Dynamics of Ice Nucleation on Water Repellent SurfacesAzar Alizadeh,*,† Masako Yamada,† Ri Li,† Wen Shang,† Shourya Otta,† Sheng Zhong,† Liehui Ge,‡

Ali Dhinojwala,‡ Ken R. Conway,† Vaibhav Bahadur,† A. Joseph Vinciquerra,† Brian Stephens,§

and Margaret L. Blohm†

†General Electric Global Research, Niskayuna, New York 12309, United States‡Department of Polymer Science, The University of Akron, Ohio 44325, United States§GE Aviation, Cincinnati, Ohio 45215, United States

*S Supporting Information

ABSTRACT: Prevention of ice accretion and adhesion onsurfaces is relevant to many applications, leading to improvedoperation safety, increased energy efficiency, and cost reduction.Development of passive nonicing coatings is highly desirable,since current antiicing strategies are energy and cost intensive.Superhydrophobicity has been proposed as a lead passivenonicing strategy, yet the exact mechanism of delayed icing onthese surfaces is not clearly understood. In this work, we presentan in-depth analysis of ice formation dynamics upon waterdroplet impact on surfaces with different wettabilities. Weexperimentally demonstrate that ice nucleation under low-humidity conditions can be delayed through control of surfacechemistry and texture. Combining infrared (IR) thermometry and high-speed photography, we observe that the reduction ofwater−surface contact area on superhydrophobic surfaces plays a dual role in delaying nucleation: first by reducing heat transferand second by reducing the probability of heterogeneous nucleation at the water−substrate interface. This work also includes ananalysis (based on classical nucleation theory) to estimate various homogeneous and heterogeneous nucleation rates in icingsituations. The key finding is that ice nucleation delay on superhydrophobic surfaces is more prominent at moderate degrees ofsupercooling, while closer to the homogeneous nucleation temperature, bulk and air−water interface nucleation effects becomeequally important. The study presented here offers a comprehensive perspective on the efficacy of textured surfaces for nonicingapplications.

1. INTRODUCTIONIce accretion on surfaces of aircraft, wind turbine blades, oil andgas rigs, heat exchangers, transmission lines, boats, buildings,and other infrastructure presents long recognized problemswith respect to safety, efficiency, and cost of operation.1−7

Current active ice mitigation approaches are often based onmelting or breaking of already formed ice layers. In addition totheir undesired weight and design complexity, active antiicingapproaches require substantial energy for their operation.3−7

Passive icephobic (low ice adhesion) coatings have also beenproposed during the past 5 decades, yet their performance hasoften been suboptimal.4,8−13 Both passive and activeapproaches rely on the removal of a finite ice layer; measuringand controlling the thickness of this layer are yet otherchallenges.Development of coatings that limit and ultimately prevent ice

accretion on their surfaces is desirable and has been the subjectof considerable recent attention.14−23 This interest has beensparked by the remarkable water repellent properties ofsuperhydrophobic surfaces.24,25 The strength of ice adhesionto a flat surface decreases with increasing hydrophobicity.26−28

Therefore, ice is expected to adhere very weakly to super-hydrophobic surfaces with water contact angles > 150°.

Paradoxically, some authors have shown that ice adhesionincreases with surface roughness,27 while others have shown upto 18-fold reduction of ice adhesion strength on super-hydrophobic surfaces.19,20 Recently, Mishchenko et al. haveshown that water droplets impinging on superhydrophobicsurfaces exhibit nonicing behavior if the time scale of dropletwetting and retraction from the surface is smaller than the icenucleation time.14 These authors offer an analysis of icingthrough visual examination of supercooled droplets and first-order modeling of icing under droplet impact. Cao et al.17 havereported that superhydrophobic surfaces exhibit icephobicproperties under water impact conditions, but only if thesurface texture dimensions fall within a critical size regime.Surprisingly, the critical size regime suggested by Cao et al.17

for nonicing behavior (<50 nm) differs significantly from thetexture dimensions used by Mishchenko et al. (severalmicrometers). Delay in freezing of static water pools onnanostructured surfaces has been reported by Tourkine et al.18

This delay has been attributed to the insulating properties of

Received: November 16, 2011Revised: January 5, 2012Published: January 11, 2012

Article

pubs.acs.org/Langmuir

© 2012 American Chemical Society 3180 dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−3186

textured superhydrophobic surfaces, which are commonlycomprised of a large fraction of air pockets. The discrepancyin icing behavior as reported by various authors may be due tothe lack of robustness of superhydrophobic surfaces, due tohumidity, insufficient pressure stability, and hysteresis duringconsecutive icing/deicing cycles.15,29−33

