Applied Surface Science 256S (2009) S46–S53

Fabrication of superhydrophobic surfaces with controlled topography andchemistry

N. Blondiaux a,*, E. Scolan a, A.M. Popa a, J. Gavillet b, R. Pugin a

a Centre Suisse d’Electronique et Microtechnique (CSEM SA), Jacquet Droz 1, CH-2002 Neuchatel, Switzerlandb LITEN\DTNM\LTS, Commissariat a l’Energie Atomique (CEA), 17, rue des Martyrs, 38054 Grenoble Cedex 9, France

A R T I C L E I N F O

Article history:

Available online 6 May 2009

PACS:

81 Material Science

Keywords:

Superhydrophobicity

Polymer demixing

Nanolithography

Contact angle hysteresis

Cassie–Baxter

Wenzel

A B S T R A C T

We report the fabrication of sub-micrometer large silicon pillars with controlled aspect ratios by

combining thin polymer film structuring and dry etching. A wide library of structures was achieved

thanks to the tunability of the process both concerning lateral and vertical dimensions. The structures

were further used to create superhydrophobic surfaces. Depending on the aspect ratio of the pillars,

different superhydrophobic wetting states were observed. Special attention was also paid to the

influence of surface structuring on the contact angle hysteresis.

� 2009 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Applied Surface Science

journal homepage: www.e lsev ier .com/ locate /apsusc

1. Introduction

The effect of surface roughness on wettability properties hasbeen receiving increasing research-interest during the last decade.An accurate control of surface topography can indeed dramaticallyenhance their wetting properties. Depending on their surfacechemistry, rough surfaces will become either ‘‘superhydrophilic’’or ‘‘superhydrophobic’’.

From a technological point of view, many potential applicationsof such effects have been demonstrated. Superhydrophilic surfacesmay be used for their anti-fogging properties while super-hydrophobic surfaces may be used as self-cleaning surfaces, themost well-known example being found in nature when observingwater droplets rolling off lotus leaf surface. Another application ofsuperhydrophobicity is the control of the slip-length of liquidflowing in microfluidic channels [1]. Due to the reduced liquid–solid interfacial area, the slip-length can be drastically increased,which results in a decrease of the drag [2]. On a more fundamentallevel, most studies have been focusing on the effect of surfaceroughness on hydrophobic surfaces [3,4]. Pioneering investiga-tions showed that there are actually two possible metastable

* Corresponding author at: CSEM SA, Nanoscale Technology, 1 Jaquet Droz, 2000

Neuchatel, Switzerland. Tel.: +41 32 720 55 38; fax: +41 32 720 57 50.

E-mail address: nbx@csem.ch (N. Blondiaux).

0169-4332/$ – see front matter � 2009 Elsevier B.V. All rights reserved.

doi:10.1016/j.apsusc.2009.04.194

superhydrophobic states [5,6]. In the first (non-composite orWenzel state), the liquid droplet completely wets the topographyof the surfaces. In the other case (composite or fakir state), thedroplet remains on top of the asperities and air pockets are trappedbetween the asperities. Between these two situations, there areintermediate states where the droplets are partially impaled on thestructures [7,9]. The type of wetting state actually has a directimpact on the dynamic contact angles of the surface and moreespecially its contact angle hysteresis. For non-composite states,an increase in advancing contact angle and hysteresis is generallyreported [18,20]. For composite states, the situation is lessstraightforward: many studies report an increase in contact angleand a decrease in hysteresis as the solid–liquid interfacial area isdecreased [14,18,12]. However, the opposite situation was alsoreported as physical defects on the surface lead to additionalpinning of the triple line, which increases the hysteresis [8,16].

Many techniques have been developed to produce super-hydrophobic surfaces having micro- and nanostructures. On oneside, some techniques have been used to create structured surfacesover large areas but with structures generally random in term ofsize and morphology. Contact angles as high as 1658 have forinstance been reported on structured alumina coating prepared bya sol–gel method and functionalised with fluoroalkylsilane [10].Another simple and inexpensive method is the fabrication ofporous polymer layer by casting a polymer solution as proposed byErbil et al. [11]. A suitable selection of solvents and temperature

