C

LI

a

ARRAA

KMPDIL

1

cdaacoi

o2hptce

0d

Biosensors and Bioelectronics 27 (2011) 64– 70

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

jou rn al h om epa ge: www.elsev ier .com/ locate /b ios

apillary-driven multiparametric microfluidic chips for one-step immunoassays

uc Gervais, Martina Hitzbleck, Emmanuel Delamarche ∗

BM Research – Zurich, Säumerstrasse 4, CH-8803 Rüschlikon, Switzerland

r t i c l e i n f o

rticle history:eceived 25 March 2011eceived in revised form 1 June 2011ccepted 13 June 2011vailable online 21 June 2011

eywords:icrofluidics

oint-of-careiagnostics

mmunoassaysab-on-a-chip

a b s t r a c t

Here we present a capillary-driven microfluidic chip for “one-step” immunoassays. The chip allows foreasy modification of several assay parameters such as the flow rates of sample, the volumes of samplesfor tests, and the type of reagents and receptors for detecting analytes. We therefore term such a chipa multiparametric chip and illustrate this concept with the integration and release of anti-C-reactiveprotein (CRP) detection antibodies (dAbs) together with splitting flow of samples containing CRP acrosslines of anti-CRP capture antibodies (cAbs). The microfluidic chip is fabricated in Si and is sealed withpolydimethylsiloxane (PDMS) patterned with cAbs. The microfluidic chip is ∼1.7 × 3.4 cm2 and is capableof analyzing 20 �L of human serum in 6 parallel flow paths with a range of flow rates from 3.3 nL s−1 to0.46 nL s−1. An inkjet spotter was used to deposit 10.6 nL of dAb solution in a structure vicinal to themain flow path of the chip. The consequent asymmetric release of dAbs in a stream of human serumis compensated by a Dean flow mixer having 9 mixing loops and a footprint of 2.8 mm × 0.78 mm. Thequantity of dAb present in the half of the flow path close to the spotting region decreases from 83% at theentrance of the mixer to 52% in the region after the mixer. The sample is then equally split into 6 reaction

chambers and proceeds via connecting channels to 2 �L capillary pumps. The hydraulic resistance ofthe connecting channels is designed to vary flow rates, and therefore the kinetics of capture of CRP-dAbcomplexes, from 10 min to 72 min. The increased incubation time leads to a fourfold increase in detectionsignal in the reaction chamber with the longer incubation time. The concept presented here is flexible andsuited for implementing various surface fluorescence immunoassays on a capillary-driven microfluidicchip.

© 2011 Elsevier B.V. All rights reserved.

. Introduction

Immunoassays are central to research in the life sciences andlinical diagnostics (Wild, 2005); they are routinely performed toetect disease markers and pathogens in patients as well as for ther-py monitoring (Durner, 2010). Portable devices for immunoassaysre increasingly used in healthcare for decentralized and point-of-are testing (Bissonnette and Bergeron, 2010), providing benefitsf a shorter time to result and accessibility where there is limitednfrastructure.

Originally, lateral flow assays on porous membranes were devel-ped to provide a portable format for immunoassays (O’Farrell,009). The immunoassay strip test, such as the one used in aome pregnancy test, was the first commercially successful exam-le of point-of-care immunodiagnostics. Recent developments of

his technology have focused mostly on improving biochemi-al receptors (antibody fragments, aptamers, engineered proteins,tc.), detection labels (nanoparticles, quantum dots, fluorescent

∗ Corresponding author.E-mail address: emd@zurich.ibm.com (E. Delamarche).

956-5663/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2011.06.016

dyes, etc.) and peripheral detection systems (readers, light emit-ting diodes, lasers, CCD chips, etc.). Flow paths for immunoassaysallowing routing and mixing have been made of paper (Martinezet al., 2010), thread (Safavieh et al., 2009), and polymers (Beckerand Gärtner, 2007). One aspect of these approaches is the increasingchallenges of scaling down the volumes of sample that are needed,miniaturizing signals for multiplexed analysis, accurately control-ling flow conditions for precise reaction protocols, and the difficultyof providing a range of quantifiable detection signal intensities.

