Wettability patterning in microfluidic devices using thermally-enhanced hydrophobic recovery of PDMS
Marc Pascual, Margaux Kerdraon, Quentin Rezard, Marie-Caroline Jullien, Lorène Champougny
aa r X i v : . [ phy s i c s . a pp - ph ] O c t Wettability patterning in microfluidic devices using thermally-enhancedhydrophobic recovery of PDMS
Marc Pascual , Margaux Kerdraon , Quentin Rezard , Marie-Caroline Jullien , and Lor`eneChampougny Gulliver, CNRS, ESPCI Paris, PSL University, 10 rue Vauquelin, 75005 Paris, France Institut de Physique de Rennes, UMR CNRS 6251, Bˆat. 11A, Campus de Beaulieu, 263 avenue du G´en´eralLeclerc, 35042 Rennes CEDEX, France Grupo de Mec´anica de Fluidos, Departamento de Ingenier´ıa T´ermica y de Fluidos, Universidad Carlos III deMadrid, Av. Universidad 30, 28911 Legan´es (Madrid), Spain
Abstract
Spatial control of wettability is key to many applications of microfluidic devices, ranging from double emulsiongeneration to localized cell adhesion. A number of techniques, often based on masking, have been developed to pro-duce spatially-resolved wettability patterns at the surface of poly(dimethylsiloxane) (PDMS) elastomers. A majorimpediment they face is the natural hydrophobic recovery of PDMS: hydrophilized PDMS surfaces tend to returnto hydrophobicity with time, mainly because of diffusion of low molecular weight silicone species to the surface.Instead of trying to avoid this phenomenon, we propose in this work to take advantage of hydrophobic recovery tomodulate spatially the surface wettability of PDMS. Because temperature speeds up the rate of hydrophobic recov-ery, we show that space-resolved hydrophobic patterns can be produced by locally heating a plasma-hydrophilizedPDMS surface with microresistors. Importantly, local wettability is quantified in microchannels using a fluorescentprobe. This “thermo-patterning” technique provides a simple route to in situ wettability patterning in closedPDMS chips, without requiring further surface chemistry.
An increasing number of applications require hydrophilic/ hydrophobic surface patterning, i.e. space-dependentwettability properties on a single surface [1, 2]. If thesewettability patterns can be useful at macroscale, for in-stance in droplet deposition for printing techniques [3]or control of heterogeneous nucleation of water [4], theyfind a considerable number of applications at microscale.These include pumpless fluid actuation [5, 6], doubleemulsion generation [7], wall-free flow control in open[8] and closed [9, 10] microfluidic devices, microdropletsgeneration and control [11, 12] and biological cell pat-terning [13, 14, 15], to name a few.Poly(dimethylsiloxane) (PDMS), which is one of themost commonly used materials in the fabrication of mi-crofluidic systems [16], offers naturally hydrophobic wet-ting conditions, therefore complicating the handling ofaqueous solutions. As a consequence, a number of tech-niques have been developed to hydrophilize PDMS sur-faces. They mostly follow two main routes: on the onehand, direct chemical modification of the surface throughO plasma treatments [17] or UV-ozone exposition [18]or, on the other hand, surface coating with additionalmaterial layers. Physical techniques have also been pro- posed, involving for instance surface roughness modifi-cation via laser ablation redeposits [19] or oxygenatedgroups removal using physical contact treatment [20].For the two main hydrophilization routes, a popularapproach to achieve spatially-resolved wettability pat-terns on PDMS surfaces relies on masks or stencils toselect the areas where the hydrophilizing treatment is ap-plied. For example, wettability patterns on PDMS canbe obtained using a photomask [21, 22], or simply epoxyglue [23] or marker [24] deposits, to shield portions of thesurface from plasma treatment before chip sealing. Somemethods require more advanced chemistry, like selectivegraft polymerization of hydrophilic polymers, either ini-tiated by UV exposure through a mask [25, 26, 27, 28] orinhibited by the O contained in integrated gas reservoirs[29]. Alternatively, mask-free techniques for wettabilitypatterning include flow confinement of chemical treat-ment using an inert solution [30] and spatially-controlledO plasma oxidation by playing on the geometry of thePDMS chip and plasma parameters [31].Unless permanent grafting of hydrophilic polymers isachieved [25, 26, 27, 28, 29], the hydrophobic recovery ofPDMS surfaces, i.e. their natural ability to restore theirhydrophobicity after an oxidizing treatment (air/oxygenplasma [32, 33, 34], UV-ozone [35, 21] or electric dis-1harge [36, 37]), is usually seen as a major impedimentto wettability patterning. This phenomenon is attributedto a combination of different mechanisms, including (i)reorientation of surface polar groups towards the bulkand of non-polar groups towards the surface [32, 33],(ii) elimination of hydroxyl groups by condensation ofsilanols [32] (iii) diffusion of low molecular weight sili-cone species (either preexisting or produced in situ bythe oxidizing treatment) from the bulk to the surface[33, 38, 37, 39]. The latter mechanism (iii) is usuallythought to provide the major contribution to hydropho-bic recovery [33, 39]. The rate of hydrophobic recoverycan therefore be decreased by removing low molecularweight chains (through solvent extraction [33] or extracuring time before plasma [40]) or by slowing down thediffusion of low molecular weight species (for exampleby storage at low temperature [24]). Conversely, studieson PDMS macroscopic samples showed that hydropho-bic recovery is accelerated when increasing temperature[32, 34, 37, 41]. Additionally, oxidized PDMS surfaceshave been observed to be covered with a thin brittlesilica-like layer, in which cracks provide preferential chan-nels for the diffusion of low molecular weight species to-wards the surface [42, 38, 34, 43, 44].In this work, instead of considering hydrophobic recoveryas an impediment, we propose to take advantage of itstemperature dependence to locally tune the wettabilityof PDMS surfaces. The idea consists in performing localheating of a plasma-hydrophilized PDMS surface witha microresistor [45], thus accelerating the hydrophobicrecovery where a hydrophobic pattern is desired, whilekeeping the rest of the surface hydrophilic. To our knowl-edge, such a space-dependent tuning of hydrophobic re-covery for patterning purposes has never been reportedin the literature. We show that our method allows in-situ patterning of sealed microchannels without need ofchemically advanced surface treatments, and with a res-olution of a few hundreds of microns. In general, wetta-bility properties are probed at macroscale using contactangle measurements. Inspired by the work of Gucken-berger et al. [46] we develop and validate the use of afluorescent protein (BSA-FITC) as a quantitative wet-tability probe at in microchannels. Our experimentalmaterials and procedures are described in section 2. Insection 3, we then present our results on the tempera-ture dependence of hydrophobic recovery at macroscale(using contact angle measurements) and at microscale(using fluorescent protein adsorption as a local wettabil-ity probe), before finally demonstrating hydrophobicitypatterning in microfluidic chips with our technique. Three experimental configurations are studied, goingprogressively from macroscale to microscale: (i) PDMSmacroscopic surfaces submitted to a global heating, (ii)