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53 S47

gives a direct control of surface roughness and allows thefabrication of surfaces with water contact angles of 1608. Durablesuperhydrophobic surfaces have also been obtained by plasmafluorination of polybutadiene films [12]. The plasma processresults in both a decrease of surface energy due to fluorination anda roughening of the surface. This allows the authors to obtainadvancing water contact angles as high as 1738 with hysteresiswithin the error measurement. Another route proposed by Sporiet al. combines titanium sandblasting and wet etching [19]. Usingthis library of surfaces with roughness at the micro- and nanoscale,the effect of hierarchical roughness on superhydrophobicity wasstudied. Still concerning randomly structured surfaces, Dorrer et al.recently investigated the wettability of silicon nanograss coatedwith polymers of varying surface tensions. For surfaces coated withhigh surfaces energy polymers, non-composite states wereobserved with an increase of advancing contact angle (uadv = 1248)and a drastic increase of hysteresis (Du = 1248). For the mosthydrophobic surfaces in composite states advancing water contactangles close to 1808 and no detectable hysteresis were measured[13]. Other techniques have been developed to fabricate extremelywell-defined structures. Generally, photolithography is used tofabricate microstructures with controlled shape and with dimen-sions down to 1 mm. McCarthy et al. and Bico et al. created forinstance surfaces with microridges, micropillars or microholes[14,15] and studied their wettability properties according to thelateral size, height, density and shape of the structures [17,18].Using E-beam lithography, Martines et al. created well-controllednanostructures in silicon and showed that the different wettingstates can be obtained depending on the type of nanostructure[20]. For the highest aspect ratio nanopillars, the authors measuredadvancing contact angle as high as 1648 and a hysteresis within theerror of measurement. The advantage of such top-down approachis the perfect control on the nanostructures produced but they aregenerally limited to small surfaces and are relatively costly.

We describe here an alternative technique to fabricate siliconsurfaces having arrays of sub-micrometer large pillars over 4 in.wafers. The structuring technique proposed is a combination ofpolymer thin film nanostructuring and dry etching. A polymerblend solution is made and spin-coated on a silicon substrate,resulting in a structured, phase-separated thin film on the surface[22]. This first step of the process gives a control on the lateraldimensions of the structures (diameter, density of pillars). Thepolymer thin film is then used to fabricate an etch-mask and thepattern is transferred into the silicon substrate using dry etchingtechniques, allowing a precise control of the height of the pillars. Insome cases, the resulting silicon pillars have been subjected to anadditional dry etching step in order to sharpen them and reducetheir lateral size. The etched surfaces were made hydrophobicusing different surface-treatments (silanisation, SiOC depositionand PTFE deposition). For each structure, the dynamic contactangles have been characterized and special attention was paid tothe contact angle hysteresis obtained on the different structures.Wettability properties of the surfaces were further characterizedby measuring the rolling angles of droplets of well-definedvolumes.

Table 1Equations proposed in literature to model contact angle. DuCB: contact angle hysteresis fo

for drops in the non-composite state (or Wenzel state).

Model name Composite state

McHale [28] DuCB ¼ GCBð f ; u flatÞDu flat

GCBð f ; u flatÞ ¼f sinðu flat ÞsinðuCBÞ

Linear model (Extrand) [26] DuCB ¼ lpðDu flat þvÞ

1.1. Theoretical background

Various models have been developed to understand the effect ofsurface roughness on wettability. Pioneering work was carried outby Wenzel and Cassie–Baxter who proposed a model to predict thecontact angle of droplet in the non-composite and composite statesrespectively [5,6]. For the non-composite state, Wenzel proposed amodel where the contact angle on rough surface depends on aroughness parameter r and on the contact angle on a flat surface.The roughness parameter r is defined as the ratio of the actualsurface-area over the projected area of the structures. The finalrelation between the apparent contact angle uwuW and the contactangle on a flat surface uflat is given by:

cosðuWÞ ¼ r cosðu flatÞ

r: roughness parameter = Areal/Aprojected.For the composite state, Cassie and Baxter derived the equation

proposed by Wenzel and considered only the area fraction of solidsin contact with the liquid when the droplet is suspended on thestructures. In that case, the relation between the apparent contactangle uCB and the contact angle on a flat surface uflat is given by:

cosðuCBÞ ¼ f � ðcosðu flatÞ þ 1Þ � 1

f: area fraction occupied by the structures.Other models have later been proposed either derived from

Wenzel and Cassie–Baxter equations [23,25–27] or by usingthermodynamic [24,25] or phenomenological approaches [26].