Microfluidic technology has the opportunity to extend thecapabilities of lateral flow assays. For example, one commer-cially successful test performs one-step immunoassays for thedetection of three cardiac makers (Buechler, 1995; Apple et al.,1999). Here, further miniaturization and multiplexing is challeng-ing due to the large footprint required for detection zones andbecause all reagents are integrated into one channel leading tocrosstalk when the number of reagents is increased. Liquids canbe passively driven in microfluidic immunoassay devices using

several approaches such as capillary forces that are generated bymicrostructures (Juncker et al., 2002; Jönsson et al., 2008), gravitydriven flow in tilted microchannels (Rossier et al., 2008) and liq-uid suction from degassed PDMS dissolving air from microchannels

nd Bio

(was(

mptcarubvtdawta

tdctitokctgur

2

2

Bswss(agcSPosc2CCaebaPFt

L. Gervais et al. / Biosensors a

Hosokawa et al., 2006). In particular, capillary driven microfluidicsere used for one-step immunoassays from human serum (Gervais

nd Delamarche, 2009) and immunoassays using multiple succes-ive pipetting steps for the detection of eleven markers from bloodWang et al., 2010).

Despite important efforts to implement lateral flow assays onicrofluidic formats, major limitations remain. For example, flow

aths tend to be designed in a “linear” manner where areas con-aining reagents for a test are placed one after another. Reactiononditions are also preset or necessitate peripheral equipment toctivate valves, release reagents, and achieve flexibility in flowates (Lee et al., 2009; Zimmermann et al., 2005). Different markerssually require specific reaction conditions for optimal detectionecause of the range of clinically relevant concentrations and theariability in binding characteristics between analytes and recep-ors. Additionally, the variability of clinical samples, such as theifference in viscosity between samples from different patients,nd changes in ambient conditions (temperature and humidity)hen the test is done can challenge tests that have preset parame-

ers. For these reasons, it is difficult to implement the detection of panel of markers on a passive device.

Here, we present one-step microfluidic chips for immunoassayshat have split flow paths that can impose different flow rate con-itions on samples added to the chips. The flow of sample in thesehips is governed by capillary forces. Reagents are integrated inhe chips and are released in a stream of sample so as to enablemmunoreactions on miniaturized capture areas. This addresseshe above mentioned limitations and builds on our earlier workn one-step immunoassays. Additional goals of this work are toeep the chip compact, simple to use, and working with the prin-iples of one-step immunoassays as well as to provide a solutionhat can be low cost to manufacture, both by preventing the inte-ration of active components and active packaging methods and bysing designs that can be transferred from Si technology to plasticeplication.

. Experimental

.1. Chemicals, biochemicals and PDMS

Water was deionized with a Simplicity 185 system (Millipore,illerica, MA). PBS was prepared by dissolving phosphate bufferedaline (PBS) tablets (Sigma–Aldrich, St. Louis, MO) in deionizedater and the resulting solution was filtered using a 0.20 �m

yringe filter (Sartorius, Epsom, U.K.). Solution of 1% (w/v) bovineerum albumin (BSA, Sigma–Aldrich) was prepared in PBS. Ethanolpuriss ≥ 99.8%, Fluka, Sigma–Aldrich) was used to rinse Si chipsnd PDMS substrates. PDMS layers were prepared by mixing Syl-ard 184 pre-polymers (Dow Corning, Midland, MI) at a ratio ofuring agent to polymer of 1:10 using a DOPAG mixer (Cham,witzerland). The mixture was poured onto planar polystyreneetri dishes (Greiner BioOne), and cured overnight at 60 ◦C in anven. The approximately 1-mm-thick cured PDMS was cut using acalpel into pieces having approximately the size of a microfluidichip and patterned with cAbs as previously published (Ziegler et al.,008). Samples for immunoassays were prepared using humanRP-free serum (8CFS, HyTest, Turku, Finland) that was spiked withRP (8C72, HyTest). Anti-CRP-C2 Abs (4C28-C2, HyTest) were useds cAbs. Anti-CRP-C6 (4C28-C6, HyTest) were used as dAbs. To thisnd, these Abs were labeled with Alexa Fluor 647 (Invitrogen, Carls-ad, CA) following the instructions from the supplier. Each dAb had

pproximately 4 fluorophores as measured using an Eppendorf Bio-hotometer (Hamburg, Germany). A solution of 1% (w/v) Pluronic108 (BASF, Ludwigshafen, Germany) in deionized water was usedo treat the Si chips. A solution of 2% (v/v) Hellmanex II (Hellma,

electronics 27 (2011) 64– 70 65

Mullheim, Germany) in deionized water was occasionally used toclean the inkjet glass pipettes. Solutions of dAbs were preparedusing 250 �g mL−1 Alexa 647 labeled anti-CRP-C6, 68 mg mL−1 d-(+)-trehalose (Fluka) and 700 �g mL−1 l-phenylalanine (Fluka) inPBS. Solutions of cAbs were prepared in PBS at a concentration of125 �g mL−1.

