Clinical Chemistry 51:10
1923–1932 (2005)
Oak Ridge Conference
Microfluidic Tool Box as Technology Platform for
Hand-Held Diagnostics
Michael J. Pugia,1* Gert Blankenstein,2 Ralf-Peter Peters,2 James A. Profitt,1
Klaus Kadel,2 Thomas Willms,2 Ronald Sommer,1 Hai Hang Kuo,1 and
Lloyd S. Schulman1
Background: Use of microfluidics in point-of-care testing (POCT) will require on-board fluidics, self-contained reagents, and multistep reactions, all at a low
cost. Disposable microchips were studied as a potential
POCT platform.
Methods: Micron-sized structures and capillaries were
embedded in disposable plastics with mechanisms for
fluidic control, metering, specimen application, separation, and mixing of nanoliter to microliter volumes.
Designs allowed dry reagents to be on separate substrates and liquid reagents to be added. Control of
surface energy to ⴞ5 dyne/cm2 and mechanical tolerances to <1 ␮m were used to control flow propulsion
into adsorptive, chromatographic, and capillary zones.
Fluidic mechanisms were combined into working examples for urinalysis, blood glucose, and hemoglobin A1c
testing using indicators (substances that react with analyte, such as dyes, enzyme substrates, and diazonium
salts), catalytic reactions, and antibodies as recognition
components. Optical signal generation characterized
fluid flow and allowed detection.
Results: We produced chips that included capillary
geometries from 10 to 200 ␮m with geometries for
stopping and starting the flow of blood, urine, or buffer;
vented chambers for metering and splitting 100 nL to 30
␮L; specimen inlets for bubble-free specimen entry and
containment; capillary manifolds for mixing; microstructure interfaces for homogeneous transfer into separation membranes; miniaturized containers for liquid
storage and release; and moisture vapor barrier seals for
1
Diagnostic Division, Bayer Healthcare LLC, Tarrytown, NY.
microParts GmbH, Boehringer Ingelheim, Dortmund, Germany.
*Address correspondence to this author at: Diagnostic Division, Bayer
Healthcare LLC, 1884 Miles Ave., Elkhart, IN 46515-0070. E-mail michael.
[email protected].
Received April 28, 2005; accepted June 2, 2005.
Previously published online at DOI: 10.1373/clinchem.2005.052498
easy use. Serum was separated from whole blood in <10
s. Miniaturization benefits were obtained at 10 –200 ␮m.
Conclusion: Disposable microchip technology is compatible with conventional dry-reagent technology and
allows a highly compact system for complex assay
sequences with minimum manual manipulations and
simple operation.
© 2005 American Association for Clinical Chemistry
The idea that microfluidic chips can perform all of the
various types of point-of-care testing (POCT)3 in one
system is attractive (1 ). A platform for immunoassay,
chemistry, blood gas, electrolyte, nucleic acid, hematology, and coagulation testing is easily envisioned. POCT
methods currently use a variety of formats and materials
as means of flow propulsion to achieve fluidics that
minimize assay steps and improve ease of use (2 ). Flow
propulsion occurs simply by absorbent, flow-through,
chromatographic, or capillary actions. Designs are typically single-use disposables made of inexpensive combinations of hydrophilic and hydrophobic materials and are
capable of detecting 1 or more analytes. Alternatively,
mechanical mixing or centrifugal, pressurized, or electrophoretic forces can be applied, but these systems are
generally more costly (3–7 ).
Absorbing and drying chemical and biological reagents into a matrix such as paper was one of the first
technologies to drive point-of-care devices (8, 9 ). Substrates have since been expanded into membranes, films,
and polymer surfaces. Adsorbent substrates became useful in formats to move fluid laterally along a strip or
cassette base (10, 11 ). This principle of lateral flow has
been particularly useful for chromatographic separation
in immunoassay strips. The majority of POCT devices
2
3
Nonstandard abbreviations: POCT, point-of-care testing; Hb, hemoglobin; PBS, phosphate-buffered saline; FITC, fluorescein isothiocyanate; BSA,
bovine serum albumin; and LED, light-emitting diode.
1923
1924
Pugia et al.: Tool Box for Hand-Held POCT Devices
now use combinations of substrates with dry reagents and
cassettes. Generally, treatment with additional liquid reagents is needed to reduce sample matrix effects and
improve results.
Capillaries and sample wells have also long been used
as disposable formats. Cassettes containing molded channels on the millimeter to centimeter scale are formed by
injection molding, embossing, printing, and layering (12–
18 ). These devices use capillary passageways to transfer a
sample into reagent areas or to draw off excess sample.
Precise control of timing and metering microliter volumes
is not possible because flow into an absorptive area is
difficult to control (13, 14 ). No means of accurately distributing (metering and splitting) small volumes was
succesful even after accounting for the impact of venting.
Attempts to uniformly distribute the sample over the reagent region led to placing the reagent inside millimetersized capillary passages (16 –18 ). Although this lowered
sample volumes, the ability to control flow for additional
processing (mixing and separating) was not achieved. Centrifugal devices were identified as an alternative format and
extended the usage of capillaries, but they require personal
computer–sized devices to control fluid flow (19 –21 ).
