Hybrid Hairy Janus Particles for Anti-Icing and De-Icing Surfaces

---

Article pubs.acs.org/cm

Hybrid Hairy Janus Particles for Anti-Icing and De-Icing Surfaces: Synergism of Properties and Effects Alina Kirillova,†,‡ Leonid Ionov,§ Ilia V. Roisman,∥ and Alla Synytska*,†,‡ †

Leibniz-Institut für Polymerforschung Dresden e.V., Hohe Str. 6, 01069 Dresden, Germany Fakultät Mathematik und Naturwissenschaften, Technische Universität Dresden, 01062 Dresden, Germany § College of Engineering, College of Family and Consumer Sciences, University of Georgia, Athens, Georgia 30602, United States ∥ Technische Universität Darmstadt, Alarich-Weiss-Str. 10, 64287 Darmstadt, Germany ‡

S Supporting Information *

ABSTRACT: A novel route for the design of functional surfaces with effective anti-icing and de-icing capability based on hybrid Janus particles is presented. The heterogeneous surfaces formed by Janus particles exhibit special surface “edge” morphologies. Water first condenses on the hydrophilic portion of the surfaces, occupying relatively large hydrophilic clusters. It is pinned at the boundary between the hydrophilic and the hydrophobic regions and thus cannot penetrate in the cavities between the particles. Further condensation leads to the fast coalescence of the water clusters, which after freezing yields a fast appearance of large ice crystals, dendrites, in the shape of the agglomeration of sector plates. The mechanism of the dendrite formation is proved experimentally and by Monte Carlo simulations. Moreover, a dry band is formed around the large crystals due to the evaporation of small drops in the vicinity of the large water clusters and the subsequent ice crystals. The synergism of both effects, the area free of ice and the large unstable dendrites at the edges of heterogeneities, leads to an extremely low ice adhesion of ca. 56 kPa. The presented approach opens a new avenue for the rational design of ice-free coatings using Janus particles as building blocks.

1. INTRODUCTION The development of surfaces with reduced icing or easy deicing is of paramount importance for the wind turbine technology as well as automotive and aircraft industries. In fact, inhibition of the ice layer formation allows not only a consequent reduction of costs but also decreases the number of emergency situations. The commonly used methods for the prevention of icing and promotion of de-icing are based either on electrical heating of the surfaces, which results in simple ice melting, or on the use of antifreeze substances, which reduce the water freezing temperature. Nonetheless, the most favorable solution of this problem, which is now broadly explored, is the design of passive anti-icing coatings, i.e., coatings that prevent icing or reduce the ice accretion rate,1−3 which allows lowering the required power consumption for the ice removal.4 There are few main strategies for the design of such passive anti-icing coatings. One of them is based on the lowering of the ice adhesion strength. Reduced ice adhesion can be achieved by the fabrication of hydrophobic5 or superhydrophobic surfaces,6−13 as well as the use of hydrophobic lubricants.14 In the first case (hydrophobic and superhydrophobic surfaces), the contact area between the surface and ice is simply reduced.15−17 In fact, superhydrophobic materials demonstrate good anti© 2016 American Chemical Society

icing properties, but once an ice layer is formed, it can hardly be removed. Moreover, the removal of the ice layer leads to the loss of the superhydrophobic properties. In some cases superhydrophobic surfaces are not ice-repellent, especially when the size of the features is large and water droplets can penetrate between them.18 In the second case (hydrophobic lubricants), ice can easily be removed because of the hydrophobic lubricant softness.9,14,19,20 Another strategy is based on the inhibition of the ice growth. Reduced ice growth can be achieved in two ways. The first way is to utilize the colligative properties of solutions.21,22 For example, hydrophilic polymers reduce the freezing point of water, and ice crystals can simply slide off due to the presence of an unfrozen water layer.23 The second approach is to employ antifreeze proteins, which kinetically decrease the point of ice crystal formation.24 Nevertheless, the disadvantage of such hydrophilic materials is the very fast growth of ice crystals on their surface. Received: July 7, 2016 Revised: September 15, 2016 Published: September 19, 2016 6995

