ISSN 1473-0197

Lab on a ChipMiniaturisation for chemistry, physics, biology, materials science and bioengineering

www.rsc.org/loc Volume 13 | Number 21 | 7 November 2013 | Pages 4149–4274

PAPERRoozbeh Safavieh and David Juncker Capillarics: pre-programmed, self-powered microfluidic circuits built from capillary elements

OFC COVER SCAN

TO FIT INTO THIS BOX

www.rsc.org/locRegistered Charity Number 207890

Featuring work from the Single Molecule Biophysics

Laboratory of Hermann E. Gaub, Center for Nano-

Science, Ludwig-Maximilians-Universität München,

Germany.

Title: Protein–DNA force assay in a microfluidic format

Protein–DNA interaction forces are studied using a miniaturized and multiplexed molecular force assay on a microfluidic MITOMI chip and with a new confocal analysis method.

As featured in:

See Marcus Otten et al.,Lab Chip, 2013, 13, 4198.

LC013021_cover_PRINT.indd 2LC013021_cover_PRINT.indd 2 9/26/2013 10:39:27 AM9/26/2013 10:39:27 AM

Cite this: Lab Chip, 2013, 13, 4180

Capillarics: pre-programmed, self-powered microfluidiccircuits built from capillary elements3

Received 8th June 2013,Accepted 31st July 2013

DOI: 10.1039/c3lc50691f

www.rsc.org/loc

Roozbeh Safaviehab and David Juncker*abc

Microfluidic capillary systems employ surface tension effects to manipulate liquids, and are thus self-

powered and self-regulated as liquid handling is structurally and chemically encoded in microscale

conduits. However, capillary systems have been limited to perform simple fluidic operations. Here, we

introduce complex capillary flow circuits that encode sequential flow of multiple liquids with distinct flow

rates and flow reversal. We first introduce two novel microfluidic capillary elements including (i) retention

burst valves and (ii) robust low aspect ratio trigger valves. These elements are combined with flow resistors,

capillary retention valves, capillary pumps, and open and closed reservoirs to build a capillary circuit that,

following sample addition, autonomously delivers a defined sequence of multiple chemicals according to a

preprogrammed and predetermined flow rate and time. Such a circuit was used to measure the

concentration of C-reactive protein. This work illustrates that as in electronics, complex capillary circuits

may be built by combining simple capillary elements. We define such circuits as ‘‘capillarics’’, and introduce

symbolic representations. We believe that more complex circuits will become possible by expanding the

library of building elements and formulating abstract design rules.

Introduction

Lab on a Chip (LOC) devices have emerged as a powerful toolfor a variety of applications including bio-analysis1,2 and pointof care diagnosis.3 Central to microfluidic applications is thecontrol of liquid flow within microconduits. Most microfluidicsystems depend on peripheral equipment to effect and controlthe flow of liquid. As the complexity of these circuitsincreased, the similarities to electronic circuits were exploited,most notably in electrokinetic and electrophoretic microflui-dics. In addition, centrifugal and hydrophobic microfluidicsystems were also built where the advancement of the liquidwas controlled by a combination of external control, and pre-embedded restrictions and hydrophobic patches.4 The mostcomplex microfluidic circuits developed to date are the onesmade by multilayer soft lithography.5 More recently, digitalmicrofluidics that use electrical fields and capacitances tomove droplets on arrays of electrodes, effectively create aconnection between electronics and microfluidics, which mayfacilitate subsequent scaling up.6 Owing to the great complex-

ity of active microfluidic systems, abstract level of representa-tion were developed.7,8 Passive microfluidics, which have notbeen developed to the same level of complexity, have not beenrepresented at a symbolic level. Here, we first briefly reviewpassive microfluidics in this introduction, and then presentnovel capillary fluidic elements and their use for making anadvanced capillary microfluidic circuit. Moreover, we proposea symbolic representation of capillary elements (see Table 1) tofacilitate the representation and design of capillary circuits.

Passive and capillary microfluidics

So-called passive microfluidic systems circumvent the need forperipheral equipment and extract the energy to move theliquids from the system, thus only requiring minimal userinterventions. The flow in passive microfluidics is typicallydriven by capillary effects, and they have in fact long been usedin diagnostics in lateral flow tests.9 Recently, paper–basedmicrofluidics have attracted renewed interest from academiaand microfluidic circuits with channels and reservoirspatterned in a paper or nitrocellulose substrate usingphotolithography, printing or laser cutting are being devel-oped.10 To advance the functionality of these circuits, variousvalves including (i) flow rate regulators by patterning dilutesolutions of a water soluble wax,11 or sugar barriers12 in paper,and (ii) metering valves using sugar bridges12 or expandablepolymer actuators13 combined in paper have been developedto control the timing of the fluid delivery. In addition, we andothers showed that thread and yarn can be used to buildfluidic circuits from the bottom up14 and can be used forimmunoassays.15

aBiomedical Engineering Department, McGill University, 740 Dr Penfield Avenue,

Montreal, QC H3A 0G1, Canada. E-mail: david.juncker@mcgill.ca; Fax: +1

514-398-1790bMcGill University and Genome Quebec Innovation Centre, McGill University, 740 Dr

Penfield Avenue, Montreal, QC H3A 0G1, CanadacDepartment of Neurology and Neurosurgery, McGill University, 740 Dr Penfield

Avenue, Montreal, QC H3A 0G1

3 Electronic supplementary information (ESI) available. See DOI: 10.1039/c3lc50691f

Lab on a Chip

PAPER

4180 | Lab Chip, 2013, 13, 4180–4189 This journal is � The Royal Society of Chemistry 2013

The heterogeneity of fibrous microfluidics, however, makesminiaturization difficult, implying that larger samples arerequired, and multiplexing will be limited. Furthermore, thelimit of detection of paper and thread based microfluidicsdon’t match the performance of classical ELISA and micro-fabricated microfluidic systems until now.