Recent trends toward miniaturized devices such as
micro total analysis systems (TAS), biological microelectromechanical systems (BIO-MEMS), or Lab-on-a-Chip
devices are becoming common for applications outside of
POCT (22, 23 ). Several attempts to develop POCT devices
have produced commercial products (24 –27 ). Such systems have one or more structures of micrometer-sized
geometry down to 0.1 ␮m produced by a variety of
microfabrication techniques. Microfabricated surfaces for
separation and capture have been described (28, 29 ).
These approaches are currently being to develop microarray systems for high-throughput screening and singlenucleotide polymorphism identification (30, 31 ), but in
general, these systems are too costly for use as disposable
hand-held devices.
Our goal was to test a disposable format based on
micrometer-sized structures as a POCT device. We
wanted a format that uses separable reagent substrates
but is capable of the complex processing needed for
quantitative analysis of blood and urine specimens. Our
approach was to first mold a generic chip with inlets,
wells, and vents and then form microstructures such as
connecting channels and geometries by micromachining
and laser ablation. Surface energies and geometries were
varied and characterized by testing with optical reagents.
Useful microstructures were then molded into multistep
reaction designs and reexamined. Finalized elements
were combined into complex protocols directed toward
chemistry and immunoassay testing.
Materials and Methods
chip fabrication
Because of its tolerance to cracking and chemical inertness, all first models used Makrolon® polycarbonate
(Bayer AG) as the surface for embedding fluidic geometries by use of injection molding, laser ablation, and
diamond milling (microParts). Polycarbonate is hygroscopic, and its surface energy decays when exposed to
water, adhesives, and polar molecules. The next round of
working models used polystyrene (145D grade; BASF)
and were produced by injection molding and laser ablation but without machining. Molds prepared by lithography achieved 0.1 ␮m tolerances, whereas machined molds
achieved 1 ␮m, as determined with a Dektak surface
profiler (Veeco Corp.). In all cases, chip surfaces were
made hydrophilic to obtain water contact angles (⌽water)
from 20 to 80 degrees (sessile drop method). Surface
treatment was accomplished by use of a high-energy gas
of hydrophilic polymer plasma (microParts). Polystyrene
exhibited high surface stability (⫾ 5 dynes/cm2) as determined by water contact angle measurements over a year
(CV ⫽ 14%).
Reagents were dried on nitrocellulose membrane,
glass, and cellulose substrates (Whatman Ltd) and were
punched and placed as 2-mm2 areas into chip by use of a
robotic arm on an automated assembler (Kuntz Manufacturing Co., Inc.). Reagent pilots were assembled under
controlled temperatures (25–35 °C) and humidity (20%–
30% relative humidity). At separate stations on a rotary
table, liquids were filled directly into chips and pressuresensitive adhesive lids (Excel Scientific) were applied.
Lids used optically clear polypropylene films for optical
areas and aluminized Mylar films for areas needing low
moisture vapor transmission rates.
reagents
Urinalysis reagents for protein, albumin, leukocytes, hemoglobin, creatinine, glucose, nitrite, ketone, and protein
were the standard colorimetric chemistry papers (Multistix®; Bayer HealthCare LLC). Urinalysis controls (Medical Analysis Systems) were used for test solutions. Chromogenic glucose reagent placed on a nylon membrane
(Biodyne, Pall Corp.) was prepared as described elsewhere (32 ). For testing, whole blood pretreated with
heparin was incubated at 25 °C to degrade naturally
occurring glucose. The blood was supplemented with 0,
500, 1000, 2000, 4000, and 6000 mg/L glucose and assayed
on the glucose reference assay instrument (YSI Life Sciences)
Dry and liquid reagents were prepared for a hemoglobin (Hb) A1c immunoassay based on an immunoaffinity
approach using conjugated blue latex particle assays (Fig.
1). Assays for both high (8%–13% Hb A1c) and low
(3%– 8% Hb A1c) concentrations were prepared and
shared the same lysis solution (Cellytic-M; Sigma-Aldrich), waste substrate (Drikette paper; Multisorb), and
capture phase. The capture phase was made by striping
supported nitrocellulose substrate (5.0 ␮m pore size) by
use of a striper (IVEK Corp) to place two 4-mm bands in
a 14 ⫻ 1.6 mm capture zone. One band contained Hb A1c
agglutinator [1 g/L in phosphate-buffered saline (PBS),
1925
Clinical Chemistry 51, No. 10, 2005
optical measurement
Fig. 1. Hb A1c detection method.
(A). FITC-labeled blue latex particles were attached to Hb A1c antibody and
incubated with sample in the conjugate substrate (left). The FITC labels are
indicated by circles with points, BSA carrier are ovals, and blue latex particle are
large circles. One band on the nitrocellulose capture substrate (right) contained
Hb A1c agglutinator (gHb) and the other monoclonal anti-FITC antibody (double Y).
