Study of Supercooled Water Drop Impact on Icephobic Gradient Polymer Coatings

Abstract

Supercooled liquid water drops, with temperatures below freezing point, are common in high-altitude clouds. These drops, despite being in a metastable state, can remain liquid for extended periods if temperatures are above the homogeneous nucleation point. Impact of such liquid drops with a cold solid surface is one of the reasons for ice accretion, which in many cases can represent a safety hazard. The study of supercooled drop impact dynamics is key to developing materials that provide resistance against the formation and accumulation of ice. In this work, the impact of supercooled water drops on dry icephobic coatings based on gradient polymers deposited via initiated chemical vapor deposition (iCVD) under several conditions is analyzed. Experimental results show that coated surfaces potentially decrease the freezing probability upon impact. The gradient polymer surfaces with higher roughness and lower wettability do not increase the freezing probability upon impact but result in rebound and eventual roll off the surface, indicating that surface hydrophobic properties prevailed over the impact. The findings demonstrate the remarkable efficacy of gradient polymer coatings in inhibiting drop freezing, even under high wind velocities, and provide insights for the design of durable and effective anti-icing coatings across diverse applications.

1 Introduction

Drop impact onto solid surfaces is a fundamental multidisciplinary phenomenon intensively studied in the framework of various research fields, including fluid mechanics, heat transfer and thermodynamics, materials, and surface science.^{[1, 2]} Understanding how drops interact with surfaces, particularly under extreme conditions such as supercooling, is highly relevant for numerous applications, including ice prevention, anti-icing strategies, and the design of advanced coating materials.^[3-5] Supercooled drops are characterized by temperatures below their freezing points in a metastable liquid state. They represent a significant challenge in various practical scenarios, namely, transportation, aviation, and energy infrastructure.^{[6, 7]} Supercooled drops are prevalent in nature, particularly in high-altitude clouds where temperatures can drop below the freezing point without nucleating into ice crystals.^[8] The impact of a liquid onto a solid surface leads to the formation of a radial flow in a thinning lamella, bounded by a rim, formed by surface tension, viscous stresses in the lamella, and the capillary forces associated with the substrate wettability.^[1] In the case of a supercooled drop impact, the drop spreading is accompanied by a heterogeneous nucleation after some time interval. The nucleation leads to the expansion of a thin ice layer and the formation of a cloud of dendrites which expands through the supercooled liquid with a certain velocity.^[9] As a result, the drop dendrite freezing yields the formation of a mushy media represented by a mixture of solid ice dendrites surrounded by liquid water at temperatures close to the freezing point. The freezing delay is not a repeatable quantity since the nucleation process is a stochastic event. The ice accumulation not only compromises the performance and efficiency of engineered systems but could also mean safety hazards in critical environments. Therefore, developing strategies to mitigate or prevent the freezing of supercooled drops upon impact represents a crucial technological necessity.^[10]

Gradient polymer coatings, i.e., coating whose chemical composition changes gradually within their thickness, have emerged as promising anti-icing materials for exhibiting extraordinary icephobic properties. Nanometric gradient polymers coatings consisting of 1,3,5,7-tetravinyl-1,3,5,7-tetramethylcyclotetrasiloxane (V₄D₄) and 1H,1H,2H,2H-perfluorodecyl acrylate (PFDA) have been deposited over several surfaces via initiated chemical vapor deposition (iCVD).^[11] These coatings successfully hindered ice nucleation in the condensed microdroplets, delayed water freezing on the surface, and drastically reduced ice adhesion.^[12] Furthermore, the coatings showed outstanding compatibility with de-icing systems when they were coupled with a resonant electromechanical de-icing system conformed of piezoelectric actuators. The nanometric nature of the coatings played a key role in the ice-detachment process since they did not contribute to any damping in the system, resulting in an efficient hybrid ice mitigation system.^[13] The icephobic properties in the coatings arise from a discontinuity in the surface energy at the surface plane due to the molecular arrangement of the PFDA functional groups. Unlike conventional homogeneous coatings, gradient polymer coatings exhibit extraordinary durability and adhesive properties. Gradient polymer coatings offer a versatile approach for controlling the dynamics of drop impact and minimizing ice formation under supercooled conditions.