In this paper, we present a quantitative study of icenucleation on surfaces with various wettabilities during andsubsequent to single water droplet impact. We use acombination of high-speed photography and infrared ther-mometry to capture the temporal evolution of both surfacewetting and icing phase transition events. Our methodologyprovides insights into multiple ice nucleation delay mecha-nisms. We complement and compare our experimental findingswith the predictions of classical nucleation theory. These resultsillustrate the strong dependence of various nucleation rates onthe supercooling temperature and quantify the effectiveness ofsuperhydrophobic surfaces as a function of temperature.

2. SURFACE FABRICATION AND EXPERIMENTALSETUP

Table 1 lists the surfaces used in this study along with the staticcontact and roll off angles. The selected surfaces span a wide

range of surface wettabilities from hydrophilic to super-hydrophobic. The hydrophilic samples (Si-PEG) were preparedby vapor deposition of self-assembled monolayers of 2-[methoxy(polyethyleneoxy)propyl]trimethoxysilane (Mw =660 g/mol) on plasma-treated silicon surfaces. The hydro-phobic samples (Si−F) were prepared in a similar way by self-assembly of tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane(Mw = 482 g/mol). Nanostructured silicon surfaces composedof single and double arrays of posts (referred to as Si-Stex-Fand Si-Dtex-F) were fabricated using standard photolithog-raphy and reactive ion etching processes. Post lateraldimensions, spacing, and heights of the nanostructured siliconsurfaces were varied within 0.5−15, 1−30, and 1−15 μm,respectively. Subsequently, self-assembled monolayers oftrideca fluoro-1,1,2-tetrahydroxyl trichlorosilane (from Gelest)

were deposited on the textured surfaces. Advancing, recedingand roll-off contact angles of sessile DI water droplets weremeasured on all of the samples at room temperature using aVCA Optima setup from AST Products, Inc.

3. EXPERIMENTAL MEASUREMENTS OF FREEZINGDELAY ON SURFACES

IR thermography is used in the present work to monitor thedroplet temperature during the icing phase transition anddetect the occurrence of freezing. IR thermometry haspreviously been utilized to provide quantitative informationon the freezing of freestanding water droplets;34 however, thereis no available literature on IR imaging of water phasetransitions in the vicinity of surfaces. We demonstrate thatremote (nonintrusive) sensing of the droplet and substratetemperatures via IR thermometry is particularly advantageousfor impinging droplets and/or complex nanostructuredsubstrates, where the use of thermocouples or intrusive thermalanalysis techniques may not be feasible. The IR camera wasfocused on the top surface of the droplet and thus measured thesurface temperature (since water is opaque at IR wavelengths).Specific details about the measurement technique are providedin the Supporting Information. Figure 1 shows a representativeIR thermal analysis of a static 6 μL water droplet on a Si-PEGsubstrate while the substrate is being cooled at 20 °C/min. Theinitial substrate temperature was ∼5 °C, and all experimentswere conducted at low relative humidity (RH 2% at roomtemperature). Next to the droplet, a piece of thermallyconductive black tape was attached to the Si-PEG surface tomonitor the substrate temperature (Figure 1a). (Using basicheat conduction calculations, the temperature drop across thetape was estimated to be less than 1 °C.) IR images of thedroplet top surface and the black tape were recorded as thesubstrate was cooled. Parts b and c of Figure 1 show thetransient temperatures at the top of the droplet and on theblack tape. A sudden temperature jump occurs when thedroplet temperature reaches −19 °C (point 1 in Figure 1c). In30 ms, the droplet temperature rises from −19 to ∼−2 °C(point 2) and then to −0.1 °C (point 3) after another 30 ms;this sudden rise in temperature indicates the start of freezingand can be attributed to the latent heat of fusion released aspart of the water freezes. From point 3 to point 7 (∼5 sduration), the droplet temperature remains constant at ∼0 °C,which is the freezing temperature of water at 1 atm. At thisstage all of the liquid water has transformed to solid ice;subsequently the droplet temperature decreases to reach thesurface temperature in ∼2 s (points 8−11 in Figure 1c).In Figure 1c, point 1 represents the onset of freezing. Since