(B), after incubation in the conjugate substrate, the fluid was separated on
nitrocellulose. In the absence of Hb A1c, the particles bind to the agglutinator
band. (C), in the presence of Hb A1c the particles bind the monoclonal anti-FITC
band. As the concentration of Hb A1c increases, the first band decreases and the
second band increases. The ratio of the 2 bands is examined in an algorithm to
arrive at the result. (D), a second wash step with blue latex particles with FITC
antibody attached (smaller double Y and circle) was used to increase sensitivity
in the low-concentration range assay. This increases the amount of color
developed at the captured bands.
Use of color generated from the reagents allowed visualization of fluidics and measurement of responses. We
used 2 types of optical detectors for reading chips, based
on video capture (Coreco Imaging). Both systems were
equipped with a spinning table, programmable controller,
stepper motor, opto-electronic trigger, and dividing circuitry to allow video capture while the table was spinning
or stationary. One system used a color camera with white
light-emitting diode (LED) lighting in a ring strobe, and
the other used a monochrome camera (CVM-536 and MV
70; Jai Pulnix Corp.) with a 4-wavelength LED illuminator
(460, 525, 620, and 660 nm). An LED stopwatch and a
tachometer were added to track the time and speed
during the spin sequence. Capillary flow in the chip was
measured in a stationary position. Fluid stops were overcome either by spinning (200, 500, 1500, or 2500 rpm at a
4-cm radius) or in a stationary position by dispensing
additional liquid in 0.1-␮L increments to increase hydrostatic pressure. Fluidic protocols included timing steps for
changes in hydrostatic pressure and were controlled by a
programmed sequence. Images taken from the chip experiments were saved by a DVD recorder and included
color calibration markers. Collected video files were digitized and reviewed frame by frame to select the proper
pictures at the desired time points. The reagent areas in
the pictures were then analyzed by appropriate image
processing software to calculate the reflectance and convert to an analytical response.
Results
pH 7.4], and the other contained anti-fluorescein isothiocyanate (FITC) monoclonal antibody (3 g/L in 0.05 mol/L
borate, pH 8.5).
For the high-concentration range assay, the conjugate
substrate was made by drying blue latex conjugate on
glass fiber paper in a 1:4 dilution with casein blocking
buffer (Pierce). Blue latex conjugate was made by labeling
bovine serum albumin (BSA) with FITC and anti-Hb A1c
antibody. The labeled-BSA was attached to blue latex
particles (300 nm; 67 ␮eq of COOH/g) at a loading of 30
␮g BSA-FITC-anti-Hb A1c/mg of latex. A wash solution
of PBS containing 1 g/L BSA was used.
The conjugate pad for the low-concentration range
assay was made by drying the blue latex conjugate on a
glass fiber membrane in a 1:400 dilution with blocking
buffer. The wash solution was a 1:10 dilution of anti-FITC
antibody latex conjugate in PBS containing 1 g/L BSA.
Anti-FITC antibody conjugate was prepared by loading
blue latex particles (1 mg) with 10 ␮g of antibody.
Calibrators for Hb A1c were prepared by adding synthesized peptide reagent (Glc-Val-His-Leu-Thr-Tyr-Cys)
to assayed blood 5%–15% Hb A1c. Blood was collected
with potassium oxalate and sodium fluoride as anticoagulants. Calibrator concentrations were assigned by direct
comparison with DCA 2000⫹ results (Bayer).
A tool box of microstructures was constructed for fluidic
control, metering, liquid application, mixing, and separation. The geometric shapes used included 50-␮m diameter
posts, 100-␮m notches, and connecting capillaries with 10to 200-␮m cross-sectional diameters (Fig. 2). These structures were applied in the interfacial regions between the
larger capillaries (1–2 mm depth and width), chambers
(0.5–5 mm depth and width), and reagent substrates.
Designs were conceptualized by use of computer-aided
design (CAD) software and made by micromachining and
molding. Ideal arrangements and geometries were arrived at by iterative testing.
fluidic control
Flow propulsion was achieved by use of capillaries 10 –
100 ␮m in diameter and with various millimeter lengths,
which were embedded in the chip base. Flow through the
capillaries occurred only after the chip surface was made
hydrophilic at a ⌽water of 20 – 80 degrees. The adhesive lid
with a ⌽water of 73 degrees (CV ⫽ 14%) had only a small
impact on the propulsion. Flow for viscous fluids such as
whole blood required lower contact angles, whereas the
mobility in nonviscous fluids such as urine allowed
higher contact angles. Capillaries 50 –200 ␮m in diameter
and 0.5–5 mm in length were filled with millisecond
velocities. Flow was consistent among samples until cap-
1926
Pugia et al.: Tool Box for Hand-Held POCT Devices
Fig. 2. Microfluidic toolbox approach.
Microstructures (post and capillaries shown; top left corner) were used to create design elements (dead stop shown) that were combined into channels connecting
chambers for reagents. Multiple channel designs were fabricated after all elements were shown to work (top right corner). Fabrication using a generic molded blank
(bottom left corner) allowed fluid designs to be tested before being molded into chips with increasing complexity (bottom row, second, third, and fourth panels).
illary diameter reached 10 ␮m; in larger-diameter capillaries, flow became erratic.