In this study, a comprehensive investigation was conducted on the interaction of supercooled drops upon impact on gradient polymer coatings. Unlike the common static testing of the icephobic properties, these experiments capture the dynamic nature of outdoor environments, where several parameters influence ice formation. Through these experiments, the icephobic properties of developed materials can be validated, accounting for the dynamic interplay between supercooled drops and surfaces, ensuring the effectiveness of ice prevention measures in practical applications. In this study, we discuss the dynamics of supercooled drop impact onto gradient polymer coatings and the mechanisms governing the prevention of freezing upon impact. By correlating the surface wettability, the roughness, and the architecture of the polymeric structure with the freezing probability, we gain insights into the properties governing the ice-repellant performance exhibited by the gradient polymer surfaces. Our findings not only advance the fundamental understanding of the interfacial phenomena of icephobic surfaces but also offer practical insights for the design and optimization of ice-repellent surfaces for various engineering applications.

2 Experimental Section

2.1 Materials and Deposition Process

Gradient polymers were synthesized and deposited using a custom-made iCVD reactor.^[14] Di-tert-butyl peroxide (TBPO) was purchased from Sigma-Aldrich and used as a reaction initiator. 2,4,6,8-Tetraethenyl-2,4,6,8-tetramethylcyclotetrasiloxane (V₄D₄) and 1H,1H,2H,2H-Perfluorodecyl acrylate (PFDA) were both purchased from Sigma-Aldrich and used as monomers. The coatings were formed with a reactor pressure of 500 mTorr, filament temperature of ≈200 °C, and a substrate temperature of 40 °C. The deposition thickness was followed in situ using a He-Ne laser using the silicon substrates as a reference. The formation of gradient polymer was achieved by introducing the V₄D₄ monomer to form the bottom section. Then the monomer PFDA was gradually introduced, forming the copolymer p(V₄D₄-co-PFDA). To form a PFDA-rich section at the top, the V₄D₄ monomer was gradually stopped. The gradient polymer structure consisted of a thickness of 50 nm for the pV₄D₄ bottom, 150 nm for the p(V₄D₄-co-PFDA), and 100, 200, and 300 nm for the top pPFDA sections. The coatings were deposited on silicon substrates.

2.2 Supercooled Drop Impact Experiments

Supercooled drop impact experiments were carried out on a custom-built setup at the Technical University of Darmstadt, Figure 1. The system for controlled supercooled drop impact experiments consists of a vertical open-return wind tunnel design to achieve a uniform velocity profile at the test section, with airflow from 0 to 40 m s⁻¹ with a turbulence intensity of 0.5%. The test section in the tunnel has a square cross-section with a side length of 14 cm. A drop generator creates drops with a diameter of D₀ = 2.6 ± 0.1 mm. The wind tunnel is located inside a cold chamber that maintains the airflow, the drops, and the substrate at a constant temperature of −10 °C.

Figure 1

The wind tunnel setup for investigating the impact of supercooled drops.

When the sample was positioned in the sample holder, a supercooled drop was generated and suspended in the generator needle above the wind tunnel. After detaching the needle, the drop was initially accelerated by gravity within a shroud pipe. Upon exiting the shroud pipe, it was further accelerated by both gravity and airflow until it impacted the substrates. The impact was recorded using a high-speed camera (Phantom SA-X2) at an experimental recording rate of 3000 frames per second (fps), resulting in a total recording time of ≈50 s. Consequently, the freezing results after the impact presented below have been limited to ≈50 s. The surface of the sample before impact was free of condensation frosting, ensuring that ice nucleation was not induced by frost on the surfaces. The impact was analyzed under three conditions: without airflows (drop impact velocity is U₀ = 4.1 m s⁻¹) and with airflow speeds of 10 m s⁻¹ (U₀ = 4.8 m s⁻¹) and 20 m s⁻¹ (U₀ = 6.5 m s⁻¹), which further accelerated the drop before impact. 30 impact experiments for each sample at each airflow condition were performed in two sample replicas. The calculation of the freezing probability takes into account the drops that ended up freezing regardless of the freezing mechanism, this was directly extracted from the freezing events that occurred during the 30 experiments.

3 Results and Discussion

The impact and freezing of supercooled water drops were studied on gradient polymer coatings deposited on silicon via iCVD. The gradient polymer structure features a 50 nm pV₄D₄-rich bottom section, a 150 nm middle section where the bottom layer transitions to a p(V₄D₄-co-PFDA) copolymer, and a pPFDA-rich top section with varying thicknesses of 100, 200, and 300 nm (labeled Grad100, Grad200, and Grad300, respectively). The gradient polymer structure is schematically represented in Figure 2a. The experiments on gradient polymer coatings were compared with silicon, and the homopolymers pV₄D₄, and pPFDA.