the transient temperature is measured at the top of the droplet,the role of the water−substrate interface can be quantifiedthrough an analysis of the heat transfer within the droplet. Athermal time constant can be defined as τ = ρh2Cp/k, where ρ,h, Cp, and k are the density, center height, heat capacity, andthermal conductivity of the sessile water droplet, respectively.This time constant is based on diffusion (liquid convection isneglected) and is a measure of the time for a temperaturechange to propagate across the droplet. Using water propertiesat 0 °C and a droplet height of ∼1.1 mm, a thermal timeconstant τw ∼ 9 s is obtained. The time constant obtained forheat to diffuse through ice (using ice properties at −20 °C) is τi∼ 0.9 s. The sharp rise of temperature between points 1 and 3cannot be explained through a heat-transfer mechanism, as thetime elapsed between points 1 and 3 (Figure 1c) is 2−3 orders

Table 1. Morphological and Static WettabilityCharacteristics of Surfaces Used in Present Study

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863181

of magnitude shorter that the thermal time constants τw and τi.A plausible explanation for the ultrafast temperature rise frompoint 1 to point 3 is that, after the initiation of a critical nucleusat the water−substrate interface, freezing is triggered across theentire volume due to the percolation of the hydrogen-bondednetwork. As water is supercooled, hydrogen-bond lifetimeincreases by orders of magnitude, forming long-lived hydrogen-bonded clusters.34,35 The fraction of water that nucleates duringthis global nucleation event, about 20% for the current system,is restricted by the liberation of latent heat of fusion at thehighest allowable temperature of 0 °C. The remainder of thewater freezes subsequently; the rate and direction ofpropagation of the multiple freeze fronts depend on heat-transfer rates in various regions of the droplet. The entiredroplet freezes in about 6 s; this again depends on the transferof latent heat from the droplet to the substrate (points 3−8 inFigure 1c). This duration is remarkably close to the waterthermal time constant τw, which is a measure of how fast heatdiffuses through the droplet. The quick decrease in the droplettemperature subsequent to freezing is due to the higher thermaldiffusivity of ice, validated by the time lapse between points 8and 11 being on the order of the ice thermal time constant τi. Itis thus seen that IR thermography can help in distinguishingand understanding two distinct physical processes governing iceformation: ice nucleation and the heat transfer from droplet tosubstrate.Figure 2 compares the transient temperature profiles of water

droplets freezing on three surfaces with varying degrees ofhydrophobicity. In these experiments, 4 μL room temperatureDI water droplets were impacted on the substrates at a speed of2.2 m/s. In all cases, the droplets came to rest within 100 msafter impact. The substrates were maintained at −20 °C in alow-humidity ambient environment (<2% RH at roomtemperature) to avoid condensation. The transient temperaturecurves in Figure 2 capture both the heat transfer between the

droplet and the substrates (cooling of the droplet from room

temperature to −20 °C) and the freezing event once the

droplet has been sufficiently supercooled. The time scale for

both the heat transfer and freezing is a strong function of the

Figure 1. IR thermal-imaging analysis of a 6 μL water droplet freezing on a PEGylated silicon substrate during constant cooling at a rate of 20 °C/min: (a) IR image of water droplet and conductive black tape; (b) transient temperature of the water droplet (red) and black tape (black) during thecooling process; (c) magnification of the phase transition regime.

Figure 2. Transient temperatures of 4 μL DI water droplet freezing on(a) Si-PEG (hydrophilic), (b) Si−F (hydrophobic), and (c) Si-STex-F(superhydrophobic) substrates. In all cases, room-temperaturedroplets were impinged on the −20 °C substrates at a velocity of2.2 m/s. The corresponding high-speed video images of 4 μL DI waterdroplets impacting on the hydrophilic (Si-PEG), hydrophobic (Si−F),and superhydrophobic (Si-STex-F) substrates are also shown. Thehigh-speed images correspond to 20 ms after the impact.