2.2. Fabrication of chips and preparation

Photolithography, quartz chrome masks (Photronix, Dresden,Germany) and 4-inch Si wafers (Siltronix, Geneva, Switzerland)were used to fabricate the microfluidic chips and stencils. First, a0.5 �m thick layer of SiO2 was grown onto a Si wafer using wetoxidation in a thermal furnace (TS-6304, Tempres, Vaasen, TheNetherlands). Hexamethyldisilazane (adhesion promoter) was firstspin coated onto the wafer. Then photoresist (AZ 6612, MicroChem-icals, Ulm, Germany) was spin coated at 4000 rpm to a thicknessof approximately 1.6 �m and soft baked for 1 min at 110 ◦C. Atwo-layer-mask process was used. The first layer contained all themicrofluidic structures of the microfluidic chip. The second layercontained all the features except the reaction chambers. The firstlayer mask was brought to a “soft” contact with the resist-coatedwafer and exposed to 6.5 mW/cm2 ultraviolet light for 15 s using amask aligner (MA/BA6, Süss Microtech, Garching, Germany) anddeveloped. Buffered hydrofluoric acid (BHF, 7:1 volume ratio of87.5% NH4F in water to 12.5% HF in water; danger) was used toetch SiO2 from the microchannels uncovered by the photoresist.The resist was stripped from the wafers and HMDS was spin coated.Photoresist (AZ 4562, MicroChemicals, Ulm, Germany) was spincoated at 4000 rpm to a thickness of approximately 7 �m and softbaked for 1 min at 110 ◦C. The second layer mask was exposed toultraviolet light for approximately 45 s and developed. The secondlayer microchannels were etched into Si to a depth of 160 �m usingdeep reactive ion etching (AMS-200SE, Alcatel Micro MachiningSystems, Annecy Cedex, France). The resist was stripped from thewafer and the first layer was etched to a depth of 20 �m using reac-tive ion etching. The fabricated microfluidic chip was composed of180 �m deep microchannels and 20 �m deep reaction chambers.

Stencils consisted of 12 mm × 12 mm chips containing 4 groupsof 4 trenches. Each trench was etched through a single-side pol-ished wafer and comprised a loading pad (for pipetting a solution)connected to a 2-mm long and 100-�m-wide channel. A one maskphotolithography process was used to pattern resist and transferthe resist pattern through the wafer using deep reactive ion etch-ing to form the stencils. A pattern delineating the perimeter of eachstencil was etched as well through the wafer, which resulted withthe release of individual stencils.

Air-based plasma (Tepla microwave-plasma system 100, PVATepla, Asslar, Germany) at 200 W and 0.7 Torr was used to cleanthe microfluidic chips for 2 min. The microfluidic chips were thenincubated in a solution of Pluronic® for 30 min, rinsed with deion-ized water and dried under an N2 stream. Microfluidic chips werestored in a dark and dry environment and used within 4 h of coatingwith Pluronic®.

2.3. Capture antibody patterning on PDMS

First, air-based plasma was used to clean the stencils and makethem hydrophilic. Then, the polished side of the stencils was placedon top of the side of the PDMS substrate that had been in contactwith the Petri dish during the curing of PDMS. Capture antibod-ies were deposited on PDMS at room temperature by pipetting a

solution of cAb in the stencils and incubating for 15 min. A smallPetri dish with sheets of clean room paper soaked in water at thebottom was used as a humidity chamber to store the chips dur-ing incubation and prevent evaporation of the solution of cAb in

66 L. Gervais et al. / Biosensors and Bioelectronics 27 (2011) 64– 70

Fig. 1. Capillary-driven microfluidic chip for multiparametric immunoassays. (a) Photograph of a chip having various microfluidic functional elements (see insets) for holdingand dissolving dAbs, mixing them in a stream of sample and splitting the sample into 6 streams passing over cAb areas at flow rate conditions determined by the resistanceo el of ta differs

twuPraer

2

iztttc

abts2tiPiFafleew

f channels connecting to individual capillary pumps. (b) Equivalent electrical modnd even distribution by mixing of dAbs in a stream of sample passing through theignal in the case of a sandwich fluorescence immunoassay as show here.

he stencil. Stencils were rinsed with PBS (∼10 mL) and deionizedater (∼10 mL) before removing them from the PDMS substratessing tweezers under a bath of blocking solution of BSA. TheDMS substrates were left in the solution of BSA for 15 min beforeinsing them with PBS and deionized water, and drying under

stream of nitrogen. Stencils were rinsed in deionized water,thanol and cleaned with an air-based plasma before they wereeused.

.4. Immunoassays

An Autodrop MD-P-705-L (Microdrop, Norderstedt, Germany)nkjet spotter and an AD-K-501 piezo-driven pipette having a noz-le of 70-�m in diameter were used to deposit solutions of dAbs inhe microfluidic chips. The inkjet pipette nozzle was aligned withhe detection antibody dissolution zone and 60 drops of dAb solu-ion of 180 pL each were deposited for a total of 10.8 nL. Microfluidichips were stored in the dark and used within 4 h of deposition.