Figure 2

a) Schematic representation of the gradient polymer structure divided by sections, the red hexagons indicate the V₄D₄ groups whereas the blue rectangles indicate the PFDA fluorinated pendant groups. The pV₄D₄-rich section consists of a densely crosslinked network, as PFDA monomers are introduced into the structure the copolymer p(V₄D₄-co-PFDA) is gradually formed, as the feed of V₄D₄ monomers stops, PFDA monomers align forming a lamellar structure. b) Schematic representation of the different PFDA crystal orientations interactions with water molecules. The different domain interactions produce a surface energy discontinuity along the surface plane.

The growth of PFDA at the top section of the gradient polymer intrinsically orders the fluorinated pendant groups face-to-face in a lamellar structure, originating crystallinity. As previously reported,^[12] the crystallites' orientation is directly related to the coating's icephobic characteristics, where a predominantly random orientation (a mixture of in-plane, out-of-plane, and configurations in between) enhances the icephobicity. Different crystallographic orientations result in different interactions with water molecules, creating thus a surface energy discontinuity at the surface plane. This is schematically represented in Figure 2b. It is hypothesized that the discontinuity disturbs the arrangement of water molecules both before and during the freezing process. It was also found that the thickness of the top section influenced the crystallinity, i.e., increasing the thickness of the PFDA top section promoted the development toward a random orientation of the crystallites.

Drop spreading is characterized by a flow in a thin lamella. The spreading diameter is determined by the dynamics of the rim governed by capillary and viscous forces, as well as the inertia of the liquid in the thin lamella. If the substrate is hydrophobic, the drop spreading is followed by the rim receding. On hydrophilic substrates, the rim is pinned at the position close to the maximum spreading point. In supercooled liquids, the formation and dissolution of ice embryos is a continuous process. Occasionally, an ice embryo will grow large enough to exceed a critical radius. This critical size is associated with the onset of heterogeneous nucleation, where the ice embryo becomes stable and continues to grow. Once an ice embryo surpasses this critical size, it triggers the nucleation process.^{[15, 16]} A thin ice layer front begins to form, which marks the beginning of the solidification of the liquid. As the solidification progresses, dendritic structures start to develop. After the dendrites propagation, the drop is a solid-liquid mixture. If the surface remains cold, the liquid between the mushy layers will continue to solidify. This slower process is known as secondary solidification.^[6] Sequential snapshots shown in Figure 3, show the freezing process of a supercooled drop after the impact onto silicon (static water contact angle is 42° ± 1°, and contact water hysteresis is 20 ± 2°). On the silicon substrate, a clear solidification process upon the impact of the drop could be observed. Similar to the mechanisms observed on hydrophilic surfaces by Wong et al.,^[17] on silicon, the drop spread to a maximum diameter after the impact, and it barely receded before it started freezing, resulting in a frozen diameter near the maximum after impact. In Figure 3, the thin ice freezing front propagation started from a nucleation point after the impact at 261.00 ms. It took ≈44.33 ms for the freezing front to propagate over the drop area. From the same point, dendrites grew and propagated in the same direction.^[6] We found that by increasing the impact velocity with the higher airflow velocity, the nucleation rate rose on hydrophilic surfaces.

Figure 3

Sequential snapshots after the drop impact onto silicon without airflow. a) At t = 0 ms, the impact moment. b) At t = 261.00 ms, the nucleus (indicated by the arrow) from where the thin ice freezing front propagates (yellow dashed line) appears. c) At t = 288.67 ms, dendrites grew and propagated (red dashed line) from the same nucleation point. The propagation of the thin ice freezing front (yellow dashed line) occurred faster than the dendrite's growth. d) At t = 305.33 ms, dendrites propagation in the drop. The full video is available in the Video S1 (Supporting Information).

Figure 4 shows the most representative impact scenarios upon the gradient coatings at different wind velocities. All impacts resulted in a drop rebound followed by a secondary impact. In the sequential impact snapshots without airflow on the Grad100 surface, the breaking up of the drop into secondary drops of similar sizes that were pinned to the surface can be observed. After the second impact, the main and secondary drops stayed on the surface as a sessile drop. On Grad200 and Grad300 surfaces, noticeable differences were seen in the secondary drops after the first impact which were smaller, more scattered drops that remained sessile. The presence of gradient nanometric coatings on silicon influenced the impact dynamics, mainly due to the wettability and morphological differences compared with silicon. Gradient polymer coatings possess an apparent hydrophobic behavior displaying simultaneously a strong pinning effect,^[12] nevertheless, they can be described as surfaces with low surface energy. The hydrophobic behavior originates from the chemical composition of the top section of the gradient structure, composed of highly fluorinated groups exposed to the surface. Whereas the pinning effect from the morphological features originated from the deposition process.