PDMS chip curing Plasma treatement and chip bonding
Post-plasma baking ( (cid:1) t) Flow of BSA solution and incubationRinsing with PBSObservation with fluorescence microscopePDMS substrate spincoating and curingRemoval of chip from resistor pattern glass PDMS layerresistorPDMS chipBSA solutionPBSglasschannel
Exp. type (i) S a m p l e p r e p a r a t i on W e tt a b ili t y m eas u r e m e n t glassPDMS layer Contact angle measurement
Exp. type (ii) Exp. type (iii) wateradsorbed BSA UV light
Figure 1:
Schematics of the experimental protocol imple-mented, with three major steps: sample preparation, hy-drophobic recovery through post-plasma baking and wetta-bility measurements. These steps are addressed differently,depending on the type of experiment performed: (i) globalheating of macroscopic PDMS surfaces, (ii) global heating ofPDMS microchannels and (iii) local heating of PDMS mi-crochannels.
PDMS microchannels submitted to a global heating and(iii) PDMS microchannels submitted to a local heating.As summarized in Fig. 1, all kinds of experiments areperformed following the same succession of steps (al-though using different techniques). First, solid PDMSsurfaces are fabricated and made hydrophilic with O plasma treatment (Sec. 2.1). These surfaces are thenheated, either globally or locally (Sec. 2.2), at varioustemperatures and for various durations ∆ t to allow forhydrophobic recovery. Finally, the wettability of the hy-drophobically recovered surface is assessed (Sec. 2.3). Throughout the whole study, PDMS elastomers are madeof Sylgard 184 (Dow Corning) with a weight ratio 1/10of crosslinker/polymer. For macroscopic experiments (i),the sample consists of a 30 ± µ m-thick PDMS layerspin-coated onto a silicon wafer and subsequently curedat 70 ◦ C for 2 hours. For the purpose of microscale stud-ies (ii) and (iii), a straight channel design (400 µ m inwidth, 2 . µ m in height)was created on a silicon wafer using standard soft lithog-raphy techniques [16]. Liquid PDMS is poured onto thismold, cured at 70 ◦ C for 2 hours; inlets and outlets of0 .
75 mm in diameter are punched into the chips, whichare finally bonded onto a substrate. For experiments2 b) (c) gold connectorchromium resistorchannel xy (a) -1,0 -0,5 0,0 0,5 1,020406080100120140 -0,4 -0,2 0,0 0,2 0,44080120 w x T ( ° C ) x (mm) U = 6 VU = 6.9 VU = 7.9 VU = 8.5 V T ( ° C ) y (mm)w y z Figure 2: (a) Sketch of the local heating device and its posi-tion with respect to the microfluidic channel to be patterned.— (b) Temperature profiles along (b) the longitudinal x -axisand (c) the transverse y -axis at the surface of a R = 0 .
78 kΩmicro-resistor of dimensions w x × w y = 50 µ m × µ m.Tuning the applied voltage U allows to change the peak tem-perature T max of the profile. of type (ii) (global heating), this substrate is simply a150 µ m-thick glass cover slip. For experiments of type(iii), it consists in a 30 ± µ m-thick PDMS layer, curedat 70 ◦ C for 2 hours, on top of the resistor pattern usedfor local heating (see Fig. 2 and description in the sectionbelow). The hydrophilization of all PDMS surfaces (andchip bonding in experiments (ii) and (iii)) is achievedwith an O plasma treatment (CUTE apparatus, Femto-science). The plasma is generated by applying a power of100 W (experiments (i)) or 25 W (experiments (ii) and(iii)) at a frequency of 50 kHz during 50 s within a vac-uum chamber filled with oxygen at a pressure of 0.5 Torr(67 Pa). Note that the potential impact of using twoslightly different hydrophilization protocoles for experi-ments of type (i) and (ii)-(iii) will be further discussedin paragraph 3.2. After hydrophilization, PDMS surfaces are baked at var-ious temperatures and for various amounts of time toallow for hydrophobic recovery. For experiments of type(i) and (ii), the samples are heated globally in conven-tional gravity ovens. In order to avoid contamination ofPDMS surfaces in the case of experiments (i), the sam-ples are stored in individual PELD or PYREX petri dish,depending on the temperature of the post-plasma baking(respectively under and above 70 ◦ C).For experiments of type (iii), a microscale heating sys-tem is fabricated from a 700 µ m-thick glass wafer coveredwith a 150 nm gold layer and a 15 nm chromium layer(A.C.M. France). These layers are successively etched,using soft lithography printed S1818 photoresist (Mi- croChem) to protect the desired pattern of chromiumresistors and gold connectors. Fig. 2a shows the geom-etry of a resistor and its positioning with respect to themicrochannel to be patterned. Each chromium resistor is w x = 50 µ m in width and w y = 400 µ m in length. Heatis produced by Joule effect when a voltage difference U is applied to the resistor.The temperature profile generated at the surface of aresistor (covered with a 30 µ m insulating PDMS layer) ischaracterized with an infrared camera (FLIR - A6500SC)coupled to an infrared zoom lens (FLIR ATS). Impor-tantly, the glass wafer is placed on a 2 cm-thick aluminumblock, serving as a heat sink to guarantee a stationarytemperature profile despite the continuous application ofheating power [47]. Fig. 2b shows the temperature pro-files obtained along the longitudinal x -direction for dif-ferent applied voltages U , yielding maximum profile tem-peratures T max = 70 , ,
110 and 130 ◦ C, respectively.The inset (Fig. 2c) displays the temperature profile mea-sured across the resistor in the transverse y -directionfor T max = 130 ◦ C. The temperature remains reason-ably homogeneous ( i.e. between 110 and 130 ◦ C) within y = ± µ m from the resistor’s center but drops whenapproaching the edges, close to the gold connectors. Ad-ditionally, COMSOL simulations of heat diffusion in ourgeometry show that the maximum temperature also re-mains acceptably uniform along the vertical z -direction,dropping from 130 ◦ C at the bottom to about 110 ◦ C atthe top of a 30 µ m-height microchannel. The wettability of the thermally-aged PDMS surfaces isfinally evaluated. For macroscopic experiments (i), con-tact angles measurements are performed, while a fluo-rescent probe is used in microscale experiments (ii) and(iii), as described below.
