Journal of Colloid and Interface Science 363 (2011) 371–378

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Wettability control and patterning of PDMS using UV–ozone and water immersion

Kun Ma, Javier Rivera, George J. Hirasaki, Sibani Lisa Biswal ⇑Department of Chemical and Biomolecular Engineering, Rice University, Houston, TX, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 12 April 2011Accepted 12 July 2011Available online 22 July 2011

Keywords:Wettability patterningPolydimethylsiloxaneUV–ozoneHydrophobic recoveryDiffusive reaction model

0021-9797/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.jcis.2011.07.036

⇑ Corresponding author. Fax: +1 7133485478.E-mail address: biswal@rice.edu (S.L. Biswal).

We demonstrate a simple method to tune and pattern the wettability of polydimethylsiloxane (PDMS) togenerate microfluidic mimics of heterogeneous porous media. This technique allows one to tailor the cap-illary forces at different regions within the PDMS channel to mimic multi-phase flow in oil reservoirs. Inthis method, UV–ozone treatment is utilized to oxidize and hydrophilize the surface of PDMS. To main-tain a stable surface wettability, the oxidized surfaces are immersed in water. Additionally, the use of aphotomask makes it convenient to pattern the wettability in the porous media. A one-dimensional diffu-sive reaction model is established to understand the UV–ozone oxidation as well as hydrophobic recoveryof oxidized PDMS surfaces. The modeling results show that during UV–ozone, surface oxidation domi-nates over diffusion of low-molecular-weight (LMW) species. However, the diffusivity of LMW speciesplays an important role in wettability control of PDMS surfaces.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

PDMS (polydimethylsiloxane) is commonly employed to fabri-cate microfluidic devices because it can be molded quickly, easily,and cost-effectively using soft lithography [1]. Moreover, varioussurface modification methods for PDMS are currently available,which have been summarized in the literature [2,3]. Of these tech-niques, oxygen plasma treatment is a convenient way to quicklyrender the naturally hydrophobic PDMS surface to be hydrophilic,as well as to provide irreversible seals for the microfluidic devices[4]. This process is achieved by converting the PDMS surface intoan oxidized surface layer, where the non-polar groups (mainly –CH3) are substituted with the polar groups (mainly –OH) in thepresence of oxygen plasma [5,6]. Alternatively, a combination ofUV light and ozone, have also shown to be effective to alter thewettability of PDMS [7], similar to those treated with oxygenplasma but at a much slower rate [8] which allows for wettabilitycontrol. Typically a 10 mTorr pressure, 400 W oxygen plasmaprocess for 1 min results in an advancing water contact angle ofless than 10� [9], while UV–ozone with a 28 mW/cm2 low-pressuremercury vapor lamp at a distance of 6 mm requires about 1 h toachieve the same water contact angle [8]. A challenge arising fromthese techniques is the rapid hydrophobic recovery (typically with-in a couple hours) that occurs on the PDMS surface after oxygenplasma or UV–ozone treatment, which is dominated by the migra-tion of uncured and in situ produced low-molecular-weight (LMW)species toward PDMS surface [10,11]. To overcome this issue,efforts have been made to reduce the amount of uncured PDMS

ll rights reserved.

oligomers, either with additional curing time [12], or through aseries of solvent extraction [13]. Moreover, by keeping the oxidizedPDMS surface in contact with water, the hydrophilicity can also bemaintained for over 14 days [14,15].

In addition to direct exposure to energy, grafting a hydrophilicpolymer onto the surface was also used to alter the wettability ofPDMS channels [16,17]. A sol–gel layer, which had been function-alized with photoreactive silanes, was used to alter and pattern thewettability of PDMS surface [18]. A hydrophilic polymer polyacry-lic acid (PAA) was grafted on the sol–gel-coated PDMS, throughpolymerization process in the presence of photoinitiator-silanesand UV light. Portions of the channels which were desired to behydrophilic were exposed to UV light, resulting in patterned wet-tability with a water contact angle difference of 83�. Wettabilityhas also been spatially patterned in sol–gel-coated PDMS micro-channels through flow confinement of a reactive surface treatmentsolution, where the reaction of wettability patterning can be initi-ated by either UV or thermal energy [19].

Despite the success of these surface modification techniques,they require additional chemical modifications to be made to thesurface. UV–ozone is a facile technique to alter the wettability ofPDMS and to bond the PDMS pieces to form microfluidic devices.UV–ozone uses less energy than oxygen plasma during the surfacemodification process of PDMS [11], resulting in a slower change inthe water contact angle of the surface [8]. This feature can be ben-eficial to precisely tailor the wettability of PDMS, in combinationwith the approaches to minimize the rate of hydrophobic recoverydiscussed above.