Autonomous capillary microfluidic systems

A landmark paper by Delamarche and colleagues highlightedthe potential of self-filling microfluidics based on capillaryeffects.16 Arrays of hydrophilic 2 mm-wide microfluidicchannels, formed by sealing structured poly(dimethylsiloxane)(PDMS) onto silicon wafers, were used to spontaneously drawin liquids into the conduits and pattern proteins on surfaces;however, solutions could not be rinsed or exchanged.Common chemical and biochemical reactions do requireaddition and exchange of multiple solutions in sequence. Anautonomous microfluidic capillary system (AMCS) that wasself-powered and self-regulated – hence autonomous – wasproposed and designed by combining a capillary pump (CP)and a capillary retention valve (CRV).17 In AMCSs the capillarypressure is encoded in the geometry and surface properties(free surface energy of both solid surface and the liquid),which drives and controls the liquid flow on the chip. AMCSallows filling and flushing multiple solutions by simplydelivering them sequentially to an inlet, without any needfor removing solutions,17,18 making them well adapted forconducting immunoassays that require the delivery of multi-ple solutions.19

One-step immunoassays

The ideal devices to conduct immunoassays and point-of-carediagnostics would only require a single manipulation, namelythe addition of sample, to complete the assay, so-called one-step immunoassays.20 Efforts to realize such systems haveprogressed along two directions. One is to keep fluidicfunctionality simple and implement the sequential deliveryof chemicals by controlled dissolution or the geometry of thesystem. For example, Gervais et al.21 pre-dried reagents in adead-end conduit thus controlling the release, although thissystem was unable to sequentially deliver multiple reagentsrequired to amplify the signals and enhance the detectionlimit of the assay.21,22 Another simple approach featured threeopen and linked reservoirs that drain sequentially, althoughlimited cross-talks between reagents occurred, and flow rateswere not individually controlled, because of the flow resis-tance, the reservoirs are drained in sequence.23 The secondstrategy was to expand the functionality of AMCSs to furtherenhance assay performance and versatility and emulate thefunctionality normally obtained using active systems.

Capillary valves

An important group of elements required for making advancedcircuits are valves, but developing them is one of thechallenges for building capillary systems, as on one hand theyneed to be self-filling and moving parts are not usable, and yetthe liquid should stop. However, to stop the liquid in capillarysystems, one approach is to design an abrupt increase inmicrochannel cross-section, to make it energetically unfavor-Ta

ble

1Sc

hem

atic

and

sym

bo

licre

pre

sen

tati

on

so

fth

elib

rary

of

cap

illar

icel

emen

tsin

clu

din

g(i)

mic

roch

ann

el,(

ii)flo

wre

sist

or,

(iii)

ven

t,(iv

)C

P,(v

)TV

,(vi

)C

RV

,(vi

i)R

BVs,

(viii

)cl

ose

dre

serv

oir

,an

d(ix

)o

pen

rese

rvo

ir

(i)

Mic

roch

ann

el(i

i)Fl

owre

sist

or(i

ii)

Ven

t(i

v)C

apil

lary

pum

pw

ith

post

s(C

P)(v

)C

apil

lary

trig

ger

valv

e(T

V)

(vi)

Cap

illa

ryre

ten

tion

valv

e(C

RV

)(v

ii)

Ret

enti

onbu

rst

valv

e(R

BV

)(v

iii)

Clo

sed

rese

rvoi

r(i

x)O

pen

rese

rvoi

r

Sch

emat

ic

Sym

boli

c

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 4180–4189 | 4181

Lab on a Chip Paper

able for the liquid to flow from the narrow to the wide channeldue to the large increase in liquid-air interface at the fillingfront.24 Such geometric valves can then be triggered by flowinga sample through the wide channel. Although the concept wasproposed long ago,25 and even multi-liquid valving wasshown,26 an implementation that is robust and reliable foreveryday use has been difficult to achieve. The channelenlargement is often only produced laterally within plane,and thus the liquid tends to creep along bottom and cover ofthe channel. High aspect ratio valves mitigate this issue, butthey are difficult to microfabricate, even in Si, and cannot betransferred to other materials, while still occasionally failing.24

In a follow up study, Zimmermann et al.27 added newfunctionalities to their CPs and improved their performance.More recently, valves have been made in low aspect ratiostructures using less hydrophilic surfaces, for example using aPDMS cover that can be depressed manually for liquidactivation.28

Hybrid capillary systems

Autonomous sequential delivery of samples has been demon-strated by trapping air bubbles between liquid plugs, mirror-ing an approach initially implemented with tubes and asyringe pump,29 but using capillary effects to drive the flow.30

However, bubbles create high flow resistance, and varyingbubble sizes will affect the flow resistance and may get stuck,which will thus affect the reproducibility and reliability of suchcircuits.