Advancing contact angle.
For experiments of type (i),we determine the advancing contact angle θ a of ultrapurewater (Milli-Q) on macroscopic PDMS surfaces by di-rect image analysis using a commercial solution (PSA30-KRUSS). The sample is first left to cool down for oneminute after the end of thermal aging. A 5 µ L droplet isthen deposited onto the substrate and inflated with anadditional 5 µ L at a rate of 30 µ L / min. The advanc-ing contact angle θ a is measured just before depinningof the contact line. Each measurement is repeated fivetimes on two different samples made of different batchesof Sylgard 184. Error bars in Fig. 3 represent the stan-dard deviation of these measurements. Local wettability measurements.
Because contact an-gle measurements can hardly be performed in sub-millimetric channels [48], we turn to a fluorescence-basedmethod to assess the wettability of PDMS surfaces in mi-3roscale experiments (ii) and (iii). Only a few indirectwettability measurements using fluorescent probes havebeen performed in microchannels so far. Guckenberger etal. [46] have used the local depletion of Red Nile coatedon polystyrene to quantify the penetration of O plasmatreatment in microchannels. On PDMS surfaces, the hy-drophilic copolymer PLL-g-PEG has been employed byBodin et al. [24] to qualitatively visualize hydrophilicchannel surfaces.In order to probe the wettability of PDMS within mi-crochannels, we use Bovine Serum Albumine tagged withFluorecein Isothiocyanate (BSA-FITC, Sigma-Aldrich),as this fluorescent protein is known to preferentially ad-sorb on hydrophobic surfaces [49]. Similar protocols areused when the microchannel is heated globally (experi-ments (ii)) or locally (experiments (iii)), as sketched inFig. 1. The fluorescent protein solution is made by dis-solving BSA-FITC in Phosphate Saline Buffer (PBS; onetablet in 200 mL of deionized water: 0.01 M phosphatebuffer, 0.0027 M potassium chloride and 0.137 M sodiumchloride - pH 7.4, Sigma-Aldrich) at a concentration of0 . / L. This concentration is larger than the thresholdconcentration 0 .
02 g / L above which the native PDMSsurface becomes saturated with BSA molecules [50].After the post-plasma baking step, the microfluidicchip described in section 2.1 is first left to cool down toambient temperature. The BSA-FITC solution is flownin the microchannel for 2 min by imposing an overpres-sure of 75 mbar with a pressure controller (Fluigent -MFCS 4C). The BSA-filled channel is then left to incu-bate for 30 min, either with or without any flow. The in-cubation time was chosen long enough to reach a plateauof BSA adsorption at a given concentration, but shortenough not to make the whole process too cumbersome.It was also experimentally noticed that keeping the BSAsolution flowing during the incubation step led to a morehomogeneous BSA adsorption, maybe due to a reducedadsorption of aggregates. The non-adsorbed proteins arefinally rinsed by flushing the channel with PBS for 5 minwith an overpressure of 75 mbar.In order to quantify the amount of BSA-FITC ad-sorbed at the walls, the PBS-filled microchannel isobserved using a fluorescence microscope (Zeiss Ob-server.A1) with 5x or 20x dry objective (N.A. 0.12 or0.5). The sample is illuminated with UV light (HBO103W/2 - OSRAM lamp, filtered at 495 nm) and the flu-orescence signal is recorded at a wavelength of 517 nm,with an exposure time of 1 second (Andor Zyla camera).
Experiments of type (i) aim at quantifying how the hy-drophobic recovery of macroscopic PDMS samples de- (cid:0) t (min) θ a ( ° ) -1 ) × -3 l n ( (cid:1) ) T = 20°CT = 70°CT = 130°CT = 180°C
Figure 3:
Advancing contact angle θ a of a sessile water dropon PDMS surfaces as a function of the post-plasma bakingtime ∆ t and for various baking temperatures T = 20 (roomtemperature), 70, 130 and 180 ◦ C. For each temperature,the solid line is a fit of the data (symbols) using a stretchedexponential function. The error-bars correspond to the max-imal dispersion of the data. — Inset: Arrhenius plot of therecovery time τ , needed for the PDMS surface to go backto θ a = 70 ◦ , i.e. plot of the logarithm of τ (in minutes) asa function of the inverse of the temperature T . The solid lineis a linear fit of the data (symbols). pends on temperature. The advancing contact angle θ a of a sessile water drop deposited on the sample sur-face is measured for different baking times ∆ t after theplasma treatment. Fig. 3 shows the time evolution of θ a for PDMS samples baked at different temperatures: T = 20 (room temperature), 70, 130 or 180 ◦ C. Thedata show that the hydrophobic recovery of PDMS iscomplete after several months at room temperature, butcan be achieved in only a few hours when the bakingtemperature is increased up to 180 ◦ C. For interpolationpurposes, the hydrophobic recovery curve at each tem-perature is fitted by a stretched exponential function oftype θ a = a + b exp[ − (∆ t/c ) d ], where a , b , c and d are ad-justable parameters, as inspired by Kim et al. [37]. Thecharacteristic time τ Θ is defined as the time needed, for asample baked at a given temperature, to reach θ a = Θ (indegrees) and is deduced from the fitted functions (solidlines in Fig. 3). As an example, the inset of Fig. 3 showsan Arrhenius plot of τ , that is to say the evolution ofln( τ ) with the inverse of the temperature 1 /T .As already observed in other studies [32, 37, 36, 34],the Arrhenius plot of τ is linear. Additionally, we findthat the corresponding slope is essentially independent ofthe definition chosen for τ Θ for target angles in the rangeΘ = 60 − ◦ . The characteristic time for hydrophobicrecovery may thus be written as τ ∝ exp( E a /RT ) , (1)where we introduce the universal gas constant R =8 .