In this study, we first demonstrate a simple method to tune andpattern wettability in a PDMS-based microfluidic porous medium.It is well known that for small pore diameters where low Reynolds

372 K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378

number and low capillary numbers dominate, the wettability ofthe pore walls plays an important role in multi-phase flows[20,21]. For example, in the case of enhanced oil recoveryprocesses in petroleum reservoirs, the different wettability ofvarious mineral surfaces results in varying oil saturation, capillarypressures and relative permeability curves [22]. In our method, thedesired water contact angles of the surface are attained withUV–ozone oxidation and subsequent water immersion. Previousstudies showed that additional curing time can hinder hydropho-bic recovery by reducing LMW species in PDMS matrix [12], andwe utilize this property to control the rate of surface hydrophil-ization through UV–ozone in this study. Additionally, variouscontact angles are tuned with the proposed approach, allowingone to mimic various conditions of wettability to study fluid flowin porous media.

To better understand and control this process, a theoreticalmodel is proposed to describe the change of wettability duringthe UV–ozone-induced hydrophilization process as well as thehydrophobic recovery of PDMS. The proposed model is based onthe previous diffusion models involved in the process of hydropho-bic recovery of electrically discharged PDMS [23]. We expand theprevious models to describe and predict the changes of water con-tact angle within the diffusive reaction process of UV–ozone oxida-tion. In order to precisely tune the wettability of PDMS surface, asystematic understanding of the mechanism of this technique isstill necessary. Specifically, the effect of curing time on surfacehydrophilization of PDMS with UV–ozone method, as well as thestability of this wettability control approach will be investigatedin the following study.

2. Materials and methods

2.1. Fabrication of microfluidic devices

The channel pattern is designed using AutoCAD. The porousmedium consists of a 4000 lm (length) � 3520 lm (width) rectan-gular reservoir with a quadrilateral network of cylindrical pillarswith a radius of 150 lm aligned at a distance of 60 lm. For wettabil-ity patterning purpose, this porous medium is divided into the upperhalf and the bottom half, and the fluid inlet is designed in the middleof the two halves with a channel width of 200 lm. Two 200 lm widefluid outlets are designed at the upper and lower right corners of theporous medium. Based on the designed geometry, the overall poros-ity of this porous medium is 45.0%.

The microfluidic devices are fabricated with standard softlithography techniques [24,25] as described below. A 4-in. siliconwafer (University Wafer) is used as a substrate and cleaned withIPA (Sigma Aldrich), followed by a DI water rinse. To dehydratethe surface, the substrate is baked at 200 �C for 5 min on a hotplate.The SU-8 50 photoresist (Microchem Corporation) is spin coatedonto a 4 in. silicon wafer (University Wafer) with a spin coater(Headway Research, Inc.) at 2000 rpm for 30 s. The substrate ispre-baked on a hot plate at 95 �C for 20 min before exposure. AnSF-100 maskless lithography tool (Intelligent Micro Patterning,LLC) is used to expose the photoresist with the desired pattern.After exposure, a post-exposure bake is performed at 95 �C for5 min. The substrate is subsequently developed in SU-8 developer(Microchem Corporation) for 6 min at room temperature, whichresults in a positive microfluidic pattern relief on the substrate. Apattern thickness of 75 lm is determined by a Dektak 6M profi-lometer (Veeco Instruments), which is the resulting height of thePDMS porous medium.

To fabricate the microfluidic device, a poly(dimethylsiloxane)(PDMS) elastomer kit (Sylgard 184; Dow Corning Corp.) is used.The kit consists of a liquid silicon elastomer base (vinyl-terminated

PDMS) and a curing agent (mixture of a platinum complex andcopolymers of methylhydrosiloxane and dimethylsiloxane) thatare combined in a 10:1 ratio and poured onto the silicon masterin a Petri dish. The wafer and PDMS are degassed under vacuumfor 20 min and cured at 80 �C in a convective oven. The curing timevaries from 0.5 h to 24 h. After curing, the patterned PDMS isremoved from the silicon master and inlet and outlet holes arepunched into the PDMS using a cork borer. To pattern the wettabil-ity of the surface, the patterned PDMS is treated with UV–ozone(Novascan Technologies, Inc.) prior to bonding. As reported bythe manufacturer, a low-pressure mercury vapor lamp with anoutput of 20 mW/cm2 at a distance of 25 mm is used in theUV–ozone instrument. A photomask is used to shield part of themicrofluidic pattern from UV light (Fig. 1). Our alignment tool(ATS115 motion controller, Aerotech, Inc.) is able to achieve aresolution of 5 lm in the patterning process. The area masked bythe black regions is shielded from the UV–ozone and remainshydrophobic after UV–ozone exposure, while the unmasked regionalters its wettability depending on the time of UV–ozone exposure.Finally, the patterned PDMS piece is placed in an UV–ozone cham-ber for an additional 5 min and then immediately brought intocontact with a blank (featureless) PDMS that has been processedwith the same conditions. Note that the 5 min UV–ozone treat-ment used for bonding is not long enough to significantly alterthe wettability of the PDMS. After sealing, the microfluidic deviceis immediately saturated with DI water prior to experiments.

