Journal of Colloid and Interface Science 468 (2016) 21–33

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

journal homepage: www.elsevier .com/locate / jc is

Wetting hysteresis induced by temperature changes: Supercooledwater on hydrophobic surfaces

http://dx.doi.org/10.1016/j.jcis.2016.01.0400021-9797/� 2016 Elsevier Inc. All rights reserved.

⇑ Corresponding author.E-mail address: ghh@kth.se (G. Heydari).

Golrokh Heydari a,⇑, Maziar Sedighi Moghaddamb,c, Mikko Tuominen b, Matthew Fielden d,Janne Haapanen e, Jyrki M. Mäkelä e, Per M. Claesson a,b

aKTH Royal Institute of Technology, School of Chemical Science and Engineering, Department of Chemistry, Surface and Corrosion Science, Drottning Kristinas väg 51, SE-10044 Stockholm, Swedenb SP Technical Research Institute of Sweden, Chemistry, Materials and Surfaces, Box 5607, SE-114 86 Stockholm, SwedencKTH Royal Institute of Technology, School of Architecture and the Built Environment, Department of Civil and Architectural Engineering, Building Materials, SE-100 44Stockholm, SwedendKTH Royal Institute of Technology, School of Engineering Sciences, Department of Applied Physics, Nanostructure Physics, SE-106 91 Stockholm, Swedene TUT Tampere University of Technology, Aerosol Physics Laboratory, Department of Physics, P.O. Box 692, 33101 Tampere, Finland

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Received 26 October 2015Revised 15 January 2016Accepted 19 January 2016Available online 21 January 2016

Keywords:Wetting hysteresisContact angleSupercooled waterMorphologyHydrophobizationMulti-scale roughnessWoodSuperhydrophobicityLiquid flame spray (LFS)Plasma polymerization

a b s t r a c t

The state and stability of supercooled water on (super)hydrophobic surfaces is crucial for low tempera-ture applications and it will affect anti-icing and de-icing properties. Surface characteristics such astopography and chemistry are expected to affect wetting hysteresis during temperature cycling experi-ments, and also the freezing delay of supercooled water. We utilized stochastically rough wood surfacesthat were further modified to render them hydrophobic or superhydrophobic. Liquid flame spraying (LFS)was utilized to create a multi-scale roughness by depositing titanium dioxide nanoparticles. The coatingwas subsequently made non-polar by applying a thin plasma polymer layer. As flat reference samplesmodified silica surfaces with similar chemistries were utilized. With these substrates we test the hypoth-esis that superhydrophobic surfaces also should retard ice formation. Wetting hysteresis was evaluatedusing contact angle measurements during a freeze–thaw cycle from room temperature to freezing occur-rence at �7 �C, and then back to room temperature. Further, the delay in freezing of supercooled waterdroplets was studied at temperatures of �4 �C and �7 �C. The hysteresis in contact angle observed duringa cooling–heating cycle is found to be small on flat hydrophobic surfaces. However, significant changes incontact angles during a cooling–heating cycle are observed on the rough surfaces, with a higher contactangle observed on cooling compared to during the subsequent heating. Condensation and subsequent

22 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

frost formation at sub-zero temperatures induce the hysteresis. The freezing delay data show that the flatsurface is more efficient in enhancing the freezing delay than the rougher surfaces, which can be rational-ized considering heterogeneous nucleation theory. Thus, our data suggests that molecular flat surfaces,rather than rough superhydrophobic surfaces, are beneficial for retarding ice formation under conditionsthat allow condensation and frost formation to occur.

� 2016 Elsevier Inc. All rights reserved.

1. Introduction with water [26,47]. Many researchers report anti-icing properties,

Wetting of hydrophobic surfaces by water has received signifi-cant attention in the last decades due to the use of (super)hy-drophobic surfaces in different application, for example forcontrolling corrosion [1,2], adhesion [3], condensation [4], fouling[5], drag in microfluidic devices [6], and for achieving self-cleaning [7] properties. In addition, (super)hydrophobic surfaceshave been suggested for anti-icing and de-icing purposes [8],requiring mechanical and wetting durability [9]. This could possi-bly offer a route to combat icing issues that are affecting everydaylife as well as industrial applications [10]. We note that environ-mental conditions promoting condensation and frost formationon surfaces is common in most practical applications where icingis a problem, such as on roads and airplanes, heat exchangers, out-door structures and buildings. Though a transition in wetting stateof (super)hydrophobic surfaces could be desired and achieved byexternal stimuli [11–15] in some applications, it could be problem-atic for their function as anti-icing and de-icing coatings. Evapora-tion [16], pressure [17] and the kinetic energy of impingingdroplets [18–21] can all induce the Cassie–Baxter [22] to Wenzel[23] state transition on structured superhydrophobic surfaces.

Due to the implication of superhydrophobic surfaces as beinguseful for combating ice formation [24–34] there is a need to studythe effect of temperature on the wetting state of water droplets onsuch surfaces and possible transitions in wetting behavior inducedby temperature cycling. Though a number of works have exploredthe influence of low temperatures and surface characteristics oninteractions of sessile [26,35–39] or impinging [31,34,40] waterdroplets on (super)hydrophobic surfaces, only a limited numberof reports address the robustness of the wetting state on superhy-drophobic surfaces during cooling–heating cycles [41,42]. There islack of understanding of how the surface roughness affects thewetting reversibility during temperature cycles relevant for anti-icing applications, especially during a freeze–thaw cycle wherecondensation and subsequently frost formation may affect wettingproperties. We provide insight into this area by utilizing flathydrophobized silica surfaces as well as stochastically roughhydrophobic and superhydrophobic wood surfaces having similarchemistry to elucidate topography effects. We note that stochasti-cally rough surface, such as our wood samples, are relevant forpractical applications, and there is an increased interest in utiliza-tion of wood in various constructions as sustainability issues arebecoming more important. Though, surface modification of woodto restrict water penetration improve dimensional stability andbiological degradation [43,44], its affect on the wetting propertiesduring freeze–thaw cycles is not clarified despite being of impor-tance for outdoor applications in cold climate regions. To ourknowledge there is no previous report on interactions betweenmodified wood surfaces and water at subzero temperatures.

Despite an increased number of scientific works devoted toheterogeneous nucleation of ice from supercooled water, there isstill a discussion on whether there is a correlation between surface(super)hydrophobicity and anti-icing properties [7,27,28,30,31,33,45,46]. This is related to differences in experimental conditionssuch as temperature and relative humidity, heat transfer mecha-nisms, size of surface features and robustness of coatings in contact

such as delayed freezing of supercooled water or higher degree ofwater supercooling, of hydrophobic and in particular superhy-drophobic surfaces [24–33,45,48]. However, others report no orlimited benefits of such surfaces [38,47,49]. In addition, in a fewcases hydrophilic surfaces have been demonstrated to inducelonger freezing delay times than hydrophobic ones [46,49,50].

It is not obvious how surface morphology will affect kinetics ofice formation even though ice nucleation should occurmore readilyon concave than convex sites [47]. How surface roughness affectsfreezing of supercooled water on surfaces with similar chemistryhas been discussed in just a few recent works [24,28,31,47,51]. Inthe current manuscript we elucidate this topic in relation totemperature-wetting stability data. Our data show that the wettinghysteresis during a cooling–heating cycle (23–25 �C down to �7 �Cand back to 23–25 �C) is smaller on flat hydrophobic surfaces thanon rough ones with similar chemistry, even though introduction ofsmall roughness features reduces the wetting hysteresis on roughsurfaces. We demonstrate that (i) the observed wetting hysteresisis induced by condensation and frost formation at low tempera-tures, (ii) the coated surfaces are robust over several freeze–thawcycles, and (iii) under conditions when condensation and frost for-mation occurs the freezing delay of supercooled water on a flat sur-face is longer than on a rough surface with similar chemistry. Thiscan be rationalized by considering classical heterogeneous nucle-ation theory. Thus, stochastically rough (super)hydrophobic sur-faces do not appear to offer any anti-icing benefits compared toflat surfaces with similar chemistry during conditions of freezingand thawing in humid air.