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863182

surface hydrophilicity. Droplets cool relatively rapidly on thePEGylated substrate, and the freezing initiates while thedroplets are still cooling down. On the fluorinated siliconsubstrate, the kinetics of heat transfer and freezing are slightlydelayed with respect to the PEGylated surface, yet the freezingis still initiated during the cooling stage. A very differentbehavior is observed on the superhydrophobic nanostructuredsilicon substrates: a prolonged period of time (∼20 s) isrequired for the droplet to reach the substrate temperature.Furthermore, the droplet remains in the liquid state at thistemperature for an additional period of 60 s before freezing.The onset of freezing is estimated from the IR transienttemperatures after subtraction of the time required to reach thesubstrate temperature.Delayed freezing of water on superhydrophobic substrates

has been reported previously and mainly attributed to theinsulating properties of these surfaces.14,18 The resultspresented here suggest that in addition to a reduced heat-transfer path, other physical mechanisms play a role in thedelayed freezing. We postulate that the ice nucleation rate onthe superhydrophobic substrates is also affected due to areduction of water−substrate interfacial area and increase in thefree energy barrier to forming a critical nucleus at the interface.Water−substrate interfacial areas upon droplet impact on the

surfaces were determined using high-speed photography. Asshown in the Supporting Information (Figure S1), a high-speedcamera (Phantom V7.3 from Vision Research) was utilized toimage the droplet from the side at 3000 frames/s.Representative high-speed images of DI droplet impact on Si-PEG (hydrophilic), Si−F (hydrophobic), and Si-DTex-F(superhydrophobic) substrates at −15 °C are shown in Figure2. The high-speed images correspond to 20 ms after the impact.Upon impact, the extent of spreading is mainly determined bythe kinetic energy of the droplet (its velocity and size) and thehydrophobicity of the surface. Subsequently, the droplets recoiland contract until finally adapting a steady-spreading diameter.Limited recoil is observed for the PEGylated surface. Dropletson the Si−F substrate, while maintaining contact at all timeswith the substrate, show a stronger retraction. In contrast,droplets bounce off the textured fluorinated substrates. Eachdroplet may experience several bounces on a superhydrophobicsurface before coming to a rest [video recordings of waterdroplets impacting various surfaces are shown in theSupporting Information].Water−substrate interfacial areaswere estimated from the final resting states of the droplets bymeasuring steady-state diameters and are reported in Table 2.The water−substrate interfacial area is a strong function of

substrate chemistry, morphology, and temperature as well as

the droplet speed. The details of droplet spreading dynamics onthese surfaces will be published elsewhere. We have estimatedthe interfacial areas for small impacting velocities (gentlepositioning of the droplet on the substrate) from theexperimentally determined static contact angles assuming aspherical geometry. At low temperatures, water-substrateinterfacial area varies by almost 2 orders of magnitude whencomparing the hydrophilic and superhydrophobic substrates.The impact of this large difference on the icing kinetics isshown in Figure 3, where the variation of the freezing onsettime is plotted as a function of the water−substrate interfacialarea for two freezing conditions.

4. ESTIMATION AND COMPARISON OF SURFACEAND BULK NUCLEATION RATES

The results in Figure 3 are consistent with a change in thenucleation rate as a function of the macroscopic water−substrate interfacial area. Classical nucleation theory principlesare used to predict homogeneous and heterogeneousnucleation rates in supercooled water droplets, as a functionof the droplet size, extent of supercooling, contact area, andsurface chemistry. Classical nucleation theory has previouslybeen used to understand the freezing of freestanding waterdroplets, showing that surface or subsurface nucleation candominate over homogeneous bulk ice nucleation in specifictemperature ranges.36−39 In this paper, we build upon thisframework and the parameters developed by Zobrist et al.40

[the details of classical nucleation theory approach andparameters used here are described in the SupportingInformation], to describe nucleation rates in water dropletson supercooled substrates. We have added an additional term,Jwater−substrate to Zobrist’s formulation, to account for heteroge-neous nucleation at the substrate−droplet interface. Accord-ingly, the total nucleation rate, Jtotal (expressed in units of nucleis−1 cm−3), of a droplet can be written as

Table 2. Water−Substrate Area for a 4 μL Water DropletImpinging at 2.2 m/sa

static conditions (T = 22 °C)

sample

contactangle(deg)

interfacial area forgently placed droplet

(mm2)

interfacial area afterdroplet impact at −15 °C

(mm2)

Si-PEG 44 2.54 32.1Si−F 109 0.79 4.26Si-STex-F 150 0.03 0.1

aInterfacial contact area after droplet impact is calculated from thesteady spreading diameters obtained through high-speed videorecording. Interfacial contact areas for static droplets are calculatedon the basis of geometrical considerations of a spherical dome.

Figure 3. Variation of the freezing onset as a function of droplet−substrate contact area. Freezing onset was extracted from the transientIR curves in Figure 2 upon subtraction of the time required to reachthe substrate temperature. Contact areas were estimated from thedroplet impact experiments at T = −15 °C. The dashed lines are justfor guidance.