Figure 4

Snapshots of the supercooled drop impacts on a) Grad100, b) Grad200, and c) Grad300 without external airflow. In all the gradient polymer surfaces the impact was followed by a rebound and the drop broke into smaller pieces. After the rebound impact, the drop remained sessile in the surface without freezing. The strong pinning effect on the Grad300 surface was observed by the number of small droplets that remained static on the surface. The full videos are available in the Video S2, S3 and S4 (Supporting Information).

By increasing the airflow velocity, the impact velocity of the drop can be increased. When the drops were accelerated by the higher airflow velocity (see Figure 5), the kinetic energy contained in the drop produced larger lamella spreading areas on all the surfaces. On gradient polymer surfaces, the drop broke the drop into tiny droplets that stayed pinned after the first impact. After the secondary impact, the main drops were blown off the surface by the airflow, leaving the tiny secondary drops pinned to the surface. In the Grad200 and Grad300, the strong pinning effect was evident through the number of secondary drops that remained sessile after the drop breakage in the first impact. Higher velocities, however, led to a smaller fraction of the remaining water (see Figure 5a–c). Things are different for silicon surfaces. At an airflow velocity of 20 m s⁻¹, all the drops froze during the receding process. Compared to freezing without airflow, the frozen ice shows a distinct central thin ice region surrounded by a rim filled with dendrites. This occurs because of the rapid solidification of the lamella. The thin ice layer propagates faster than the dendrites (also can be seen in Figure 3). If freezing occurs before the lamella recedes completely, the warm region near the thin ice layer expands over the lamella before the dendrites form.

Figure 5

Snapshots of the supercooled drop impacts on a) Grad100, b) Grad200, c) Grad300, and d) silicon (the nucleus is indicated by the arrow, thin ice freezing front is indicated by the yellow dashed line and dendrites are indicated by the red dashed line) with an external airflow of 20m/s. After the rebound impact, the drops rolled off the surfaces leaving tiny droplets pinned on the gradient polymer surfaces. While all the drops froze while receding on the silicon surface. The full videos are available in the Video S5, S6, S7 and S8 (Supporting Information).

Figure 6 summarizes the freezing probability of supercooled drops upon the impact with gradient polymer coatings, and compares them with silicon and pV₄D₄ surfaces. It has been shown^[18] that the nucleation rate was enhanced when the roughness and impact velocity increased on epoxy resins and PTFE surfaces. In the study of the impact of supercooled drops impact onto aluminum surfaces,^[19] the rise of nucleation rate on the rough substrates is explained by the formation of micro-size bubbles. However, in gradient polymer surfaces, increments in roughness, WCA hysteresis, and impact velocity did not result in an increment of the nucleation probability. We hypothesized that since the coating has a partially elastic nature, its deformation can prevent the formation of micro-bubbles at the surface. Thus, the gradient surfaces significantly reduced in some cases by at least 70% of the freezing probability upon impact. When the drop impacted without airflow, the freezing probability systematically decreased as the thickness of the top section of the gradient polymer increased, following 13%, 3.3%, and 0.5% for Grad-100, -200, and -300, respectively. In other words, the freezing probability decreased as the roughness in the samples increased (Figure 6) and the mobility of the drops decreased. This revealed that the variables that usually promote ice nucleation upon the impact, such as roughness and surface wettability, in some cases do not affect or may even aid the icephobic properties of the gradient polymer coating. The combination of the nanometric surface and the surface chemistry might have reduced the formation of the nuclei. The systematic correlation of higher top-section thickness with stronger icephobic ability was consistent with the icephobic characterization previously reported.^[12]

Figure 6

a) Comparison of the probability of freezing upon impact of the gradient polymers with other surfaces at different wind velocities. b) Atomic force micrograph of the gradient polymer surfaces g100 20.5+4 g200 42.7+6 g300 53.4+9. The roughness increases proportionally to the increment in thickness of the gradient structure top-section. Freezing mechanisms observed in the gradient polymer surfaces, c) first case, d) second case, e) third case, and f) fourth case. The full videos are available in the Video S9, S10, S11 and S12 (Supporting Information).