2. Experimental section

2.1. Materials

The surface topography of stochastically rough wood surfaceswas modified utilizing a thermal aerosol based technique, LiquidFlame Spray (LFS) [52,53], for depositing titania (TiO2) nanoparti-cles. This adds sub-micro- and nano-roughness features to thewood surface that has roughness features predominantly on themicro scale. Subsequently, the surface chemistry of uncoated andTiO2-LFS-coated wood was altered using cold plasma polymeriza-tion, which is a dry technique that previously has been utilizedfor surface modification of wood [43,54]. Silicone and fluorine con-taining monomers are appropriate for depositing hydrophobiccoatings [43,55,56], and in this study we utilized hexamethyldis-iloxane (HMDSO) and perfluorohexane (PFH) monomers to makea plasma polymerized hydrophobic layer. Flat silica surfaces weremodified with the same plasma treatment. Water for samplepreparation and experiments was purified utilizing a Milli-ROP1sunit connected to a Milli-Q plus 185 system, and filtered througha 0.2 lm Millipak filter. The purified water had resistivity of18.2 MO cm and organic content of less than 3 ppb.

2.1.1. Wood surfacesKiln dried blocks of Scots pine (Pinus sylvestris L.) sapwood with

dimensions of about 30 mm in the longitudinal (L) direction, 7 mm

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 23

in the radial (R) direction and 120 mm in the tangential (T) direc-tion were prepared with a band saw and hand veneer saw. Subse-quently, wood veneers with dimensions of approximately30 � 20 � 1 mm3 (L � R � T) were prepared by cutting blocks alongthe fiber direction using a wood chisel. The wood veneers werewrapped and kept in aluminum foil for further treatment. We notethat the wetting properties of such unmodified wood veneers havebeen studied in detail [57]. The initial contact angle is around 60 �Cbut it rapidly decreases to zero as water is absorbed into the woodsample within the veneer. This precludes performing any freezing-thaw studies of the type reported in this work using unmodifiedwood veneers.

2.1.2. Silica surfacesFlat silicon wafers (waferNet, Inc., USA) with a thin silica surface

layer were utilized as flat reference substrates. They were rinsedwith water and analytical grade ethanol, and then dried under astream of N2. The wafers were further cleaned for 5 min in airplasma (created using a power of 18 W in a PDC-3XG plasma clea-ner from Harrick).

2.2. Plasma polymer deposition

Low-pressure plasma deposition was performed utilizing an in-house constructed reactor consisting of a glass vessel (volumeabout 11 L) connected to a double-stage rotary vacuum pump(Leybold-Heraeus D 65 B) [58]. The sample to be coated was posi-tioned inside the reactor between two externally wrapped, capac-itively coupled, copper electrode bands that were powered by a13.56 MHz radio-frequency power generator (ENI, Model ACG-3).Hexamethyldisiloxane (HMDSO, Fluka), or alternatively, perfluoro-hexane (PFH, Apollo Scientific) monomers were used as precursors.The harsh plasma condition breaks up the precursors moleculesinto reactive species that subsequently react to form a polymercoating on the substrate [59]. The plasma polymerization condi-tions are provided in Table 1. For each precursor, the plasmaparameters were chosen to provide a uniform and homogeneouspolymer film.

2.3. Liquid Flame Spray (LFS) coating

LFS is a thermal spray process utilized for depositing nano-sizedmetal and metal oxide particles. In LFS, the precursor is diluted inwater, alcohol or xylene and then the liquid together with the com-bustion gases are provided to a specially designed burner. The pre-cursor is atomized to micro-sized droplets and evaporated. Somefurther reactions of the precursor vapor result in formation of solidnanoparticles that are then deposited onto the substrate [52,60]. Inthis study, LFS coatings were prepared using a single nozzle typeburner. Hydrogen and oxygen with flow rates of 50 and 15 L min�1,respectively, were used as combustion gases. Titanium tetraiso-propoxide (Aldrich) diluted in isopropanol was utilized as TiO2 pre-cursor. The concentration of the precursor solutions was 50 mg ofatomic metal per mL. A feed rate of 6 mL min�1 and a depositiontime of 10 s were used. LFS-generated nanoparticles were intro-duced indirectly onto the substrate (by a tailor-made flow tube)

Table 1Plasma deposition conditions.

Precursor HMDSO PFH

Power 150 W 40WPressure 3.3 Pa 18 PaTreatment time 5 min 5 min

to avoid undesired heat effects. Detailed descriptions of our LFSequipment can be found elsewhere [52,53].

The size distribution of the LFS-generated TiO2 nanoparticlesfalls within the range of 20–70 nm (determined from AFM imagesas described in section D of the Supplementary material). This issimilar to the particle size, 5–70 nm, of TiO2 nanoparticles pro-duced utilizing the same LFS equipment but with different processparameters and determined by transmission electron microscopy[61]. These TiO2 nanoparticles have been reported to be mainlyin the form of anatase (density 3.89 g/cm3) [61,62]. A sessile dro-plet of water residing on the LFS coated (TiO2 nanoparticle-deposited) wood sample without further plasma modificationresults in complete wetting after a few minutes. Thus, the depos-ited TiO2 nanoparticle layer is not altering the hydrophilic natureof the wood surface.

2.4. Sample notation

The samples were named according to the principle substrate-LFS coating-plasma polymer coating. For instance wood-TiO2-HMDSO means a wood sample coated with TiO2 nanoparticlesusing LFS, followed by coating with a plasma polymer layer ofHMDSO. Likewise the sample Si-PFH means a silica surface coatedwith a plasma polymer layer of PFH without an intermediate LFSlayer.

2.5. Surface characterization

2.5.1. AFMTopographical surface images were captured utilizing an atomic

force microscope (AFM) Dimension Icon (Bruker, USA) operated inTapping Mode in air using antimony (n)-doped silicon rectangularcantilevers with a nominal tip radius of 8 nm (MPP-13120, Bruker).Analysis of the AFM images was performed using the NanoScopeanalysis software (Bruker). The same AFM mode and cantilevertype were utilized for determining the thickness of the plasmapolymer coatings. In order to facilitate the thickness measure-ments, a region of a silica substrate was first masked with a10 wt% solution of Poly(D,L-lactide-co-glycolide) (Resomer� RG755 S, Evonik) in acetone, and then the substrate was coated withplasma polymer as described in Table 1. Afterward, the mask waslifted off with the aid of a scalpel. Using this method a nanometersized step was made on the smooth silica substrate that enabled usto measure the thickness of the polymer layers by analyzing10 lm � 10 lm AFM height images of these steps using the ‘‘bear-ing analysis” feature in the NanoScope analysis software (Bruker).More details are provided in the supplementary information.

2.5.2. XPSPhotoelectron spectra of the surfaces were recorded using a

Kratos AXIS UltraDLD X-ray photoelectron spectrometer (KratosAnalytical, Manchester, UK). The samples were analyzed using amonochromatic Al X-ray source operated at 150W for high resolu-tion spectra. One spot with an area of approximately700 � 300 lm2 on two samples of each bare wood, Si-PFH, Si-HMDSO, wood-TiO2-PFH and wood-TiO2-HMDSO were analyzed.

2.6. Freeze–thaw contact angle measurements at temperature-controlled condition

Water contact angle measurements were made using a Data-Physics OCA40micro instrument (DataPhysics GmbH, Germany).The system is equipped with a high speed CCD camera (maximum2200 images s�1) with 20 times magnification, a computer pro-grammable droplet-dispensing unit, and a Peltier cooling stage.Image analysis was done using the SCA 20 (DataPhysics) software.

24 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

A fast response temperature sensor with a flat end part was con-nected to a high-resolution temperature logger (TC-08, Pico Tech-nology, UK) and utilized for surface temperature measurements.The droplet image capturing and the temperature logger werestarted simultaneously with the same time resolution. The mea-surements were done in a climate control room with relativehumidity of about 40% at room temperature (�23–25 �C).

Advancing (ACA) and receding (RCA) contact angles were mea-sured by dispensing a 5 lL droplet and tilting the sample stage.ACA and RCA were evaluated from the advancing and receding sideof the droplet right before sliding.