DOI: 10.1021/acs.chemmater.6b02765 Chem. Mater. 2016, 28, 6995−7005

Article

Chemistry of Materials There are several attempts to employ topographical and/or chemical heterogeneity to reduce icing.17,25−30 For instance, Mishchenko et al. demonstrated that topographical control over water condensation and freezing was achieved at the micrometer scale via deposition of hydrophilic polymers and particles on the tips of selected patterned, superhydrophobic surfaces.17 Indeed, the hydrophilic/hydrophobic heterogeneity and roughness allow a controlled nucleation and growth of ice crystals.25,26 Commonly, the photolithographic approach was used for the fabrication of such patterned surfaces with the size of the features on the microscale. There are also reports showing that block copolymers, which typically possess nanoscale heterogeneity, allow reduction of the ice adhesion.31−33 The exact mechanism of the ice adhesion reduction on a block copolymer surface was, however, not elucidated. Moreover, while the homogeneous surfaces demonstrate a relatively clear correlation between wetting and ice adhesion (the higher the contact angle, the lower the ice adhesion), the correlation between wettability and ice adhesion of composite surfaces is weak. In this work, we demonstrate for the first time that hydrophilic/hydrophobic hybrid hairy Janus particles (particles having both hydrophilic and hydrophobic sides) can be successfully employed for the design of structured heterogeneous surfaces with controlled ice nucleation and growth, as well as extremely low ice adhesion strength. As a step toward the understanding of the unusual icing properties of heterogeneous surfaces, we correlate the mechanism of the ice layer growth and the strength of ice adhesion. Previously, some works have been dedicated to the assembly of Janus particles at air/liquid interfaces.34,35 However, there are very few studies regarding their assembly at solid/air interfaces and their potential application for functional surfaces,36,37 especially with controlled wetting and adhesion.38,39 Moreover, the large advantage of the Janus particle-based surfaces compared to the lithography-based methods is that the Janus particles can easily be prepared on a large scale by the recently developed methods and can be used to cover large surface areas by simple spraying or solvent casting. Furthermore, similar to block copolymers which reproducibly form heterogeneous structures, however, only at the nanoscale, due to the phase separation between the incompatible blocks, Janus particles offer heterogeneous structures at different scales between the nano- and micrometer levels depending on the particle size. Additionally, the hybrid Janus particles with inorganic core possess higher robustness and long-term stability over the block copolymers. Therefore, the presented approach offers feasible and scalable building blocks for the rational design of ice-free surfaces and coatings in the future.

Figure 1. Representative layers prepared with P(PEGMA)/PDMS Janus particles: (a) chemical formulas of the polymers; (b) false color SEM image (orange side, P(PEGMA); green side, PDMS; the original image can be found in Supporting Information, Figure S2); (c) direct force measurements on the P(PEGMA)/PDMS Janus particle layer at 80% relative humidity (dashed line, approaching curve; solid line,retracting curve; the respective force curves measured under water are displayed in Figure S3); (d) false color mapping of the Janus particle orientation using AFM direct force measurements (orange, P(PEGMA); green, (PDMS)).

subsequent colloidosomes when the wax is solidified, where one side of the particles is exposed to the environment and the other side is immersed in the wax. Initiator groups were then immobilized onto the exposed particle surface, and after the dissolution of the wax, polymer chains of one sort (P(PEGMA)) were grafted from the initiator-modified sides of the particles. The second polymer (PDMS) was grafted using the “grafting to” approach on the opposite sides of the particles. The resulting Janus particles have a slight asymmetry in the Janus ratio (2:1), which is determined by the degree of immersion of the SiO2 particles in the wax during the synthesis. We observed the edge between the two sides of the Janus particles with scanning electron microscopy (SEM) and concluded that the smaller part is PDMS, while the larger one is P(PEGMA). In addition, we fabricated reference layers made from colloidal particles of the same size modified by each of the polymers: either hydrophilic P(PEGMA) or hydrophobic P(PDMSMA) (fully covered particles), as well as flat surfaces modified by the same polymers (Table 1). Both polymers were

