Biomedical Microdevices 3:4, 267±274, 2001
`# 2001 Kluwer Academic Publishers. Manufactured in The Netherlands.
`
`Design and Rapid Prototyping of Thin-Film Laminate-Based
`Micro¯uidic Devices
`
`Abstract. An integrated micro¯uidic design, modeling, and rapid
`prototyping process is presented. It is based on laser cutting and
`lamination of individual thin layers of plastic. The process allows
`the rapid and low-cost manufacturing of both simple and complex
`3-dimensional micro¯uidic ¯ow structures that are routinely
`designed, fabricated, and tested within the space of 24 hours. It
`has yielded micro¯uidic elements and systems, such as mixers,
`separators, and detectors, as well as complete micro¯uidic
`integrated circuits that have been used with complex biological
`samples such as whole blood. Both ``active'', machine-controlled
`micro¯uidic disposables as well as ``passive'', self-contained cards
`that do not require any external instrument are presented. Such
`devices include a disposable hematology analyzer chip, as well as
`blood separation and analysis tools.
`
`Key Words. micro¯uidic, diffusion-based separation and detection,
`microcytometry, rapid prototyping, plastic micro¯uidic disposables
`
`1.
`
`Introduction
`
`system concepts have
`Micro¯uidics-based analysis
`proliferated in the last few years, with the promise of
`enabling miniaturization, faster response times, and
`simpli®cation of analysis procedures. These devices, as
`developed in various research groups, and now beginning
`to be commercialized (e.g., by Caliper Technologies,
`Cepheid, Micronics, and Orchid Biosciences), generally
`consist of a small micro¯uidic chip, surrounded by a
`substantial, typically desk-top-sized analysis instrument
`[1 8]. At this point, almost all commercial applications
`are in non-FDA-regulated life science areas.
`One of the promises of micro¯uidic total-analysis
`systems is the capability to handle all steps of the
`analysis on-chip,
`from sampling, sample-processing,
`separation and detection steps
`to waste handling.
`Integrating all
`these functions on a chip becomes
`considerably more challenging when dealing with
`complex and variable samples such as whole blood or
`other clinical specimen.
`Several Lab-on-a-chip companies have developed
`technologies that work very well for highly predictable
`and homogeneous samples common in genetic testing
`and drug discovery processes [1,2,5]. One of the biggest
`challenges for current Labs-on-a-chip, however, is to
`
`Bernhard H. Weigl, Ron Bardell, Thomas Schulte,
`Fred Battrell, and Jon Hayenga
`Micronics, Inc. Redmond, WA, USA
`E-mail: bernhardw@micronics.net
`
`perform analysis in the presence of the complexity and
`heterogeneity of actual clinical samples such as whole
`blood. Micronics has developed Lab-on-a-Chip technol-
`ogies that can overcome some of those shortcomings.
`They are implemented as low-cost plastic disposable
`integrated micro¯uidic circuits. In this paper, we will
`discuss the micro¯uidic design, modeling, and rapid
`prototyping process that Micronics employs to develop
`these novel integrated micro¯uidic elements and inte-
`grated circuits.
`
`2. Micro¯uidic Circuit Design and Rapid
`Prototyping
`
`Micronics' ORCAmFluidicTM micro¯uidic circuits com-
`prise laminates built of several layers of individually cut
`or stamped ¯uidic circuits. While each layer can be
`manufactured very easily and inexpensively, the lamina-
`tion process yields complex 3-dimensional micro¯uidic
`structures. This allows the design, for example, of 3-
`dimensional hydrodynamic focusing channels for cell
`analysis, or of multiple separate circuits with crossing
`channels on a single card. Figure 1 shows an example of
`a Micronics disposable.
`These disposables are typically credit-card-sized, and
`most structural elements on these cards have dimensions
`ranging from about 100 micrometers to a few milli-
`meters. The laminates can comprise various kinds of thin
`plastic sheets,
`ranging in thickness from about 10
`micrometers to a few hundred micrometers. The layers
`can be bonded using adhesive or thermal bonding
`processes. In some cases, the internal surfaces of the
`laminates are chemically treated (e.g., with oxygen
`plasma) to change their wettability.
`Micronics uses a low-cost rapid prototyping process
`that allows the design and testing of new micro¯uidic
`structures in 24 hours or less. This is, we believe, a great
`advantage over other microfabrication processes such as
`silicon microfabrication, micro-injection-molding or
`chemical etching processes. Figure 2 outlines a typical
`design process.
`
`267
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:25)(cid:26)
`
`