Among those, some models permit the calculation of contactangle hysteresis. McHale et al. for instance modelled contact angleamplification or attenuation using gain factors [27]. These weredefined based on the derivatives of Wenzel and Cassie–Baxterequations. Different gain factors were proposed depending on thewetting state and allowed the calculation of the contact anglehysteresis on rough surfaces (see Table 1). In the model proposedby Extrand, the contact angle hysteresis on a rough surface isexpressed as a linear function of the contact angle hysteresis on aflat surface and a parameter lp corresponding to the linear fractionof the contact line on the asperities [26]. This contrasts with theCassie–Baxter and Wenzel models where the key parameters arelinked to the surface fraction of the structures or the surfaceenhancement due to the structures. Although it is phenomen-ological, this model was found to successfully predict contact anglehysteresis for various structures [26].

2. Materials and methods

2.1. Fabrication of the structures

2.1.1. Substrate preparation

The substrates used in this study were 4 in. silicon wafers(h1 1 0i, boron doped) purchased from Si-Mat (Landsberg, Ger-many). The silicon wafers were cleaned in piranha solution (H2SO4/H2O2) (4:1)v/v for 10 min at 120 8C and rinsed in flowing water(MilliQ 185 plus, Millipore AG, Switzerland). Attention: piranha

r drops in the composite state (or Cassie–Baxter state), DuW: contact angle hysteresis

Non-composite state Remark

DuW ¼ GW ðr; u flatÞDu flat

GW ð f ; u flatÞ ¼r sinðu flat Þ

sinðuW Þ

DuW ¼ Du flat þ 2lpv lp: linear fraction of the contact line

v: rise-angle of the asperities

Table 2Parameters used for the deposition of the SiOC coating.

Precursors Carrier Working pressure Power Process

temperature

Plates spacing Deposition rate

Hydrophobic SiOxCyHz OMCTSO (partial

pressure 0.15 mbar)

Reducing

mixture

0.25 mbar 100 W 80 8C 30 mm �1 nm/s

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53S48

solution reacts violently with all organics and should be handled with

care. The substrates were dried with nitrogen under a laminar flowjust before use.

2.1.2. Thin film preparation

Polymer solutions of poly(methyl methacrylate) (PMMA)(Mw = 106 kDa) and polystyrene (PS) (Mw = 101 kDa) were madein dioxane (dioxane absolute, over molecular sieve, puriss, Fluka)at a concentration of 15 mg/mL. Both polymers were purchasedfrom Polymer Standard Services (PSS, Germany). The solutionswere then mixed to obtain a PS/PMMA ratio of (30:70)w/w. Thepolymer blend solution was then spin-coated onto the substrateunder controlled atmosphere (T = 21 8C and RH = 35%). To get ahomogeneous coating over 400, a two step spin-coating wasperformed: 2 s at 500 rpm followed by 60 s at 2000 rpm. Onevaporation of the solvent the solution phase separated to form apolymer film with a frozen-in, non-equilibrium microphasestructure [22]. A selective solvent was then used to remove oneof the polymers. Samples were dipped in cyclohexane (for analysis,Merck) for 1 min to remove PS and subsequently dried withnitrogen. Larger polymer structures were also done using the sameprotocol except that the concentration of the polymer was 40 mg/mL instead of 15 mg/mL. This resulted in a thicker polymer filmwith larger structures.

2.1.3. Etching of the structures

To fabricate the etch-mask, the thin polymer film was firstexposed to a short oxygen plasma (O2, 50 sccm, 0.05 Torr, 15 W, for1 min 30 s) using a RIE plasma (Plasmalab 80plus, Oxfordinstruments, UK) to remove any residual layer and expose thesilicon substrate. A 10 nm thick chromium layer was thendeposited on the sample using a thermal evaporator (Edwards306 auto, UK). The lift off was done by dipping the sample inacetone in an ultrasonic bath. The silicon was then etched throughthe metal etch-mask in a deep reactive ion etcher (DRIE) (AMS 200,Alcatel). The etching was done using a Bosch-type process. Thedepth of the structures was controlled by varying the etching-time.This resulted in silicon pillars with heights ranging from 200 nm to4 mm. For the deepest pillars obtained, the samples were subjectedin some cases to an additional RIE step (CF4 88 sccm, O2 22 sccm,0.05 Torr, 150 W, 4 min) in order to sharpen the pillars.

