, kel,
`was 150 Mm. The contributions to the plate height for the
`injection plug length calculated using eq 3 are 4.81 Mm for the
`0.9-mm separation length, 2.25 Mm for the 1,6-mm separation
`length, and 0.33 Mm for the 11.1-mm separation length. The
`contributions to the plate height for the detector path length
`calculated using eq 4 are 2.50 Mm for the 0.9-mm separation
`length, 1.17 Mm for the 1.6-mm separation length, and 0.17
`pm for the 11.1-mm separation length. The sums of these
`time-independent contributions are 7.31, 3.42, and 0.50 Mm
`for the 0.9-, 1.6-, and 11.1 -mm separation lengths, respectively.
`The average plate height for the separation at the 0.9-mm
`separation length in Figure 3a is 7.25 Mm, which is comparable
`to the calculated value for the time-independent contributions
`of 7.31 Mm. For analyte linear velocities greater than 3 mm/
`the time-independent contributions to the plate height
`s,
`predominate as expected from eq 6 and seen in Figure 4; i.e.,
`
`Agilent Exhibit 1280
`Page 4 of 5
`
`

`

`format of plate height versus linear velocity, small deviations
`between the experimental and fitted data are exaggerated
`when presented in the format of plates per unit time versus
`electric field strength.
`
`CONCLUSIONS
`This work demonstrates advantages of microfabricating
`capillary electrophoresis devices. Micromachining allows four
`channels to be joined at an intersection with essentially no
`dead volume. The spatial extent of the injection plug is limited
`to the intersection volume by electroosmotic flow in all four
`channels. These well-defined injections have permitted
`extremely high speed separations with the greatest number of
`plates per second yet reported by capillary electrophoresis.19
`This injection scheme, and thus high-speed separation, is not
`limited to fluorescent species; any type of sample amenable
`In addition,
`to electrophoretic separation can be analyzed.
`this injection approach does not introduce mobility-based bias
`as with the optically gated approach. The results presented
`indicate that microchip capillary electrophoresis devices can
`provide very fast response times for sensor applications.
`In
`addition, microfabricated devices such as demonstrated here
`incorporated into two-dimensional separation strategies would
`appear attractive.
`
`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 by an appointment for S.C.J. to the
`Alexander Hollaender Distinguished Postdoctoral Fellowship
`Program sponsored by the U.S. DOE, for L.B.K. to the DOE
`Laboratory Cooperative Postgraduate Research Training
`Program, and for R.H. to the ORNL postdoctoral research
`associate program. These programs are administered by the
`Oak Ridge Institute for Science and Education and ORNL.
`The authors acknowledge many useful discussions with Drs.
`Roswitha Ramsey and Georges Guiochon.
`
`Received for
`1994.®
`
`review October 5, 1993. Accepted January 10,
`
`•Abstract published in Advance ACS Abstracts, February 15, 1994.
`
`electric field strength [kV/cm]
`Figure 5. Variation of the number of plates per unit time as a function
`of the electric field strength for rhodamine B at separation lengths of
`1.6 (circle) and 11.1 mm (square) and for fluorescein at separation
`lengths of 1.6 (diamond) and 11.1 mm (triangle). Lines are calculated
`from eq 8 using the coefficients listed in Table 2.
`
`The values of the plates per second are approximately twice
`as high for rhodamine B than for fluorescein. For the electric
`field strength of 1.5 kV/cm (see Figure 3 for electrophero-
`grams), the number of plates per second for rhodamine B and
`for fluorescein is listed in Table 1. The number of plates for
`the two analytes is nearly the same, but the retention time of
`the rhodamine B is approximately half that of the fluorescein.
`In Figure 5, the number of plates per second for the 1.6- and
`11.1 -mm separation lengths is plotted versus the electric field
`strength, and the number of plates per second quickly becomes
`a linear function of the electric field strength, because the
`plate height approaches a constant value (see Figure 4). The
`symbols in Figure S represent the experimental data collected
`for the two analytes at the 1.6- and 11.1-mm separation lengths.
`The lines are calculated using eq 8 and the coefficients from
`Table 2. A slight deviation is seen between the experimental
`data and the calculated numbers for rhodamine B at the 11.1 -
`mm separation length. This is primarily due to experimental
`error. Also, when eq 6 is fitted to the data in Figure 4 in the
`
`(19) Since submission of the manuscript, the following article appeared, producing
`a greater number of plates per second using the optical gating injection
`technique: Moore, A. W.; Jorgenson, J. W. Anal. Chem. 1993, 65, 3550.
`
`1118 Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
`
`Agilent Exhibit 1280
`Page 5 of 5
`
`