`

`268
`
`Weigl et al.
`
`Fig. 1. Micronics hematology cartridge, designed to determine red cell and platelet counts, hemoglobin concentration, a white cell differential
`count, and various derived parameters.
`
`In conjunction with our rapid prototyping method, we
`use design tools such as the ¯uid modeling package
`CFDRC ACE ‡ , and the circuit simulator SCEPTRE, as
`well as various custom software packages. CFDRC
`ACE ‡ is used for more complex ¯ow situations, where
`the modeling of the ¯ow ®eld is required (such as ¯ow in
`devices with varying cross-sections).
`This package can solve both ¯ow and diffusion of
`multiple constituents of ¯uids with various viscosities.
`However, ¯uid dynamic solutions are computationally
`intensive, so we also use simpli®ed models based on a
`series solution for transient diffusion in cases where the
`¯ow ®eld is already known.
`
`For the modeling of integrated ¯uidic systems, we
`also use the circuit simulator SCEPTRE, which can solve
`non-linear system dynamics problems. This package has
`been used, for example,
`to model
`the ¯uid ¯ow in
`multiple inlet and outlet channels of hydrostatically
`driven devices such as the passive H-Filters and T-
`Sensors described in Section 3B.
`The micro¯uidic devices are then designed in
`individual layers using AutoCAD. For example, a typical
`integrated circuit consists of anywhere from 4 12 layers.
`A single channel requires three layers (top and bottom
`cover, and the cut out channel layer). If the channel
`crosses over another channel, then a minimum of ®ve
`
`Fig. 2. Micronics micro¯uidic development process.
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:25)(cid:27)
`
`

`

`Design and Rapid Prototyping of Micro¯uidic Devices
`
`269
`
`Fig. 3. Housing and laminate stack of Micronics hematology cartridge.
`
`layers is required. The cytometer structure shown in
`Figure 5 requires seven layers to achieve 3-dimensional
`hydrodynamic focusing.
`The so-called ``cut-®le'', a collection of AutoCAD
`design drawings for each individual
`layer,
`is then
`transferred to the prototype laser cutter. This device
`consists of a plotter-like stage, on which 11 in 6 17 in
`sheets of plastic are placed. The CO2 laser head then
`moves, according to the cut-®le drawings, over the
`
`plastic sheet and cuts each individual laminate layer. One
`sheet typically holds 24 credit card sized laminate layers.
`The cut-out layers are then cleaned, and, in some cases,
`treated in an oxygen plasma chamber to change their
`wettability. This allows, for example, undiluted whole
`blood to be introduced into the card by capillary force
`only.
`laminate layers are then
`The individual cutout
`transferred to a registration frame, where, alternately, a
`
`Fig. 4. Macro micro interface, designed to interface micro¯uidic disposables with external samples and reagents.
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:25)(cid:28)
`
`