2.2. Apparatus for evaluating wettability patterning

The effect of surface wettability is tested by observing how wellgas displaces a dye solution from the porous medium. The micro-fluidic device is placed on the stage of an inverted microscope(Olympus IX 71). To saturate the porous medium with water, a syr-inge pump (Harvard Apparatus PHD 2000) is used at the outflowend to pull out the fluid. After a 48-h water immersion, a3.0 wt.% aqueous dye solution (ESCO Foods, Inc.) is injected intothe microfluidic device until the color of aqueous phase in the por-ous medium is homogeneous. Then air is injected at a volumetricflow rate of 1.0 ml/h into the system. A CCD camera (PhantomV4.3, Vision Research, Inc.) is used to record the flow patterns thatemerge once the gas thread enters the porous medium.

2.3. Monitoring surface wettability

To determine the surface wettability inside the microfluidicdevice, an unpatterned PDMS is treated with the same conditions,including UV–ozone oxidation (with or without the photomask)and water immersion, as the patterned PDMS. The thickness ofall PDMS pieces in this study is controlled uniformly at 2.8 ±0.2 mm. The water static contact angle of the unpatterned pieceof PDMS is measured by the sessile drop method with a KSVCAM 200 contact angle and surface tension meter (KSV Instru-ments, Ltd.). All the experimental data of contact angles are ac-quired by three repeated measurements. There is no significantdifference of static contact angle on PDMS between DI water andthe diluted dye solution used in our experiments.

3. Results and discussions

3.1. Wettability patterning

Table 1 shows the wettability of two microfluidic devices usedin this study. In both devices, the PDMS is subject to 1-h curingand 4-h UV–ozone treatment. In Device A, the upper half isexposed to UV–ozone and made hydrophilic with a water contact

Fig. 1. Schematic of the two-step process of wettability control for microfluidic devices. Step 1: UV–ozone treatment with a photomask to selectively expose the surface ofPDMS; step 2: bond the microfluidic device and fill it with DI water to maintain the wettability.

Table 1Wettability of the microfluidic devices shown in Fig. 2.

Microfluidic device Water static contact angle (�) Graph Remarks

Upper half Bottom half

A 7.5 96.7 Fig. 2a Wettability-patterned deviceB 7.5 7.5 Fig. 2b Hydrophilic device

K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378 373

angle of 7.5� while and the bottom half is masked and kept hydro-phobic with a water contact angle of 96.7�. As a control, Device B, isexposed completely and has a uniform hydrophilic surface with astatic water contact angle of 7.5�.

Both devices are primed with DI water for 48 h, and saturatedwith dye solution before air injection, as shown in Fig. 2. Afterair injection, in Device A, the dye solution is displaced only in partof the hydrophobic half, while the hydrophilic half remains satu-rated with dye solution. Fig. 2a shows the saturation snapshot aftera 2-min air injection. In comparison, air displaces the dye solutionwithout discrimination in Device B, due to the homogeneous wet-tability condition. In both devices, after air breaks through the out-flow end and forms a continuous gas channel, the saturationsnapshot remains unchanged.

In the control sample (Device B), the wettability is homoge-neous and the air penetrates through the porous medium in bothhalves initially. As the gas thread continues displacing the liquid,the gas will choose the path with the least resistance until itreaches one of the outlets. After gas breakthrough, the newly in-jected air will preferentially follow the path of the continuousgas channel, making it difficult to breakthrough another outlet asshown in Fig. 2b. Repeated experiments show that, in a randommanner, the air will always choose one outlet to breakthroughwhile the other outlet remains filled with dye solution, leading toan asymmetric flow path.

In porous media, the capillary pressure across the interface ofair and dye solution can be described with the Young–Laplaceequation:

Pcapillary ¼2r cos h

rð1Þ

where r is the characteristic radius of the pore throat, r is the inter-facial tension between the two phases, and h is the static contactangle of the aqueous phase. In our case, since there is no significantdifference of static contact angles on PDMS between DI water andthe dye solution, we describe the wettability in terms of water staticcontact angle. According to Eq. (1), a smaller water contact angleleads to larger value of capillary pressure, which creates larger bar-rier for air to displace dye solution in the hydrophilic region giventhe same size of pore throats. Therefore, in Fig. 2a the dye solutionis preferentially displaced the hydrophobic half in the porous med-ium. In contrast, in Fig. 2b the capillary pressure is uniform for bothhalves, and the injected air permeates through both halves of thedevice.

The experimental results shown in Fig. 2 demonstrate that thewettability of PDMS can be successfully patterned in microfluidicdevices with the proposed approach.