314 J / mol / K and an activation energy E a . The ac-4ivation energy E a = 42 ± / mol deduced from thedata in Fig. 3 is in good agreement with numericalvalues reported in the literature [32, 37, 36, 34]. Re-markably, these values all lie in the range E a = 30 −
60 kJ / mol for various hydrophilization processes (elec-trical discharges, air or oxygen plasma), exposure timesand PDMS crosslink densities.The Arrhenius behavior of the hydrophobic recoverytime and its robustness may be understood as follows.Diffusion of species in elastomers can usually be de-scribed as an activated process[51], where the diffusioncoefficient has the form D ∝ exp( − E a /RT ). Then, as-suming the hydrophobic recovery of PDMS is mainly con-trolled by the diffusion of low molecular weight species tothe surface, the recovery timescale τ and the diffusioncoefficient D are related by δ ∼ D × τ , where δ is thetypical distance over which low molecular weight specieshave to diffuse to reach the surface. The lengthscale δ islikely to be the thickness of the porous silica layer formedat the surface of the oxydized PDMS [37]. Although itdepends on parameters such as plasma intensity and ex-posure time [34, 52, 53], this thickness δ does not varywith temperature, hence the Arrhenius behavior of Eq.(1), where E a can be interpreted as the diffusion activa-tion energy of low molecular weight species in the poroussilica layer. In this section, we present results corresponding to exper-iments of type (ii), where plasma-hydrophilized PDMSmicrochannels are further baked in ovens at various tem-peratures and for various amounts of time. The wet-tability in those channels is assessed by observing thefluorescence of BSA-FITC adsorbed onto the channel’swalls, as this protein is known to preferentially adsorb onhydrophobic surfaces. Fluorescence pictures of a 400 µ m-wide PDMS microchannel are shown in Fig. 4a, after a5 min and a 150 min bake in an oven at 130 ◦ C (∆ t = 0corresponds to the time when the plasma treatment isperformed). While no fluorescence – hence no proteinadsorption – is visible on the freshly hydrophilized chan-nel (left picture), a strong fluorescence signal is observedin the channel that has been heated for 150 min (rightpicture), demonstrating the adsorption of BSA-FITC onthe hydrophobically-recovered PDMS walls.The average intensity I fluo of the fluorescence signalwithin the microchannel is systematically measured forvarious baking temperatures and baking times. Thisvalue is then corrected from the background intensity I (average intensity outside of the microchannel) and nor-malized by the asymptotic value reached after a 24-dayshydrophobic recovery (∆ t → ∞ ) at the given temper-ature. The resulting normalized intensity ∆ I , defined Δ t = 5 min Δ t = 150 min T = ° C (a)0 100 200 300 4000 ,00,20,40,60,81,0 Δ I T = 20°CT = 70°C
T = 130°C (b)
20 40 60 80 100 1200,00,20,40,60,81,0 Δ (cid:2) θ a (°) (c) Δ t (min) Figure 4: (a) Top view of the microchannel illuminatedby UV light after a 5 min and 150 min bake at 130 ◦ C fol-lowing the plasma treatment (∆ t = 0). The scale bars are200 µ m. — (b) Evolution of the average normalized fluores-cence intensity ∆ I in the channel (see Eq. (2)) as a functionof the baking time ∆ t , when the sample is baked at roomtemperature (20 ◦ C), 70 and 130 ◦ C. Error bars representthe standard deviation of the fluorescence intensity. — (c)Normalized fluorescence intensity ∆ I in microchannels as afunction of the corresponding contact angle θ a measured onmacroscopic PDMS surfaces baked in the same conditions.The dashed line is a linear fit of the data. as ∆ I (∆ t, T ) = I fluo (∆ t, T ) − I (∆ t, T ) I fluo ( ∞ , T ) − I ( ∞ , T ) , (2)is plotted in Fig. 4b as a function of time ∆ t for bakingtemperatures of 20, 70 and 130 ◦ C. The fluorescenceintensity in hydrophobically recovering microchannels isobserved to follow the same qualitative trends as contactangle measurements on macroscopic PDMS surfaces. Ata given temperature, ∆ I increases with the baking timeand the larger the temperature, the faster the recoverytowards fully hydrophobic surfaces (∆ I = 1).In order to evaluate how quantitative the fluorescenceintensity measurements are, as a wettability probe, wecorrelate the contact angle measurements presented inFig. 3 to the fluorescence intensity measurements. Foreach data point in Fig. 4b, we compute the advancingcontact angle on a macroscopic PDMS surface that hasbeen baked in the same conditions (temperature and du-ration) as the microchannel, using the stretched expo-nential functions fitted in paragraph 3.1. Fig. 4c thusdisplays the fluorescence intensity ∆ I observed within amicrochannel as a function of the contact angle θ a mea-sured on the corresponding macroscopic PDMS surface.A first important observation is that the intensitiesmeasured in microchannels baked at various tempera-tures all collapse onto the same curve when plotted as afunction of the contact angle. This shows that fluores-cence intensity does not depend on how a given contact5ngle was achieved (e.g. by baking the channel at a hightemperature for a short time or at a low temperature for along time). A second observation is that the correlationbetween the fluorescence intensity ∆ I and the contactangle is monotonous and approximately linear with θ a .Both observations lead to the conclusion that the fluo-rescence intensity of BSA-FITC proteins adsorbed ontoPDMS surfaces can be used as an indirect and quantita-tive indicator of the contact angle, and thus of the wetta-bility inside PDMS microchannels. In our experimentalconditions, ∆ I ≈ . × θ a , where the contact angle isexpressed in degrees. Note that the normalization of thefluorescence intensity by its maximal value, obtained inthe same conditions on a fully hydrophobic PDMS sur-face, is essential. The