2.2. Modification of surface chemistry

Three techniques were used to make the surfaces hydrophobic.The first one was a silanisation with an apolar silane. The

samples (flat or structured) were cleaned in piranha solution,rinsed in Millipore water and dried with nitrogen. The sampleswere then placed in a dessicator with a dish containing[Tris(trimethylsiloxy)silylethyl]-dimethylchlorosilane purchasedfrom ABCR. Vacuum was created in the dessicator. It was thensealed and let overnight to ensure a complete silanisation. Thesamples were rinsed with n-hexane (UVaSol, Merck), ethanol(UVaSol, Merck) and Millipore water.

The second one was a sputter deposition of Teflon. The sampleswere placed in a PVD chamber having a Teflon target. Sputteringwas achieved using Ar gas at a pressure of 5 mTorr with a power of100 W. The deposition time was adjusted to have a 10 nm thickcoating on the samples.

The SiOC coating was made by means of Plasma EnhancedChemical Vapour Deposition (PECVD) technique. Plasmas wereproduced inside a cylindrical stainless steel vacuum chamber witha parallel plate configuration. Precursors vapour was uniformlydistributed in the reactor by the upper showerhead. The upperelectrode is externally connected, through a semi-automatedmatching network (Dressler VM1000A), to a 13.56 MHz-RF powersupplier (Advanced Energy Cesar1 RF power supply) whichprovides a RF voltage with respect to the grounded chamber.Before operating the discharge the device is evacuated to5 � 10�3 mbar using a rotary pump (Alcatel ADS 501). Octamethylsiloxane (OMCTSO) monomer was polymerised during plasmatreatments, leading to soft coatings of SiOxCyHz with high contentof methylene and methyl groups. The deposition parameters arepresented in Table 2. Plasma deposition was carried out in areducing mixture with low plasma activation to preserve methylgroups [21].

2.3. Characterization techniques

2.3.1. Topography

Polymer films and chromium etch-masks were characterizedwith atomic force microscopy (AFM), using a Nanoscope Dimen-sion 3100 (Digital Instruments/Veeco, Santa Barbara, CA). Tappingmode AFM was used for the characterization of topography usinggold-coated silicon tips (typical force constant of 5.5 N/m)obtained from NT-MDT (the Netherlands). In the case of highaspect ratio pillars in silicon, scanning electron microscope (SEM)was used to characterize the diameter of the pillars (top view of thesample) and the height (by observing the cross-section). SEMmeasurements were performed using a Philips XL-30 ESEM-FEGinstrument. The AFM and SEM images were further analysed usingan image analysis software (AnalySIS software, Soft ImagingSystem GmbH) to calculate the dimensions of the structures.

2.3.2. Contact angle and rolling angles

The wettability changes of the surfaces were characterized bymeasuring the contact angle of sessile water droplets deposited onthe sample. Advancing and receding water contact angles weredetermined using a Drop Shape Analysis System DSA10 providedby Kruss (Hamburg, Germany). Standard deviations werecalculated using four measurements made on two samples andused to calculate 95% of the confidence intervals whichcorrespond to the error bars shown on the graphs. Rolling angleswere measured on the DSA10 using an additional sample-holderthat can be tilted. A drop of Millipore water of well-definedvolume was deposited on the sample and the holder was tilteduntil the drop rolled off. The volume of the droplets was variedfrom 40 mL down to 2.5 mL.

3. Results

3.1. Fabrication of the superhydrophobic surfaces

As mentioned before, silicon pillars were fabricated bycombining polymer thin film structuring and dry etching. A thinstructured polymer film was done by means of polymer demixingand used to fabricate an etch-mask and transfer the structures intothe underlying silicon wafer by deep reactive ion etching (DRIE).

Fig. 1. (a) Topography image (AFM, tapping mode) of the thin film with the smallest structures fabricated by polymer demixing. (b) Topography image (AFM, tapping mode) of

the metal etch-mask before DRIE.