A PDMS substrate was placed on the microfluidic chip in such way that perpendicular lines of cAbs crossed the reaction cham-ers of the chip and that the chip was covered from the pumps upo half of the loading pad. The microfluidic chip was placed on thetage of a fluorescence microscope. Immunoassays started when0 �L of human serum spiked with CRP at 100 ng mL−1 was pipet-ed into the loading pad of the microfluidic chip. The fluorescencen the different chip regions was monitored directly through theDMS cover. Fluorescence images of different parts of the microflu-dic chip were captured at various times during the experiment.luorescence images were captured along the Dean flow mixernd stitched together using an image processing software. The

ow distributor was monitored throughout the experiment. Afterxperiments, microfluidic chips were rinsed with deionized water,thanol, and dried under a stream of N2. The microfluidic chipsere cleaned with an air-based plasma and ready for reuse.

he chip used to calculate flow rates and filling times. (c) Illustration of the releaseent components of the microfluidic chip and that can result in variable amount of

2.5. Instrumental setup and data acquisition

Fluorescence measurements were done using a fluorescencemicroscope (Eclipse 90i, Nikon, Japan) and images were typicallyacquired using a 10× objective. The microscope had a 100 W mer-cury arc lamp and a black and white CCD camera (DS-1QM, Nikon)cooled to −30 ◦C. Fluorescence images were analyzed using NIS-Elements software (Nikon).

3. Results and discussion

3.1. Chip design and filling

The selected strategy to vary assay conditions uses parallel flowpaths on a microfluidic chip, Fig. 1a. The sample initially followsa common flow path that includes a loading pad, a dAb dissolu-tion zone and a Dean flow mixer and then splits into six individualreaction chambers. Six connecting channels join the end of eachreaction chamber (at an area denoted J in Fig. 1a) to an individualcapillary pump. These connecting channels have dissimilar depthsand lengths and therefore modulate flow rates throughout the 6parallel flow paths via their specific flow resistance characteristics,Fig. 1b.

A test begins when a sample is pipetted onto the loading pad,Fig. 1c. The dAb dissolution zone is a side channel having a circu-lar reservoir of branching channels that accommodate dissolveddAbs spotted using an inkjet spotter and dried. The dAbs redissolveinto one side of the central channel mostly by passive diffusion.The mixer distributes the concentration of dAbs across the channelbefore splitting into six reaction chambers. The analyte–dAb com-

plex binds to lines of cAbs on PDMS covering the reaction chamber.With all the reaction chambers seeing the same concentration ofanalytes and dAbs passing through, we expect to see a gradient offluorescence detection signal as a result of having dissimilar flow

L. Gervais et al. / Biosensors and Bio

Fig. 2. Steps involved in the fabrication of the microfluidic chips. Two lithographicmasks are used to define structures that are 20 �m deep (reaction chambers) or180 �m deep (all other structures). Structures are not drawn to scale and the topvsp

rhttgspettpd

opFTotataaiipcttrav

TF

iew schematic of the chip (shown below (h)) helps visualizing how the depths oftructures relate to elements such as the loading pad, reaction chamber and capillaryump.

ates in the reaction chambers. The detection signal should be theighest when flow rates are reduced leading to increased incuba-ion times. Other parameters than the flow rate can be varied withhe chip shown in Fig. 1 such as, for example, the types of dAbs inte-rated in the chip and receptors patterned on PDMS. The fabricationteps for producing the microfluidic chips are shown in Fig. 2. Theresence of structures having different depths requires using sev-ral photolithographic masks. Here, it is advantageous to use a SiO2hin mask and a thicker resist at the end of the fabrication processo perform two consecutive dry etching steps after all photoresistrocessing steps (spin coating, exposure, development) have beenone.

The flow resistances, flow rates and filling times of the flow pathsf the microfluidic chip shown in Fig. 1 were calculated followingrior work (Delamarche et al., 1998) and are summarized in Table 1.irst, we calculated the capillary pressure of a single capillary pump.his was done using a value of 56.2 mN m−1 for the surface tensionf the serum (Lewin, 1972) and 20◦ and 116◦ for the respective con-act angles of the serum on the walls of the microfluidic flow pathsnd the PDMS. The pump is an array of microstructures with separa-ion distances that generate two levels of capillary pressures (−2.8nd −7.2 kPa) depending on the position of the meniscus in therray (Zimmermann et al., 2007). The weakest pressure (−2.8 kPa)n the pumps was used to calculate approximate flow rates of thendividual flow paths. For this, the hydraulic resistance of each flowath between the loading pad and its corresponding pump was cal-ulated by including the contribution of each element composinghe flow path. These resistances correspond to the loading pad (RLP),

he dAb dissolution zone (RdAb), the Dean flow mixer (Rmix), oneeaction chamber (RRC1–6 ), a specific connection channel (RCC1–6 )nd a capillary pump (Rpump). Taking a value of 1.8 mPa s for theiscosity of the serum at 20 ◦C (McMillan, 1974), we estimated the

able 1low characteristics of the microfluidic chip.