3.2. Effect of curing time on surface modification by UV–ozone

In order to precisely tune the wettability of PDMS surface, anunderstanding of how curing time influences the PDMS surface isnecessary. After exposure to UV–ozone or oxygen plasma, oxidizedPDMS surfaces gradually recover their hydrophobicity with time,mainly because low molecular weight (LMW) species in the PDMSmove from bulk to the surface causing the generated hydrophilicgroups to migrate into the bulk PDMS to achieve a new equilib-rium. Thermal curing can eliminate low molecular weight (LMW)species in PDMS thereby reducing the rate of hydrophobic recoveryof plasma-treated hydrophilic PDMS surface [12]. Moreover, wedemonstrate that additional curing time can also accelerate the

Fig. 2. The effect of patterned wettability on displacement efficiency of aqueous dye solution by air in homogeneous porous media. Left: initially saturated with dye solution;right: after 2 min air injection at a volumetric flow rate of 1.0 ml/h. The red scale bar at the upper left corner represents 500 lm. The wettability conditions for both devicesare shown in Table 1. (a) Top view of the porous medium in Device A. The masked area is indicated in a white dashed box. (b) Top view of the porous medium in Device B. (Forinterpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. The effect of curing time on surface hydrophilization process of PDMS at80 �C.

374 K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378

surface hydrophilization process of PDMS. As shown in Fig. 3 for afixed curing temperature of 80 �C, the static water contact angle atvarying UV–ozone exposure times varies with curing time. In thefirst hour after UV–ozone treatment, there is little difference inthe water contact angle for the PDMS cured for various timesexcept for the one with 0.5-h curing. The differences emerge afterthe 1 h UV–ozone exposure, where PDMS pieces with longer curingtimes exhibit a faster surface hydrophilization process. TheUV–ozone oxidation process, which makes the PDMS surface

hydrophilic, competes with the migration of LMW species towardsthe surface, which makes it hydrophobic. Thus, the existence oflarge amount of LMW species weakens the efficiency of UV–ozoneoxidation. Meanwhile, when the curing time is extended from3 days to 7 days, the surface hydrophilization results show no sig-nificant difference from that with 24-h curing time (data notshown). A possible explanation is that the amount of LMW speciesreaches its lower limit after 24-h thermal aging at 80 �C, with thegiven mixing ratio of the silicon elastomer base and the curingagent. These results suggest that one can shorten the UV–ozonetreatment time by extending the curing time of PDMS.

In addition to the curing time, varying the ratio of elastomerbase and curing agent will also affect the rate of surface hydroph-ilization process. It is thought that the low molecular weight(LMW) species, which dominates the hydrophobic recovery pro-cess of PDMS, is a result of either uncrosslinked linear PDMS chainsor residual crosslinking agent [12]. Based on the absorbed massconcentration of LMW species in PDMS, a similar effect on surfacehydrophilization is expected if the ratio of elastomer base and cur-ing agent is varied.

3.3. Wettability maintenance by water immersion

Previous studies have shown that water immersion is an effec-tive way to achieve relatively stable wettability of a plasma-trea-ted hydrophilic PDMS surface with a static water contact angle of

Fig. 4. Wettability maintenance by keeping UV/ozone-treated PDMS (1-h curing at80 �C) surface in contact with DI water.

K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378 375

less than 10� [14]. In this study, we extend the previous conclu-sions to a partially water-wet system after UV–ozone oxidation.

As shown in Fig. 4, after different periods (1–4 h) of UV–ozonetreatment, the PDMS surfaces obtain different water contact angles(78–6�). The curing conditions chosen are 80 �C for 1 h as typicaloperations for fabricating PDMS microfluidic devices, and the oxi-dized PDMS surfaces are kept in contact with water up to 7 days.The static water contact angles are recorded to monitor the wetta-bility change during this process. Initially, a decrease of water con-tact angle is observed for all of PDMS pieces, presumably due towater absorption of the PDMS matrix [14,26]. Interestingly, forthe 4-h UV–ozone treated piece, there is a slight increase in thewater contact angle after 24-h water immersion indicating thathydrophobic recovery occurs and competes with the effect ofwater absorption, and the rate of hydrophobic recovery is influ-enced by degree of surface hydrophilization. After 48 h of watercontact, the wettability of oxidized PDMS surface becomes rela-tively stable up to 7 days, which enables one to precisely controlthe wettability of an air/water system in the microfluidic devices.

Hydrophobic recovery in PDMS is influenced by ambient condi-tions, such as humidity, temperature and the water that is retainedinside the polymer matrix. At room temperature, the presence ofwater provides a much more polar environment than air at the oxi-dized PDMS surface, and the Si–OH functional groups in the treatedPDMS can hydrogen bond with water. It is thought that the largedifference in dielectric constants between PDMS and water slowsdown the migration of LMW species toward the surface and ham-pers the reorientation of the hydrophilic groups inward bulk PDMSin the presence of aqueous solution [14]. Thus, we observe stablewettability when the PDMS is immersed in water compared toexposure in air.

Meanwhile, for non-aqueous applications, it is important torealize that many organic solvents can swell PDMS [27] and arenot suitable for PDMS microfluidic devices. Nevertheless, severaltypes of oil which exhibit good compatibility with PDMS can beused in microfluidic experiments, such as vegetable oils and heavymineral oil [28]. Although not discussed in this paper, it would beinteresting to investigate how these oils may affect the migrationof LMW species in PDMS and the wettability at the oxidized sur-face, which is helpful to evaluate the reliability of our proposedtechnique in the presence of oil.