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863183

= +

+− −

− −

J J V J S

J Stotal bulk water air water air

water substrate water substrate (1)

where Jbulk, Jwater−air, and Jwater−substrate are the correspondingtemperature-dependent bulk, water−air, and water−substrateinterface nucleation rates, respectively. V, Swater−air, andSwater−substrate represent the droplet volume, water−air, andwater−substrate interfacial areas, respectively. Details of theestimations of all of these nucleation rates are provided in theSupporting Information.Figure 4a shows the predicted total nucleation rate, Jtotal, as a

function of water−substrate contact area for a droplet radius of1 mm at two different temperatures. In both cases, a significantdecline in nucleation rate is observed, as the surface becomesmore hydrophobic. In this temperature range, the totalnucleation rate is increased with decreasing the supercoolingtemperature for the hydrophobic substrates. Both observationsare consistent with our experimental results: fast nucleation ofwater droplets on hydrophilic substrates is observed, whereasthe freezing time of water droplets on hydrophobic andsuperhydrophobic substrates is reduced to a few seconds at−25 °C (see Figure 3).Although it is tempting to compare nucleation times derived

via classical nucleation theory relative to those measuredexperimentally, the theoretical nucleation time and experimen-tal “onset of freezing” time cannot be directly compared, as theydo not represent the same physical state. The equivalent of“nucleation time” cannot be measured directly, as the criticalnucleus is only a few nanometers in size and cannot bedetected. By the time the onset of freezing can be measuredusing the IR camera, the nucleus has already grown tomacroscopic dimensions [the spatial resolution of the IRcamera in the current setup is ∼100 μm]. The rate of crystalgrowth from a critical nucleus size to a detectable size is

impacted by the diffusivity of the molecules and heat removalduring the phase transition. These additional factors are notaccounted for in the calculation of theoretical nucleation time.Therefore a direct comparison between the theoreticalnucleation rates and the inverse of freezing onset time is notjustified.In Figure 4b−d, the contributions of the bulk, air−water, and

substrate−water nucleation rates to Jtotal for various water−substrate contact areas are shown (these areas correspond tohydrophilic, hydrophobic, and superhydrophobic surfaces,respectively). For hydrophilic substrates, nucleation rates arepredominantly higher at the water−substrate interface (Figure4b). Previous studies of static water freezing in the vicinity ofhydrophilic substrates also indicate that nucleation is initiated atthe water−substrate interface and not in the bulk or at thewater−air surface18,41 with the exceptions of cases where thewater structure at the interface or further in to the bulk isstrongly disturbed by the presence of a substrate.16 In contrast,on hydrophobic substrates (Figure 4b), the contributions ofbulk, water−air, and water−substrate to Jtotal can all be equallyimportant (see Figure 4b−d). Moreover, each of these termscan become more significant depending on the supercoolingtemperature. For instance, at lower temperatures, the substratenucleation rates are less than bulk nucleation rates. The classicalnucleation theory analysis used here shows a crossovertemperature below which substrate effects become lessimportant due to the enhanced rates of bulk and air−waterinterface nucleation. This crossover temperature is a strongfunction of substrate wettability. As expected, for the extremecase of superhydrophobic substrates (contact angle > 150°), thecontribution of water−substrate nucleation term to Jtotal isnegligible. For modest supercoolings, the ice nucleation rate isdramatically reduced.40 For instance, under current exper-imental conditions, at temperatures ≥ −15 °C, the onset offreezing for most substrates was longer than 2.5 min. Under

Figure 4. (a) Variation of total nucleation rate as a function of water−substrate contact area at −20 and −25 °C. Contributions of bulk (blacksymbols), air−water (red symbols), and substrate−water (blue symbols) nucleation rates to Jtotal for (b) hydrophilic substrate, (c) hydrophobicsubstrate, and (d) superhydrophobic substrate.

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863184

current experimental conditions of low humidity, during thisextended induction period, droplets may evaporate leading toeither evaporation-induced crystallization43 or convolutedeffects due to droplet volume reduction. Therefore, experi-ments at temperatures above −20 °C, which will lead toextended induction times before freezing, have not beenincluded in this study. It should be noted that all of the classicalnucleation theory calculations were presented for a dropletradius of 1 mm. Droplet size has a profound effect on the totalnucleation rate, as well as the prominence of substrate effectsrelative to bulk or air−water interface effects. It should also benoted that our experiments were conducted under lowhumidity, to replicate idealized freezing conditions. Preexistingcondensate or frost on the substrate would add additionalcomplexity to the picture, due to the time-dependent variabilityof many factors, including varying distribution of condensatedrop size and contact angles.