`

`270
`
`Weigl et al.
`
`Fig. 5. Microcytometer structure, designed to hydrodynamically and geometrically focus particles (e.g., blood cells) into a single ®le, and
`allowing for the laser optical interrogation of these particles.
`
`regular Mylar (or other plastic layer) and a Mylar layer
`with additional adhesive layers on both sides are placed
`on top of each other in order such that the micro¯uidic
`circuit is formed. The fully assembled micro¯uidic card
`is then transferred to a conveyor, where it is pressed
`between two constant-pressure rollers.
`The cards are now ready to be used. This process
`yields prototype micro¯uidic cards with a depth
`tolerance of less than 1 mm, and a width tolerance of
`roughly 10 mm. Typical channel dimensions on these
`cards range from 100 3,000 mm in width, and from 50
`400 mm in depth. These dimensions are selected
`according to the
`requirements of
`the
`structure.
`Typically, lower limits are de®ned by the size of the
`largest particles to be passed through the channel (e.g.,
`white blood cells have diameters of up to 30 mm),
`whereas upper limits are set by the requirements for
`laminar ¯ow, and the need to provide suf®ciently small
`diffusion dimensions between adjacent streams ¯owing
`in parallel next to each other. While other processes such
`as hot embossing, micro-injection molding, and,
`in
`particular, silicon or glass lithographic techniques yield
`signi®cantly better dimensional
`tolerances
`[9],
`the
`described method has its major advantages in turn-
`around time, cost, and the ease of generating 3-
`dimensional structures, as well as incorporating hybrid
`elements into the design (such as electrodes, ®lter
`membranes, sensors, etc.).
`
`3. Results of the Rapid Prototyping Process
`
`3.1. Machine-controlled micro¯uidicsÐ
`MicroCytometryTM technology
`As an example of how our devices can interface with a
`machine-controlled analysis system, we have developed
`a hematology cartridge that is designed to determine red
`
`cell and platelet counts, hemoglobin concentration, a
`white cell differential count, and various derived
`parameters (see Figure 1). For this system, the laminate
`card is inserted into a rigid injection-molded housing that
`comprises functionality such as sample and reagent
`introduction, waste storage, and valve actuation. Figure 3
`shows the laminate and housing. The laminate contains
`all micro¯uidic elements, whereas the injection-molded
`housing contains all ¯uidic macro-micro-interfaces for
`introducing reagents, as well as valve actuator interfaces
`for isolating certain subcircuits on the laminate during
`measurement (Figure 4).
`For this card, a number of sample processing issues
`had to be resolved. This cartridge contains a variety of
`sample preprocessing structures,
`including mixers,
`diluters, and chemical reactors. Micronics is currently
`developing, in cooperation with Beckman Coulter, Inc., a
`point-of-care hematology instrument based on dispo-
`sable micro¯uidic circuits.
`Several drops of whole blood are injected into the
`card. The sample is split into three essentially separate
`circuits. In the hemoglobin circuit, red blood cells are
`lysed, and the extracted hemoglobin is converted to
`cyanmethemoglobin, which is then quanti®ed in an
`optical window element of the circuit.
`In the second circuit, whole blood is diluted (but not
`lysed) and the cells are focused in all three dimensions in
`a microcytometer structure (Figure 5), where red cells
`and platelets are counted and identi®ed using light
`scattering. In the third circuit, whole blood is diluted and
`lysed with soft lyse, which renders red cells invisible to
`the detector but leaves white cells largely unaffected.
`The white cells are focused in a microcytometer and
`classi®ed into subpopulations using multi-angle light
`scatter detectors.
`To determine the exact geometry for the cytometer
`structures, we model the ¯ow of the core stream and the
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:26)(cid:19)
`
`