3.4. Model analysis of surface hydrophilization and hydrophobicrecovery process

3.4.1. Description of the diffusive reaction modelIn order to understand the surface hydrophilization process of

PDMS with UV–ozone and the subsequent hydrophobic recovery

process, we extend the model developed by Kim and coworkers[23] to describe hydrophobic recovery to include the effect ofUV–ozone surface hydrophilization. It is observed that an oxidizedsurface layer (Fig. 5) with SiOx structures is formed at the surface ofPMDS during the UV–ozone process [7], and the composition ofbulk PDMS stays unchanged [8]. In a similar way, an oxidized sur-face layer is formed at the PDMS surface during active electricaldischarges, such as oxygen plasma. In one of the models developedby Kim and coworkers, the hydrophobic recovery of electricallydischarged PDMS surface is described using a one-dimensionalhomogeneous diffusion equation:

@Cnp

@t¼ Dnp

@2Cnp

@x2 ð2Þ

where Cnp is the adsorbed mass concentration of LMW species, Dnp

is the diffusivity of LMW species in the oxidized surface layer, and xis the direction toward the surface of PDMS. They also used the Cas-sie equation [29] to correlate the adsorbed mass of LMW specieswith the water static contact angle:

cos h ¼ fpo cos hpo þ fnp cos hnp ð3Þ

where the subscripts po and np represent polar and non-polar(LMW species) groups in the oxidized surface layer, and f denotesthe areal fraction of the component. An assumption made is thatthe areal fractions can be estimated with the fraction of the ad-sorbed mass concentration of the LMW species [23,30], and thenon-polar groups are essentially the same as LMW species basedon this assumption. Thus, Kim and coworkers derived the watercontact angle ht at time t as shown in

cos ht �C0 � Cnp

C0cos hpo þ

Cnp

C0cos hnp

¼ cos hnp þ ðcos hpo � cos hnpÞ 1� Cnp

C0

� �ð4Þ

where C0 is the maximum adsorbed mass concentration of LMWspecies in the oxidized surface layer.

Extending the above model to a UV–ozone process requires anunderstanding of the kinetics involved in this process. Unlike theoxygen plasma, the UV–ozone surface hydrophilization takes placeover several hours so that there is a competition between the sur-face hydrophilization and the hydrophobic recovery. It is generallyaccepted that the dominating reaction during PDMS surface mod-ification process is the conversion of methyl groups into hydroxyland bridging oxygen species [8,31,32]. However, the oxidation rateof PDMS surface decreases significantly as the reaction progresseswith UV–ozone exposure [31]. The results of sum frequency vibra-tional spectroscopy showed that the concentration of methylgroups on PDMS surface decreased exponentially with time, whichwas described with first-order reaction kinetics of removal andin situ regeneration of methyl groups [33]. However, slowing-down of the oxidation rate may also be caused by the rate-limitingdiffusion of oxidative groups into PDMS network, according to thedepth profile with Auger electron spectroscopy [34].

The complex nature of this oxidation reaction and limiting dataof kinetic characterization make it challenging to quantitativelydescribe this process. We tentatively employ a power-law kineticmodel to represent the reaction of PDMS surface oxidation withUV–ozone, and the transport of LMW species can be describedwith:

@Cnp

@t¼ Dnp

@2Cnp

@x2 � kCnnp ð5Þ

where k is the reaction rate constant and n is the reaction order.We program a finite difference algorithm with a grid point for-

mulation in MATLAB to calculate Eq. (5), which contains 50 grid

Fig. 5. The schematic of migration of LMW species in PDMS (a) during UV–ozone treatment; (b) upon air exposure; (c) upon water immersion. The dimensions are not toscale. Modified from Kim and coworkers [23].

Table 2List of the parameters in the model.

Parameter Value Unit Remark

Dnp 3.67 � 10�12 cm2/s Diffusivity of LMW speciesk 6.0 � 10�5 (g/

cm3)1�n/sReaction rate constant

n 0.6 – Reaction orderCnp0 0.1 g/cm3 Initial LMW species concentration

after curinghpo 0 degree Contact angle of polar groupshnp 100 degree Contact angle of LMW speciesL 1.0 � 10�3 cm Thickness of the oxidized surface

layer

376 K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378

points equally spaced in the oxidized surface layer along the direc-tion perpendicular to the PDMS surface. The parameters of thesimulation are listed in Table 2. The thickness of the PDMS layeroxidized by UV–ozone treatment (10 lm) is based on the experi-mental observation of a cross-section of a microchannel [7], whichis independent of the time of UV–ozone exposure. The initialadsorbed mass concentration of LMW species Cnp0 beforeUV–ozone oxidation is based on a Freundlich adsorption isothermwhich has been used to model a hydrophobic recovery process ofPDMS in the previous work [23]. The contact angles of polar andnon-polar groups are 0� and 100�, which correspond to completewater-wet condition and unoxidized condition of PDMS surface,respectively. In Table 2, the diffusivity Dnp, the reaction rate con-stant k and the reaction order n are obtained through a processof parameter estimation of the experimental data described below.