Flow stopped at the end of a micrometer-sized capillary when it reached an opening to a wider and deeper
passageway, called a “drop stop” (Fig. 2). A substantial
increase (⬎5-fold) in cross-sectional area was needed to
bring flow to a stop. Liquid (up to 20 ␮L) was held back
in large chambers when micrometer-sized exit capillaries
were used with a drop stop (Figs. 2 and 3). Chambers,
capillaries, and small wells millimeters in size all served
as deeper passageways. Stops were allowed on a hydrophilic chip base surface (⌽ of 20 – 80 degrees) and were not
passive but complete stops until additional pressure was
applied. The flow was always away from the source and
required venting for air inlets and outlets positioned
above and below the stops. Vents in contact with liquids
required larger capillaries to prevent clogging.
The pressure needed to overcome the fluid stop was
directly dependent on the cross-sectional areas of the
capillary and the stop. Capillary diameters of 200, 100,
and 50 ␮m had increasing stop pressures at a fixed surface
energy and stop diameter. Stop strength also increased
with hydrophilicity as did capillary force. Once fluid
passed the stop, flow resumed. Absorbent substrates in
the reagent well were found to break the stop if brought
into contact with liquid in the micrometer-sized capillaries. Stops therefore needed to be distanced from the
absorbent substrates to preventing false triggering.
metering
Metering was achieved by use of fixed volume areas with
micrometer-sized capillaries at the beginning and end of
the metered area (Fig. 3). The exit capillary had flow held
back with a stop. The design was capable of measuring
volumes down to 50 nL with a CV of 3%. A variety of
fixed volume shapes were used as metering areas. Use of
micrometer-sized capillaries as inlets and outlets allowed
little fluid to be trapped outside of the metering area. A
capillary passage 100 ␮m wide and 50 ␮m deep had a low
dead volume of ⬃ 5 nL per mm of length. The small
sample losses allowed the liquid splitting needed for
multiplexing.
Variations in the stop strength allowed emptying of
fixed volume areas at different times. For example, use of
⌽water of 25–35, 35– 45, and 45–55 degrees allowed a 200
␮m ⫻ 150 ␮m ⫻ 1.5 mm capillary opening into a stop to
be overcome at decreasing pressures. Although this
approach worked, creating separate hydrophilic areas
Clinical Chemistry 51, No. 10, 2005
1927
Fig. 3. Chemistry chips.
Scale perspective and details of microstructures are shown for 2 chemistry chips: one for glucose (A and B) and one for urinalysis with complex specimen processing
(C and D). The 48-assay glucose disk (A and B) includes design elements for fluidic control into a reagent dried on a substrate, and liquid application of sample (A and
B). The glucose disk uses only capillary forces without any centrifugal force. The 16-assay urine chip (C and D) includes design elements for fluidic control, metering,
and liquid application of sample and reagents in a multistep assay format. The urine disk uses capillary force to move liquids to stops. Stops emptied into reaction
and waste wells and out of liquid wells and were triggered at various spinning speeds.
was avoided by holding ⌽water at 25–30 degrees and
varying the cross-section area of the capillary from 50 to
300 ␮m.
liquid application
The design used for entry of biological samples into the
chip consisted of an inlet hole and a capillary feeding a
containment chamber from the back or top side (Figs.
2– 4). To achieve a uniform amount of specimen leaving
the chamber, one or more 100 ⫻ 100 ␮m grooves were
placed across the exit path. These acted to force alignment
of the fluid front and displace air through the vents. An
additional overflow chamber was also useful for lowering
operator dependence and allowed overfilling by 1 to 20
␮L. Finally, conical inlet holes were added to aid sample
transfer from a fingerstick drop or a transfer capillary
(Aqua Cap®; Drummond Scientific). Conical inlets reduced dependence on alignment and back pressure. Com-
plete filling was observed with urine (specific gravity,
1.005–1.030) and blood (hematocrit, 44%– 60%).
Reagent fluids (e.g., dilution or wash buffers) were
loaded on the chip either at the time of assay or before,
within liquid-holding wells. Liquid wells with micrometer-sized exit capillaries and stops at the exits were used
to prevent leakage (Fig. 3). Pressure prompted release of
liquids as long as the chamber was vented. Preventing
vapor diffusion through an exit capillary into other assay
areas was a problem over 30 days unless liquid could be
sealed from the capillary. Sealed designs to store several
microliters with moisture vapor transmission rates of 1
g/m2-day could be achieved with foils and breakaway
designs.
separation
Separation of particles from fluids occurred in chambers 6
mm in length and 1 mm in width at depths from 25 to 500
1928
Pugia et al.: Tool Box for Hand-Held POCT Devices
␮m. One end of the chamber was closed and the other
open to inlet and outlet capillaries. Chambers allowed
separation of erythrocytes (5 ␮m diameter) and latex
particles (10 ␮m). The time to complete separation was
dependent on the depth of the separator, the particle
diameter, and the hydrophilicity. In all cases, having a
surface energy equal to or less than the surface energy of
the particles facilitated separation. Separation of 1 ␮L of
plasma from 10 ␮L of blood was obtained in 10 s with a
capillary depth to particle diameter ratio of 10:1 with no
additional pressure. Applied pressure (⬎2 g) was needed
for the complete separations required for hematocrit
determinations. Under these conditions, good accuracy
(95%) and low imprecision (CV ⬍10%) were achieved for
hematocrit measurements (vs HemoCue) when the area of
the optical image of blood compared with plasma was
used for calculation.