`

`Design and Rapid Prototyping of Micro¯uidic Devices
`
`271
`
`Fig. 6. Predicted dye concentration in the focusing chamber of a microcytometer structure. The dye enables visualization of the core stream
`during focusing.
`
`surrounding sheath ¯uid in variants of the structure using
`the computational ¯uid dynamics software package
`CFDRC ACE ‡ . To identify the core stream it
`is
`marked with a dye in both computational simulations
`and lab tests of the prototype designs. Good correlation
`between the predictions of the simulations and the lab
`measurements is essential to ensure the physics of the
`¯ow is understood and the prototype designs will
`consistently meet design speci®cations.
`The cytometer structure consists of several ¯ow
`layers, and is symmetrical in both width and height
`dimension. Figure 6 shows the concentration of dye
`predicted by a computational simulation of the focusing
`chamber of the cytometer as the core is focused in both
`dimensions by the surrounding sheath ¯uid.
`Figure 7 shows the predicted velocity vector ®eld,
`illustrating the ¯ow speed in various parts of
`the
`structure. Both images represent
`top views of
`the
`structure. The top image shows the ¯uid velocity in the
`laminate layer in which the core stream is introduced
`(``sample layer''). The bottom image shows the ¯uid
`velocity in the sheath layers,
`located symmetrically
`above and below the sample layer (an intermediate ¯ow
`layer between sheath and sample layer is not shown).
`Note how the sheath ¯uid reorients its ¯ow outward to
`expedite its arrival in the deeper focusing chamber. Both
`images show this design avoids ¯ow separation and
`recirculation regions near
`the walls thus providing
`optimal geometric focusing.
`Figure 8 shows the formation of a core stream in the
`MicroCytometerTM card. In the left
`image the core
`stream is entering the focusing chamber and narrowing
`
`as it is accelerated by the surrounding sheath ¯ow. In the
`right image the core is approaching the ®nal stage of
`focusing into a single ®le of cells. Both geometric
`(structural) and hydrodynamic (¯uid pressure) focusing
`are used for lining up and spacing cells in a single ®le, as
`they ¯ow past a light scatter detector
`for
`further
`characterization. Since the ¯ow is laminar, slightly
`uneven wall surfaces are not signi®cant to the ¯ow,
`thus enabling lower-cost manufacturing techniques.
`
`3.2. Passive micro¯uidicsÐSelf-contained cards for
`diffusion-based micro¯uidic sample preparationÐ
`H-FilterTM and T-SensorTM
`As an example of micro¯uidic devices that do not require
`external
`instrumentation, devices for diffusion-based
`separation and detection are discussed [6]. While these
`devices can be operated with interfaces to external
`pumps as implemented for MicroCytometryTM cards,
`they can also be powered by readily available forces such
`as hydrostatic pressure, capillary action, and absorption
`or evaporation effects [11].
`An example is an absorption-driven disposable
`micro¯uidic detector card based on the T-Sensor
`(diffusion-based detection) method [1,4,8,10,11,13 15]
`(Figure 9) that combines the ease of use of a paper test
`strip with the versatility of a micro¯uidic system. Here,
`an absorption-driven integrated T-Sensor design is
`shown to the left; the ¯uid is initially aspirated using
`capillary action, and then ¯ow-controlled by the
`absorption pad (right). The cartridge is made self-
`wetting by plasma-treating the channel surfaces. A
`sample is put into the top left reservoir, a reagent (e.g.,
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:26)(cid:20)
`
`

`

`272
`
`Weigl et al.
`
`Fig. 7. Predicted velocity vector ®elds in a microcytometer structure. Arrow length and color re¯ect ¯uid velocity magnitude.
`
`Fig. 8. Focusing of blood cells in the MicroCytometer laminate; initial geometric and hydrodynamic core formation (left), and approaching the
`®nal stage of focusing into a single ®le core (right).
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:26)(cid:21)
`
`

`

`Design and Rapid Prototyping of Micro¯uidic Devices
`
`273
`
`an indicator dye) is put into the middle left reservoir, and
`a reference solution with a known concentration of
`analyte is put
`into the lower
`left
`reservoir. The
`comparison of the intensity and position of the two
`diffusion interaction zones allows a semi-quantitative
`analyte determination. The photograph on the right
`shows such a T-Sensor in operation as it determines the
`pH of a buffer solution [4,8,10]. The shape, geometry,
`and directional preference of these absorbent pads can be
`such that it controls the ¯ow speed of the ¯uid that is
`being absorbed by the pad.
`Another passive device (Figure 10) can separate
`chemical compounds by their diffusion coef®cients
`(based on the H-Filter method [4]) and produce several
`microliters of clean sample in one minute for further
`processing. The ®gure shows an H-Filter experiment in
`progress, showing the separation of ¯uorescein from a
`solution containing ¯uorescein and 2,000 kD dextran
`labeled with blue dye. This device, which requires no
`power or moving parts, can be a simple and cheap
`replacement for a centrifuge.
`
`4. Conclusions
`
`rapid prototyping and integrated design
`Micronics'
`process has yielded a number of innovative and diverse
`micro¯uidic devices and elements for microcytometry,
`separation, and detection. It was shown that
`these
`concepts apply to both machine-controled devices for
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:26)(cid:22)
`
`Fig. 9. Absorption driven integrated T sensor device.
`
`Fig. 10. Hydrostatic pressure driven integrated H ®lter design.
`
`