Flow path Total resistance (m−3) Flow rate D (L s−1)

1 (RCC1 ) 3.8 × 1017 6.3 × 10−9

2 (RCC2 ) 7.5 × 1017 3.2 × 10−9

3 (RCC3 ) 1.5 × 1018 1.6 × 10−9

4 (RCC4 ) 1.8 × 1018 1.4 × 10−9

5 (RCC5 ) 2.0 × 1018 1.2 × 10−9

6 (RCC6 ) 2.2 × 1018 1.1 × 10−9

electronics 27 (2011) 64– 70 67

flow rate to range from 1.1 to 6.3 nL s−1 for the slowest and fastestflow paths, respectively. The corresponding time needed to fill the2 �L volume of each capillary pump is consequently estimated torange from 8 to 47 min.

The resistance RCC1 was designed to have a filling time of thepump of approximately 10 min, which is suited for a fast diagnos-tic test. Shorter times where not considered here, although testsperformed in just a few minutes might be beneficial in situationssuch as the diagnostics of heart disease and stroke. Additionally,the resistance of this connecting channel cannot be easily loweredwithout a major redesign of the microfluidic chip. For example, RCC1was kept small by shifting all the structures (including the load-ing pad) before the connecting channels to one edge of the chip,and by shortening the first reaction chamber by 50% and keep-ing the depth of the first connection channel at 180 �m. In theeventuality that a diagnostic test of just a few minutes would bedesirable, capillary pumps having a smaller volume can be used.The next resistances (RCC2–6 ) were designed to increase the fillingtime incrementally to reach an upper limit of about 1 h, which is apractical upper time limit for diagnostics tests. The experimentalfilling times, shown in Table 1, were measured by observing thetime for a pump to fill completely. The experimental filling timeof the first 4 flow paths are in line with the calculated ones andthe last two are, respectively, 42% and 53% slower than the calcu-lated ones. Possible reasons for this could be the difference betweenthe viscosity and surface tension of human serum used for cal-culations and experiments. Some evaporation of sample throughthe PDMS (Randall and Doyle, 2005) covering the microfluidic chipand in the loading pad may both increase the viscosity of the sam-ple over time and compete with the filling of the pumps linked tothe slowest flow path. Covering the PDMS layer with a thin trans-parent layer of material non-permeable to water and gases (suchas e.g. cyclic-olefin copolymer) might provide a simple solutionto lower this evaporation. Placing the loading pad over a Peltierthermoelement and cooling the pad slightly above the dew pointtemperature can also minimize evaporation (Zimmermann et al.,2005).

The microfluidic chip has a factor seven in filling time betweenthe fastest and slowest flow paths, which can be used to shiftthe advancement of immunobinding reactions (Wild, 2005), dis-solution profiles of reagents (García et al., 2004), and can alsocompensate for changes in sample characteristics. Filling time canbe further increased if needed by routing resistances in the arealeft free in the chip. The viscosity of body fluid samples can varygreatly. For example, at 37 ◦C in healthy adults, whole blood vis-cosity can be in a range of 2.83–6.3 mPa s (Rosenson et al., 1996)and whole saliva can be in a range 2.75–15.51 mPa s (Dodds et al.,2005). The viscosity of a sample is also affected by tempera-ture. A whole blood sample can vary in viscosity in a range from5.48 to 14.41 mPa s when temperature varies from 38 ◦C to 5 ◦C(Langstroth, 1919). These viscosities can vary six fold between dif-

ferent samples and conditions. Since the flow rate is proportionalto the viscosity of the sample, resistances in microchannels canbe used to encode the same flow rates with samples of differentviscosities.