Capillary microfluidics can also operate with positivepressures formed by hemispheric droplets dispensed atop ofthe inlet.31 The simplicity of this approach makes themappealing for generating concentration gradients for example;however, the flow rate is low, and changes as the droplets areshrinking, thus only affording limited control over the flowrate. Recently, positive pressure fluidics have also beencombined with capillary stop valves for more advanced fluidicoperations;32 however, the timing and triggering of liquid flowwas not fully pre-programmed, and in fact required multipletimed user interventions, reminiscent of the constraints of theoriginal AMCS.

Capillary elements and capillarics

Over the last few years, the functionality of capillary micro-fluidic systems has been expanded significantly. Here, weintroduce additional capillary components, namely a robusttrigger valve (TV) and a retention burst valve (RBV).Furthermore, we introduce capillarics, in analogy to electronics,denoting both the complex capillary microfluidics as such,and the modularity of their architecture allowing them to bedesigned and assembled hierarchically by combining basicbuilding elements selected from a library. The elementalbuilding blocks are called capillary elements or capillaricelements depending on the context and their complexity. Tounderline this idea, we introduce a symbolic representationakin to the standardized symbolic representation widely usedin electronics. To illustrate potential of this concept, acapillaric circuit was designed that upon flowing a sample,reverses the flow and flushes four different chemicals in apredetermined sequence with a different flow rate. This circuit

is then applied to measure the concentration of C-reactiveprotein.

Materials and methods

Chemicals and materials

Sylgard1 184 silicon elastomer kit (PDMS) was purchasedfrom Dow Corning (Midland, MI, USA). Prepolymers, i.e.curing agent and polymer base, were manually mixed at a ratioof 1 : 10, and cured for 8 h in an oven (Lindberg Blue M, FisherScientific) at 60 uC. A 1 mg mL21 solution of Fluoresceinsodium salt (C20H10Na2O5) (fluorescein dye, Sigma-Aldrich,USA) in water (Milli-Q purified water, Millipore, USA) was usedfor characterizing the retention burst valves. Food dyes(McCormick & Co., MD, USA) were purchased from localstores and 50% diluted solutions were used to test thecapillaric circuit. Experiments were performed at the roomtemperature of 23 ¡ 2 uC. Phosphate buffered saline (PBS)tablets (Sigma-Aldrich, USA) were reconstituted in deionized(DI) water to form 1% PBS in water. Bovine serum albumin(BSA) (Jackson ImmunoResearch, PA, USA) was dissolved at aconcentration of 3% in DI water. Human C-reactive protein(CRP) antigen, capture anti-CRP, and biotinylated detectionanti-CRP were purchased from R&D Systems (Minneapolis,MN, USA). Streptavidin-Alexa Fluor 488 (Invitrogen,Burlington, ON, Canada) was used as a label in theimmunoassay.

Chip design and fabrication

Circuits were designed in a layout editor software, CleWin(CleWin 5, WieWeb software, Netherlands), and the designswere printed in a 799 6 799 chrome mask with 65 000 dots perinch resolution (Thin metal parts, Colorado Springs, USA). Asoft lithography technique was followed as described pre-viously elsewhere.33 Briefly, 2 level moulds were fabricated inSU-8 50 (Microchem, Massachusetts, USA), and replicated intoPDMS. The elements and circuits are 100 mm deep, except thetwo level trigger valves, which are only 50 mm deep. Thereaction chamber was designed in an oval shape with 200 mmwidth and 2000 mm length. A razor blade (single edged razorblade, Fisher Scientific, Canada) was used to cut the chipsfrom the replica. The vents of the chip were fabricated using abiopsy punch (1.5 mm puncher, Ted Pella Inc., USA). The chipwas rendered hydrophilic with an air plasma (Plasmaline 415,Tegal Inc., US) for 45 s with 250-mTorr pressure and 150-mWpower. A flat 2 mm thick PDMS cover was used to seal the chip.

Capture antibody patterning on PDMS

We used a microfluidic capillary system (CS)34 with 16 parallelmicrochannels to pattern capture anti-CRP on the PDMScover. Each microchannel was 2000 6 100 6 100 mm3 in size,and was separated from the neighboring microchannel by a200 mm gap. The CS was activated with air plasma to render ithydrophilic, and sealed reversibly to the PDMS cover,orthogonal to the future reaction chamber that was outlinedwith a pen. 2 mL of anti-CRP capture antibody (250 mg mL21)were pipetted into the loading port and spontaneously filled

4182 | Lab Chip, 2013, 13, 4180–4189 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

the microchannels. The PDMS was incubated for 45 min atroom temperature in a humidified closed Petri dish contain-ing a wet tissue to prevent evaporation. The liquid was thendrawn out using a clean room paper, the PDMS removed,rinsed with PBS for 15 s and blocked with a 3% BSA solutionfor 30 min to avoid nonspecific binding, rinsed with DI water,dried, and stored for later use.