For freeze–thaw measurements, a 5 lL sized drop was dis-pensed on a sample mounted on the cooling stage. The tempera-ture was then lowered from room temperature to �7 �C. Thecooling rate for the wood samples was 4.5 ± 1 �C/min. The siliconwafer has higher thermal conductivity and lower thickness com-pared to the wood samples, resulting in a faster cooling rate of12.1 ± 2.4 �C/min. The samples were left at �7 �C until the sessiledrop was freezing, and the process was captured with the high-speed camera. ‘‘Freezing onset” was considered as the momentthe droplet started to become cloudy. Subsequently, the dropletwas left to freeze completely. Next, the wood sample was heatedat a rate of 6.9 ± 1.9 �C/min to room temperature and the frozendrop was melted. The heating rate was 23.2 ± 5.7 �C/min for siliconsurfaces due to the reasons mentioned above.

2.7. Freezing measurements

Freezing experiments, performed with the DataPhysicsOCA40micro instrument, utilized a room temperature 5 lL sizedwater droplet that was dispensed on a surface that subsequentlywas cooled from room temperature to either �4 �C or �7 �C at arate of 4.5 ± 1 �C/min and 12.1 ± 2.4 �C/min for wood and silica sur-faces, respectively. In addition to following the freezing processwith a CCD camera, an IR camera (Optris PI230, Optris, Germany)was also utilized for monitoring the temperature of the water dro-plet and the sample surface. The time to ‘‘freezing onset” after thesurface reached the desired temperature will be referred to as‘‘freezing delay” at the particular temperature.

3. Results and discussion

3.1. Surface characterization

Some relevant XPS data is summarized in Table 2. The Si/Catomic ratio determined from high resolution-spectra reveals sim-ilar chemistry for plasma polymerized HMDSO on flat silica andTiO2-coated wood substrates. In contrast, the F/C atomic ratiofound in the plasma polymer coating is significantly smaller onthe TiO2-coated wood substrate as compared to on the flat silicasubstrate. Inspection of the C1s spectra (see Tables S1 and S2 inSupplementary material) show that the wood-TiO2-PFH samplecontains more carbon bound to only other carbon and hydrogen

Table 2Atomic ratios measured by XPS.

Surface Si/C F/C

Sample 1 Sample 2 Sample 1 Sample 2

Bare wood 0 0 0 0Si-HMDSO 0.25 0.28 n.a. n.a.Wood-TiO2-HMDSO 0.29 0.28 n.a. n.a.Si-PFH n.a. n.a. 1.82 2.00Wood-TiO2-PFH n.a. n.a. 0.96 1.17

n.a. = not applicable.

than the Si-PFH sample. This carbon could originate from TiO2-induced catalytic degradation of the PFH, or being due to migrationof wood extractive [57] through a porous PFH layer. Whatever theorigin, the surface chemical difference between the wood-TiO2-PFH and Si-PFH sample is significant, and thus these two samplesdo not have similar surface chemistry. However, according to theXPS analysis the wood-TiO2-HMDSO and Si-HMDSO samples dohave similar surface chemistry, and using these two samples theeffect of surface roughness on the wetting and freezing of super-cooled water droplets will be discussed.

The coating thickness of the HMDSO and PFH plasma polymerlayers on silica substrates was determined to be 29 ± 6 nm and31 ± 2 nm, respectively (see Fig. S2 in the Supplementary material).

AFM height and phase images of bare silica, Si-HMDSO, and Si-PFH are shown in Fig. S1 in the Supplementary material. The phaseimages show contrast depending on variation in energy dissipationduring tapping and this is affected by material properties such astip-sample adhesion, surface viscoelasticity and stiffness [63,64].The root-mean-square (RMS) values of the surface roughnessobtained from analysis of the height images are reported in Table 3and illustrate that the silica surface remains smooth after plasmapolymerization, with RMS-values below 1 nm. These surfaces arethus suitable flat reference samples to which we compare ourrough wood surfaces that are treated with the same plasma poly-mer coating. The mean value of the phase angles obtained fromthe phase images are also reported in Table 3. We note the largechange in phase angle that follows upon deposition of the softplasma polymer layer on the hard silica surface.

AFM height and phase images of bare and modified wood sam-ples are shown in Fig. 1. These surfaces are very rough and for suchsurfaces topographical features also affect the energy dissipation[65], explaining much of the contrast observed in the phaseimages. The wood surfaces are inherently inhomogeneous so it isonly appropriate to mention major differences between these sam-ples. From the AFM images we see that polymerizing HMDSO andPFH directly on wood hides some of the smaller morphological fea-tures observed on the bare wood surface. The remaining surfacefeatures are broader when HMDSO is used as the last layer com-pared to when PFH is added last (compare Fig. 1b with c andFig. 1d with e). In addition, the local nano-scale features are morepronounced on LFS-coated surfaces (compare Fig. 1d with b andFig. 1e with c). We note that in addition to these nano-scale struc-tures observed in the AFM images, the wood surface also containsmicro-sized roughness features. Comparing AFM images of thebare wood and LFS-coated surfaces demonstrates that thenanoparticles cover the whole surface (compare Fig. 1a with dand e) and that they tend to form aggregates.

3.2. Temperature-dependent wetting properties

Static water contact angles on modified silica and wood sur-faces measured during a freeze–thaw cycle are presented inFig. 2. The measurement started with a liquid droplet held at roomtemperature and then proceeded down to �7 �C. During this cool-ing session the liquid state of the droplet remained also below 0 �C.Next, the temperature was maintained at �7 �C until the dropletunderwent freezing and the last data point during the cooling

Table 3Surface roughness and phase angle of reference samples.

Surface RMS roughness (nm) Phase angle (�)

Bare silica 0.2 ± 0.02 94.5 ± 0.6Si-HMDSO 0.7 ± 0.03 20.1 ± 1.4Si-PFH 0.3 ± 0.03 19.5 ± 3.0

Fig. 1. AFM height images (left column) and phase images (right column) recorded in air of (a) bare wood, (b) wood-HMDSO, (c) wood-PFH, (d) wood-TiO2-HMDSO and(e) wood-TiO2-PFH.

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 25

Table 4Droplet shape change associated with phase transitions at �7 �C.

Droplet Frozen (solid state) Few seconds after melting(liquid state)

Base diametera 1.03 ± 0.01 1.08 ± 0.02Heighta 1.09 ± 0.02 1.03 ± 0.01

a Normalized by the liquid droplet dimension right before freezing onset.

26 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

session on each plot in Fig. 2 was evaluated for solid ice formed atthis temperature. Subsequently, the temperature was increased toroom temperature where the frozen droplet melts at 0 �C. Asexpected and obvious in Fig. 2, the contact angle of solid ice isnot changing during heating from �7 �C to its melting point, butthen the ice melts the contact angle decreases significantly onthe rough surfaces. It should be noted that the droplet shape ischanging during the phase change from liquid to solid state atfreezing and from solid to liquid state at melting. These phasechanges affect the height of the droplet and the base diameter asreported in Table 4. We also note that droplets held at �7 �C slowlygrow due to condensation and frost formation from water vapor,explaining why the droplet size is larger after melting than itwas at the beginning of the experiment.

The initial and final contact angles measured in the beginningand at the end of freeze–thaw cycles are reported in Table 5, andsignificantly lower contact angles were observed on the rough sur-faces after the freeze–thaw cycle. The advancing and receding con-tact angles were determined using the tilting method. We notethat they could be determined with good precision even thoughthe tilting angle that initiated sliding varied significantly, in therange 40 ± 10� prior to the heating–cooling cycle and above 70�after this cycle, due to the heterogeneous topography of the

Fig. 2. Static contact angles measured during a freeze–thaw cycle on (a) Si-PFH, (b) Si-HMDSO. The measurements were done in contact with ambient air with about 40% relatleast 3 measurements.

modified wood surfaces. The larger tilt angle observed at the endof the experiment is a consequence of the lower contact angle,which, as we argue below, is due to water vapor condensationand melting of frost that facilitates droplet spreading.