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53 S49

The polymer film was made by dissolving two polymers in acommon solvent and the polymer blend solution was spin-coatedon the silicon substrates. The resulting thin polymer film wasstructured due to the phase separation occurring during theprocess [22]. In order to increase the aspect ratio of the polymerstructures, the sample was dipped in a selective solvent to removeone of the phases. As can be seen on the AFM images presented inFig. 1a, randomly distributed pits with a noticeable dispersity in pitdiameter were obtained on the surface. An image analysiscombined with statistics permitted the determination of theaverage pit diameter (d = 370 nm for smaller structures andd = 900 nm for the larger ones) and their area fraction (w = 0.27for smaller structures and w = 0.23 for the larger ones). Due to thehydrophilicity of the silicon surface, a thin residual layer of themost polar polymer (PMMA) was covering the substrate [22,28]. Ashort oxygen plasma treatment was thus performed to expose thesilicon surface. A direct use of such structured polymer films asetch-mask is however not very interesting since they are ratherthin (tens of nanometers) and the polymer does not have a veryhigh etch-selectivity with silicon. To improve this, a thinchromium layer was evaporated through the polymer film and alift off of the polymer was done to get dots of metal on the surface.An AFM image of the resulting metal etch-mask can be seen inFig. 1b. The last fabrication step was the transfer of the pattern intothe silicon substrate. A Bosch-type process was used and sixdifferent etching-times were performed to obtain structures withheights ranging from 200 nm up to 4 mm. Longer etching-timeswere also tried to get deeper structures but this resulted in a maskfailure. The structures obtained were further characterized bymeans of scanning electron microscopy (SEM). As shown in Fig. 2, alarge database of structures was achieved thanks to the tunabilityof the process regarding both lateral and vertical dimensions. Theonly drawback arose from the Bosch process which induced anoticeable roughness on the side of the silicon pillars. Half of thesamples were kept as such and the remaining structures weresubjected to an additional RIE step in order to remove the sideroughness and to sharpen the pillars. This resulted in slightlyconical pillars as presented in Fig. 2c.

The structured silicon surfaces were then made hydrophobicusing either silanisation, PTFE or SiOC deposition. The advancing(ua) and receding (ur) contact angles on flat surfaces weremeasured and the results are presented in Table 3. The silanisedand SiOC coated coatings surfaces were fairly hydrophobic but themost interesting point was their low contact angle hysteresis. PTFEcoated wafers exhibited much higher advancing contact angle butalso a huge contact angle hysteresis.

3.1.1. Wettability of structured surfaces

The changes in the water contact angles value were monitoredas a function of the nanostructure height, as well as the surfacechemistry. The dynamic water contact angles as a function of thedepth of the pillars is shown in Fig. 3 for the smallest pillars(d = 370 nm).

As can be seen, all advancing contact angles are increased onstructured surfaces. There are however two distinct situations: for200 nm high pillars, the receding contact angle is lower than thaton flat surfaces and the drops were in a non-composite state. Inthat case, the contact angle hysteresis is thus much higher than onflat surface (Du = 628; 828 and 1008 for the silanised, SiOC and PTFEcoated surfaces respectively). For pillars having heights from500 nm to 4 mm, the results are approximately the same and thereceding contact angles are higher than on flat surfaces, suggestingthat the drops were in a composite state. Concerning contact anglehysteresis, there is a substantial increase for silanised and SiOCcoated surfaces (Du = 318 and 268 compared to Du = 78 and 88 onflat surfaces). For PTFE coated surfaces, the contact angle hysteresisis similar to that on flat surface.

The wettability of sharpened pillars was also characterized andthe results are presented in Fig. 4. The main difference comparedwith the non-sharpened pillars is that the non-composite state wasobserved for the 200 and 500 nm high pillars. In this regime,advancing contact angle increased when increasing the height ofthe pillars while receding decreased when increasing the height ofthe pillars. For the other structures, the drops were in thecomposite state and similar results were obtained for all pillarheights. It is worth pointing out that sharpening did not affect somuch the advancing contact angle but lead to a significantreduction of contact angle hysteresis. While it has a rather highhysteresis on flat surfaces, PTFE treated structured surfacesshowed a clear decrease in hysteresis since it dropped down toDu = 188 compared with Du = 488 on flat surfaces.

Other experiments were also made using non-sharpened pillarswith larger diameters (d = 900 nm) and using the same surfacechemistry treatments but no significant effect was observedcompared with the smaller pillars (data not shown).

To sum up these observations, we can say that there is a heightthreshold below which water drops completely wet the structures(non-composite state) and above which the drops remain on top ofthe pillars (composite state). The position of this threshold isbetween 200 and 500 nm for the normal silicon pillars andbetween 500 nm and 1 mm for the sharpened pillars. In all cases, anincrease in advancing water contact angle is observed. When anon-composite state is obtained, there is a larger contact angle

Fig. 2. (a) Top view SEM images of the large (left) and small (right) silicon pillars. (b) SEM images of the small pillars after different etching-times. (c) Tilted view of the normal

pillars (left) and the pillars after the sharpening process (right).