One step immunoassay

For the sandwich immunoassay, we positioned a plasmaactivated chip on a patterned cover. The orientation of thereaction chamber was orthogonal to the capturing stripespatterned on the PDMS cover. We then preloaded 4 reagentsincluding: (i) 1% PBS in DI water, (ii) biotinylated anti-CRPdetection antibody (200 mg mL21), (iii) 1% PBS in DI water, and(iv) streptavidin-Alexa Fluor 488 (500 mg mL21) in fourreservoirs in the chip by contacting the tip of a pipette tothe side channel vents. Subsequently, we loaded 2.5 mL of abuffer with CRP protein, and repeated the entire experimentwith a series of different concentrations of CRP. Upon thecompletion of the assay, we separated the cover from the chip,rinsed it with DI water for 15 s, and dried it using a nitrogengun, followed by fluorescence imaging of the PDMS coverusing a microscope.

Imaging analysis and signal quantification

To characterize the fabricated capillaric elements and thecircuits, we used both optical (LV150 industrial microscope,Nikon, Japan) and scanning electron microscopy (S-3000Nvariable pressure SEM, Hitachi, Japan). Images of the liquidflow in the chip were captured using a stereomicroscope (LeicaMZ8, Leica Microsystems, Switzerland) outfitted with a CCDcamera (DS-Fi1, Nikon, Japan). As for the immunoassay, weused a customized fluorescence confocal microscope (C1siNikon Inverted Confocal microscope, Nikon, Japan) connectedto a CCD camera (CoolSNAP HQ2, Photometrics, USA) toquantify the binding of the assay. The CRP binding curve wascalculated from three independent experiments. Averagecolour intensities of the fluorescence images were extractedusing Image J (NIH, Bethesda, MD), and binding curves fittedwith a four-parameter logistic curve (GraphPad Prism 5,GraphPad Software Inc., USA), Fig. 6. In each experiment, wetested six chips with six different concentrations of CRPantigen and measured the average fluorescent intensities oftwo stripes.

Results and discussion

Library of capillaric elements

Although active fluidic and microfluidic systems benefitedfrom conceptualization and symbolic representation of thevarious elements,7,8 no symbolic representation has beendeveloped for capillary and passive microfluidics to date.Together with technological advances proposed here, weintroduce a symbolic representation for capillary elementsincluding (i) microchannels, (ii) fluidic resistors, (iii) vents, (iv)capillary pumps (CPs), (v) trigger valves (TVs), (vi) capillary

retention valves (CRVs), (vii) novel retention burst valves(RBVs), (viii) closed reservoirs as well as (ix) open reservoirs.The elements are shown in both schematic and symbolicforms in Table 1.

Microchannels are transporting the liquids and while theyalso generate a flow resistance, it can often be neglected, andthey thus represent the equivalent of electrical lines inelectrical circuits. Fluidic resistors are typically microchannelswith reduced cross-sections. Because the flow resistance scalesas R24 (radius) in channels with circular cross-section, smallchanges in size can have significant effects, and a few smallsections along the flow path can dominate the overall flowresistance of the circuit, as in electrical circuits. Vents areconduits that are connected to the atmosphere. It isnoteworthy that the capillary pressure is proportional to R21

and in practice becomes negligible when the smallestdimension in a circular or rectangular conduit is . 1 mm;thus vents should be big enough and hydrophobic to minimizesurface tension.

CPs generate a constant capillary pressure similar to avoltage source in an electronic circuit. Microstructuredreservoirs with posts serve as CPs and the gaps between theposts define the capillary pressure; the smaller the gap, thelarger the pressure drop and the stronger the pump.27,35 TVsstop the flow of a first liquid at a point until it is triggered bythe flow of a second liquid entering through a second conduit,as discussed previously.24 RBVs are a modification of theCRVs, and both are formed by a localized reduction of thechannel cross-section, which prevents the liquid from drainingdue to a large capillary pressure. CRVs are meant to retain theliquid permanently, while the RBVs act as release valves, oncethe capillary pressure exceeds a threshold value. A series ofRBVs with increasing threshold can thus be used to control thesequence of liquid being delivered.

Previously, we and other researchers have introduced someof these capillary elements, shown in Table 1, including CPs,flow resistors, vents, closed and open reservoirs.35,36 Weintroduce a symbolic representation that follows the repre-sentation from electronics for some elements. Also we presentnovel symbols for some elements such as capillary pumpsalong with numerical indication of the strength and both openand closed reservoirs. The symbolism of the valves wasdeveloped on the ones commonly used for macroscopicvalves,7 while introducing an arc representing the curvedcapillary filling front. For the capillary TVs, the curve issuperposed on a diode that represents the unidirectionality ofthe valve. The CRVs are shown as a classical valve symbol withan added curve indicating the dewetting front and an ‘

symbol conveying the fact that they never burst under normaloperation. The RBV symbol is similar to the CRV one, but the‘ symbol is replaced by a scale bar representing the burstthreshold; the scale is arbitrary, specific to each circuit, andwas chosen to represent the pressure range of the circuit.Whereas most other elements are self-explanatory, the CPs areshown as a rectangle with a symbol mimicking a voltagesource along with a scale bar indicating their strength. In thefollowing sections we explain the functionality of the two novelcapillaric elements introduced here, namely the two-levelcapillary TVs and the RBVs.