Affinity separation using a membrane inside the chip
required microstructures to facilitate homogeneous transfer and lateral flow of liquid. Without the interface, flow
was nonuniform, and the liquid flowed around or under
the substrate, forming air pockets. Reagents, such as
antibodies and other biomolecules, were immobilized on
the substrate, which allowed affinity assays to be performed and the separation to be viewed. Imaging of the
solvent fronts of colored particles allowed comparison of
the flow properties to the ideal (10 ). The entry structure
was the primary determinate, with post structures (50 ⫻
75 ␮m posts) on a platform just above the bottom of the
substrate providing ideal uniform flow distribution. Once
flow was initiated, the washing flow was dependent only
on the exit structure and capillary strength.
mixing
Mixing via fluid passage through one or more parallel
micrometer-sized capillaries was sufficient for cell lysis or
reagent mixing. Shear forces to mix components when
they left the receiving manifold could be varied by
adjusting the capillary lengths (2–5 mm) and diameters
(50 –200 ␮m) to cause a localized increase in flow velocity.
Complete mixing was facilitated by longer and shallower
capillaries. Momentary turbulence at the point of confluence was confirmed by high-speed video. Mixing efficiencies were observed optically by the uniform color after
mixing of 10 ␮L of 250 g/L polyethylene glycol (Mr
20 000) in 0.5 mol/L NaOH with 10 ␮L of phenol red dye
in 50 mmol/L phosphate buffer (pH 4). Instantaneous
lysing of 1 ␮L of erythrocytes with 100 ␮L of lithium
thiocyanate buffer was observed by examination for intact
erythrocytes by blood reagent (MULTISTIX). In cases with
incomplete mixing, incomplete lysis appeared as dark
green spots on the pad.
chemistry and immunoassay chips
Combining of elements produced chips for blood glucose,
urinalysis, and immunoassays (Figs. 3 and 4). The glucose
chips shown in panels A and B of Fig. 3 were used for 48
whole-blood glucose measurements. Each assay area consists of an inlet port, a lead channel, a fluidic interface area
for sample and reagent substrate, and a venting structure
through which the air is removed. The array of microstructure posts required for color uniformity is shown in
Fig. 3A. Speed grooves and microstructure placement in
the prechamber and the reagent chamber were adjusted to
allow bubble-free filling and uniform wetting of 300 nL of
sample across the glucose reagent substrate (not shown).
The glucose chip gave results after 10 s by use of reflectance membranes (CV ⫽ 4.2%) with good correlation (R2
⬎0.99). An overfill tolerance ⬎1 ␮L was observed.
The MULTISTIX urine chemistries were read on a
16-channel chip (Fig. 3, C and D). This chip was capable of
multiple steps. After addition of 50 ␮L of urine to the inlet
port, 16 separate 2.0-␮L aliquots were split and metered
within the chip and, when triggered, transferred into the
first chemistry reaction area. The reaction was read optically after 15 s, and then a second pressure triggered the
volume into a second chemistry area for another reaction
and optical reading. After 60 s, washing of the second
chemistry area was initiated, and wash fluid was collected
in a waste well. The urine chip produced 16 separate
assays with calibration curves comparable to those obtained with dip-and-read reagents. The urine chip provided a sample pretreatment step in the first reaction area.
Hydrophobic reagents were retained in paper substrates,
whereas water-soluble indicators were washed into later
wells. Complete washing of all color from the pads was
observed for these with water-soluble dyes. Passing of
multiple volumes through the substrate allowed amplification while washing removed interference. The latter
was key to providing quantitative results for these reagents by reducing sample matrix bias.
Two examples of immunochips are shown in Fig. 4:
one for liquid reagents dispensed at the time of assay (Fig.
4, A and B) and the other with on-board liquid reagents
(Fig. 4, C and D). Heterogeneous competitive immunoassays for Hb A1c were tested with both (Fig. 1). The same
chips were equally useful for heterogeneous sandwich
immunoassays. A homogeneous immunoassay for Hb A1c
would not require the separation elements included in
these designs but rather a different combination of tool
box elements.