For the flat modified silica surfaces we notice that the plasmapolymer coating formed by PFH is slightly more hydrophobic thanthat formed by HMDSO. However, for both these plasma polymersurfaces the contact angles are lower than observed on pure fluo-rocarbon or siloxane surfaces, respectively. The reason is that poly-merization under harsh plasma conditions also introduces somepolar groups due to reactions with e.g. water and oxygen speciespresent in the plasma [66]. There is a small but noticeable wettinghysteresis on the modified silica surfaces during the cooling–heating cycle, with a higher contact angle observed on cooling than

HMDSO, (c) wood-PFH, (d) wood-HMDSO, (e) wood-TiO2-PFH, and (e) wood-TiO2-ive humidity at 23 �C and the error bars correspond to the standard deviation of at

Table 5Initial and final water contact angles on modified silica and wood surfaces measured at room temperature in the beginning and at the end of a freeze–thaw cycle.

Surface Initial Final

RCA (�) Static CA (�) ACA (�) RCA (�) Static CA (�) ACA (�)

Si-PFH 99 ± 2 90 ± 2Si-HMDSO 92 ± 1 88 ± 1Wood-PFH 129 ± 1 65 ± 5Wood-HMDSO 127 ± 3 60 ± 27Wood-TiO2-PFH 114 ± 5 146 ± 4 157 ± 2 92 ± 2 107 ± 9 113 ± 4Wood-TiO2-HMDSO 116 ± 3 143 ± 4 162 ± 3 94 ± 5 102 ± 21 107 ± 4

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 27

on the subsequent heating. In particular, the contact angledecreases slightly with decreasing temperature during cooling. Inorder to demonstrate how the temperature-dependence of thewater surface tension [67] affects the observed contact anglereduction during cooling, the equilibriumwork of adhesion definedby Eq. (1) [68] is plotted versus temperature in Fig. 3.

We ¼ cLV ð1þ cos hsÞ ð1Þ

where hs is the static equilibrium contact angle and cLV is the watersurface tension. For the modified silica samples the work of adhe-sion increases slightly with decreasing temperature during cooling,and this effect is slightly larger than expected from thetemperature-dependence of the surface tension of water as illus-trated in Fig. 3. We note that no change in contact angle is observedwhen the water droplet is left on the modified silica surface at roomtemperature for a period of time that equals that of the freeze–thawcycle. Thus, the temperature cycling, rather than any change in ori-entation of chemical groups at the solid surface [66], is the primarycauses of the small hysteresis reported for modified silica surfacesin Fig. 2a and b.

Significantly larger temperature effects on the wetting proper-ties are observed for the rough modified wood surfaces. Since thechemistry of the outer HMDSO plasma polymer layer is very sim-ilar on the wood surfaces and on the flat silica surfaces, this isclearly due to the larger roughness of the wood samples. Thisresults in a higher water contact angle at room temperature atthe beginning of the experiment, and the multi-scale roughnessintroduced by the LFS coating increases the contact angle further.For the plasma polymer coated wood surfaces the contact anglestarts decreasing significantly (Fig. 2) and the work of adhesion

Fig. 3. Equilibrium work of adhesion as a function of temperature measured on cooling otension, for situations corresponding to temperature-independent contact angles equal

increases (Fig. 3) at a temperature of about 10 �C. This is due towater vapor condensation and, at subzero temperatures, frost for-mation on the surfaces. For all modified wood surfaces a highercontact angle was observed on cooling than on the subsequentheating (Fig. 2). In particular, we note the large decrease in contactangle during melting at 0 �C, which shows that the presence ofwater in surface depressions now leads to a transition in wettingfrom the superhydrophobic state. The degree of contact angledecrease after melting varies between samples and positions,resulting in large standard deviation for data collected during theheating cycle. This suggests an important effect of the local surfacestructure on the spreading that occurs as the frost and ice particleare melted. The wood-TiO2-plasma polymer surfaces present smal-ler hysteresis compared to the wood-plasma polymer surfaces,which we suggest is due to the multi-scale roughness introducedby the LFS treatment. Thus, the LFS treatment enhances thehydrophobicity also after a freeze–thaw cycle as quantified inTable 5. Clearly, the temperature dependence of the water surfacetension cannot rationalize the temperature-dependent wettingobserved for the modified wood samples as illustrated in Fig. 3.

In contrast to the large changes in wetting observed during afreeze–thaw cycle, leaving a water droplet at a constant tempera-ture on the modified wood surface for a period of time similar tothat of the freeze–thaw cycle results in only a small decrease incontact angle (about 2�) at room temperature, and no decrease incontact angle at subzero conditions, see Fig. 4.

We note that the droplets were cooled to �4 �C at the normalrate, i.e. 4.5 ± 1 �C/min for wood samples and 12.1 ± 2.4 �C/minfor silica samples. Thus, we conclude that the different cooling ratefor the modified silica samples and modified wood samples is not

n (a) HMDSO and (b) PFH coated samples. The effect due to changes in water surfaceto those determined at 23 �C, are shown as lines.

Fig. 4. Contact angle as a function of time measured at 25 �C and at �4 �C using awater droplet on wood-TiO2-PFH and Si-PFH surfaces.

28 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

affecting the measured contact angles because the values obtainedat constant temperatures (equivalent to zero cooling rate) are timeindependent even though the cooling rate used to reach the giventemperature was different.

In order to quantify droplet evaporation we have determinedthe droplet volume and base diameter at constant temperatureand during freeze–thaw cycles. Representative graphs are pre-sented for the wood-TiO2-PFH sample in Fig. 5. No detectable vol-ume or base diameter change is observed when the droplet is keptat subzero-temperatures (such as �4 �C as seen in Fig. 5). However,both volume and base diameter are reduced due to evaporation at25 �C, which results in a small reduction in the measured contactangle. In addition, water vapor condensation results in increasein base diameter and droplet volume during the cooling stage ofa freeze–thaw cycle. The net result is a decrease in the contactangle during cooling as illustrated in Fig. 2. However, decreasingthe temperature further below 0 �C induces frost formation (ratherthan condensation) on the surface that reduces the rate of the

Fig. 5. (a) Measured volume and (b) base diameter of a sessile droplet on a wood-TiO2

triangles represent measurements during a freeze–thaw cycle, the circles measurementemperature of 25 �C. The measurements were done in ambient air with about 40% rela

change in the droplet volume and base diameter. During theisothermal stage of the cycle (at constant temperature of �7 �C)and before the freezing onset, the volume and the base diameterare not changing in agreement with the data obtained for dropletsheld at �4 �C (Fig. 4). We note that from the freezing onset to thebeginning of the heating stage it is difficult to calculate the volumeof the droplet due to the shape change (see Fig. 7). However, theincrease in the base diameter can be followed and this is reportedin Fig. 5b (and Table 4). Finally, during the heating stage melting ofthe formed frost results in increase in base diameter and dropletvolume (see Fig. 5 and Table 4). The net result is a decrease inthe contact angle during heating as illustrated in Fig. 2. Interest-ingly, even though evaporation from the droplet is significant atroom temperature it is not significant during freeze–thaw cycles,but rather condensation of water vapor from the air increases thedroplet volume. Thus, the wetting hysteresis observed for roughmodified wood samples in Fig. 2 is not related to evaporation butaffected by condensation which results in a slight increase in dro-plet volume, a large increase in droplet base diameter and a largedecrease in contact angle.

The robustness of the modified wood surface was investigatedby measuring the contact angle of a droplet placed on a carefullydried spot after it was evaluated in a freeze–thaw cycle. The con-tact angle of the droplet placed on the dried spot after 10 freeze–thaw cycles was found to be very similar to the initial contact angleon the sample, and thus distinctly different to the rapid waterabsorption that occurs when a water droplet is placed on anunmodified wood veneer. Clearly, the hysteresis reported inFig. 2 is not caused by damage of the coatings during the freeze–thaw cycle, but due to a transition in the wetting state facilitatedby water vapor condensation and frost formation as quantified inFig. 5. A representative freeze–thaw cycle for evaluating therobustness of wood-TiO2-PFH sample is presented in Fig. S4 inthe Supplementary material. We also disregard the possibility thatthe hysteresis is related to droplet or surface contamination duringthe freeze–thaw cycles by noting the following observations. First,no significant hysteresis during a freeze–thaw cycle was observedon modified flat silica samples with similar chemistry to the mod-ified rough wood samples (see Fig. 2a and b). Secondly, contactangle measurements as a function of time at constant temperaturepresented in Fig. 4 suggest no effect of contamination. Third,

-PFH sample versus time. The data are normalized by the value at time zero. Thets at constant temperature of �4 �C, and the squares measurements at constanttive humidity at 23 �C and the lines are connecting the data points.