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 4180–4189 | 4183

Lab on a Chip Paper

Two-level capillary trigger valve (TV)

Conventional capillary TVs that expand only laterally lackrobustness, and are prone to spontaneous triggering.24 Asingle layer TV with an aspect ratio of depth/width = 3 wasmade of plasma-activated, hydrophilic, PDMS and closed withnative, hydrophobic, PDMS cover. This leaked in less than twoseconds, Fig. 1A and Movie S1 in electronic supplementaryinformation (ESI3). To overcome the lack of reliability, wepropose a two-level capillary TV, which consists of a shallowconduit intersecting a deep one and a hydrophobic cover,Fig. 1B. The abrupt enlargement in the cross-section of themicrochannel occurs laterally and vertically at the bottom,while the hydrophobic cover prevents the liquid from creepingalong the top. A two level TV with the aspect ratio, depth/width= 1.5, stopped the liquid for more than 20 min, Fig. 1C andMovie S2 in ESI.3 These valves robustly stopped liquids for over20 min, and we did not observe a single failure in 50experiments.

Retention burst valves (RBVs)

CRVs and RBVs are formed by constrictions in the micro-channel that produce high capillary pressure. The CRVs andRBVs shown here were all engineered in deep channels, whilevarying the width to gradually increase the capillary pressure.A schematic illustrating a step by step operation of an RBV ispresented in Fig. 2A. If the capillary pressure of a CP is weakerthan the one of the constriction, a CRV is formed. If thecapillary pressure of the pump is stronger than the constric-tion, then an RBV is formed that will retain the liquid until thepressure at a point within the circuit drops below the capacityof the valve. Thus, as long as there is flow in the circuit, thepressure drop across the microchannel makes the hydrody-namic pressure of any point in the vicinity of the RBV weakerthan the receding capillary pressure of the RBV itself, |DP|X ,

|DP|RBV. As a result the liquid stops at the valve, but once theflow stops, the circuit acts as a hydraulic system, and thecapillary pressure of the pump is transmitted throughout,causing |DP|X . |DP|RBV, and eventually the RBV bursts. Thus,if multiple retention burst valves with different thresholds areincluded, the weakest one will burst first and the liquid stored

downstream will be drained, and then the second weakest onewill burst and the liquid will be drained, and so on.

The burst pressure of the valve is related to its dimensionsaccording to the rules of capillary pressure. However, while theadvancing contact angle is present during filling, it is thereceding contact angle that arises during draining. Hysteresisbetween advancing and receding contact angles thus requiresthat the dimensions of the CP be significantly smaller than theone of the RBVs to burst it. Moreover, although not reflected inthe equations, we found that the length of the RBV alsocontributed to the strength of the RBV. We then designed aseries of RBVs with varying width and length, and integratedthem in a simple circuit to illustrate sequential drainage of 6valves, Fig. 2B&C. The RBVs are located at the extremity of eachside arm. CRVs with a capillary pressure exceeding the one ofthe CP were included at the junction of each side arm and themain conduit, as well as on the inlet side to prevent drainageof the main conduit. Fig. 2D and Movie S3 in ESI3 show time-lapse images of the filling and sequential drainage of the sixside arms with a solution containing fluorescein. The liquidwas added to the loading port (not visible) and then startedfilling the microchannels and the CP. A flow resistor in front ofthe CP ensured that all side-arms were filled despite the largecapillary pressure of the CP. Next, as the liquid in the loadingport is drained, it is pinned and stopped at the CRV0. TheCRVs are numbered in the schematic to facilitate thediscussion, but they have no functional difference except thefact that they are activated at different time points. As all flowstops at this moment, the pressure of the CP acts throughoutthe entire circuit, and bursts the RBV1, followed by drainage ofthe first conduit until the liquid is pinned at the CRV1 next tothe main channel. Then, the RBV2, is burst, and so on, until all6 side arms are drained.

Capillaric circuit with flow reversal

To illustrate the possibility of making complex systems using avariety of capillaric elements, we designed a capillaric circuitfor performing one-step immunoassays, Fig. 3A&B. The chip issealed with a hydrophobic PDMS cover with vents and loadingports to add the reagents and a sample. The circuit comprises4 side-arms, each comprising an RBV at the extremity close to

Fig. 1 Two-level TV used to passively stop the liquid. The microstructures are hydrophilic (plasma activated PDMS). The sealing layer is hydrophobic PDMS. (A) Timelapse images showing that the abrupt enlargement of the one–level TV fails after only 1.5 s; (B) SEM micrograph of the two-level TV. (C) Images showing that theabrupt enlargement of the two level TV together with the use of a hydrophobic cover were effective to stop the liquid for periods of 20 min, and could be triggered atany time by flowing a sample in the deep conduit. The scale bars are 200 mm.