`

`274
`
`Weigl et al.
`
`high performance applications, as well as to self-
`contained passive micro¯uidics devices. A tight integra-
`tion of design, computational modeling, prototype
`testing, and manufacturing has enabled a rapid pace of
`product innovation and an ability to quickly produce
`micro¯uidic-based designs providing innovative solu-
`tions to particular assay requirements.
`
`References
`
`1. L. Bousse, B. Dubrow, and K. Ulfelder, in Micro Total Analysis
`Systems '98, D.J. Harrison and A. van den Berg, (eds), (Kluwer
`Academic Publishers, Dordrecht, 1998), 271.
`2. R. Mathies, P. Simpson, and A. Woolley, in Micro Total Analysis
`Systems '98, D.J. Harrison and A. van den Berg, (eds), (Kluwer
`Academic Publishers, Dordrecht, 1998), 1.
`3. F. Pourahmadi, M. Taylor, G.A. Kovacs, K. Lloyd, S. Sakai, T.
`Schafer, B. Helton, L. Western, S. Zaner, J. Ching, W.A. McMillan,
`P. Belgrader, and M.A. Northrup, Clinical Chemistry 46(9), 1511
`1513 (2000).
`4. B.H. Weigl and P. Yager, Science 283, 346 (1998).
`
`and M.T. Boyce-Jacino,
`5. V.L. Bogdanov, Y.H. Rogers,
`Ultrasensitive Biochemical Diagnostics II, SPIE 2985, 129 137
`(1997).
`6. L.J. Kricka, Clin. Chem. 44, 2008 2014 (1998).
`7. N.H. Chiem and D.J. Harrison, Clin. Chem. 44, 591 598 (1998).
`8. P. Yager, D. Bell, J.P. Brody, D. Qin, C. Cabrera, A. Kamholz, and
`B.H. Weigl, Micro Total Analysis Systems '98, D.J. Harrison and A.
`van den Berg, (eds), (Kluwer Academic Publishers, Dordrecht,
`1998), 207 212.
`9. H. Becker, W. Dietz, and P. Dannberg, in Micro Total Analysis
`Systems '98, D.J. Harrison and A. van den Berg, (eds), (Kluwer
`Academic Publishers, Dordrecht, 1998), 253.
`10. A.E. Kamholz, B.H. Weigl, B.A. Finlayson, and P. Yager, Anal.
`Chem. 71(23) 5340 5347 (1999).
`11. These devices are now available in form of an Evaluation Kit at
`Micronics, Inc.
`12. J.P. Brody, T.D. Osborn, F.K. Forster, and P. Yager, Sensors and
`Actuators (A), A54 Proceedings of Transducers '95, (1±3), 704
`708 (1996).
`J.P. Brody, M.A.
`13. M.R. Holl, P. Galambos, F.K. Forster,
`Afromowitz, and P. Yager, Proc. ASME Meeting 189 195 (1996).
`14. A.E. Kamholz, B.H. Weigl, B.A. Finlayson, and P. Yager, Anal.
`Chem. 71(23), 5340 5347 (1999).
`15. K. MacounovaÂ, C.R. Cabrera, M.R. Holl, and P. Yager, Anal.
`Chem. submitted.
`
`(cid:36)(cid:37)(cid:54)(cid:3)(cid:42)(cid:79)(cid:82)(cid:69)(cid:68)(cid:79)(cid:15)(cid:3)(cid:44)(cid:81)(cid:70)(cid:17)(cid:3)(cid:68)(cid:81)(cid:71)(cid:3)(cid:42)(cid:72)(cid:81)(cid:88)(cid:86)(cid:3)(cid:83)(cid:79)(cid:70)(cid:3)(cid:177)(cid:3)(cid:40)(cid:91)(cid:17)(cid:3)(cid:20)(cid:19)(cid:20)(cid:26)(cid:15)(cid:3)(cid:83)(cid:17)(cid:3)(cid:21)(cid:26)(cid:23)
`
`