The immunochip shown in panels A and B of Fig. 4
consisted of 4 millimeter-sized wells connected by micrometer-sized capillaries and microstructures. A separate liquid dispenser added lysis and wash buffer. The
first well served as the specimen inlet with a 300-nL meter
volume and as the target for dispensing. The second area
contained conjugate pad and was filled with sample when
3.0 ␮L of lysis buffer was dispensed. Addition of wash
fluid at 2 min overcame the stop and started the flow from
the conjugate area into the capture substrate in the third
well. As 12 ␮L of wash fluid was added, flow exited the
capture area into the waste area containing absorbent in
1929
Clinical Chemistry 51, No. 10, 2005
Fig. 4. Immunoassay chips.
Scale perspective and details of microstructures in 2 immunoassay chips adapted to the Hb A1c immunoassay: a 4-step chip for use with a separate liquid dispenser
(A and B), and an 11-step chip with on-board liquid reagents (C and D). The 4-step chip (A and B) uses only capillary forces and stops at the end of wells that are triggered
by additional dispensing of wash into entry port. The 11-step chip (C and D) contains all elements for fluidic control, metering, on-board liquid containers, liquid
application of sample and reagents, separation, and mixing in a multistep assay format. This complex chip uses capillary forces to move fluids without spinning and
flow stops at the end of wells triggered by various centrifugal forces.
⬍1 min. At the end of the wash, the bands in the capture
area were read.
The immunochip shown in panels C and D of Fig. 4 is
an example of a chips with on-board liquids and consists
of meter, capture, conjugate, and waste wells connected to
2 liquid holders, a mixer, and 2 liquid overfill chambers.
Again, micrometer-sized capillaries and structures were
used to interface areas. In practice, a whole blood specimen was added by the user into the inlet (0.5–2 ␮L), and
the 0.3-␮L meter area was filled by capillary action. The
flow sequence started mixing of the lysis buffer (3.0 ␮L)
from a liquid well with sample, causing the mixture to
enter the mixing manifold and pass through a mixing
capillary followed by transfer of 2 ␮L into the conjugate
pad well. After incubation for 2 min, the liquid in the
conjugate chamber was loaded in the capture chamber,
starting wicking into the nitrocellulose capture zone by
lateral flow. After antigen binding, the capture zone was
washed by 12 ␮L of additional fluid released from a
second liquid well. The wash fluid contained an additional conjugate to increase sensitivity for the low-range
assays. Waste containment was added for excess liquids
from the specimen port, mixer, and capture phase.
Both chips were tested with the same Hb A1c reagents.
The best results produce a response of 3%–13% Hb A1c
with an overall SD of 0.18% Hb A1c (Fig. 5). Washing of
the nitrocellulose membrane was very complete in both
examples, and reaction band color was uniform.
Discussion
Fluid control, or the pumping and valving of fluid, is a
key requirement for bioassays in which fluids must be
retained and then dosed into designated areas at desired
times (e.g., metering, incubation, reaction, and washing).
Fluidic control is achievable in several different ways
(23–29, 33, 34 ), but the use of micrometer-sized structures
with separable reagent substrates works well for disposable POCT chips and eliminates electrical, mechanical, or
sacrificial valves. The number of times that the flow can
be started and stopped is more limited, but complex assay
protocols can still be achieved.
Although the use of capillary channels that connect to
1930
Pugia et al.: Tool Box for Hand-Held POCT Devices
Fig. 5. Hb A1c results.
Results obtained with reagents used in the immunoassay
chip for low- (A) and high-concentration range (B) Hb A1c
analyses. (A), y ⫽ ⫺3.15x2 ⫹ 20.44x ⫹ 41.44 (R2 ⫽
0.9831); (B), y ⫽ ⫺2.34x ⫹ 34.73 (R2 ⫽ 0.994). Error
bars, SD.
reagent wells is well established, the use of devices
possessing micrometer-sized structures to control flow
volumes away from and into reagent substrates is less
well known (13–18 ). Tight control of surface energies and
dimensions allows pressure differences for triggers to be
better tolerated by low-cost systems. In our chips, the
length of time that the pressure needs to be applied is
small and once stops are broken, the overall hydrophilicity of the chip continues the flow. Large capillaries
(millimeter-sized) require tight control of pressure
changes (e.g., centrifugal) and were unsuitable for our
chips. There are considerable advantages to the use of
micrometer-sized capillaries to feed dead stops because
liquids can be kept from contacting dry reagents, and flow
rates changes are ON/OFF.
A key distinction is that the designs shown do not
require any centrifugal force to operate. Only capillary
force is used to move fluid through the chip. Although
centrifugal force can be used to overcome the flow stops,
stops were also overcome by dispensing of additional
liquid upstream from the stop. All other design elements
were purely capillary driven (no spinning), including the
metering, sample application, separation designs, and
mixing designs. Flow stops were used at the end of the
on-board liquid containers and again were adjusted for
triggering either by various centrifugal forces or by additional dispensing volumes without centrifugal forces. The
capillary separation chamber operated with or without
centrifugal force.