Calculated filling time Experimental filling time

8 min 3 s 10 min16 min 10 s 19 min32 min 23 s 36 min37 min 1 s 43 min42 min 11 s 60 min47 min 5 s 72 min

6 nd Bioelectronics 27 (2011) 64– 70

3

atitcfctonsad

acmrueieaccetfim

tawtmesc

D

waodcmoacic1t

awidaqh

Fig. 3. Fluorescence profile of sample containing fluorescently labeled dAbs flowingthrough a Dean flow mixer. (a) Fluorescence image of the mixing of dAb solutiondiffusing into the lateral wall of the channel (top of the image) from the dAb dis-solution zone and going through a series of mixer loops. The fluorescence intensityis measured from cross-sections of the microchannel denoted by the numberedarrows for the number of Dean flow loops. (b) Graph of the fluorescence intensity

8 L. Gervais et al. / Biosensors a

.2. Mixing

The dAb reagents enter the laminar stream of sample flowinglong the lateral wall of the channel. They need to be evenly dis-ributed across the channel before equal quantities of dAbs are splitnto 6 individual reaction chambers. If only passive diffusion is used,he dAbs, which are 150 kDa immunoglobulins G, would need a 2.9-m-long channel to mix across a channel. The 125 kDa pentamericorm of CRP, would need a 2.7-cm-long channel to mix across ahannel. A channel this long would not only add significant resis-ance but would also occupy significant area on the chip. Even forther analytes commonly detected using immunoassays, the chan-el length needed for mixing would be very long: 2.2-cm-long forerum albumin (64 kDa), and 9.7-mm-long for insulin (51 aminocids, 5.8 kDa), a small amino acid fragment and one of the smallestetectable molecules using immunoassays.

We therefore implemented a mixer on the chip to mix dAbscross the channel while using a smaller footprint than a straighthannel. Many strategies have been developed for mixing inicrofluidics. Active mixers are not considered here because they

equire external energy input, are complex and expensive to man-facture on microfluidic chips. Several passive mixers, requiring noxternal energy input and using simple geometries, can easily bentegrated into diagnostic devices. For example, porous polymermbedded in microchannels (Mair et al., 2009) can mix liquids in

short microchannel at the cost of increased complexity of fabri-ation. Staggered groove structures in the floor of a microchannelan induce chaotic flows and mix liquids (Stroock et al., 2002). Mix-rs requiring only one lithography step for fabrication are simplero manufacture and lower cost. For example, a mixer using modi-ed Tesla structures can divide flow into two substreams that areerged and collide in opposite flow directions (Hong et al., 2004).Among passive mixers, we found the Dean flow mixer to be par-

icularly attractive because it is simple to fabricate and composed of geometry of looping channels that do not entrap air when fillingith sample using capillary forces. The Dean flow mixer switches

he sample between flow vortices of opposite direction that allowolecules to explore the channel cross section chaotically (Jiang

t al., 2004; Ahn et al., 2008). The Dean number, which is a dimen-ionless number used to quantify the efficiency of mixing in curvedhannels, is

e = Re

√dH

R

here Re is the Reynold’s number, dH is the hydraulic diameternd R is the radius of curvature. The Reynold’s number dependsn the flow rate encoded in the microfluidic chip. The width andepth, with which the hydraulic diameter is calculated, were keptonstant at 100 �m and 180 �m, respectively, so that the Dean flowixer could be integrated with the channels of the same dimension

n the microfluidic chip. The radius of curvature was kept as smalls possible, 200 �m, in order to increase the Dean number. Thealculated Dean flow number for the mixing loops in the microflu-dic chip was 31, which is above 15 where enhanced mixing viahaotic flows begins. Moreover, the Dean flow mixer resistance of.0 × 1015 m−3 contributes minimally to the overall resistance ofhe flow paths.

Microscope fluorescence images of dAbs in sample flowing in anbove-described Dean flow mixer are shown in Fig. 3a. The imagesere captured 3 min after the sample was introduced into the load-

ng pad, once the microchannels were filled and the capillary pumps

etermined the flow rate. The mixing effect is observed by lookingt the signal cross section after every Dean flow loop, Fig. 3b, anduantified by measuring the portion of dAbs that remain in the topalf of the channel after diffusing from the dAb dissolution zone.

of cross sections showing concentrated dAbs along the lateral wall of the channelbeing gradually homogenized to the middle of the channel once they reach the endof the mixer.

At the entrance of the mixer (0th loop) there is a peak of 83% ofdAbs in the top half of the channel. After the 9th loop, only 52% ofdAbs remain in the top half of the channel. At the outlet, a bit morefluorescence is in the center of the channel. This could be due to flu-orescent antibody agglomerates and antibody–antigen complexesbeing focused towards the middle of the channel by alternatingDean flow loops (Di Carlo et al., 2007).