Fig. 6. A representative water and surface temperature measurement recorded byutilizing an IR camera. The data on the water plot represents the surfacetemperature of a 5 lL water droplet measured from the side. Point 1 correspondsto the ‘‘freezing onset” when the ice nuclei are formed and start to grow, and thefreezing process progresses between point 2 and 3.

Table 6Freezing time, i.e. the time from freezing onset to completely frozen, for 5 lL sizedsupercooled water drops.

Surface Freezing time at �4 �C Freezing time at �7 �C

Hydrophobized wood 360 ± 230 s 240 ± 90 sHydrophobized silica 33 ± 11 s 21 ± 10 s

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 29

repeating the freeze–thaw measurements on the same spotresulted in the same results within the error bars (Fig. S4). If theobserved contact angle change during a freeze–thaw cycle hadbeen related to contamination or disruption of the coating, thenplacing a fresh droplet on a spot that had been through a freeze–thaw cycle and subsequently dried would have resulted in a signif-icantly lower contact angle. However, this was not the case.

3.3. Freezing of supercooled water droplets

3.3.1. IR cameraThe surface temperature of the droplets and the samples were

measured simultaneously utilizing an IR camera during coolingfrom room temperature to freezing occurrence at the sample sur-face temperature of �7 �C. The plot in Fig. 6 shows a representativefreezing event on a wood-TiO2-PFH sample. The surface tempera-tures of the sample and water droplet were observed to changeat a similar rate, with the droplet being slightly warmer due to

Fig. 7. Snapshots of water droplets on Si-PFH as an example of hydrophobized silica (top)before ‘‘freezing onset” (corresponding to point 1 in Fig. 6), (b) and (e) at ‘‘freezing onset(corresponding to point 3 in Fig. 6).

the ambient air. This is in agreement with our previous workwhere we utilized a sensor inside the droplet [47]. Freezing ofwater is initiated by ice nucleation and then followed by ice crystalgrowth. As shown in Fig. 6, at point 1 on the droplet temperatureplot a sudden temperature increase is observed due to release ofthe latent heat of fusion. This moment corresponds to the start ofice crystal formation and is referred to as ‘‘freezing onset”. Subse-quently the droplet freezes completely between point 2 and 3.These observations are in qualitative agreement with the findingsof Alizadeh et al. [24].

3.3.2. CCD cameraWhen ice crystals start to form at ‘‘freezing onset” the droplet

becomes ‘‘cloudy” as recorded by a CCD camera. The white partin the middle of the images of the droplets is caused by light reflec-tion from the liquid (Fig. 7a and d) and it is eliminated at ‘‘freezingonset” (Fig. 7b and e). The freezing front moves from the solid-water interface and a small protrusion appears on the top of thefrozen droplet (Fig. 7c and f). After freezing a well-defined spotdue to light reflection appears again for ice formed on wood butnot for ice formed on silica (Fig. 7c and f). The visual observationof the ‘‘freezing onset” corresponds to the point where a suddenincrease is observed in the droplet temperature, i.e., point 1 inFig. 6. At this point the droplet absorbs the latent heat of fusionand its temperature increases temporarily as also observed anddiscussed in our previous work [47]. The nucleation phase of thefreezing process is kinetically fast [49], and the time intervalbetween the snapshots in Fig. 7a and b and between those inFig. 7d and e is 500 ms that corresponds to the interval betweenpoint 1 and 2 in Fig. 6. The subsequent crystal growth, leading tocomplete freezing of the droplet, occurs at a lower rate (reportedin Table 6) and it is accompanied by heat conduction to the surfaceand heat convection to the surrounding air. This corresponds to the

and wood-TiO2-PFH as an example of hydrophobized wood (bottom) (a) and (d) just” 500 ms later (corresponding to point 2 in Fig. 6), and (c) and (f) completely frozen

Fig. 8. Freezing time characteristics for 5 lL sized supercooled water dropletsmeasured on modified silica and wood surfaces. The filled bars represent thefreezing delay (the time to the ‘‘freezing onset” after the surface reaches the desiredtemperature of (a) �7 �C, and (b) �4 �C, where the error bars correspond tostandard deviation of 10 different measurements, and the striped bars represent thecooling time from 0 �C to the desired temperature. The complete bars represent theoverall time the droplet is below 0 �C in the supercooled state prior to freezing.

30 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

interval between point 2 and 3 in Fig. 6 where the droplet temper-ature reaches a plateau after the initial fast increase.

It took 130 ± 60 s and 230 ± 60 s to cool all types of hydropho-bized wood samples from 0 �C to �4 �C and from 0� to �7 �C,respectively. The corresponding times for both types of hydropho-bized silica samples were 26 ± 4 s and 50 ± 7 s. We suggest that thedifference in appearance of the frozen drops on wood and silica isrelate to differences in ice crystallization due to the difference inthe heat release rate, which also is reflected in the freezing timesreported in Table 6. It seems likely that the longer freezing timeon wood facilitates formation of larger ice crystals.

3.4. Freezing delay time

The freezing time characteristics on flat modified silica andrough modified wood samples are reported in Fig. 8. The filled barsin this figure represent the ‘‘freezing delay” at a certain tempera-ture calculated as the time interval from the moment that the sur-face reaches the target temperature (�4 �C or �7 �C) to the time offreezing onset. The cooling time from 0 �C to the target tempera-ture, that is longer for wood samples than for silica, is added tothe ‘‘freezing delay” time as striped bars. According to our previousstudy [47], and based on the heterogeneous ice nucleation theory,an ice embryo [69] starts to grow when it reaches a critical size.The critical radius can be calculated to be 11.3 nm at �4 �C and6.4 nm at �7 �C [47,69]. An activation energy, which decreaseswith the degree of supercooling, needs to be overcome to reachthe critical nucleation size. This explains the shorter freezing delaytime at �7 �C (Fig. 8a) compared to at �4 �C (Fig. 8b). The activa-tion energy increases with the contact angle of the ice embryoon the surface. Due to the small size of the ice embryo it is the localcontact angle that matters, and for a flat chemically heterogeneoussurface ice nucleation is expected to occur most readily on themost hydrophilic areas [47,69]. For a rough chemically homoge-neous surface, ice nucleation occurs most readily on concave sites(i.e. in surface depressions) [47,70] and, due to the small size of theice embryo, it is the contact angle measured on a flat chemicallyidentical surface that should be used in calculations of the energybarrier rather than the contact angle measured on the rough sur-face (which is affected by surface features that are orders of mag-nitude larger than the size of the critical ice embryo).

In our study we have prepared HMDSO-coated surfaces withsimilar chemistry, see Table 2, but different roughness characteris-tics. Based on the arguments above one would expect ice to freezeslowest on the most flat surfaces, and indeed we observe longestfreezing delay times (filled bars) on the flat modified silica surfacesboth at �4 and �7 �C. Further, despite the higher cooling rates forthe flat modified silica samples in comparison with the rough mod-ified wood samples the longest time for the droplets in the super-cooled state (the complete bars in Fig. 8) are observed on the flatsurfaces.