4184 | Lab Chip, 2013, 13, 4180–4189 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

the loading ports, and a TV that connects each arm to themain channel, and that simultaneously acts as a CRV. The sidearms are preloaded with 4 different reagents that are appliedto the respective loading ports, Fig. 3C, (t1–t2). A filled pipette

tip is brought into contact with the chip, and the reagents arespontaneously drawn by capillary pressure into the side-armsup to the TVs1–4 (t3). A sample applied to the main loading portflows via a channel, a CRV, fills a short side-conduit stopping

Fig. 2 Operation of the RBVs. (A) schematic illustrating a step by step operation of an RBV as used in this demonstration. (t1) First, the RBV is filled by capillary flow. (t2)While the pressure in the capillary circuit in the vicinity of the RBV (indicated by X) is weaker than the capillary burst pressure of the RBV (|DP|X , |DP|RBV), the liquidremains pinned by the RBV. (t3) if and when the pressure in the circuit rises beyond the withholding burst pressure, |DP|X . |DP|RBV, the liquid starts draining thenarrow channel of the RBV until it eventually bursts (t4), resulting in rapid drainage of the (closed) reservoir downstream of the RBV. (B) symbolic representation of acircuit comprising 6 RBVs with incremental burst thresholds, 7 CRVs, a flow resistance, a CP and an inlet (not shown). SEM micrographs of RBV3 (weaker) and RBV5

(stronger) are shown on the left. (C) Micrograph of the circuit to test the RBVs. (D) Time lapse images showing the sequential filling and draining of reservoirsaccording to the pre-defined sequence dictated by the incremental burst threshold of the RBVs. Each of the serpentine side-reservoirs comprises an RBV and a CRV.The CRVs were placed at the junctions of the side-arm and the main conduit, and have a stronger capillary retention pressure than that of the CP. The RBVs were atthe end of each side arm, and have a weaker capillary pressure than that of the CP. By adjusting the length and the width of each RBV, the sequence of draining of 6side-reservoirs follows the pre-defined sequence. The depth of the microchannels is 100 mm. The scale bar in (B) is 400 mm and the scale bars in (C) and (D) are 3 mm.

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 4180–4189 | 4185

Lab on a Chip Paper

at a TV5 (t4), and progresses to the reaction chamber. Theliquid then flows past the 4 side-arms, through a resistor andflows into the incubation CP that draws a precise volume,flushing it through the reaction chamber. Upon filling theincubation CP (t5), the sample moves continuously, andactivates the TV5, which is located upstream of the reactionchamber. Following activation, excess sample flows throughthe TV into the waste CP until the entire excess sample isdepleted and the flow stops at CRV5 (t6). The capillary pressurein the circuit then rises to the level of the waste CP, andtriggers the first RBV (t7) and drains the reagent from the firstside-arm, which flows back through the main channel and thereaction chamber to the CP until the CRV stops the drainage.Each arm is drained in sequence according to the threshold ofeach RBV (t8–t10). The flow path through the incubation CPremains open but, because the flow resistance is y55 timeshigher than that through the reaction chamber, virtually all

reagents flow back through the reaction chamber at apreprogrammed flow rate. This circuit was designed forperforming immunoassays.

The capillaric circuits were made out of PDMS using a twolevel SU8 mould as a 19 6 21 mm2 chip. The chips wereplasma activated to render them hydrophilic, and a hydro-phobic PDMS with vents and loading ports was used as thecover layer, Fig. 4.

To validate the design, solutions of food dyes and DI waterwere filled as outlined in Fig. 3, and recorded, Fig. 5 and MovieS4 in the ESI.3 The side-arms were filled by simply contactingthe extremity of the channel with a pipette tip. 1 mL of a blackfood dye, serving as a sample, was introduced into the mainloading port. The flow pattern replicated the steps outlinedabove accurately.

Fig. 3 Capillaric circuit for flowing a metered amount of sample through a reaction chamber followed by flow reversal and sequential flowing of 4 pre-loadedreagents at a different flow rate. (A) Symbolic and (B) schematic overviews of the circuit comprising a reaction chamber, 4 side-arms with RBVs and combined, flowresistors, an incubation CP, as well as a waste CP. (C) Schematic drawings outlining the operation of the capillaric circuit step-by-step. (t1) Initially, the cover is sealedwith the patterned lines of capture antibodies orthogonal to the reaction chamber. (t2) Next, various reagents (green, yellow, black and red) are dispensed into thelateral openings into the side-reservoirs and (t3) stop at the TVs1–4 at the intersection with the main channel. (t4) Then the sample is introduced through the loadingport, flowing via the main channel to (t5) the incubation CP which draws a predetermined amount at a controlled flow rate. (t6) Subsequently, when the liquid passesthe TV5 at the junction upstream of the reaction chamber, the excess amount of the sample flows through the shortcut directly into the waste CP. (t7–t10) Finally, theside-reservoirs are drained sequentially, and flow in reverse direction through the reaction chamber via the shortcut into the waste CP.

4186 | Lab Chip, 2013, 13, 4180–4189 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

Sandwich immunoassay in a capillaric circuit

To show the potential applications of capillaric circuits, asandwich immunoassay for CRP was performed as outlined inFig. 6A. We first patterned anti-CRP capture antibody on thecover of the chip using a CS perpendicular to the reactionchamber.34 Later on, biotinylated anti-CRP antibody was filledinto side-arm 1, washing buffer in side-arms 2 and 4, andstreptavidin conjugated to Alexa Fluor 488 in side-arm 3. Next,a 1 mL of phosphate buffer saline spiked with CRP was appliedto the main loading port, flowing and triggering the flow of allthe other reagents. Fig. 6B shows the assay results for triplicatemeasurements of the CRP concentrations between 0.01 and 10mg mL21 as well as a negative control. Two fluorescencemicrographs (vertical stripes) in a reaction chamber, corre-sponding to the signals of 10 mg mL21 concentration of CRPantigen were also illustrated as an inset of the graph. The scalebar of the micrograph is 100 mm. To achieve higher sensitivity,it will be necessary to increase the incubation time of thesample, and optimize the flushing time for the reagents, whilealso improving the stability of the surface, because the PDMSused slowly reverted to a hydrophobic state.