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53S50

hysteresis, the hysteresis being greater for rougher surfaces. Fordrops in a composite state, the hysteresis is larger than on flatsurfaces but it can be effectively decreased by sharpening thepillars.

The last part of the wettability study concerns rolling angles(RA), which were characterized as follow: the measurements weredone using water drops of volumes ranging from 2.5 to 40 mL, eachdata point corresponding to an average value measured using fivedrops. The rolling angles are presented in Fig. 5a and b for non-sharpened and sharpened pillars respectively. For structuresleading to composite states, low rolling angles were measuredfor large droplets (20 mL and above) and RA were increasing whendecreasing the drop volume. When samples with sharpened pillarswere used, very low rolling angles (<108) were obtained for alldrop volumes. The situation was very different for drops in a non-composite state: rolling angles were much greater and the smallestdrops were even sticking to the surface (RA = 908).

4. Discussion

As seen in the previous section, the technique proposed in thisstudy permits the fabrication of superhydrophobic surfacesleading to different wettability states (composite, non-composite)depending on their structures. For non-sharpened pillars, thetransition between both states was observed for pillar height

Table 3Dynamic contact angles on flat surfaces for the different surface chemistries.

Contact angle Silane SiOC PTFE

ua 1078 1098 1258ur 1008 1018 778

between 200 and 500 nm, while for sharpened pillars, it wasobserved between 500 nm and 1 mm. Interestingly, for non-composite states contact angle hysteresis was always larger thanon flat surfaces but for composite states it was either larger orlower depending on the type of structure (sharpened or non-sharpened pillars). Contact angle hysteresis actually arises fromthe imperfections of the surface both in terms of chemistry andtopography. When the triple line moves onto the surfaces andencounters a chemical heterogeneity (change in surface energy) orphysical heterogeneity (asperities), it is pinned, which affects themacroscopic contact angle measured. The final contact anglehysteresis is thus governed by pinning on both chemical andphysical heterogeneities on the surface.

4.1. Contact angle hysteresis of drops in a non-composite state

Other studies also reported similar increase in contact anglehysteresis for droplets in the non-composite state. Bico et al. andMartines et al. used micro- and nanopillars in silicon functiona-lised with a silane and observed that for low aspect ratiostructures, the droplets were in the non-composite state[15,20]. The reported contact angles follow the same trend asin our study: the advancing contact is increasing compared withthat on flat surface and the receding one is decreasing, resulting inan increase in contact angle hysteresis. A potential explanation islinked to the increase of the solid/liquid interfacial area whichincreases the number of chemical heterogeneities encountered bythe triple line during advancing and receding motion. Moreoverthe presence of structures on the surface also increases thenumber of asperities encountered by the contact line. Thisincreases in both cases the pinning effect, resulting in largerhysteresis.

Fig. 3. Water contact angles (advancing: blue squares, receding: red squares) for structures with different heights. (For interpretation of the references to color in this figure

legend, the reader is referred to the web version of the article.)

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53 S51

4.2. Contact angle hysteresis of drops in a composite state

Concerning droplets in the composite state, the increase incontact angle hysteresis observed for non-sharpened pillars maybe surprising since this state is generally associated with self-cleaning effect and thus low hysteresis. However, some studiesalso reported situations where high hystereses were obtained.Using microfabricated silicon pillars, Dorrer and Ruhe, Oner et al.as well as Barbieri measured high hysteresis for composite droplets

Fig. 4. Water contact angles (advancing: blue squares, receding: red squares) for surfac

legend, the reader is referred to the web version of the article.)

depending on the type of structure [18,16,29]. Higher hystereseswere more especially observed for surfaces having a high density ofpillars/features. As mentioned before, contact angle hysteresis isgoverned by the interplay between the pinning of the triple lineonto chemical and physical heterogeneities. Since the solid/liquidinterface is reduced for composite droplets, the triple lineencounters less chemical heterogeneities which should lowerthe hysteresis. However, the presence of more physical hetero-geneities (pillars) increases pinning effect and thus hysteresis.

es with different depths. (For interpretation of the references to color in this figure

Fig. 5. Rolling angle versus drop volume for pillars with different depths for normal and sharpened pillars respectively.