Without airflow, the gradient polymer coatings reduced the freezing probability by at least 25% compared to silicon, interestingly, pV₄D₄ exhibited a freezing probability similar to Grad100. The hydrophobic nature of the pV₄D₄ prevented the ice nucleation under these conditions. The high roughness of Grad100, compared with pV4D4, could have favored the freezing in the Grad100 surface. However, when the drops were accelerated with the airflow, the scenario drastically changed. For silicon and pV₄D₄ surfaces, the freezing probability substantially increased with almost every experiment leading to freezing, these results were coherent with the literature. However, gradient polymer surfaces showed no straight correlation between the freezing probability and impact velocity. In Grad100 the freezing probability achieved a maximum freezing probability of only 23% with the fastest drop impact velocity, which represents at least 70% less than for Silicon and pV₄D₄. Grad200 and Grad300 showed fluctuation of values with the acceleration, therefore, no trend in freezing probability was distinguished. However, it can be seen that the freezing probability on gradient polymer coatings is much lower than on silicon and pV₄D₄. At a wind velocity of 20 m s⁻¹, the freezing probability was 3.3% and 6.7%, for Grad200 and Grad300 respectively. This represents a reduction of 96.7% and 93% compared to silicon and 91.7% and 88.3% compared with pV₄D₄. Regardless of the lack of correlation, the effect of the coatings on preventing freezing upon impact was evident. A complex contribution between the hydrophobicity, the roughness, and the coating surface energy discontinuity resulted in an effective mechanism to inhibit supercooled drops from freezing upon the impact.

It is worth analyzing the behavior of pV₄D₄ through the experiments and comparing it with Grad100: without the airflow, pV₄D₄ performed better than Grad100 but the performance drastically decayed as the speed of the airflow increased. This exemplifies how hydrophobic coatings might display apparent icephobicity only at specific conditions but beyond the material fails to mitigate ice nucleation. As previously demonstrated, water repellency does not necessarily translate to icephobicity.^{[20, 21]}

Four different freezing mechanisms were identified in the rare event of freezing in the gradient polymer surfaces, shown in Figure 6. In the first case, after the first impacted and rebounded, the drop landed and stayed pinned to the surface preserving its spherical shape, and eventually froze. In the second case, part of the drop (see the region within the red dashed line) froze in the air. The freezing continued with the connected drop and then impacted as a fully frozen drop. In the third case, which is similar to the second one, part of the bouncing drop froze not in the air but upon contact with the surface. This initial freezing spread to the upper part, inhibiting the bounce and causing the drop to be “dragged” back to the surface, where it then froze completely. The fourth mechanism happened when a tiny droplet produced after the impact froze and then when in contact with the main drop during the second impact, triggered the main drop to freeze.

It is important to note that the four different freezing mechanisms occurred randomly on any of the surfaces. The cases shown in Figure 6 are merely representative and can happen at any time on any surface. However, we found that the first and fourth cases were the most common. The first case occurred particularly at low airflow velocities. After the secondary impact, the drop stayed on the surface, and after a delay, the drop froze. This phenomenon is typical on hydrophobic surfaces due to their tendency to delay freezing. In the fourth case, due to the pinning effect of the gradient polymer surfaces, a secondary tiny droplet is often left after the rebound. Due to its small size, the tiny droplets freezes more easily. When the larger rebounding drop contacts the small frozen droplet, the freezing can propagate to the larger drop.

As previously reported,^[11] the main disadvantage of fluorinated coatings is the instability and weak adhesion to substrates leading to fast degradation or delamination. The tests in this study served as an indirect measure of durability against water impact and erosion, since after at least 90 experiments the coatings showed no visual delamination or erosion damage, or any decrease in functionality, this, based on the full rebound observed consistently through all the experiments.

4 Conclusion

An experimental investigation in a cold wind tunnel was conducted to examine the impact of supercooled water drops on gradient polymer coating substrates at different wind speeds. The icephobicity of gradient polymer coatings was characterized. Without airflow, the freezing probability systematically decreased as the thickness of the top section increased on gradient polymer coating substrates. As the drop accelerated by the airflow, the freezing probability barely reached 14% on average, demonstrating that gradient polymer coatings reduced the freezing probability, even with the high impact velocities.

The roughness and wettability of the gradient polymer surfaces were not observed to promote freezing upon impact, as reported in the literature. When considering the dynamics of supercooled drop impact, especially at high impact velocities, the drop will come into contact with the surface with immense impact force. At this point, the microstructure, roughness, wettability, and even more properties of the surface need to be comprehensively considered and studied to determine their effects on hydrophobicity and icephobicity.

Acknowledgements

G.H.R. and M.D. contributed equality to this work and shared the first authorship. This project has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 956703 (SURFICE Smart surface design for efficient ice protection and control).

Conflict of Interest

The authors declare no conflict of interest.

Open Research

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.