3.3. Distribution and detection

Once the distribution of dAbs is improved by the Dean flowmixer, the sample and dAbs proceed to a region where the cen-tral channel is split into 6 reaction chambers and where individualimmunoassays are performed. The distribution channels, Fig. 4a,and the detection areas in the reaction chambers, Fig. 4b, wereimaged at 30 min when flow paths 1–3 were filled, and flow paths4–6 were still filling and contributing to the signal. The cross sec-tion intensity profile of the flow distributor, Fig. 4c, shows the sameamount of dAbs flowing through each channel. Fig. 4d correspondsto the detection of CRP at a concentration of 100 ng mL−1 in humanserum. This concentration corresponds to the lower clinically rel-evant concentration of CRP in serum for the diagnosis of coronaryheart diseases. Interestingly, the areas between the receptors pro-vide a reference for the background signal and can be used as acontrol to verify that there were no issues with non-specific bind-ing of analytes or dAbs on PDMS and that PDMS did not generateautofluorescence. The intensity profile of the signal areas shows afour-fold increase in signal in the slow filling flow path 6 comparedto the fast filling flow path 1.

Frequently, in the context of medical diagnostics, a few ana-lytes occurring in different concentration ranges may be detected.

Antibodies of different affinities can be used to vary the strengthof the generated signal based on the desired concentration rangeof detection of analytes. Conversely, when keeping antibodies andantigens the same, the affinities cannot be changed, however the

L. Gervais et al. / Biosensors and Bio

Fig. 4. Fluorescence images of the distribution of fluorescently labeled dAbs (a)before and (b) at the sites with receptors and (c, d) corresponding intensity profiles(measured from top to bottom). (c) Intensity profile showing a similar quantity ofdAbs entering RC6 to RC1. (d) Intensity profile measured across the middle signalareas and revealing a fourfold increase in signal in RC6 compared to RC1. Channelswith slower flow rates, such as RC6, provide longer incubation times for analyte-dAbcomplexes to bind to cAb lines and increase signal compared to channels with fasterflow rates and shorter incubation times, such as RC . The sample analyzed here is2

itetwdmigttedsnc

ictdbwcc

1

0 �L of human serum spiked with 100 ng mL−1 of CRP.

ncubation time of analytes with receptor areas can be varied withhe flow rate to vary the signal. An increased signal can be gen-rated with increased incubation times due to slower flow rateshat result in molecules having more time to diffuse and interactith the detection areas. A decreased signal can be generated withecreased incubation times due to faster flow rates that do not giveolecules much time to diffuse to the detection areas. A microflu-

dic chip with multiple channels providing a range of signals for aiven concentration of analyte, increases the detectable range ofhis analyte. This allows tuning for the optimal flow rate neededo detect an analyte antibody interaction of a given affinity. Forxample, using microfluidic chips with active pumping of sample,Ab solution, rinsing, and flow rates of 33 nL min−1, a detectableignal for 10 pg mL−1 of CRP in human serum was found (resultsot shown). In this way, low flow rates can be used to detect lowoncentrations of analytes.

Multiplexing could be implemented on a microfluidic chip sim-larly to micromosaic immunoassays (Wolf et al., 2004). On such ahip, a cocktail of dAbs would be inkjet deposited in the dAb dissolu-ion zone. Alternatively, specific dAbs can be deposited in individualAb dissolution zones to minimize the risk of cross-reactivity

etween dAbs and cAbs. Six independent reaction chambers onhich parallel lines of different cAbs can be patterned on the same

hip would detect for clinically relevant markers in different con-entration ranges. We also note that the concepts presented here

electronics 27 (2011) 64– 70 69

are compatible with the multiplexed detection of various analytesbased on the method reported by Wang et al. (2010).

4. Conclusion

A microfluidic chip for effecting multiparametric immunoassayswas developed. The microfluidic chip can accommodate variousvolumes of sample and dAb reagents, varying flow conditionsand incubation times in 6 independent reaction chambers, andas a consequence can vary the detection signal strength basedon incubation times, and has the possibility of multiplexing thedetection of several analytes by inkjet depositing a cocktail ofdAbs in the dAb dissolution zone and different lines of cAbsover the reaction chambers. We believe that such microfluidicchips will become increasingly attractive for decentralized diag-nostics due to their flexibility, performance, simplicity of use andcompetitive low cost of fabrication when produced in polymericmaterials.

Acknowledgements

We thank colleagues Richard Stutz and Ute Drechsler for exten-sive technical support. We are grateful to Robert Lovchik, GovindKaigala, and Heiko Wolf for discussions and to Michel Despont,Walter Riess, Janos Vörös (ETHZ) and Nico de Rooij (EPFL) for theircontinuous support.

References

Ahn, Y.C., Jung, W., Chen, Z., 2008. Lab Chip 8, 125–133.Apple, F.S., Christenson, R.H., Valdes, R., Andriak, A.J., Berg, A., Duh, S.H., Feng, Y.J.,

Jortani, S.A., Johnson, N.A., Koplen, B., Mascotti, K., Wu, A.H., 1999. Clin. Chem.45, 199–205.