It is more difficult to predict if ice should freeze faster on wood-TiO2-plasma polymer or on wood-plasma polymer surfaces. Theorypredicts that ice will form most easily in the concave sites thathave the highest curvature (smallest radius), but this is only trueprovided water can penetrate these concave sites. Practically, wenote that the multi-scale roughness offered by the wood-TiO2-plasma polymer surfaces results in a longer freezing delay timeat �4 �C compared to on wood-plasma polymer surfaces(Fig. 8b). In contrast, no clear difference is observed between thesetwo types of surfaces at �7 �C (Fig. 8a). This suggests that water atthe supercooling temperature of �7 �C can enter surface depressionsto similar extent on these two surface types due to significantfrost formation facilitating local wetting. However, the super-hydrophobicity observed at room temperature for the wood-TiO2-plasma polymer surfaces still provides a benefit at�4 �C during

the time scale of ourmeasurement (� 640 ± 360 s below 0 �C), whichwe suggest is due to less frost formation at this temperature.

3.5. Do superhydrophobic surfaces have favorable anti-icingproperties?

There is general agreement that superhydrophobic surfaceshave important water repellent and self-cleaning properties attemperatures above 0 �C. However, it is debated if such surfacesalso have favorable icephobic properties such as long freezingdelay times and low ice adhesion. Likewise there is no agreementin the literatures on whether a wetting transition occurs on(super)hydrophobic surfaces at low temperatures [26,31,34–40,71].One major reason for the contradictory conclusions drawn indifferent studies is that the experimental conditions have varied.In the following discussion we will first consider freezing delaystudies that have utilized sessile droplets, as in our case, and thendiscuss some experiments that have utilized impinging droplets.

3.5.1. Experiments with sessile dropletsIn this work and in our previous work [47] we have found no

benefit of surfaces that are superhydrophobic at room temperaturein freezing delay studies as compared to flat surfaces with similarchemistry. This contrasts to works by Eberle et al. [28] and Boino-vich et al. [26] that demonstrated extremely long, several hours,freezing delay times of water droplets on superhydrophobic sur-faces at temperatures down to �20 �C, whereas shorter freezing

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 31

delay times were reported on surfaces with similar chemistry thatdid not display superhydrophobicity. The reason for this dramaticdifference is that our experiments were performed with a temper-ature gradient between the cold surface and the environmentwhich facilitates water vapor condensation and frost formationwhereas the experiments that demonstrated very long freezingdelay times on superhydrophobic surfaces were performed in ther-mally homogeneous chambers where the ambient air never wassupersaturated. It appears that superhydrophobic surfaces delaydroplet freezing when the Cassie–Baxter state can be maintainedat low temperatures, which is facilitated by avoiding water vaporcondensation and frost formation. Consistent with this Boinovichet al. [26] also utilized a thermally homogeneous environment tostudy contact angle stability during cooling of superhydrophobicsurfaces at saturated water vapor, and their data suggested thatthe wetting state was unaffected by temperature. Similar resultswere obtained by Yeong et al. [41] in a thermally homogeneousenvironment (where the surface, air and water droplet are keptat equal temperature) at low humidity. In contrast, in the openenvironment used by us, water vapor condensation and frost for-mation results in a transition from a superhydrophobic statetoward lower hydrophobicity as the temperature is lowered. Sucha transition was also observed by Yeong et al. at higher humidity[41]. Similar results were also obtained by Furuta et al. [42] whostudied the effect of temperature on the wetting state at a relativehumidity of 30% (related to the temperature 23 �C) in a cooling–heating cycle from about 23 �C to �6 �C and back to about 23 �C.They noted that the contact angle on both smooth and roughhydrophobic surfaces decreased with decreasing temperature andthat it was recovered during heating of smooth surfaces whereaswetting hysteresis was observed on rough surfaces. This is qualita-tively similar to our observations even though we allowed the dro-plet to freeze prior to heating.

Some studies have also compared freezing delay times onsuperhydrophobic surfaces with those found on other surfaceswith different chemistries, and concluded no benefits [38] or mod-erately longer freezing delay times on the superhydrophobic sur-face [30]. It is difficult to judge if the benefit reported in suchcases is due to the superhydrophobicity since no flat reference sur-face with similar chemistry was studied.

3.5.2. Experiments with impinging dropsAmong the experiments that have utilized impinging water

droplets the ones reported by Mishchenko et al. [31] using a cham-ber with controlled humidity demonstrate a clear benefit of super-hydrophobic surfaces, compared to surfaces with similar chemistrybut lower hydrophobicity, where the droplet was shown to bounceoff the superhydrophobic surface before it had time to freeze. Incontrast, Jung et al. [49], utilizing a range of different surface che-mistries and surface topographies report that a superhydrophobicsurface does not show a significantly longer droplet freezing delaytime than many other types of surfaces. In the work by Alizadehet al. [24] longer freezing delay on superhydrophobic surfaces,compared to on hydrophobic ones, was reported at �20 �C,whereas no benefit of the superhydrophobicity was observed at�25 �C, presumably due to frost formation. (Their relative humid-ity was 2% at room temperature, meaning that the dew point isclose to �25 �C).

We conclude that surfaces that are superhydrophobic at roomtemperature can prolong the freezing delay provided the wettingstate can be maintained at subzero temperatures. This requiresprevention of water vapor condensation and frost formation, whichis more easily achieved under dry conditions than under humidconditions, as found in outdoor environments and in e.g. heatexchangers.

4. Conclusions

The stability or transition of wetting states on hydrophobic sur-faces is important for their potential applications. In particular,temperature-dependent wetting of these surfaces is important inapplications related to icing issues. In this work we have elucidatedthe influence of sub-zero temperatures on the wetting characteris-tics of hydrophobic surfaces with different roughness scales, utiliz-ing contact angle measurements during a freeze–thaw cycle underhumid conditions that induces condensation and subsequentlyfrost formation. To this end we utilized flat surfaces as well asmodified wood surfaces with similar chemistry and differenttopography.

The smallest wetting hysteresis during a freeze–thaw cycle wasfound for the flat surface. Introduction of multi-scale roughness onthe wood surfaces increased the initial contact angle and its stabil-ity during a freeze–thaw cycle as compared to wood with mainlylarge roughness features. This is in qualitative agreement withthe report by Furuta et al. [42]. However, we also observe a largereduction in contact angle as the droplet and the frost is meltedduring heating.

When comparing surfaces with similar chemistry we observethe longest freezing delay on the most smooth hydrophobic sur-face, which is consistent with predictions based on the heteroge-neous nucleation theory [47]. Our results suggest that at lowsupercooling (�4 �C) water freezes less readily on LFS treatedand hydrophobized wood surfaces than on wood surfaces that isjust hydrophobized. This suggests that the multi-scale roughnessintroduced by the LFS method provides some benefit by decreasingthe penetration of supercooled water into surface depressions.However, at higher supercooling (�7 �C) this effect was negligible.We note that these results, showing longer freezing delay times fora smooth surface than for a rough superhydrophobic surface, wereobtained under humid conditions. In contrast, experiments carriedout with sessile droplets under temperature homogeneous condi-tions, where vapor condensation and frost formation is of lessimportance, have demonstrated beneficial icephobic properties ofsuperhydrophobic surfaces [26,28]. Thus, the usefulness of a sur-face that is superhydrophobic at room temperature for a givenanti-icing or de-icing application depends not only on the surfacecharacter but also on the environment of the particular application.

Acknowledgments

The authors thank the Top-level Research Initiative and NordicInnovation for financial support within the TopNANO project. M.S.M and M.T acknowledge support from the Nils and Dorthi Tröeds-son Foundation and M.S.M acknowledges the Swedish ResearchCouncil FORMAS within EnWoBio (2014-172) project. Marie Ern-stsson (SP) and Mikael Sundin (SP) are acknowledged for runningthe XPS and helping in analysis. Johan Andersson is thanked forperforming the plasma treatments at SP.

Appendix A. Supplementary material

XPS C1s detailed data of the samples, AFM height and phaseimages of silica samples, an AFM height image and the height his-togram of a step made for thickness measurement of the plasmacoatings, size determination of LFS-generated TiO2 nanoparticlesby AFM, and a control measurement to evaluate the robustnessof the coatings in a freeze–thaw cycle. Supplementary data associ-ated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcis.2016.01.040.