2. RESULTS AND DISCUSSION Structure of the Janus Particle Layers. In the present study, we fabricated robust (Figure S1) surfaces with hydrophilic/hydrophobic heterogeneity using 1 μm large core−shell poly(poly(ethylene glycol) methyl ether methacrylate)/polydimethylsiloxane (P(PEGMA)/(PDMS)) Janus particles (Figure 1a). The core of the Janus particles is a solid SiO2 particle obtained by the Stöber method,40 while the shell is formed by two polymers at the opposite sides of the core. The details on the synthesis of the Janus particles are given elsewhere.41 Briefly, 3-aminopropyltriethoxysilane (APTES)modified SiO2 particles were mixed with wax and water to form a Pickering emulsion at an elevated temperature and

Table 1. List of the Prepared Samples for the Icing Experiments sample ID

description

P(PEGMA) flat P(PDMSMA) flat SiO2−P(PEGMA)

P(PEGMA) polymer brush on a Si wafer P(PDMSMA) polymer brush on a Si wafer surface prepared from fully covered 1 μm large SiO2 particles with a P(PEGMA) shell surface prepared from fully covered 1 μm large SiO2 particles with a P(PDMSMA) shell surface prepared from SiO2-based core−shell Janus particles with two polymer shells

SiO2−P(PDMS) Janus

6996

DOI: 10.1021/acs.chemmater.6b02765 Chem. Mater. 2016, 28, 6995−7005

Article

Chemistry of Materials grafted to grafted by “grafting from” procedure. The particle layers were prepared by a simple solvent casting method, that is, by depositing and drying particle dispersions on premodified silicon wafers (see Experimental Section). We found that all particle layers are mechanically stable, and particles are not removed during the ice test experiments (surfaces were tested before and after the ice adhesion measurements, Figure S1). The layers formed by bicomponent Janus particles (JPs) are typically disordered (Figure 1b). In the SEM image, all kinds of orientations can be observed: JPs turned with their hydrophobic side up, JPs with their hydrophilic side facing up, and JPs partially exposing both sides to different extents. In order to gain statistical insight into the orientation of Janus particles in the prepared layers, their distribution and orientation was investigated by probing the adhesion properties of the top layer of the JPs using atomic force microscopy (AFM) direct force measurements (Figure 1c,d). The measurements were carried out under normal air pressure and increased humidity (80%). These are the conditions which were used for the subsequent ice adhesion testing. Under these conditions, the P(PEGMA)modified side of the particles is less adhesive than the P(PDMSMA)-modified one (Figure 1c). For a given particle, only its overall contribution to the hydrophobicity/hydrophilicity of the layer was determined, an approximation justified by the particles’ small size (Figure 1d). After statistical analysis of the particles in five 20 × 20 μm topographical AFM images, we found that the distribution of the particles oriented with their hydrophobic or hydrophilic side to the top is ca. 55%:45%, respectively. Thus, we fabricated mechanically robust layers of Janus particles, and the orientation of the Janus particles in the layers was random. For simplicity, we consider that the Janus particles are oriented upward either with their hydrophilic or their hydrophobic parts. Randomly distributed Janus particles, whose hydrophilic sides are directed upward with a probability ph, form irregular hydrophilic clusters. To better understand the statistics of particle distribution in the layer, percolation of these hydrophilic clusters on a honeycomb lattice is simulated using Monte Carlo simulations. In Figure 2a an example of such

estimated average size of a hydrophilic cluster is Laveragecluster = 7.9 μm. Wetting Properties. We found that flat P(PEGMA)- and P(PDMSMA)-modified surfaces are hydrophilic (ΘA = 30°, ΘR = 10°) and hydrophobic (ΘA = 120°, ΘR = 90°), respectively (Figure 3a). Importantly, roughness enhances the intrinsic

Figure 3. Summary of the wetting properties of rough surfaces made from colloidal particles modified by P(PEGMA), P(PDMSMA), and Janus particles (a−d). Representative cryo-SEM images of frozen water droplets on the surfaces formed by 1 μm P(PDMSMA) (e), P(PEGMA) (f), and Janus particles (g).