Fluid metering is critical to performing quantitative
assays because it allows measured volumes to be reacted
and split for multiplexing. The techniques generally used
in capillary systems, such as overflow metering, in which
a chamber is filled to an excess that is overflowed, did not
have the accuracy needed for nanoliter volumes
(13, 19, 26 ). Overflow areas were used for the containment of waste but not metering in this design. Passive
stops also could not be used for metering; instead, dead
stops were required to prevent contact between liquid
and the absorbent reagent substrate (35 ). The low dead
volumes in micrometer-sized capillaries provided the
minimal volume loss needed for metering and splitting
for multiplexing.
Inlet and outlet ports are critical in POCT chips,
because microchips contain air that must be expelled (36 ).
Underfilling resulting from trapped air leads to underestimation. At the same time, sample not completely entering the inlet port and remaining on the surface of the
device causes carryover and contamination. Air inside the
capillary must be controlled because fluid flows or
trapped air will cause flow issues. One solution has been
to expel the air before or while liquid is entering the inlet
port, such as with a pipette with a plunger (4 5, 7 ). We
found a means of allowing air to escape the chip by
placing micrometer-sized structures in the direction of the
flow. As the fluid gathers through the structures, air is
expelled away from the solvent front. The containment
area for samples in the design was important for preventing sample carryover while allowing for overfilling and
thus improved the overall ease of use.
Mixing is important in bioaffinity assays because sample dilution and incubations are needed for binding.
Centrifugal mixing, in which components are mixed in a
chamber by tidal movements or oscillation cycles, is well
known (19 ). Perturbation structures in capillaries have
been shown to facilitate mixing (37, 38 ). In the capillary
mixer design tested, momentary turbulence was produced by use of high-velocity micrometer-sized capillaries at the point of confluence exiting the feed chamber.
The resulting mixture then quickly hits the bottom chamber, and the fluid is further mixed. Running the mixture
through high-velocity micrometer-sized capillaries allow
the rapid mixing needed in POCT assays.
Centrifugal separation is a well-understood principle
(21 ). Small chambers with tight wall-to-particle ratios
have been less studied. In our design, separation effi-
Clinical Chemistry 51, No. 10, 2005
ciency was controlled by the ratios of particle diameter to
chamber depth and surface energies, suggesting that
attraction of particles to the wall is the driving force in this
separation. Without adjustment of wall surface energy,
adhesion was either irreversible or separation did not
occur. For POCT applications, separation at exceedingly
low forces and times is critical, making this approach
attractive.
Separation with affinity-capture substrates allows fluidic exchanges between the detector area and the reacted
fluid (10 ). Approximately 50 exchange volumes are typically seen with nitrocellulose membranes, which helps
amplification. The greater volume of nitrocellulose membrane compared with a surface-immobilized capture
monolayer allows less fluidic exchange. Both substrate
and surface immobilization face problems of uniform
solvent flow across the detector area. Although separate
reagent substrates are not typically applied to microfluidic chips, the use of microstructures to draw fluids into
and away from substrates provided unexpectedly uniform affinity reactions. Posts, tiers, and grooves were key
geometric shapes impacting the direction of fluid into the
substrate.
In conclusion, disposable chips based on microstructures
and a hydrophilic surface energy work well with reagents
on substrates and provide a possible POCT platform.
Areas for substrate preparation can be separated from
chip surfaces, allowing cost savings. No lithography was
required for molding. Although capable of complex multistep protocols, the design is fundamentally simple, and
the inner workings can be easily kept transparent to the
user. A tool box of useful microgeometries allows easy
reconfiguring. These chips require no pressurization or
electrokinetic or centrifugal forces and provide results
typical of those expected in POCT devices with less
reagent and smaller samples.
References
1. Pugia MJ, Price CP. Technology of handheld devices for point care
testing. In: Price CP, St John A, Hicks JM, eds. Point of care
testing. Washington DC: AACC Press, 2004:13–30.
2. Tudos AJ, Besselink GAJ, Schasfoort RBM. Trends in miniaturized
total analysis systems for point-of-care testing in clinical chemistry. Lab Chip 2001;1:83–95.
3. Volles DF, McKenney JM, Miller WG, Ruffen D, Zhang D. Analytic
and clinical performance of two compact cholesterol-testing devices. Pharmacotherapy 1998;18:184 –92.
4. John WR, Edwards R, Price CP. Laboratory evaluation of the DCA
2000 HbA1c immunoassay analyzer. Ann Clin Biochem 1994;31:
367–70.
5. Parsons MP, Newman DJ, Newall RG, Price CP. Validation of a
point-of-care assay for the urinary albumin:creatinine ratio. Clin
Chem 1999;45:414 –7.
6. Schembri C, Burd TL, Kopf-Sill AR, Shea LR, Braynin B. Centrifugation and capillarity integrated into a multiple analyte whole
blood analyzer. J Automat Chem 1995;3:1799 – 804.
7. Erickson KA, Wilding P. Evaluation of a novel point-of-care system:
the i-STAT® Portable Clinical Analyzer. Clin Chem 1993;39:283–7.
1931
8. Zipp A, Hornby WE. Solid phase chemistry: its principles and
application in clinical analyses. Talanta 1984;31:863–77.
9. Walter B. Dry reagent chemistries. Anal Chem 1983;55:498A–
514A.