32 G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33

References

[1] L. Ejenstam, L. Ovaskainen, I. Rodriguez-Meizoso, L. Wågberg, J. Pan, A. Swerin,P.M. Claesson, The effect of superhydrophobic wetting state on corrosionprotection – the AKD example, J. Colloid Interface Sci. 412 (2013) 56–64.

[2] J.D. Brassard, D.K. Sarkar, J. Perron, A. Audibert-Hayet, D. Melot, Nano-microstructured superhydrophobic zinc coating on steel for prevention of corrosionand ice adhesion, J. Colloid Interface Sci. 447 (2015) 240–247.

[3] H. Zhu, Z. Guo, W. Liu, Adhesion behaviors on superhydrophobic surfaces,Chem. Commun. 50 (2014) 3900–3913.

[4] R. Enright, N. Miljkovic, A. Al-Obeidi, C.V. Thompson, E.N. Wang, Condensationon superhydrophobic surfaces: the role of local energy barriers and structurelength scale, Langmuir 28 (2012) 14424–14432.

[5] J. Arnott, A.H.F. Wu, M.J. Vucko, R.N. Lamb, Marine antifouling from thin air,Biofouling 30 (2014) 1045–1054.

[6] N.M. Oliveira, A.I. Neto, W. Song, J.F. Mano, Two-dimensional open microfluidicdevices by tuning the wettability on patterned superhydrophobic polymericsurface, Appl. Phys. Exp. 3 (2010) 085205.

[7] Z. Guo, W. Liu, B.-L. Su, Superhydrophobic surfaces: from natural to biomimeticto functional, J. Colloid Interface Sci. 353 (2011) 335–355.

[8] J. Lv, Y. Song, L. Jiang, J. Wang, Bio-inspired strategies for anti-icing, ACS Nano 8(2014) 3152–3169.

[9] E. Celia, T. Darmanin, E. Taffin de Givenchy, S. Amigoni, F. Guittard, Recentadvances in designing superhydrophobic surfaces, J. Colloid Interface Sci. 402(2013) 1–18.

[10] T.M. Schutzius, S. Jung, T. Maitra, P. Eberle, C. Antonini, C. Stamatopoulos, D.Poulikakos, Physics of icing and rational design of surfaces with extraordinaryicephobicity, Langmuir 31 (2015) 4807–4821.

[11] E. Bormashenko, R. Pogreb, G. Whyman, M. Erlich, Cassie–Wenzel wettingtransition in vibrating drops deposited on rough surfaces: is the dynamicCassie–Wenzel wetting transition a 2D or 1D affair?, Langmuir 23 (2007)6501–6503

[12] N. Verplanck, Y. Coffinier, V. Thomy, R. Boukherroub, Wettability switchingtechniques on superhydrophobic surfaces, Nanoscale Res. Lett. 2 (2007) 577–596.

[13] V. Bahadur, S.V. Garimella, Electrowetting-based control of droplet transitionand morphology on artificially microstructured surfaces, Langmuir 24 (2008)8338–8345.

[14] G. Manukyan, J.M. Oh, D. van den Ende, R.G.H. Lammertink, F. Mugele,Electrical switching of wetting states on superhydrophobic surfaces: a routetowards reversible Cassie-to-Wenzel transitions, Phys. Rev. Lett. 106 (2011)014501.

[15] M. Wåhlander, P.M. Hansson-Mille, A. Swerin, Superhydrophobicity: cavitygrowth and wetting transition, J. Colloid Interface Sci. 448 (2015) 482–491.

[16] Y.C. Jung, B. Bhushan, Wetting transition of water droplets onsuperhydrophobic patterned surfaces, Scripta Mater. 57 (2007) 1057–1060.

[17] R. Hensel, A. Finn, R. Helbig, S. Killge, H.-G. Braun, C. Werner, In situexperiments to reveal the role of surface feature sidewalls in the Cassie–Wenzel transition, Langmuir 30 (2014) 15162–15170.

[18] M. Reyssat, A. Pépin, F. Marty, Y. Chen, D. Quéré, Bouncing transitions onmicrotextured materials, EPL (Europhys. Lett.) 74 (2006) 306.

[19] Z. Wang, C. Lopez, A. Hirsa, N. Koratkar, Impact dynamics and rebound ofwater droplets on superhydrophobic carbon nanotube arrays, Appl. Phys. Lett.91 (2007).

[20] T. Deng, K.K. Varanasi, M. Hsu, N. Bhate, C. Keimel, J. Stein, M. Blohm,Nonwetting of impinging droplets on textured surfaces, Appl. Phys. Lett. 94(2009).

[21] Y.C. Jung, B. Bhushan, Dynamic effects of bouncing water droplets onsuperhydrophobic surfaces, Langmuir 24 (2008) 6262–6269.

[22] A.B.D. Cassie, S. Baxter, Wettability of porous surfaces, Trans. Faraday Soc. 40(1944) 546–551.

[23] R.N. Wenzel, Resistance of solid surfaced to wetting by water, Ind. Eng. Chem.28 (1936) 988–994.

[24] A. Alizadeh, M. Yamada, R. Li, W. Shang, S. Otta, S. Zhong, L. Ge, A. Dhinojwala,K.R. Conway, V. Bahadur, A.J. Vinciquerra, B. Stephens, M.L. Blohm, Dynamicsof ice nucleation on water repellent surfaces, Langmuir 28 (2012) 3180–3186.

[25] F. Arianpour, M. Farzaneh, S.A. Kulinich, Hydrophobic and ice-retardingproperties of doped silicone rubber coatings, Appl. Surf. Sci. 265 (2013) 546–552.

[26] L. Boinovich, A.M. Emelyanenko, V.V. Korolev, A.S. Pashinin, Effect ofwettability on sessile drop freezing: when superhydrophobicity stimulatesan extreme freezing delay, Langmuir 30 (2014) 1659–1668.

[27] L. Cao, A.K. Jones, V.K. Sikka, J. Wu, D. Gao, Anti-Icing superhydrophobiccoatings, Langmuir 25 (2009) 12444–12448.

[28] P. Eberle, M.K. Tiwari, T. Maitra, D. Poulikakos, Rational nanostructuring ofsurfaces for extraordinary icephobicity, Nanoscale 6 (2014) 4874–4881.

[29] H. Li, Y. Zhao, X. Yuan, Facile preparation of superhydrophobic coating byspraying a fluorinated acrylic random copolymer micelle solution, Soft Matter9 (2013) 1005–1009.

[30] W. Li, X. Zhang, J. Yang, F. Miao, In situ growth of superhydrophobic andicephobic films with micro/nanoscale hierarchical structures on the aluminumsubstrate, J. Colloid Interface Sci. 410 (2013) 165–171.

[31] L. Mishchenko, B. Hatton, V. Bahadur, J.A. Taylor, T. Krupenkin, J. Aizenberg,Design of ice-free nanostructured surfaces based on repulsion of impactingwater droplets, ACS Nano 4 (2010) 7699–7707.

[32] G. Momen, M. Farzaneh, R. Jafari, Wettability behaviour of RTV silicone rubbercoated on nanostructured aluminium surface, Appl. Surf. Sci. 257 (2011)6489–6493.

[33] P. Tourkine, M. Le Merrer, D. Quéré, Delayed freezing on water repellentmaterials, Langmuir 25 (2009) 7214–7216.

[34] T. Maitra, M.K. Tiwari, C. Antonini, P. Schoch, S. Jung, P. Eberle, D. Poulikakos,On the nanoengineering of superhydrophobic and impalement resistantsurface textures below the freezing temperature, Nano Lett. 14 (2014) 172–182.

[35] L. Yin, L. Zhu, Q. Wang, J. Ding, Q. Chen, Superhydrophobicity of natural andartificial surfaces under controlled condensation conditions, ACS Appl. Mater.Interfaces 3 (2011) 1254–1260.

[36] R. Karmouch, G.G. Ross, Experimental study on the evolution of contact angleswith temperature near the freezing point, J. Phys. Chem. C 114 (2010) 4063–4066.

[37] B. Mockenhaupt, H.-J. Ensikat, M. Spaeth, W. Barthlott, Superhydrophobicity ofbiological and technical surfaces under moisture condensation: stability inrelation to surface structure, Langmuir 24 (2008) 13591–13597.