properties of the polymers and makes the hydrophobic surfaces even more hydrophobic and the hydrophilic ones even more hydrophilic. In particular, the advancing and receding contact angles on the layers formed by P(PDMSMA)-modified particles (ΘA = 150°, ΘR = 135°) are higher than those on a respective flat surface. The contact angles on the layers formed by P(PEGMA)-modified particles (ΘA = 20°, ΘR = 10°) are slightly lower than those on a respective flat surface (Figure 3a,b). This finding agrees with our previous observations.42,43 The values of advancing and receding water contact angles on the layers formed by Janus particles are in between the values measured for the layers of particles modified either by pure P(PEGMA) or by pure P(PDMSMA). Additionally, the values of sliding angles for liquid water drops of various volumes (5, 10, and 20 μL) were also used to assess the wetting properties of the particle layers (Figure 3c,d). Particularly, we observed no sliding of the water droplets on the P(PEGMA)-modified surfaces, which is due to their high hydrophilicity and strong spreading effects. The values of the sliding angles on pure P(PDMSMA)-modified surfaces are very high, and water droplets are pinned, which indicates a high contact angle hysteresis. In addition, the values of sliding angles on the layers made of Janus particles are slightly higher than those on the P(PDMSMA)-modified surfaces (Figure 3d). Thus, both types of measurements (advancing/receding contact angles as well as sliding angles) clearly show that the wetting properties of the bicomponent Janus particle layers are intermediate between those of the homogeneous surfaces with particles modified by only one polymer. Further, we microscopically investigated the behavior of water droplets on the layers of particles modified by one polymer and the layers of Janus particles by means of cryo-SEM measurements (Figure 3e−g). In a typical measurement,

Figure 2. Monte Carlo simulation of a substrate coated by randomly distributed Janus particles: (a) example of a site percolation on a honeycomb lattice, where the green sites correspond to the apparently hydrophobic particles and the orange to the hydrophilic (with the probability ph = 0.45); (b) estimated probability density function of the apparent sizes of the hydrophilic clusters.

a distribution is shown for the probability ph = 0.45, the same as in our experiments. The image of the sites and the clusters formed from the sites with the same wettability is very similar to the map shown in Figure 1d. We have measured the apparent sizes Lcluster of the hydrophilic clusters. The probability density function for the size Lcluster is shown in Figure 2b. The 6997