Table 4Contact angle hysteresis measured and calculated for two different samples. The value of uflat was 1078. c: composite state; nc: non-composite state.

Topography Chemistry State f r lp Measurements McHale model Linear model (Extrand)

Du Du Du

Flat Silane nc – – – 7 – –

1: 370 nm, h = 4 mm Silane c 0.27 – 0.43 30 3 34

1: 370 nm, h = 0.2 mm Silane nc – 1.56 0.43 62 12 84

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53S52

Depending on their morphologies and densities, physical struc-tures may therefore become more important for the hysteresisthan surface chemistry. Our study gives a good illustration of thishypothesis. For silanised and SiOC coated surfaces bearing highaspect ratio, non-sharpened pillars, the hysteresis was higher thanthat of flat surfaces. An explanation may be that on flat surfaces,hysteresis was only governed by chemical heterogeneities (nophysical defects encountered by the triple line), while onstructured ones, hysteresis was restrained by physical hetero-geneities due to increased pinning on the pillars.

Regarding sharpened pillars, lower hystereses were measuredcompared with non-sharpened pillars. In that case, the top of theconical pillars obtained after the sharpening process was only200 nm in diameter (compared with 370 nm for non-sharpenedpillars). This leads to a smaller liquid/solid interfacial area whichreduces pinning on chemical heterogeneities. Moreover, as thedensity of pillars remains the same, it also results in a decrease incontact-line length, which reduces the pinning on physicalheterogeneities. Both of these effects tend to a decrease in contactangle hysteresis as observed experimentally. A last observation,which was not investigated into further details in this study, is theshift of the composite/non-composite transition for sharpenedpillars. When sharpening is carried out, the diameter of the pillarsis reduced but the sidewall roughness arising from the Boschprocess is also removed. The presence of this additional roughnesson the non-sharpened pillars prevents the impalement of thedroplet on the pillars and makes the composite state more stable.This may explain why the transition occurs at a higher pillar heightfor non-sharpened pillars.

4.3. How do the measurements correlate with the theoretical

predictions?

The contact angle hysteresis measurements were also com-pared with the theories of McHale and Extrand presented in theintroduction. The key parameters (area fraction, roughness factor,linear fraction of the contact line) were calculated from the AFMand SEM images of the surfaces. The parameters used wereobtained from a statistical analysis since the structures producedare random in terms of diameters and distribution. The results for

the 200 nm and 4 mm high non-sharpened pillars are presented inTable 4. For droplets in the composite state (on 4 mm high pillars),the hysteresis is not very well predicted by the model proposed byMcHale, which foresees a decrease of the hysteresis. This may bedue to the fact that the only area fraction of solid is considered andnot the pinning effect on physical asperities. As mentioned in theirarticle, this model is better suited to surfaces having smoothvariations such as spherical or hemispherical structures [28]. Thelinear model gives better results and the value calculated agreesfairly well with the measurements. For droplets in the non-composite state, both theories foresee higher hysteresis butMcHale model tends to underestimate the hysteresis while thelinear model overestimates it. These discrepancies may also belinked to the sidewall roughness due to the Bosch process whichare not taken into account in the calculation and creates additionalroughness.

5. Conclusions

Sub-micrometer large silicon pillars were fabricated bycombining polymer nanostructuring (polymer demixing) anddeep reactive ion etching. The lateral and vertical dimensions ofthe pillars were tuned independently via the polymer demixing orthe dry etching step respectively. The resulting structures weresilanised with an apolar silane or coated with a low energymaterial to make them hydrophobic. Depending on the type ofstructures, different superhydrophobic wetting states areobserved. For low aspect ratios pillars, water droplets completelywet the structures (non-composite state) while composite statesare observed for high aspect ratio pillars. The height at which thetransition occurred is between 200 and 500 nm. This transition isshifted between 500 nm and 1 mm when the pillars are sharpenedwith an additional dry etching step. For all surfaces, dynamiccontact angles have also been measured. For drops in the non-composite state, an increase in advancing contact angle and adecrease in receding contact angles is observed in all cases,resulting in a larger hysteresis. For drops in the composite state,larger advancing and receding contact angles are observed. Thehysteresis is also larger in this case but it can be effectively reducedby sharpening the silicon pillars.