3.4.2. Parameter estimationAs Cnp0 is dependent on the curing time of PDMS at elevated

temperature, we estimate the parameters with a fixed curing time(1-h), and will discuss the effect of curing time on surface hydroph-ilization in the following section. Based on our experimental dataof hydrophobic recovery of PDMS after 1-h curing and 4-hUV–ozone treatment, we estimate the diffusivity of LMW speciesin the oxidized surface layer via the golden section search method

[35]. In a hydrophobic recovery process, Eq. (5) reduces to Eq. (2)due to the absence of the reaction term, and it is solved with thefollowing initial and boundary conditions:

IC : Cnpðt ¼ 0; xÞ ¼ Cis � Cnp0

Lxþ Cnp0

BCs : Cnpðt; x ¼ 0Þ ¼ Cnp0 ðinner boundary as shown in Fig: 5aÞ

@Cnpðt; x ¼ LÞ@t

¼ 0 ðouter boundary as shown in Fig: 5aÞ

where Cis is the initial adsorbed mass concentration of LMW speciesat the outer surface of the oxidized surface layer, which iscalculated through Eq. (4) with an initial contact angle of the hydro-phobic recovery right after UV–ozone treatment.

The initial condition is a linear LMW concentration profilethrough the oxidized surface. At the interface of the oxidizedsurface (x = L), the concentration of LMW species is determinedfrom the UV–ozone treatment. This initial condition can vary as afunction of UV exposure time, oxygen content, and initial LMWconcentration in the PDMS. The linear profile accounts for in situchain scission of PDMS at high doses of energy [11], whichincreases the LMW species in the sub-surface region [23], leadingto a faster hydrophobic recovery in air. At the inner boundary,Cnp is assumed to be a constant Cnp0 due to the large amount ofLMW species residing in PDMS matrix, while at the outer boundaryit is assumed to be no flux for Cnp such that the LMW speciesaccumulate at the surface of PDMS.

The results in Fig. 6a yield an optimized diffusivity Dnp of3.67 � 10�12 cm2/s for the LMW species, which is several orders ofmagnitude lower than the measured diffusivity (�10�8–10�7

cm2/s) of silicone fluid in a silicone elastomer at room temperature[36–38]. This indicates that LMW species diffuses slower in theoxidized surface layer near the surface than in the bulk PDMS ma-trix. The existence of hydroxyl groups, generated during UV–ozonetreatment, increases the dielectric constant of the oxidized layercompared with bulk PDMS.

The next step is to determine the kinetic parameters for theUV–ozone process. Before the UV–ozone treatment, the initial

Fig. 6. (a) Experimental data for hydrophobic recovery in air of PDMS (after 1-hcuring and 4-h UV–ozone treatment). The diffusion coefficient Dnp is estimated byfitting Eq. (2) to the experimental data. (b) Experimental data of how time for UV–ozone exposure changes PDMS wettability using a 1-h curing time. The derivedreaction rate constant k and the reaction order n are derived by fitting Eq. (5) to theexperimental data.

Fig. 7. Simulation of how changes in the initial concentration of LMW species affectthe PDMS surface hydrophilization process with UV–ozone. The unit of Cnp0 shownin the legend is g/cm3. All other parameters are consistent with those in Table 2.

100

Dnp=3.67e-12 Dnp=3.67e-13 Dnp=3.67e-14 Dnp=3.67e-16 Dnp=3.67e-18

K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378 377

condition of Cnp(t = 0, x) = Cnp0 applies for Eq. (5) in the oxidizedsurface layer, due to the homogeneous nature of cured PDMS.The boundary conditions are identical to those in the hydrophobicrecovery process. Thus, for any reaction rate constant, k, andreaction order, n, Eq. (5) can be solved numerically with our finitedifference algorithm.

A series of two-parameter contour plots are performed to obtainthe optimal k and n using the experimental data from 1-h curingPDMS shown in Fig. 3. According to the results of this approach,the parameters that best fit the experimental data are k = 6.0 �10�5 (g/cm3)1�n/s and n = 0.6, respectively. Although the kineticsassociated with the UV–ozone oxidation process still needs to beunderstood and developed, our power-law reaction model fits wellto the experimental data through the one-dimensional diffusivereaction equation, as shown in Fig. 6b.

0 20 40 60 80 100 120 140 1600

20

40

60

80

Time (hours)

Con

tact

ang

le (d

egre

es)

Fig. 8. Simulation of how diffusivity of non-polar groups influences the wettabilityof the PDMS surface. The unit of Dnp shown in the legend is cm2/s. All otherparameters are consistent with those given in Table 2 without the reaction term.