Conclusion

We introduced the concept of capillarics, in analogy toelectronics, as a more complex form of capillary systemsintroduced a decade ago. Whereas one can draw a generalconceptual analogy to electronics, there are also deepdifferences, because with capillaric circuits the fluid logic isrealized with the advancement and retreat of the wetting frontof the liquid in different parts of the circuit, in a time andspace dependent manner. We demonstrated that capillariccircuits can be made from a number of capillary and capillaricelements including CRVs, RBVs, resistors, vents, TVs, CPs,

closed and open reservoirs. We notably designed closedreservoirs with RBVs at the extremity and a combined TV/CRV at the intersection with the main channel, so thatreagents filled in each reservoir would be drained according toa pre-programmed sequence, regardless of the order withwhich they were filled. These elements were integrated into acapillaric circuit for a one-step immunoassay with flowreversal implementing distinct flow rates for the sample andsubsequent reagents, and comprising two CPs for samplemetering, and one acting as a final waste collector.Considering that the maximum aspect ratio in the capillaryelements and circuit presented in this paper is 3, thesecapillary components can be replicated in polymers usinginjection molding for making low cost microfluidic liquidhandling platforms. The concept of flow reversal, and thecapillary valves including retention burst valves (RBVs) and thecapillary trigger valves (TVs) presented in the paper can beintegrated with other microfluidic platforms including cen-trifugal microfluidics and vacuum assisted filling to enhancetheir functionalities. The combination of various elementsallows creating new fluidic logic operations and circuits, whichmay be designed at an abstract level, using symbols such asthe ones introduced in this manuscript, and then realizedphysically according to design rules yet to be established. Webelieve that many more fluidic circuits may be built both asmicrofabricated structures that afford great control, but alsousing porous supports such as paper that are inexpensive, orpossibly by combining the best of both worlds so as to obtain abetter control at a reduced cost. However, challenges remain.For example the number of RBVs that can be reliably operatedin parallel are limited by imperfections and hysteresis inwetting, which commands the use of significant differences inretention pressure, while requiring CPs with significantlyhigher capillary pressure than the highest RBV to ensure its

Fig. 4 Optical micrograph of a microfluidic capillaric circuit with 4 side-reservoirs and flow reversal made in PDMS. The size of the circuits is 19 6 21 mm2. The PDMScover includes vents and a loading port.

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 4180–4189 | 4187

Lab on a Chip Paper

drainage. Scaling up of capillaric circuits will depend onovercoming such limitations. However, as the examples shownhere demonstrate, circuits with moderate complexity canalready be designed and built. Whereas we demonstrated a

one-step immunoassay, improvement such as long termreagent storage are still needed to permit pre-filling of chipsindependently of usage. Furthermore, a more convenient assayreadout is also needed, and may be implemented using forexample silver amplification,37 permitting assay readout usinga cell phone, or an electrochemical detection using simpleelectronics only. We thus believe that capillarics will be usefulfor point-of-care diagnostic applications, as well as for manychemical and biochemical processes that require the sequen-tial addition and removal of multiple reagents.

Acknowledgements

We would like to acknowledge financial support from NSERC,CIHR, MITACS and CFI, and the assistance of the McGillNanotools Microfab Laboratory (funded by CFI, NSERC andNanoquebec). We also acknowledge Elizabeth Jones for accessto her lab. We wish to thank Hanadi Sleiman, Ali Tamayol,Mohammadali Safavieh, Mohammad Qasaimeh, VeroniqueLaforte, Sebastien Bergeron, Andy Ng, Jeffrey Munzar, and

Fig. 5 Video images illustrating the step-by-step operation of the capillariccircuit with flow reversal. Four food dyes, blue, red, orange, and green weredispensed to the side-reservoirs using a pipette. 1 mL of a black food dye servingas a sample was introduced into the main loading port. It flowed for y150 sthrough the metering pump (800 nL), flowed back via the small channel, andactivated the TV upstream of the reaction chamber. The excess sample was thusflushed via the shortcut directly into the waste CP (t6). Next, the reagents in theside-reservoirs are drained sequentially from bottom to top (blue, red, orange,and green, t7–t10) and flow in the reverse direction through the reactionchamber and the shortcut into the waste CP as well. The scale bar is 3 mm in theoverviews and 200 mm in the close-up views.

Fig. 6 Sandwich immunoassay for CRP carried out using a capillaric circuit. (A)The sandwich assay consists of five steps. First, CRP antigen is captured on thereaction chamber by the pre-immobilized capture antibodies. Next, biotinylateddetection CRP, washing buffer, fluorescent streptavidin, and second washingbuffer were flushed from the side channels sequentially. Finally, the stamp wasimaged with a fluorescence microscope. (B) Standard curves of the fluorescentsignals obtained from three independent sets of experiments. In eachexperiment, we tested six chips and measured the average intensities of thefluorescent signals in two patterned reaction zones in each chip. We then fitteda curve the set of data. The inset illustrates fluorescence micrographs (verticalstripes) correspond to the signals of 10 mg mL21 concentration of CRP antigen.The scale bar is 100 mm.