N. Blondiaux et al. / Applied Surface Science 256S (2009) S46–S53 S53

On an application point of view, the main drawback of thetechnique is the mechanical fragility of the structures produced,which prevents their use for demanding applications (self-cleaningsurfaces with scratch resistance for instance). However, other morerelevant applications may be found such as the control of wettabilityin microfluidic systems or their use in biological microsystems. Thestrength of the technique developed is the possibility to fabricatesub-micrometer large features with controlled length-scale, over 400

wafers and at lower cost than conventional top down techniquessuch as e-beam or focused ion beam.

Acknowledgements

This project was funded by the European project Napolyde, wethank them for their support. Christian Santschi, Andre Meister andVladislav Spassov are gratefully acknowledged.

References

[1] P. Joseph, C. Cottin-Bizonne, J.M. Benoit, C. Ybert, C. Journet, P. Tabeling, L.Bocquet, PRL 97 (2006) 156104.

[2] C.H. Choi, C.J. Kim, PRL 96 (2006) 066001.[3] N.J. Shirtcliffe, S. Aqil, C. Evans, G. McHale, M.I. Newton, C.C. Perry, P. Roach, J.

Micromech. Microeng. 14 (2004) 1384–1389.[4] B. He, N.A. Patankar, J. Lee, Langmuir 19 (2003) 4999–5003.

[5] R.N. Wenzel, Ind. Eng. Chem. 28 (1936) 988–994.[6] A.B.D. Cassie, S. Baxter, Trans. Faraday Soc. 40 (1944) 0546–0550.[7] S. Moulinet, D. Bartolo, Eur. Phys. J. E 24 (2007) 251–260.[8] D. Quere, Annu. Rev. Mater. Res. 38 (2008) 71–99.[9] C. Ran, G. Ding, W. Liu, Y. Deng, W. Hou, Langmuir 24 (2008) 9952–9955.

[10] K. Tadanaga, N. Katata, T. Minami, J. Am. Ceram. Soc. 80 (1997) 3213–3216.[11] H. Erbil, A.L. Demirel, Y. Avci, O. Mert, Science 299 (2003) 1377–1380.[12] I. Woodward, W.C.E. Schofield, V. Roucoules, J.P.S. Badyal, Langmuir 19 (2003)

3432–3438.[13] C. Dorrer, J. Ruhe, Adv. Mater. 20 (2008) 159–163.[14] L. Gao, T.J. McCarthy, Langmuir 22 (2006) 6234–6237.[15] J. Bico, C. Marzolin, D. Quere, Europhys. Lett. 47 (1999) 220–226.[16] D. Oner, T.J. McCarthy, Langmuir 16 (2000) 7777–7782.[17] M. Callies, Y. Chen, F. Marty, A. Pepin, D. Quere, Microelectron. Eng. 78–79 (2005)

100–105.[18] C. Dorrer, J. Ruhe, Langmuir 22 (2006) 7652–7657.[19] D.M. Spori, T. Drobek, S. Zurcher, M. Ochsner, C. Sprecher, A. Muhlebach, N.D.

Spencer, Langmuir 24 (2008) 5411–5417.[20] E. Martines, K. Seunarine, H. Morgan, N. Gadegaard, C.D.W. Wilkinson, M.O.

Riehle, Nano Lett. 5 (2005) 2097–2103.[21] D. Hegemann, H. Brunner, C. Oehr, Plasmas Polym. 6 (2001) 221.[22] S. Walheim, M. Boltau, J. Mlynek, G. Krausch, U. Steiner, Macromolecules 30

(1997) 4995.[23] B. Bhushan, M. Nosonovsky, Y.C. Jung, J. R. Soc. Interface 4 (2007) 643–648.[24] W. Li, A. Amirfazli, J. Colloid Interface Sci. 292 (2005) 195–201.[25] G. Whyman, E. Bormashenko, T. Stein, Chem. Phys. Lett. 450 (2008) 355–359.[26] C.W. Extrand, Langmuir 18 (2002) 7991–7999.[27] G. McHale, N.J. Shirtcliffe, M.I. Newton, Langmuir 20 (2004) 10147–10149.[28] R.A.L. Jones, L.J. Norton, E.J. Kramer, F.S. Bates, P. Wiltzius, PRL 66 (1991) 1326–1329.[29] L. Barbieri, Phd n 83692, EPFL.