non-normalized intensity I fluo mayindeed depend on the concentration of the protein solu-tion and the PBS flushing flow rate, but has also beenobserved to vary with the pH and ionic strength of thesolution, as well as with the polarization of light [54].Finally, we briefly comment on the potential effectsof using different plasma parameters for experiments (i)and (ii) (see paragraph 2.1) on the calibration curve ∆ I vs θ a (Fig. 4c). Following Bayley et al. [52], increasingthe plasma power from 25 W (ii) to 100 W (i) in ourexperimental conditions increases the thickness of thesurface oxidized layer from about 5 to 20 nm, thus slow-ing down the diffusion of low molecular weight chainsto the surface. Concomitantly, the appearance of cracksin the oxydized layer at high plasma power[44] may in-stead accelerate this diffusion. Owing to these two an-tagonistic effects, we observed (data not shown) that thetimescale of hydrophobic recovery changes only by a fac-tor of 2 upon decreasing the plasma power from 100 Wto 25 W. Shifting the contact angle data of Fig. 3 in timeby the corresponding amount barely affects the calibra-tion curve of Fig. 4c, thereby showing the robustness ofour conclusions. We now use the wettability probing technique describedin the previous paragraph to demonstrate that hydropho-bic regions can be patterned in a microchannel by locallyincreasing the temperature in the channel using a mi-croresistor (see paragraph 2.2).As an example, Fig. 5a shows the fluorescence signalobserved in a microchannel heated locally by a 50 µ m-wide microresistor at a peak temperature T = 130 ◦ Cduring ∆ t = 30 min. Note that, because the tempera-ture decreases sharply in the y -direction at the edges ofthe resistor (see Fig. 2c), the fluorescent intensity is onlyconsidered up to y = ± µ m from the microchannelcentral line. The corresponding normalized intensity pro-file ∆ I ( x ) (averaged across the y -direction) is displayedin Fig. 5b, along with the temperature profile T ( x ) gen-erated by the resistor. In this case, the definition of thenormalized intensity ∆ I (Eq. (2)) becomes a function of
250 µm (a)(b)(c) (d) w xy -1,0 -0,5 0,0 0,5 1,00,00,10,20,30,40,5 ∆ I x (mm) 20406080100120140T max ∆ I max T ( ° C ) ∆ I m a x (cid:3) t (min) θ a , m a x ( ° ) w / ( mm ) (cid:4) t (min) Figure 5: (a) Fluorescence image of a microchannel heatedlocally by a 50 µ m wide microresistor at a peak temperature T max = 130 ◦ C during ∆ t = 30 min. — (b) Correspondingnormalized intensity profile ∆ I ( x ) as defined in Eq. (3) (solidblue line) and temperature profile T ( x ) generated by the re-sistor (dashed red line). — Evolution of the intensity peakwidth at half-maximum w / (c) and maximum value ∆ I max (d) as a function of the baking time ∆ t using the temperatureprofile displayed in (b). Open circles are data obtained forseveral repetitions of the experiments and filled symbols showthe corresponding average value and standard deviation. the space coordinate x along the channel:∆ I (∆ t, T ( x )) = I fluo (∆ t, T ( x )) − I (∆ t, T ( x )) I fluo ( ∞ , T max ) − I ( ∞ , T max ) , (3)where I fluo (∆ t, T ( x )) is the raw fluorescence intensityprofile within the microchannel and I (∆ t, T ( x )) is theraw background intensity profile outside the microchan-nel. The normalization is done using the intensity I fluo ( ∞ , T max ) − I ( ∞ , T max ) obtained when the PDMSsurface has fully recovered its hydrophobicity after hav-ing been heated at a temperature T max corresponding tothe peak value of the temperature profile.Fig. 5a and b show that local heating of a PDMS sur-face with a stationary temperature profile T ( x ) indeedgives rise to a fluorescence peak centered on the microre-sistor ( x ≈
0) and characterized by its maximum value∆ I max and width at half-maximum w / . Repeating theexperiment using the same temperature profile but var-ious baking times, we obtain the width w / and am-plitude ∆ I max of the peak as a function of the bakingtime ∆ t , reported in Fig. 5c and d, respectively. Asestablished in paragraph 3.2, this fluorescence enhance-ment corresponds to a local increase in hydrophobicitythat can be quantified using the linear correspondence∆ I ∝ θ a . Hence, in Fig. 5d, the maximum intensity value6 I max of the fluorescence profile is converted into the cor-responding maximum advancing contact angle θ a, max .Although the temperature profile T ( x ) remains sta-tionary, the wettability profile evolves with the bakingtime ∆ t because low molecular weight chains diffusemuch slower than heat in PDMS. Both the amplitude∆ I max (or equivalently θ a, max ) and the width w / ofthe fluorescence peak exhibit increasing trends with ∆ t ,as shown in Fig. 5c and d. Note that the widening ofthe fluorescence peak with time is most likely due tothe lateral diffusion of low molecular weight species ( i.e. in directions parallel to the PDMS surface). Despitethe scattering of the measured peak widths w / , onecan estimate from Fig. 5c the typical resolution of thepattern to be a few hundreds of microns for a resistorwidth w x = 50 µ m. From Fig. 5d, the time τ at which θ a, max = 70 ◦ is found to be about 20 ±
10 min. This valueis comparable to, albeit somewhat smaller than, the time τ = 60 ± T = T max = 130 ◦ C (seeparagraph 3.1). The reason for this discrepancy may liein the fact that PDMS elastomers dilate upon increas-ing temperature. Temperature gradients, such as thosegenerated by our local heating device, cause a localizeddeformation of the PDMS [55, 56] that may result inlarger mechanical stresses as compared to when PDMSis heated (and therefore dilates) globally. This may pro-mote the apparition of cracks in the silica-like layer at thePDMS surface [38, 36] and, as a consequence, acceleratethe hydrophobic recovery of locally heated samples.