5. CONCLUSIONS

In summary, this work shows that icing kinetics of impingingwater droplets is impacted by substrate wetting dynamics. Ourwork not only extends previous observations of significantwater roll-off, reduced water/ice adhesion, and limited heat-transfer path on superhydrophobic substrates but also revealsnew fundamental insights on the kinetics of ice nucleation onthese surfaces. It is seen that, for intermediate supercoolingtemperatures, nucleation at the water−substrate interfacebecomes the dominant factor controlling icing; in thisnucleation regime; the contact angle affects both the interfacialsurface area (for nuclei formation) and the energy barrier, orprobability, of forming a single nucleus. The results presentedhere suggest that high contact angles lead to both the reductionof the water−substrate interfacial area and an increase innucleation activation energy, which can lead to a drasticreduction of the macroscopic nucleation rate (delayedfreezing). However, at lower supercooling temperatures it isseen that water−substrate interface nucleation is not thedominant factor (bulk and air−water nucleation dominate)controlling icing; this reduces the attractiveness of usingsuperhydrophobic surfaces at such temperatures. It can beconcluded that the observed delay in nucleation withsuperhydrophobic surfaces has tremendous promise, yet,there exist challenges associated with designing nonicingtextured surfaces that are effective in all temperature ranges.

■ ASSOCIATED CONTENT

*S Supporting InformationFigures showing the experimental apparatus and variation ofthe freezing onset as a function of droplet−substrate contactarea, text describing IR thermometry, freezing of supercooleddroplets, room-temperature droplet impact dynamics, andestimation of nucleation rates, and videos showing room-temperature impact of a water droplet on hydrophilic,hydrophobic, and superhydrophobic surfaces. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: Alizadeh@research.ge.com.

■ ACKNOWLEDGMENTS

The authors are very thankful to James Ruud, Chris Keimel,Oliver Boomhower, Peter Morley, Lauraine Denault, and TaoDeng for their assistance of this work and helpful discussions.The Nanotechnology Advanced Technology Program at GEGlobal Research, the Department of Energy (Grant Award No.DE-AC07-05ID14517), and the National Science Foundation(Grant No. 1006764) are acknowledged for financial support.