DOI: 10.1021/acs.chemmater.6b02765 Chem. Mater. 2016, 28, 6995−7005

Article

Chemistry of Materials micrometer-sized frozen droplets were formed on the particle layers through condensation and subsequent instantaneous freezing. It was observed that frozen water droplets are spread, and their contact angle is small on a hydrophilic surface with P(PEGMA)-modified particles (Figure 3f). It could be concluded that water penetrates into the pits between the particles due to the hydrophilicity of P(PEGMA). On the other hand, water droplets are compactly shaped and spherical on a hydrophobic surface with P(PDMSMA)-modified particles (Figure 3e). Water does not penetrate between the particles due to the hydrophobicity of P(PDMSMA). Droplets formed on the surface made from Janus particles have different shapes (Figure 3g). Some of them are spherical and look like droplets on the P(PDMSMA)-modified layers. Some droplets are flattened, and their shape is similar to the shape of the droplets on the P(PEGMA)-modified layers. Other droplets are nearly spherical but have clear pinning points (“mixed” droplets) where they are “pinned” to the particles. This indicates that the mechanism of water condensation on the Janus particle coatings is completely different. The liquid volumes under action of the capillary forces are collected on the hydrophilic regions. These irregular hydrophilic regions are composed of many particles with the hydrophilic spots directed upward. The size of the majority of the small drops is approximately 6−10 μm. This size is of the same order as the size of the hydrophilic cluster, 7.9 μm, estimated from Figure 2b. Larger droplets in Figure 3g of 70− 80 μm size occupy several hydrophilic clusters. This result is in agreement with the theoretical predictions, shown in Figure 2b. These larger liquid volumes are thus obtained by coalescence of the smaller liquid drops. The irregular shapes of these drops are caused by the fact that the contact line mainly coincides with the cluster boundary. The shapes of such liquid drops, despite a constant mean curvature, are not always spherical. Examples in Figure 3g are more similar to the constant mean curvature unduloid shapes.44 Thus, microscopic investigations by means of cryo-SEM analysis again confirmed that the wetting properties of surfaces formed by the Janus particles are intermediate between those of surfaces formed by P(PEGMA)and P(PDMSMA)-modified particles. Mechanisms of the Frost Layer Formation. First, we observed the frost layer formation on pure hydrophobic and pure hydrophilic particle-based surfaces using optical microscopy. On pure hydrophobic P(PDMSMA)-modified surfaces, we typically discovered a condensation of water droplets, which freeze and form slowly growing ice crystals due to the condensation of water molecules from vapor (Supporting Information: Movie_S3_PDMS_ acc12x). The formation of large rapidly growing crystals was very seldom. The observed mechanism of icing is explained by the growth of small dendrites (or bridges) on the frozen drops toward the nearest neighboring liquid drops. If the velocity of the dendrite (or the ice bridge between the drops) formation is higher than the speed of the liquid drop evaporation, this process leads to the chain (or percolation) expansion of icing from drop to drop.45 On the contrary, on pure hydrophilic P(PEGMA)-modified surfaces, we noticed that water condenses and forms a continuous layer, which freezes and slowly grows due to the condensation (Supporting Information: Movie_S4_PEG_acc24x). The mechanisms of freezing on hydrophilic and hydrophobic particle-based surfaces are shown schematically in Figure 4a,b, respectively.

Figure 4. Mechanisms of the ice layer formation on rough hydrophilic (a), hydrophobic (b), and Janus surfaces (c).

Next, we studied the frost layer formation on the coating made from bicomponent Janus particles (Figure 5, Figure 6). We found that the mechanism of icing on such surfaces differs significantly from the icing on the uniform layers of fully covered particles.46 Three main stages have been observed on the Janus surfaces: (i) water nucleation and condensation where liquid water appears as small drops, whose size increases during condensation, and the growing drops are also merging

Figure 5. Representative optical microscopy images of the P(PEGMA)/P(PDMSMA) Janus particle-based surface during icing: (a) native surface; (b) surface after the water droplet condensation; (c) growth of large liquid clusters and their solidification; (d, e) freezing of the condensed water droplets (different scales) and formation of dry bands around large crystals/dendrites; and (f) thawing of ice after finishing of the icing test. 6998

DOI: 10.1021/acs.chemmater.6b02765 Chem. Mater. 2016, 28, 6995−7005

Article

Chemistry of Materials

apparently hydrophilic clusters the water contact line is pinned at the hydrophobic/hydrophilic boundary at each particle. The sessile liquid drops are therefore in the Wenzel state.51,52 This state is rather stable. The impalement pressure53 on spherical particles can be roughly estimated through pimp =

upon contact; (ii) spontaneous coalescence and fast spreading of liquid water on the hydrophilic clusters;and (iii) solidification of the liquid clusters which leads to the appearance of relatively large irregular ice crystals (dendrites). These ice crystals look like agglomerates of sector plates which are usually observed at the temperatures around −20 °C.47,48 Further ice crystal deposition is accompanied by the evaporation of the neighboring liquid drops and the appearance of a dry band around the large single dendrites (Figure 4c). i. Water Nucleation on the Janus Surface. Cooling first leads to the condensation of micrometer-sized water droplets from air (Figure 5a,b; a video of the process is available in Supporting Information: Movie_S1_Janus_ acc12x, Movie_S2_Janus_acc4x).46 Then their size gradually increases, as more water condenses on the surface. Further cooling leads to the supercooling of the liquid droplets. They remain unfrozen until the temperature reaches ca. −20 °C. The nucleation rate J for water condensation during the stage (i) depends on the substrate wettability49 J = A1F 3/2(1 − cos θ ) exp[−A 2 F 3], 2 − 3 cos θ + cos3 θ 4