The wettability of PDMS is a key parameter in manymicrofluidics applications, yet its control can be chal-lenging. PDMS elastomers indeed naturally recovertheir original hydrophobicity after an oxidizing treat-ment, mainly because of the migration of low molec-ular weight chains towards the surface. Because thisdiffusion-controlled mechanism is strongly acceleratedby temperature, we proposed to take advantage of hy-drophobic recovery to pattern hydrophobic patches inotherwise hydrophilic microchannels by locally heatingthem. We first characterized the rate of hydrophobic re-covery of PDMS macroscopic surfaces as a function ofbaking time and temperature, using (advancing) contactangle measurements. The same experiments were thencarried out at microscale, using adsorption of a fluores-cent protein (BSA-FITC) as a hydrophobicity marker inPDMS microchannels. Comparison between the two setsof data revealed a linear correlation between contact an-gle and fluorescence intensity in our experimental condi-tions. This fluorescent wettability probe allowed us tocharacterize quantitatively the space-resolved hydropho-bic patterns created in hydrophilized PDMS channels lo-cally heated with a rectangular microresistor. This “thermo-patterning” method could be easily ex-tended to create more elaborated wettability patterns bychanging the microresistor geometry and choosing ad-equate temperature and baking time. Because the lo-cal resistance depends on the local width of the resistor,smooth wettability gradients may also be obtained us-ing resistors of varying width. It should be stressed thatthe obtained wettability patterns are ephemeral, as thenon-heated surfaces will slowly recover their original hy-drophobicity with time. However, the durability of thepattern may be improved by simply storing the treatedchips at low tempreature[24] or filling the channel withwater or culture medium[57].One of the main interests of “thermo-patterning” isthat hydrophobic patches of a few hundreds of mi-crons in size can be patterned on demand in sealedPDMS chips, without need of any further chemicaltreatment. Space-resolved in situ wettability pattern-ing techniques indeed usually require advanced polymerchemistry[25, 27, 28, 29]. Additionally, simpler meth-ods to pattern sealed chips, using for example spatiallycontrolled oxydation[31] or flow confinement of chemicaltreatement[30], can only be applied to entire channels.Because it simply takes advantage of the natural polydis-persity of commercial PDMS elastomers, our techniquedoes not raise any biocompatibility issues. “Thermo-patterning” therefore stands as a valuable complementto existing in situ wettability patterning techniques.
Acknowledgments
The authors are grateful to St´ephanie Descroix for intro-ducing them to Bovine Serum Albumine and to JavierRodr´ıguez Rodr´ıguez for fruitful discussions regardingthe manuscript.This work was supported by CNRS, ESPCI Paris,Agence Nationale de la Recherche (ANR) under thegrant 13-BS09-0011-01, IPGG (Equipex ANR-10-EQPX-34 and Labex ANR-10-LABX-31), PSL (Idex ANR-10-IDEX-0001-02) and Dim-NanoK R´egion Ile de France.
References [1] D. Xia, L. M. Johnson, and G. P. L´opez,“Anisotropic wetting surfaces with one-dimesionaland directional structures: Fabrication approaches,wetting properties and potential applications,”
Ad-vanced Materials , vol. 24, no. 10, pp. 1287–1302,2012.[2] E. Ueda and P. A. Levkin, “Emerging appli-cations of superhydrophilic-superhydrophobic mi-cropatterns,”
Advanced Materials , vol. 25, no. 9,pp. 1234–1247, 2013.73] D. Tian, Y. Song, and L. Jiang, “Patterning of con-trollable surface wettability for printing techniques,”
Chemical Society Reviews , vol. 42, no. 12, pp. 5184–5209, 2013.[4] K. K. Varanasi, M. Hsu, N. Bhate, W. Yang, andT. Deng, “Spatial control in the heterogeneous nu-cleation of water,”
Applied Physics Letters , vol. 95,no. 9, p. 094101, 2009.[5] A. A. Darhuber and S. M. Troian, “Principlesof microfluidic actuation by modulation of surfacestresses,”
Annu. Rev. Fluid Mech. , vol. 37, pp. 425–455, 2005.[6] A. Ghosh, R. Ganguly, T. M. Schutzius, and C. M.Megaridis, “Wettability patterning for high-rate,pumpless fluid transport on open, non-planar mi-crofluidic platforms,”
Lab on a Chip , vol. 14, no. 9,pp. 1538–1550, 2014.[7] A. Aserin,
Multiple emulsion: technology and appli-cations , vol. 1. John Wiley & Sons, 2008.[8] N. M. Oliveira, A. I. Neto, W. Song, and J. F. Mano,“Two-dimensional open microfluidic devices by tun-ing the wettability on patterned superhydrophobicpolymeric surface,”
Applied Physics Express , vol. 3,no. 8, p. 085205, 2010.[9] B. Zhao, J. S. Moore, and D. J. Beebe, “Surface-directed liquid flow inside microchannels,”
Science ,vol. 291, no. 5506, pp. 1023–1026, 2001.[10] S. Wang, T. Wang, P. Ge, P. Xue, S. Ye, H. Chen,Z. Li, J. Zhang, and B. Yang, “Controlling flow be-havior of water in microfluidics with a chemicallypatterned anisotropic wetting surface,”