Becker, H., Gärtner, C., 2007. Anal. Bioanal. Chem. 390, 89–111.Bissonnette, L., Bergeron, M.G., 2010. Clin. Microbiol. Infect. 16, 1044–1053.Buechler, K.F., 1995. US patent 5,458,852.Delamarche, E., Bernard, A., Schmid, H., Bietsch, A., Michel, B., Biebuyck, H., 1998. J.

Am. Chem. Soc. 120, 500–508.Di Carlo, D., Irimia, D., Tompkins, R.G., Toner, M., 2007. Proc. Natl. Acad. Sci. U.S.A.

104, 18892–18897.Dodds, M.W.J., Johnson, D.A., Yeh, C.-K., 2005. J. Dent. 33, 223–233.Durner, J., 2010. Angew. Chem. Int. Ed. 49, 1026–1051.García, E., Kirkham, J.R., Hatch, A.V., Hawkins, K.R., Yager, P., 2004. Lab Chip 4, 78–82.Gervais, L., Delamarche, E., 2009. Lab Chip 9, 3330–3337.Hong, C.C., Choi, J.W., Ahn, C.H., 2004. Lab Chip 4, 109–113.Hosokawa, K., Omata, M., Sato, K., Maeda, M., 2006. Lab Chip 6, 236–241.Jönsson, C., Aronsson, M., Rundström, G., Pettersson, C., Mendel-Hartvig, I., Bakker,

J., Martinsson, E., Liedberg, B., MacCraith, B., Öhman, O., Melin, J., 2008. Lab Chip8, 1191–1197.

Jiang, F., Drese, K.S., Hardt, S., Kuepper, M., Schoenfeld, F., 2004. AIChE J. 50,2297–2305.

Juncker, D., Schmid, H., Drechsler, U., Wolf, H., Wolf, M., Michel, B., de Rooij, N.,Delamarche, E., 2002. Anal. Chem. 74, 6139–6144.

Langstroth, L., 1919. J. Exp. Med. 30, 597–606.Lee, B.S., Lee, J.-N., Park, J.-M., Lee, J.-G., Kim, S., Cho, Y.-K., Ko, C., 2009. Lab Chip 9,

1548–1555.Lewin, S., 1972. Br. J. Haematol. 22, 561–566.Mair, D.A., Schwei, T.R., Dinio, T.S., Svec, F., Fréchet, J.M.J., 2009. Lab Chip 9,

877–883.Martinez, A.W., Phillips, S.T., Whitesides, G.M., Carrilho, E., 2010. Anal. Chem. 82,

3–10.McMillan, D.E., 1974. J. Clin. Invest. 53, 1071–1079.O’Farrell, B., 2009. Lateral Flow Immunoassay. Humana Press, Totowa, NJ,

pp. 1–33.Randall, G.C., Doyle, P.S., 2005. Proc. Natl. Acad. Sci. U.S.A. 102, 10813–10818.Rosenson, R.S., McCormick, A., Uretz, E.F., 1996. Clin. Chem. 42, 1189–1195.Rossier, J., Baranek, S., Morier, P., Vollet, C., Vulliet, F., Dechastonay, Y., Reymond, F.,

2008. J. Assoc. Lab. Autom. 13, 322–329.Safavieh, R., Mirzaei, M., Qasaimeh, M.A., Juncker, D., 2009. Proc. 13th Int. Conf.

on Miniaturized Systems for Chemistry and Life Sciences. Presented at the Thir-teenth International Conference on Miniaturized Systems for Chemistry and LifeSciences , Jeju, Korea, pp. 685–687.

Stroock, A.D., Dertinger, S.K.W., Ajdari, A., Mezic, I., Stone, H.A., Whitesides, G.M.,2002. Science 295, 647–651.

Wang, J., Ahmad, H., Ma, C., Shi, Q., Vermesh, O., Vermesh, U., Heath, J., 2010. LabChip 10, 3157–3162.

Wild, D., 2005. The Immunoassay Handbook. Gulf Professional Publishing.

7 nd Bio

W

Z

0 L. Gervais et al. / Biosensors a

olf, M., Juncker, D., Michel, B., Hunziker, P., Delamarche, E., 2004. Biosens. Bioelec-tron. 19, 1193–1202.

iegler, J., Zimmermann, M., Hunziker, P., Delamarche, E., 2008. Anal. Chem. 80,1763–1769.

electronics 27 (2011) 64– 70

Zimmermann, M., Bentley, S., Schmid, H., Hunziker, P., Delamarche, E., 2005. LabChip 5, 1355–1359.

Zimmermann, M., Schmid, H., Hunziker, P., Delamarche, E., 2007. Lab Chip 7,119–125.