(1)

where the constants A1 and A2 are determined by the thermodynamic properties of the vapor and liquid as well as temperature. For small contact angles, the nucleation rate increases when the contact angle is larger, 3 3 8

A1(1 − cos θ )4 . For hydrophobic substrates (θ ≈ A A 90°), the nucleation rate approaches J ≈ 81 exp⎡⎣ − 82 ⎤⎦. Moreover, the energy barrier for heterogeneous nucleation of water condensation is also the function of the substrate wettability J∼

ΔG =

4γπrc 2F 3

(3)

where θ ≈ 130° is the contact angle on the hydrophobic surface and d = 1 μm is the particle diameter. On our Janus surfaces the predicted impalement pressure is approximately pimp = 150 kPa. ii. Spontaneous Coalescence and Fast Spreading of Liquid Water on Hydrophilic Clusters. The volume of the liquid water which can be collected on the initially dry hydrophilic clusters of the Janus particle coatings is limited. The maximum volume per unit of projected area can be roughly estimated as Vmax ≈ phLaveragecluster, since the average height of the liquid water is comparable with the cluster size. If the volume of the condensed water exceeds Vmax, the drops start to quickly coalesce leading to the formation of relatively large liquid areas occupied by several hydrophilic clusters. At the temperature of about −20 °C we observed the nucleation and fast growth of ice crystals,45,49,54−58 the size of which approaches hundreds of micrometers (Figure 5c). At some point, the remaining small water droplets freeze and form small ice crystals (Figure 5d). To better understand the mechanism of the formation of large and small crystals, we have measured the evolution of the area occupied by these crystals (Figure 6). At small times (t < 5 s), the characteristic size of the dendrite (shown in Figure 6c) grows as L ∼ t0.71. Beysens and Knobler54 have observed the average drop radius growth during dew condensation in the coalescence regime as L ∼ t0.75. This result is close to our observations. Therefore, the fast liquid area growth at the initial stage of the freezing process can be most probably explained by the coalescence of the liquid condensed on the clusters. The simulation of the liquid coalescence process on the hydrophilic clusters is shown exemplarily in Figure 7. The sketches of a site and its neighbors for the simple percolation and for the percolation associated with the cluster coalescence are shown in Figure 7a,b, respectively. In Figure 7c four steps of a theoretically predicted growth of an arbitrary primary dendrite are shown. The predicted irregular cluster shape is similar to the observed form of a fast growing ice dendrite shown in Figure 7d. The size of the observed small ice crystals is 10−15 μm, which is of the same order of magnitude as the size of a single primary hydrophilic cluster, Figure 2b. During stage (iii), the liquid in large liquid spots on hydrophilic clusters solidifies and the rate of growth of the obtained ice crystal reduces drastically, in comparison to the rate of the liquid coalescence during the first 5 s. At times t > 5 s (Figure 6c), the typical length of the large dendrite follows the scale L ∼ t0.34 (obtained by the best fitting to the experimental data). This scaling is rather similar to the predicted and observed rate of growth45,54 of a single small dendrite accompanied by the evaporation of the neighboring liquid drops, L ∼ t1/3. Evaporation of the liquid drops in the neighborhood of the growing ice dendrite is caused by the fact that the saturation pressure in a vapor at the ice crystal surface is usually smaller than at the surface of a liquid supercooled drop.55 The evaporation of the small liquid droplets leads to the formation of a thin expanding dry band around a dendrite. This band is seen as a bright region around each dark dendrite in the

Figure 6. Growth of ice crystals on rough P(PEGMA)/P(PDMSMA) Janus particle-based surfaces: (a) large crystals; (b) small crystals; and (c) comparison of the surface area increase under ice crystals marked by dashed circles in (a) and (b).