Langmuir ,vol. 31, no. 13, pp. 4032–4039, 2015.[11] K. Handique, B. P. Gogoi, D. T. Burke, C. H. Mas-trangelo, and M. A. Burns, “Microfluidic flow con-trol using selective hydrophobic patterning,” in
Mi-cromachined Devices and Components III , vol. 3224,pp. 185–196, International Society for Optics andPhotonics, 1997.[12] T. Kobayashi, K. Shimizu, Y. Kaizuma,and S. Konishi, “Novel combination of hy-drophilic/hydrophobic surface for large wettabilitydifference and its application to liquid manipula-tion,”
Lab on a Chip , vol. 11, no. 4, pp. 639–644,2011.[13] J. L. Tan, W. Liu, C. M. Nelson, S. Raghavan, andC. S. Chen, “Simple approach to micropattern cellson common culture substrates by tuning substratewettability,”
Tissue Engineering , vol. 10, no. 5-6,pp. 865–872, 2004. [14] N. Patrito, C. McCague, P. R. Norton, and N. O.Petersen, “Spatially controlled cell adhesion via mi-cropatterned surface modification of poly (dimethyl-siloxane),”
Langmuir , vol. 23, no. 2, pp. 715–719,2007.[15] J.-P. Frimat, H. Menne, A. Michels, S. Kittel,R. Kettler, S. Borgmann, J. Franzke, and J. West,“Plasma stencilling methods for cell patterning,”
Analytical and Bioanalytical Chemistry , vol. 395,no. 3, pp. 601–609, 2009.[16] Y. Xia and G. M. Whitesides, “Soft lithography,”
Angewandte Chemie International Edition , vol. 37,no. 5, pp. 550–575, 1998.[17] J. R. Hollahan and G. L. Carlson, “Hydroxylation ofpolymethylsiloxane surfaces by oxidizing plasmas,”
Journal of Applied Polymer Science , vol. 14, no. 10,pp. 2499–2508, 1970.[18] Y. Berdichevsky, J. Khandurina, A. Guttman,and Y.-H. Lo, “Uv/ozone modification of poly(dimethylsiloxane) microfluidic channels,”
Sensorsand Actuators B: Chemical , vol. 97, no. 2-3, pp. 402–408, 2004.[19] S. van Pelt, A. Frijns, R. Mandamparambil, andJ. den Toonder, “Local wettability tuning with laserablation redeposits on pdms,”
Applied Surface Sci-ence , vol. 303, pp. 456–464, 2014.[20] D. J. Guckenberger, E. Berthier, E. Young, and D. J.Beebe, “Induced hydrophobic recovery of oxygenplasma-treated surfaces,”
Lab on a Chip , vol. 12,no. 13, pp. 2317–2321, 2012.[21] K. Ma, J. Rivera, G. J. Hirasaki, and S. Biswal,“Wettability control and patterning of pdms usinguv–ozone and water immersion,”
Journal of Colloidand Interface Science , vol. 363, no. 1, pp. 371–378,2011.[22] B. J. Larson, S. D. Gillmor, J. M. Braun, L. E. Cruz-Barba, D. E. Savage, F. S. Denes, and M. G. Lagally,“Long term reduction of poly(dimethylsiloxane) sur-face hydrophobicity via cold-plasma treatments,”
Langmuir , vol. 29, pp. 12990–12996, 2013.[23] S. Li, X. Gong, C. S. Mc Nally, M. Zeng, T. Gaule,C. Anduix-Canto, A. N. Kulak, L. A. Bawazer, M. J.McPherson, and F. C. Meldrum, “Rapid prepara-tion of highly reliable pdms double emulsion mi-crofluidic devices,”
RSC Advances , vol. 6, no. 31,pp. 25927–25933, 2016.[24] N. Bodin-Thomazo, F. Malloggi, and P. Guenoun,“Marker patterning: a spatially resolved methodfor tuning the wettability of pdms,”
RSC Advances ,vol. 7, no. 73, pp. 46514–46519, 2017.825] S. Hu, X. Ren, M. Bachman, C. E. Sims, G. Li,and N. L. Allbritton, “Surface-directed, graft poly-merization within microfluidic channels,”
Analyticalchemistry , vol. 76, no. 7, pp. 1865–1870, 2004.[26] N. Patrito, C. McCague, S. Chiang, P. R. Norton,and N. O. Petersen, “Photolithographically pat-terned surface modification of poly (dimethylsilox-ane) via uv-initiated graft polymerization of acry-lates,”
Langmuir , vol. 22, no. 8, pp. 3453–3455,2006.[27] A. R. Abate, A. T. Krummel, D. Lee, M. Marquez,C. Holtze, and D. A. Weitz, “Photoreactive coatingfor high-contrast spatial patterning of microfluidicdevice wettability,”
Lab on a Chip , vol. 8, no. 12,pp. 2157–2160, 2008.[28] M. H. Schneider, H. Willaime, Y. Tran, F. Rezgui,and P. Tabeling, “Wettability patterning by uv-initiated graft polymerization of poly (acrylic acid)in closed microfluidic systems of complex geometry,”
Analytical Chemistry , vol. 82, no. 21, pp. 8848–8855,2010.[29] M. B. Romanowsky, M. Heymann, A. R. Abate,A. T. Krummel, S. Fraden, and D. A. Weitz, “Func-tional patterning of pdms microfluidic devices usingintegrated chemo-masks,”
Lab on a Chip , vol. 10,no. 12, pp. 1521–1524, 2010.[30] A. R. Abate, J. Thiele, M. Weinhart, and D. A.Weitz, “Patterning microfluidic device wettabilityusing flow confinement,”
Lab on a Chip , vol. 10,no. 14, pp. 1774–1776, 2010.[31] S. C. Kim, D. J. Sukovich, and A. R. Abate,“Patterning microfluidic device wettability withspatially-controlled plasma oxidation,”
Lab on aChip , vol. 15, no. 15, pp. 3163–3169, 2015.[32] M. Morra, E. Occhiello, R. Marola, F. Garbassi,P. Humphrey, and D. Johnson, “On the aging of oxy-gen plasma-treated polydimethylsiloxane surfaces,”
Journal of Colloid and Interface Science , vol. 137,no. 1, pp. 11–24, 1990.[33] A. T´oth, I. Bertoti, M. Blazso, G. Banhegyi, A. Bog-nar, and P. Szaplonczay, “Oxydative damage andrecovery of silicone rubber surfaces. i. x-ray pho-toelectron spectroscopic study,”