3.4.3. Effect of the concentration of LMW speciesWe have already shown in Fig. 3 that additional curing time

accelerates the surface hydrophilization process of PDMS withUV–ozone. As the initial adsorbed mass concentration of LMWspecies Cnp0 is dependent on the curing time of PDMS, we varythe value of Cnp0 to investigate this effect on surface hydrophil-ization in our model.

In Fig. 7, Cnp0 ranges from 0.02 to 0.10 g/cm3, with other param-eters staying unchanged. The results in Fig. 6 show that less initialadsorbed mass concentration of LMW species leads to fast surfacehydrophilization of PDMS, which is consistent with the trend ob-served in Fig. 3. We also find that with the parameters in Table2, the process of UV–ozone treatment of PDMS in Fig. 7 is in areaction-dominated regime. Decreasing the diffusivity does not

improve the rate of surface hydrophilization. However, increasingthe diffusivity by an order of magnitude would significantly slowdown the surface hydrophilization process at all Cnp0 levels (datanot shown). Thus, the effect of Cnp0 on PDMS hydrophilizationmainly contributes to the kinetics of the oxidation process ofLMW species in the oxidized surface layer.

The small differences in the water contact angle for variouscuring time in the first hour observed in Fig. 3 is not predictedby our simulation results in Fig. 6, indicating that there are otherfactors that influence surface hydrophilization, such as the concen-tration-dependent diffusion of LMW species. Other kinetic modelsneed to be investigated in the future in order to improve themodel.

3.4.4. Effect of the diffusivity of LMW speciesWe vary the diffusivity of LMW species in the hydrophobic

recovery process. Fig. 8 shows the effect of the diffusivity on therate of hydrophobic recovery of PDMS, and the lower diffusivityresults in better wettability maintenance of the oxidized PDMSsurface. If the diffusivity is five orders of magnitude lower thanthat was calculated in Fig. 5a, the wettability is expected to be verystable over a week, according to the simulation results in Fig. 8.

One way to control the diffusivity of LMW species is to decreasethe temperature. The absorption of silicone fluid in a siliconeelastomer was experimentally measured from 0 �C to 150 �C, andthe variation of calculated diffusivity with temperature was foundto be well represented by the Arrhenius relationship [38]. It wasalso shown that the hydrophobic recovery in air of plasma-treated

378 K. Ma et al. / Journal of Colloid and Interface Science 363 (2011) 371–378

PDMS surfaces was much slower in 4 �C or 20 �C than that in 70 �C[14], as a result of low diffusivity of LMW species at low tempera-tures. Therefore, low-temperature operations are able to slowdown the rate of hydrophobic recovery of oxidized PDMS surfaces.

As indicated in Fig. 4, exposing an oxidized PDMS surface to anaqueous environment is an effective method to control the wetta-bility of PDMS. Other than decreasing the diffusivity of LMWspecies via temperature, water molecules residing in the PMDSprovide additional hydrogen bonding between the hydrophilicgroups, which resists the migration of LMW species toward thesurface. For this reason, immersing the PDMS in water preventsthe water from evaporating from the oxidized surface and retardsthe hydrophobic recovery. This effect is beneficial to wettabilitycontrol of PDMS and may be used in combination with diffusivitycontrol (for example, decreasing the temperature) to slow themigration of LMW species. Other fluids that strengthen the dipoleinteractions between the hydrophilic groups, as well as possibleadditives used in the uncured PDMS to inhibit hydrophobicrecovery, will be studied in the future to extend the proposed tech-nique for wettability control and patterning of PDMS.

4. Conclusions

By utilizing UV–ozone oxidization approach to modify surfaces,we have successfully patterned and tuned the wettability in PDMS-based microfluidic devices. The rate of surface hydrophilization ofPDMS with UV–ozone oxidation is accelerated by extending curingtime at elevated temperature. This improves the efficiency ofmicrofluidic fabrication without losing the advantage of precisewettability tuning of this approach.

Through a one-dimensional diffusive reaction model, it is foundthat the initial concentration of LMW species residing in curedPDMS significantly affects the rate of surface hydrophilization dur-ing the UV–ozone oxidation process which agrees with the trendobserved in our experimental results. The initial concentration ofLMW species is minimized through sufficient curing time of PDMSat elevated temperature. The key issue of wettability control ofUV–ozone-oxidized PDMS is to control the diffusivity of LMW spe-cies as shown by the experimental and modeling results of thehydrophobic recovery process. Low temperature and water immer-sion are effective ways to hinder the migration of LMW species inPDMS and achieve better wettability control.

Acknowledgments

This work was financially supported by Abu Dhabi National OilCompany (ADNOC), Abu Dhabi Company for Onshore Oil

Operations (ADCO), Zakum Development Company (ZADCO), AbuDhabi Marine Operating Company (ADMA-OPCO) and the Petro-leum Institute (PI), U.A.E.

References

[1] J.C. McDonald, D.C. Duffy, J.R. Anderson, D.T. Chiu, H.K. Wu, O.J.A. Schueller,G.M. Whitesides, Electrophoresis 21 (2000) 27.