[38] L. Yin, Q. Xia, J. Xue, S. Yang, Q. Wang, Q. Chen, In situ investigation of iceformation on surfaces with representative wettability, Appl. Surf. Sci. 256(2010) 6764–6769.

[39] S.A. Kulinich, S. Farhadi, K. Nose, X.W. Du, Superhydrophobic surfaces: are theyreally ice-repellent?, Langmuir 27 (2011) 25–29

[40] A. Alizadeh, V. Bahadur, S. Zhong, W. Shang, R. Li, J. Ruud, M. Yamada, L. Ge, A.Dhinojwala, M. Sohal, Temperature dependent droplet impact dynamics onflat and textured surfaces, Appl. Phys. Lett. 100 (2012).

[41] Y. Han Yeong, A. Steele, E. Loth, I. Bayer, G. De Combarieu, C. Lakeman,Temperature and humidity effects on superhydrophobicity of nanocompositecoatings, Appl. Phys. Lett. 100 (2012) 053112.

[42] T. Furuta, M. Sakai, T. Isobe, A. Nakajima, Effect of dew condensation on thewettability of rough hydrophobic surfaces coated with two different silanes,Langmuir 26 (2010) 13305–13309.

[43] A.R. Denes, M.A. Tshabalala, R. Rowell, F. Denes, R.A. Young,Hexamethyldisiloxane-plasma coating of wood surfaces for creating waterrepellent characteristics, Holzforschung (1999) 318.

[44] G. Sèbe, M.A. Brook, Hydrophobization of wood surfaces: covalent grafting ofsilicone polymers, Wood Sci. Technol. 35 (2001) 269–282.

[45] T.V.J. Charpentier, A. Neville, P. Millner, R.W. Hewson, A. Morina, Developmentof anti-icing materials by chemical tailoring of hydrophobic textured metallicsurfaces, J. Colloid Interface Sci. 394 (2013) 539–544.

[46] K. Li, S. Xu, J. Chen, Q. Zhang, Y. Zhang, D. Cui, X. Zhou, J. Wang, Y. Song,Viscosity of interfacial water regulates ice nucleation, Appl. Phys. Lett. 104(2014) 101605.

[47] G. Heydari, E. Thormann, M. Järn, E. Tyrode, P.M. Claesson, Hydrophobicsurfaces: topography effects on wetting by supercooled water and freezingdelay, J. Phys. Chem. C 117 (2013) 21752–21762.

[48] H. Wang, G. He, Q. Tian, Effects of nano-fluorocarbon coating on icing, Appl.Surf. Sci. 258 (2012) 7219–7224.

[49] S. Jung, M. Dorrestijn, D. Raps, A. Das, C.M. Megaridis, D. Poulikakos, Aresuperhydrophobic surfaces best for icephobicity?, Langmuir 27 (2011) 3059–3066

[50] K. Li, S. Xu, W. Shi, M. He, H. Li, S. Li, X. Zhou, J. Wang, Y. Song, Investigating theeffects of solid surfaces on ice nucleation, Langmuir 28 (2012) 10749–10754.

[51] J.M. Campbell, F.C. Meldrum, H.K. Christenson, Is ice nucleation fromsupercooled water insensitive to surface roughness?, J Phys. Chem. C 119(2015) 1164–1169.

[52] J.M. Mäkelä, M. Aromaa, H. Teisala, M. Tuominen, M. Stepien, J.J. Saarinen, M.Toivakka, J. Kuusipalo, Nanoparticle deposition from liquid flame spray ontomoving roll-to-roll paperboard material, Aerosol Sci. Technol. 45 (2011) 827–837.

[53] M. Stepien, J.J. Saarinen, H. Teisala, M. Tuominen, M. Aromaa, J. Kuusipalo, J.M.Mäkelä, M. Toivakka, Adjustable wettability of paperboard by liquid flamespray nanoparticle deposition, Appl. Surf. Sci. 257 (2011) 1911–1917.

[54] M. Kim, H.S. Kim, J.Y. Lim, A study on the effect of plasma treatment for wastewood biocomposites, J. Nanomater. 2013 (2013) 6.

[55] L. Podgorski, C. Bousta, F. Schambourg, J. Maguin, B. Chevet, Surfacemodification of wood by plasma polymerisation, Pigm. Resin Technol. 31(2002) 33–40.

[56] Z. Stefano, R. Claudia, O. Marco, F. Valeria, C. Maria Perla, D. Dorina Ines, L.Stefano, P. Vincenzo, Wood coated with plasma-polymer for water repellence,Wood Sci. Technol. 42 (2007) 149–160.

[57] M. Sedighi Moghaddam, M.E.P. Wålinder, P.M. Claesson, A. Swerin, Multicyclewilhelmy plate method for wetting properties, swelling and liquid sorption ofwood, Langmuir 29 (2013) 12145–12153.

[58] M. Pykönen, K. Johansson, M. Dubreuil, D. Vangeneugden, G. Ström, P. Fardim,M. Toivakka, Evaluation of plasma-deposited hydrophobic coatings onpigment-coated paper for reduced dampening water absorption, J. Adhes.Sci. Technol. 24 (2010) 511–537.

[59] H.K. Yasuda, Plasma Polymerization, Academic Press, Orlando, FL, 1985.[60] H. Teisala, M. Tuominen, M. Aromaa, J.M. Mäkelä, M. Stepien, J.J. Saarinen, M.

Toivakka, J. Kuusipalo, Development of superhydrophobic coating onpaperboard surface using the liquid flame spray, Surf. Coat. Technol. 205(2010) 436–445.

[61] M. Aromaa, H. Keskinen, J.M. Mäkelä, The effect of process parameters on theliquid flame spray generated titania nanoparticles, Biomol. Eng. 24 (2007)543–548.

G. Heydari et al. / Journal of Colloid and Interface Science 468 (2016) 21–33 33

[62] M. Stepien, J.J. Saarinen, H. Teisala, M. Tuominen, J. Haapanen, J.M. Mäkelä, J.Kuusipalo, M. Toivakka, Compressibility of porous TiO2 nanoparticle coatingon paperboard, Nanoscale Res. Lett. 8 (2013). 444–444.

[63] S.N. Magonov, V. Elings, M.H. Whangbo, Phase imaging and stiffness intapping-mode atomic force microscopy, Surf. Sci. 375 (1997) L385–L391.

[64] P.J. James, M. Antognozzi, J. Tamayo, T.J. McMaster, J.M. Newton, M.J. Miles,Interpretation of contrast in tapping mode AFM and shear force microscopy. Astudy of nafion, Langmuir 17 (2000) 349–360.

[65] G. Bar, Y. Thomann, R. Brandsch, H.J. Cantow, M.H. Whangbo, Factors affectingthe height and phase images in tapping mode atomic force microscopy. Studyof phase-separated polymer blends of poly(ethene-co-styrene) and poly(2,6-dimethyl-1,4-phenylene oxide), Langmuir 13 (1997) 3807–3812.

[66] J.H. Wang, P.M. Claesson, J.L. Parker, H. Yasuda, Dynamic contact angles andcontact angle hysteresis of plasma polymers, Langmuir 10 (1994) 3887–3897.

[67] M.A. Floriano, C.A. Angell, Surface tension and molar surface free energy andentropy of water to �27.2.degree.C, J. Phys. Chem. 94 (1990) 4199–4202.

[68] A.J. Meuler, J.D. Smith, K.K. Varanasi, J.M. Mabry, G.H. McKinley, R.E. Cohen,Relationships between water wettability and ice adhesion, ACS Appl. Mater.Interfaces 2 (2010) 3100–3110.

[69] N.H. Fletcher, Size effect in heterogeneous nucleation, J. Chem. Phys. 29 (1958)572–576.

[70] H.R. Pruppacher, J.D. Klett, Microphysics of Clouds and Precipitation, Springer,1997.

[71] M. He, H. Li, J. Wang, Y. Song, Superhydrophobic surface at low surfacetemperature, Appl. Phys. Lett. 98 (2011) 093118.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具