10. Qian S, Bau HH. A mathematical model of lateral flow bioreactions
applied to sandwich assays. Anal Biochem 2003;322:89 –98.
11. Oberhardt BJ. Thrombosis and homeostasis testing at the point of
care. Am J Clin Pathol 1995;104:S72– 8.
12. Ashworth L, Gibb I, Alberti KGMM. HemoCue: evaluation of a
portable photometric system for determining glucose in whole
blood. Clin Chem 1992;38:1479 – 82.
13. Blatt JM, Heiland R, Morris JR, Pugh J, Wogoman FW, inventors.
Volume metering capillary gap device for applying a liquid sample
onto a reactive surface. US Patent 4,761,381; Aug 2, 1988.
14. Charlton SC, Rebec VR, Ruetten C, inventors. Self-metering fluid
analysis device. US Patent 5,208,163; May 4, 1993.
15. Moore CE, Wilson K, inventors. Capillary flow apparatus for
inoculation of a test substrate. US Patent 5,110,555; May 5,
1992.
16. Musho MK, Noell JO, Tse HS, inventors. Use of conductive
sensors in diagnostic assays US Patent 5,250,439; Oct 5, 1993.
17. Nilsson SE, Lija J, inventors. Cuvette. US Patent 5,286,454; Feb
15, 1994.
18. Shanks IA, Smith AM, Nylander CI, inventors. Device for the use in
chemical test procedures. US Patent 5,141,868; Aug 25, 1992.
19. Burtis CA, Bostick WD, Johnson WF. Development of a multipurpose optical system for use with a centrifugal fast analyzer. Clin
Chem 1975;21:1225–33.
20. Kellogg G, Kieffer-Higgins SG, Carvalho BL, Davis GA, Willis JP,
Minior T, et al; inventors. Devices and methods for using centripetal acceleration to drive fluid movement on a microfluidic system.
US Patent 6,063,589; May 16, 2000.
21. Johnson RD, Badr IHA, Barrett G, Lai S, Lu Y, Marc J, et al.
Development of a fully analysis integrated system for ions based
on ion-selective optodes and centrifugal microfluidics. Anal Chem
2001;73:3940 – 6.
22. Wang J. Survey and summary. From DNA biosensors to gene
chips. Nucleic Acid Res 2000;26:3011– 6.
23. Figeys D, Pinto D. Lab-on-a-chip: a revolution in biological and
medical sciences. Anal Chem 2000;71:330A–5A.
24. Schulte TH, Bardell RL, Weigl BH. Microfluidic technologies in
clinical diagnostics. Clin Chim Acta 2002;32:1–10.
25. Kricka LJ. Miniaturization. In: Price CP, St. John A, Hicks JM, eds.
Point of care testing, 2nd ed. Washington DC: AACC Press,
2004:71– 83.
26. Cunningham DD. Fluidics and sample handling in clinical chemical
analysis. Anal Chim Acta 2001;429:1–18.
27. LJ Kricka. Miniaturization of analytical systems. Clin Chem 1998;
44:2008 –14.
28. MacBeath G. Proteomics comes to the surface. Nat Biotechnol
2001;19:828 –9.
29. Cahill DJ. Protein and antibody arrays and their medical applications. J Immunol methods 2001;250:81–91.
30. Moutereau S, Narwa R, Matheron C, Vongmany N, Simon E,
Goossens M. An improved electronic microarray-based diagnostic
assay for identification of MEFV mutations. Hum Mutat 2004;23:
621– 8.
31. Wu G, Irvine J, Luft C, Pressley D, Hodge CN, Janzen B. Assay
development and high-throughput screening of caspases in microfluidic format. Comb Chem High Throughput Screen 2003;6:
303–12.
32. Bell DE, Chu AH, inventors. Preparation of diagnostic test strips
containing tetrazolium salt indicators. US Patent 5,360,595; Nov
1, 1994.
33. PaulPH, Garguilo MG, Rakestraw DJ. Imaging of pressure- and
1932
Pugia et al.: Tool Box for Hand-Held POCT Devices
electrokinetically driven flows through open capillaries. Anal Chem
1998;70:2459 – 67.
34. Chiem NH, Harrison DJ. Microchip systems for immunoassay: an
integrated immunoreactor with electrophoretic separation for serum theophylline determination. Clin Chem 1998;44:591– 8.
35. McNeely MR, Oliphant AR, Spute MK, inventors. Fluidic circuit
components based upon passive fluid dynamics. US Patent
6,296,020; Oct 2, 2001.
36. Columbus RL, inventor. Capillary transport device having speed
and meniscus control means. US Patent 4,618,476; Oct 21,
1986.
37. He B, Tan L, Regnier F. Microfabricated filters for microfluidic
analytical systems. Anal Chem 1999;71:1464 – 8.
38. Kamholz AE, Weigl BH, Finlayson BA, Yager P. Quantitative
analysis of molecular interaction in a microfluidic channel: the
T-sensor. Anal Chem 1999;71:5340 –7.