[2] H. Makamba, J.H. Kim, K. Lim, N. Park, J.H. Hahn, Electrophoresis 24 (2003)3607.

[3] J.W. Zhou, A.V. Ellis, N.H. Voelcker, Electrophoresis 31 (2010) 2.[4] S. Bhattacharya, A. Datta, J.M. Berg, S. Gangopadhyay, J. Microelectromech.

Syst. 14 (2005) 590.[5] J.R. Hollahan, G.L. Carlson, J. Appl. Polym. Sci. 14 (1970) 2499.[6] J.L. Fritz, M.J. Owen, J. Adhes. 54 (1995) 33.[7] Y. Berdichevsky, J. Khandurina, A. Guttman, Y.H. Lo, Sens. Actuators, B 97

(2004) 402.[8] K. Efimenko, W.E. Wallace, J. Genzer, J. Colloid Interface Sci. 254 (2002) 306.[9] K. Tsougeni, A. Tserepi, G. Boulousis, V. Constantoudis, E. Gogolides, Jpn. J. Appl.

Phys. 1 46 (2007) 744.[10] J. Kim, M.K. Chaudhury, M.J. Owen, T. Orbeck, J. Colloid Interface Sci. 244

(2001) 200.[11] A. Olah, H. Hillborg, G.J. Vancso, Appl. Surf. Sci. 239 (2005) 410.[12] D.T. Eddington, J.P. Puccinelli, D.J. Beebe, Sens. Actuators, B 114 (2006) 170.[13] J.A. Vickers, M.M. Caulum, C.S. Henry, Anal. Chem. 78 (2006) 7446.[14] I.J. Chen, E. Lindner, Langmuir 23 (2007) 3118.[15] X.Q. Ren, M. Bachman, C. Sims, G.P. Li, N. Allbritton, J. Chromatogr., B: Anal.

Technol. Biomed. Life Sci. 762 (2001) 117.[16] Y.L. Wang, H.H. Lai, M. Bachman, C.E. Sims, G.P. Li, N.L. Allbritton, Anal. Chem.

77 (2005) 7539.[17] M.H. Schneider, H. Willaime, Y. Tran, F. Rezgui, P. Tabeling, Anal. Chem. 82

(2010) 8848.[18] A.R. Abate, A.T. Krummel, D. Lee, M. Marquez, C. Holtze, D.A. Weitz, Lab Chip 8

(2008) 2157.[19] A.R. Abate, J. Thiele, M. Weinhart, D.A. Weitz, Lab Chip 10 (2010) 1774.[20] A.M. Barajas, R.L. Panton, Int. J. Multiphase Flow 19 (1993) 337.[21] T. Cubaud, U. Ulmanella, C.M. Ho, Fluid Dyn. Res. 38 (2006) 772.[22] G.J. Hirasaki, SPE Form. Eval. 6 (1991) 217.[23] J. Kim, M.K. Chaudhury, M.J. Owen, J. Colloid Interface Sci. 293 (2006) 364.[24] Y.N. Xia, G.M. Whitesides, Angew. Chem., Int. Ed. 37 (1998) 551.[25] G.C. Kini, J. Lai, M.S. Wong, S.L. Biswal, Langmuir 26 (2010) 6650.[26] J.D. Andrade, Surface and Interfacial Aspects of Biomedical Polymers, Plenum

Press, New York, 1985.[27] J.N. Lee, C. Park, G.M. Whitesides, Anal. Chem. 75 (2003) 6544.[28] A. Spiendiani, C. Nicolella, A.G. Livingston, Biotechnol. Bioeng. 83 (2003) 8.[29] A.B.D. Cassie, S. Baxter, Trans. Faraday Soc. 40 (1944) 0546.[30] S.R. Holmesfarley, R.H. Reamey, R. Nuzzo, T.J. Mccarthy, G.M. Whitesides,

Langmuir 3 (1987) 799.[31] C.L. Mirley, J.T. Koberstein, Langmuir 11 (1995) 1049.[32] Y.J. Fu, H.Z. Qui, K.S. Liao, S.J. Lue, C.C. Hu, K.R. Lee, J.Y. Lai, Langmuir 26 (2010)

4392.[33] H.K. Ye, Z.Y. Gu, D.H. Gracias, Langmuir 22 (2006) 1863.[34] V.Z.H. Chan, E.L. Thomas, J. Frommer, D. Sampson, R. Campbell, D. Miller, C.

Hawker, V. Lee, R.D. Miller, Chem. Mater. 10 (1998) 3895.[35] J. Kiefer, Proc. Am. Math. Soc. 4 (1953) 502.[36] A.N. Gent, R.H. Tobias, J. Polym. Sci., Part B: Polym. Phys. 20 (1982) 2317.[37] S.L. Rice, A.F. Diaz, J.C. Minor, P.A. Perry, Rubber Chem. Technol. 61 (1988)

194.[38] D.E. Mathison, B. Yates, M.I. Darby, J. Mater. Sci. 26 (1991) 6.