Journal of AppliedPolymer Sciences , vol. 52, pp. 1293–1307, 1994.[34] H. Hillborg, M. Sandelin, and U. W. Gedde, “Hy-drophobic recovery of polydimethylsiloxane after ex-posure to partial discharges as a function of crosslinkdensity,”
Polymer , vol. 42, no. 17, pp. 7349–7362,2001. [35] H. Hillborg, N. Tomczak, A. Ol`ah, H. Sch¨onherr,and G. J. Vancso, “Nanoscale hydrophobic recov-ery: A chemical force microscopy study of uv/ozone-treated cross-linked poly (dimethylsiloxane),”
Lang-muir , vol. 20, no. 3, pp. 785–794, 2004.[36] H. Hillborg and U. W. Gedde, “Hydrophobic-ity recovery of polydimethylsiloxane after exposureto corona discharges,”
Polymer , vol. 39, no. 10,pp. 1991–1998, 1998.[37] J. Kim, M. K. Chaudhury, and M. J. Owen, “Hy-drophobic recovery of polydimethylsiloxane elas-tomer exposed to partial electrical discharge,”
Jour-nal of Colloid and Interface Science , vol. 226, no. 2,pp. 231–236, 2000.[38] J. L. Fritz and M. J. Owen, “Hydrophobic recoveryof plasma-treated polydimethylsiloxane,”
The Jour-nal of Adhesion , vol. 54, no. 1-4, pp. 33–45, 1995.[39] J. Kim, M. K. Chaudhury, M. J. Owen, and T. Or-beck, “The mechanisms of hydrophobic recovery ofpolydimethylsiloxane elastomers exposed to partialelectrical discharges,”
Journal of Colloid and Inter-face Science , vol. 244, no. 1, pp. 200–207, 2001.[40] D. Eddington, J. P. Puccinelli, and D. J. Beebe,“Thermal aging and reduced hydrophobic recoveryof polydimethylsiloxane,”
Sensors and Actuators B:Chemical , vol. 114, no. 1, pp. 170–172, 2005.[41] I.-J. Chen and E. Lindner, “The stability of radio-frequency plasma-treated polydimethylsiloxane sur-faces,”
Langmuir , vol. 23, no. 6, pp. 3118–3122,2007.[42] M. J. Owen and P. J. Smith, “Plasma treatment ofpolydimethylsiloxane,”
Journal of Adhesion Scienceand Technology , vol. 8, no. 10, pp. 1063–1075, 1994.[43] D. Bodas and C. Khan-Malek, “Hydrophilizationand hydrophobic recovery of pdms by oxygenplasma and chemical treatmentan sem investiga-tion,”
Sensors and Actuators B: Chemical , vol. 123,no. 1, pp. 368–373, 2007.[44] S. B´efahy, P. Lipnik, T. Pardoen, C. Nascimento,B. Patris, P. Bertrand, and S. Yunus, “Thick-ness and elastic modulus of plasma treated pdmssilica-like surface layer,”
Langmuir , vol. 26, no. 5,pp. 3372–3375, 2010.[45] V. Miralles, A. Huerre, H. Williams, B. Fourni´e, andM.-C. Jullien, “A versatile technology for droplet-based microfluidics: thermomechanical actuation,”
Lab on a Chip , vol. 15, pp. 2133–2139, 2015.[46] D. J. Guckenberger, E. Berthier, E. Young, and D. J.Beebe, “Fluorescence-based assessment of plasma-induced hydrophilicity in microfluidic devices via9ile red adsorption and depletion,”
Analytical chem-istry , vol. 86, no. 15, pp. 7258–7263, 2014.[47] B. Selva, J. Marchalot, and M.-C. Jullien, “An opti-mized resistor pattern for temperature gradient con-trol in microfluidics,”
Journal of Micromechanicsand Microengineering , vol. 19, no. 6, pp. 960–971,2009.[48] Y. Zhu and K. Petkovic-Duran, “Capillary flowin microchannels,”
Microfluidics and Nanofluidics ,vol. 8, no. 2, pp. 275–282, 2010.[49] M.-E. Vlachopoulou, P. Petrou, S. Kakabakos,A. Tserepi, K. Beltsios, and E. Gogolides, “Effect ofsurface nanostructuring of pdms on wetting prop-erties, hydrophobic recovery and protein adsorp-tion,”
Microelectronic Engineering , vol. 86, no. 4-6,pp. 1321–1324, 2009.[50] B. K. Lok, Y.-L. Cheng, and C. R. Robertson, “Pro-tein adsorption on crosslinked polydimethylsiloxaneusing total internal reflection fluorescence,”
Jour-nal of Colloid and Interface Science , vol. 91, no. 1,pp. 104–116, 1983.[51] L. W. McKeen,
Permeability properties of plasticsand elastomers . William Andrew, 2016.[52] F. A. Bayley, J. L. Liao, P. N. Stavrinou, A. Chiche,and J. T. Cabral, “Wavefront kinetics of plasma ox-idation of polydimethylsiloxane: limits for sub- µ mwrinkling,” Soft Matter , vol. 10, no. 8, pp. 1155–1166, 2014.[53] B. Sarrazin, R. Brossard, P. Guenoun, and F. Mal-loggi, “Investigation of pdms based bi-layer elastic-ity via interpretation of apparent young’s modulus,”
Soft Matter , vol. 12, no. 7, pp. 2200–2207, 2016.[54] Y. L. Cheng, S. A. Darst, and C. R. Robertson,“Bovine serum albumin adsorption and desorptionrates on solid surfaces with varying surface prop-erties,”
Journal of Colloid and Interface Science ,vol. 118, no. 1, pp. 212–223, 1987.[55] B. Selva, V. Miralles, I. Cantat, and M.-C. Jullien,“Thermocapillary actuation by optimized resistorpattern: bubbles and droplets displacing, switchingand trapping,”
Lab on a Chip , vol. 10, pp. 1835–1840, 2010.[56] M. Kerdraon, J. D. McGraw, B. Dollet, and M.-C.Jullien, “Self-similar relaxation of confined microflu-idic droplets,”
Physical Review Letters , vol. 123,no. 2, p. 024501, 2019.[57] L. H. Zhao, J. Lee, and P. N. Sen, “Long-term re-tention of hydrophilic behavior of plasma treatedpolydimethylsiloxane (pdms) surfaces stored underwater and luria-bertani broth,”