F=

πγ |cos θ| d

(2)

where γ is the surface tension and rc is the critical radius for stable nucleation. Therefore, in the case of the substrates coated by the Janus particles, liquid water is nucleated first on the hydrophilic clusters, where the nucleation rate is high in comparison to that on the hydrophobic parts of the substrate. This phenomenon has been recently used for the creation of controlled arrays of condensed drops on surfaces with patterned wettability.50 At the initial stage of water condensation the hydrophobic regions remain dry. Also on the 6999

DOI: 10.1021/acs.chemmater.6b02765 Chem. Mater. 2016, 28, 6995−7005

Article

Chemistry of Materials

Figure 7. Simulation of a large hydrophilic cluster (dendrite) growth on a honeycomb lattice. (a) Scheme of the site neighbors for simple percolation, illustrated in Figure 2, and (b) scheme of the site neighbors for the cluster growth due to an increase of a condensed liquid volume. Orange illustrates a hydrophilic site; white sites are neighbors in a cluster. (c) Four first steps of an arbitrary cluster growth. (d) Experimental observation of the initial fast growth of a large irregular liquid cluster on a Janus surface.

Figure 8. Summary of the ice weight (a) and the ice growth rate (b) measurement results on flat surfaces and on surfaces prepared with particles modified by one polymer (either P(PEGMA) or P(PDMSMA)), as well as on surfaces made from bicomponent hybrid Janus particles.

videos shown in Supporting Information: Movie_S1_Janus_acc12x, Movie_S2_Janus_ acc4x. Evaporation rate of a liquid drop of diameter d is proportional to the surface area, ∼d2, and the gradient of the vapor concentration. The square of the drop diameter of an evaporating drop, d2, decreases linearly in time. The vapor concentrations at the ice and liquid water surfaces are determined only by the temperature and the surrounding pressure; therefore, the concentration gradient at given ambient conditions is inversely proportional to the dry band thickness h. The velocity u of the propagation of the outer boundary of the dry band is estimated as the ratio of the typical interdrop distance to the evaporation time ∼ h−1. This scaling in the limit h ≪ L and dL/dt ≪ u yields a rough estimation for the scaling of the band thickness h ∼ t1/2. The least squares fit of the measurements of the dry band thickness from Supporting Information Movie_S1_Janus_acc12x gives the dependence in the form h ∼ t0.59, which is rather close to the predicted scaling. Characterization of the Icing Process. Next, we tested the icing properties of the Janus particle layers and compared them to the icing of the layers formed by fully covered particles modified with either hydrophilic P(PEGMA) or hydrophobic P(PDMSMA), as well as with flat surfaces modified with the same polymers. The ice growth kinetics experiments were carried out by using seven identical samples from each set and freezing them gradually on a cooling plate in a temperature and humidity controlled chamber (Figure 8a). The weight of the surfaces was measured after certain time intervals. The weight difference of the samples before and after frost formation corresponds to the amount of ice formed with time (Figure 8a). Ice thickness was calculated from the ice weight assuming uniform ice coverage. The ice growth rate (Figure 8b) was calculated from the slope of the linear parts of the curves in Figure 8a. It was found that the highest and lowest rates of ice formation (ice weight and ice layer thickness) are intrinsic to the hydrophilic P(PEGMA)- and the hydrophobic P(PDMSMA)-modified surfaces, respectively (Figure 8a,b). This difference can readily be explained by the known fact that polar substances (water) “like” polar P(PEGMA) and

“dislike” nonpolar P(PDMSMA). Furthermore, the Janus particle-modified surfaces occupy an intermediate position between the hydrophobic and the hydrophilic ones, indeed demonstrating an intermediate ice growth rate in accordance with our expectations. Finally, we investigated the adhesion of ice on the layers of Janus particles as well as on the reference ones. Ice adhesion measurements were performed on a homemade centrifuge apparatus mounted in the temperature and humidity controlled chamber. Ice cylinders were placed on the investigated samples and mounted on the centrifuge beams. The samples were spun on the centrifuge to determine the rotational speed at which the detachment of the ice cylinder occurred, subsequently recalculating it into the ice adhesion strength (see Experimental Section for details). It was found that the ice adhesion depends not only on the properties of the polymers but also on the roughness of the surface (flat surface or particle layer, Figure 9). In fact, the flat P(PEGMA)-modified surface demonstrated very low ice adhesion (