4188 | Lab Chip, 2013, 13, 4180–4189 This journal is � The Royal Society of Chemistry 2013

Paper Lab on a Chip

Huiyan Li for their help and discussion. DJ acknowledgessupport from Canada Research Chair.

References1 R. Safavieh, M. P. Roca, M. A. Qasaimeh, M. Mirzaei and

D. Juncker, J. Micromech. Microeng., 2010, 20, 055001.2 J. Moorthy, R. Burgess, A. Yethiraj and D. Beebe, Anal.

Chem., 2007, 79, 5322–5327.3 P. Yager, G. J. Domingo and J. Gerdes, Annu. Rev. Biomed.

Eng., 2008, 10, 107–144.4 L. Gervais, N. de Rooij and E. Delamarche, Adv. Mater.,

2011, 23, H151–H176.5 M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer and S.

R. Quake, Science, 2000, 288, 113–116.6 K. Choi, A. H. C. Ng, R. Fobel and A. R. Wheeler, Annu. Rev.

Anal. Chem., 2012, 5, 413–440.7 http://www.roymech.co.uk/Useful_Tables/Drawing/

Flow_sheets.html, Retrieved 2013/07/25.8 M. Rhee and M. A. Burns, Lab Chip, 2008, 8, 1365–1373.9 S. Haeberle and R. Zengerle, Lab Chip, 2007, 7, 1094–1110.

10 X. Li, D. R. Ballerini and W. Shen, Biomicrofluidics, 2012, 6,11301–1130113.

11 H. Noh and S. T. Phillips, Anal. Chem., 2010, 82, 8071–8078.12 E. Fu, S. A. Ramsey, P. Kauffman, B. Lutz and P. Yager,

Microfluid. Nanofluid., 2011, 10, 29–35.13 B. J. Toley, E. Fu and P. Yager, A powerless valving system for

fluid flow in paper networks, Okinawa, Japan, 2012.14 R. Safavieh, G. Z. Zhou and D. Juncker, Lab Chip, 2011, 11,

2618–2624.15 G. Zhou, X. Mao and D. Juncker, Anal. Chem., 2012, 84,

7736–7743.16 E. Delamarche, A. Bernard, H. Schmid, B. Michel and

H. Biebuyck, Science, 1997, 276, 779–781.17 D. Juncker, H. Schmid, U. Drechsler, H. Wolf, M. Wolf,

B. Michel, N. de Rooij and E. Delamarche, Anal. Chem.,2002, 74, 6139–6144.

18 M. Mirzaei, M. Pla-Roca, R. Safavieh, E. Nazarova,M. Safavieh, H. Li, J. Vogel and D. Juncker, Lab Chip,2010, 10, 2449–2457.

19 M. Wolf, D. Juncker, B. Michel, P. Hunziker andE. Delamarche, Biosens. Bioelectron., 2004, 19, 1193–1202.

20 M. Zimmermann, P. Hunziker and E. Delamarche, Biomed.Microdevices, 2009, 11, 1–8.

21 L. Gervais, M. Hitzbleck and E. Delamarche, Biosens.Bioelectron., 2011, 27, 64–70.

22 M. Hitzbleck, L. Gervais and E. Delamarche, Lab Chip,2011, 11, 2680–2685.

23 J. Wang, H. Ahmad, C. Ma, Q. Shi, O. Vermesh, U. Vermeshand J. Heath, Lab Chip, 2010, 10, 3157–3162.

24 M. Zimmermann, P. Hunziker and E. Delamarche,Microfluid. Nanofluid., 2008, 5, 395–402.

25 P. F. Man, C. H. Mastrangelo, M. A. Burns and D. T. Burke,presented in part at the MEMS’98, 1998.

26 J. Melin, N. Roxhed, G. Gimenez, P. Griss, W. van derWijngaart and G. Stemme, Sens. Actuators, B, 2004, 100,463–468.

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

28 M. Hitzbleck, L. Avrain, V. Smekens, R. D. Lovchik, P. Mertensand E. Delamarche, Lab Chip, 2012, 12, 1972–1978.

29 V. Linder, S. K. Sia and G. M. Whitesides, Anal. Chem.,2005, 77, 64–71.

30 P. Novo, F. Volpetti, V. Chu and J. P. Conde, Lab Chip, 2013,13, 641–645.

31 G. M. Walker and D. J. Beebe, Lab Chip, 2002, 2, 131–134.32 S.-J. Kim, S. Paczesny, S. Takayama and K. Kurabayashi,

Lab Chip, 2013, 13, 2091–2098.33 A. Papra, A. Bernard, D. Juncker, N. B. Larsen, B. Michel

and E. Delamarche, Langmuir, 2001, 17, 4090–4095.34 D. Juncker, H. Schmid, A. Bernard, I. Caelen, B. Michel,

N. de Rooij and E. Delamarche, J. Micromech. Microeng.,2001, 11, 532–541.

35 R. Safavieh and D. Juncker, Serpentine and leading edgecapillary pumps, Seattle, Washington, USA, 2011.

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

37 C. D. Chin, T. Laksanasopin, Y. K. Cheung, D. Steinmiller,V. Linder, H. Parsa, J. Wang, H. Moore, R. Rouse andG. Umviligihozo, Nat. Med., 2011, 17, 1015–1019.

This journal is � The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 4180–4189 | 4189

Lab on a Chip Paper