■ REFERENCES(1) Laforte, J. L.; Allaire, M. A; Laflamme, J. Atmos. Res. 1998, 46,143.(2) Petrenko, V. F.; Whitworth, R. W. Physics of Ice; OxfordUniversity Press: Oxford, U.K., 1999.(3) Gent, R. W.; Dart, N. P.; Cansdale. Philos. Trans. R. Soc., A 2000,358, 2873.(4) Mulherin, N. D.; Haehnel R. B. Ice Engineering, ERDC/CRRELTechnical Note 03−04, 2003.(5) Lasher-Trapp, S.; Anderson-Bereznicki, S.; Shackelford, A.;Twohy, C. H.; Hudson, J. G. J. Appl. Meteorol. Clim. 2008, 47, 2659.(6) Ryerson, C. C. U.S. Army Corps of Engineers, ERC/CRREL TR-08-14, 2008.(7) Laakso, T.; Baring-Gould, I.; Durstewitz, M.; Horbaty, R.;Lacroix, A.; Peltola, E.; Ronsten, G.; Tallhaug, R.; Wallenius, T. State ofthe Art of Wind Energy in Cold Climates; VTT Technical ResearchCentre of Finland: Espoo, Finland, 2010; http://www.vtt.fi/publications/index.jsp(8) Porte, H. A.; Nappier, T. E. U.S. Naval Civil Engineering Report1963.(9) Bascoma, W. D.; Cottingtona, R. L.; Sinleterry, C. R. J. Adhes.1969, 1, 246.(10) Yoshida, M.; Ochichi, T.; Konno, K.; Gocho, M. Cold RegionsTechnology Conference, Sapporo, Japan, 1991.(11) Croutch, V. K.; Hartley, R. J. Coat. Technol. 1992, 64, 41.(12) Laforte, C.; Carriere, J. C.; Laforte, J. L. International Workshopon Atmospheric Icing of Structures (IWAIS 2000), Brno, Czech Republic,2002.(13) Petrie, E. M. Met. Finish. 2009, 107, 56.(14) Mishchenko, L.; Hatton, B.; Bahadur, V.; Taylor, J. A.;Krupenkin, T.; Aizenberg, J. ACS Nano 2010, 4, 7699.(15) Meuler, A. J.; McKinley, G. H.; Cohen, R. E. ACS Nano 2010, 4,7048.(16) Ehre, D.; Lavert., E.; Lahav, M.; Lubomirsky, I. Science 2010,327, 672.(17) Cao, L.; Jones, A. K.; Sikka, V. K.; Wu., J.; Gao, D. Langmuir2009, 25, 12444.(18) Tourkine, P.; LeMerrer, M.; Quere, D. Langmuir 2009, 25,7214.(19) Daton, A.; Dodiuk, H.; Laforte, C.; Kenig, S. J. Adhes. Sci.Technol. 2009, 23, 1907.(20) Kulinich, S. A.; Farzaneh, M. Appl. Surf. Sci. 2009, 255, 8153.(21) Jung., S.; Dorrestijn, M.; Raps, D.; Das, A.; Megaridis, C. M.;Poulikakos, D. Langmuir 2011, 27, 3059.(22) Bahadur, V.; Mishchenko, L.; Hatton, B.; Taylor, J. A.;Aizenberg, J.; Krupenkin, T. Langmuir 2011, 27 (23), 14143−14150.(23) Zheng, L.; Li, Z.; Bourdo, S.; Khedir, K. R.; Asar, M. P.;Ryerson, C. C.; Biris, A. S. Langmuir 2011, 27, 9936.(24) Cao, L.; McCarthy, T. J. Langmuir 2009, 25, 14105.(25) Quere, D. Annu. Rev. Mater. Res. 2008, 38, 71.(26) Petrenko, V. F.; Peng, S. Can. J. Phys. 2003, 81, 387.(27) Saito, H.; Takai, K.; Yamauchi, G. A. Surf. Coat. Int., Part B2003, 80, 168.(28) Meuler, A. J.; Smith, J. D.; Varanasi, K. K.; Mabry, J. M.;McKinley, G. H.; Cohen, R. E. ACS Appl. Mater. Interfaces 2010, 2,3100.(29) Kulinich, S. A.; Farhadi, S.; Nose, K.; Du, X. W. Langmuir 2011,27, 25.

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863185

(30) Varanasi., K. K.; Deng, T.; Smith, J. D.; Hsu, M.; Bhate, N. Appl.Phys. Lett. 2010, 97, 234102.(31) Nosonovsky, M.; Bhushan, B. J. Phys.: Condens. Matter 2008, 20,225009.(32) Jung, Y. C.; Bhushan, B. Langmuir 2008, 24, 6262.(33) Reyssat, M.; Pepin, A.; Marty, F.; Chen, Y.; Quere, D. Europhys.Lett. 2006, 74, 306.(34) Bauerecker, S.; Ulbig, P.; Buch, V.; Vrbka, L.; Jungwirth, P. J.Phys. Chem. C 2008, 112, 7631.(35) Matsumoto, M.; Saito, S.; Ohmine, I. Nature 2002, 416, 409.(36) Earle, M. E.; Kuhn, T.; Khalizov, A. F.; Sloan, J. J. Atmos. Chem.Phys. 2010, 10, 7945.(37) Shaw, R. A.; Durant, A. J.; Mi, Y. J. Phys. Chem. B 2005, 109,9865.(38) Tabazadeh, A.; Djikaev, Y. S.; Reiss, H. Proc. Natl. Acad. Sci. U. S.A. 2005, 99, 15873.(39) Kjellander, R.; Marcelja, S. Chem. Phys. Lett. 1985, 120, 393.(40) Zobrist, B.; Koop, T.; Luo, B. P.; Marcolli, C.; Peter, T. J. Phys.Chem. C 2007, 111, 2149.(41) Furuta, T.; Sakai, M.; Isobe, T.; Nakajima, A. Chem. Lett. 2010,39, 1090.(42) Vali, G. Ice Nucleation Theory, A Tutorial, NCAR/ASP, 1999.(43) Durant, A. J.; Shaw, R. A. Geophys. Res. Lett. 2005, 32, 1.

Langmuir Article

dx.doi.org/10.1021/la2045256 | Langmuir 2012, 28, 3180−31863186