Elevating thermal-deicing efficiency via regulating interfacial-ice induced by microstructure under flow condition Authors: Jiawei Jiang, Yizhou Shen, Yangjiangshan Xu, Zhen Wang, Senyun Liu, Lingfeng Zhao, Ying Pan, Jie Tao, Zhong Chen Published: 2026 ABSTRACT Regulating ice type to optimize the failure behavior of the solid-ice interface during the melting process presents a viable approach for enhancing thermal de-icing efficiency, nevertheless, the prevailing notion that ice type is exclusively determined by flow conditions restricts the advancement of solid-ice interface control technologies. Here, we propose a strategy that leverages non-wetting microstructures to induce spontaneous formation of near-wall rime (interfacial-rime) under flow conditions. Thicker interfacial-rime can significantly enhance the de-icing efficiency by modifying conventional solid-liquid-ice interactions into direct solid-ice contact during the melting process. Interestingly, pre-frozen droplets adhering to defined mathematical criteria are fragmented by microstructures with tailored dimensions with the assistance of flow field, significantly reducing the initial ice crystal size. The blocking effect induced by ice crystals with smaller initial size effectively elevates the thickness of interfacial-rime layer, significantly reduces the energy consumption of thermal deicing by roughly 56% without introducing additional energy consumption. Introduction Currently, icing on critical aircraft components frequently results in loss of lift and increased drag, directly impacting aircraft stability and posing significant risks to flight safety1–3. To enhance aircraft safety under icing conditions encountered at high altitudes, several active de-icing strategies have been investigated, including mechanical4,5, electrical pulse6,7, electrothermal8–10, and ultrasonic methods11,12. Therein, electrothermal de-icing technology has emerged as the most widely used solution in aviation due to its minimal aerodynamic disruption, limited surface damage, and effectiveness in completely disrupting the solid-ice interface13–15. Unfortunately, the additional weight and energy consumption brought by the electrothermal de-icing system significantly limit the fuel efficiency and aircraft range16,17. Hence, various strategies have been proposed to further reduce energy consumption, such as phase change material (PCM) assisted energy storage18,19, photothermal-electrothermal material composites20–22, and superhydrophobic surface modification23,24. Therein, superhydrophobic materials that preserve air-pockets within microstructures can effectively minimize solid-ice contact area in icy environments by facilitating droplet rolling or bouncing off the surface, becoming an optimal strategy for enhancing electrothermal deicing efficiency25–29. Recent researches highlight the significant impact of non-wettability on de-icing efficiency in a static environment. It was confirmed that superhydrophobic graphene materials demonstrated an 86.1% reduction in ice melting and removal time compared to hydrophilic graphene counterparts30. Similarly, an ice-melting experiment on the polytetrafluoroethylene (PTFE)@fluorinated epoxy resin surface confirmed that improved non-wettability effectively reduced the de-icing time by ~35 s21. However, several studies have indicated that improved non-wettability does not necessarily correlate with enhanced thermal de-icing efficiency. For instance, complete melting of a 6 μL frozen droplet on a hydrophilic silicon wafer required only 135 s, whereas a hydrophobic silicon wafer took 205 s31. It was primarily due to the fact that surfaces with higher wettability facilitate greater solid-liquid contact, significantly enhancing the heat flux32. This controversy also extends to the inflow environment. The thermal de-icing time for hydrophobic surfaces was 50% longer than that for superhydrophobic surfaces at a flow velocity of 10.5 m s−1 and a heating power of 27.7 W33, while it was reduced by 16.8% to 20.5% with a flow velocity of 50 m s−1 and a heating power ranging from 40 to 60 W34. Thus, the mode of ice shedding dominated by the interaction at the solid-ice interface during melting critically influenced de-icing efficiency and energy consumption. Particularly, the rime with porous structure and fewer solid-ice interface can directly detach from the superhydrophobic surface during melting, resulting in a 27.2% reduction in the melting time35. Precisely controlling the formation of fewer solid-ice interface to optimize its interaction during melting presented a viable approach to further enhance electrothermal de-icing efficiency on superhydrophobic surfaces. Previous research indicated that water was preferentially condensed onto hydrophilic regions and nucleated into dendritic ice through the modulation of surface wettability in a static environment, and then forming a porous ice layer with fewer solid-ice interfaces with the assistance of solidification latent heat released by neighboring droplets36. However, the condensation effect induced by the solidification latent heat was always neglected in flow environments, and the characteristics of ice structures were universally considered to ultimately determined by environmental factors (wind velocity, temperature, and liquid water content)37–40. Given the substantial influence of surface non-wettability and microstructure on droplet freezing behavior41–43, an important unresolved issue was whether inherent surface properties could modulate ice growth to control solid-ice interface interactions effectively under flow conditions. In this work, stable hydrophobic/superhydrophobic surfaces with varying non-wetting characteristics and structural scales are fabricated using a straightforward spraying technique of PTFE nanoparticles. Subsequently, the freezing dynamics and melting behavior of microdroplets show that contrary to conventional understanding, the microstructure on superhydrophobic surface not only modulates ice structure but also expands both the thickness and generation environmental range of interfacial rime. Notably, the interfacial-rime induced by the microstructure effectively transforms conventional solid-liquid-ice contacts into direct solid-ice interactions during melting, substantially lowering energy consumption during electrothermal de-icing without requiring additional energy consumption under broader icing conditions. Results PTFE surfaces preparation with various non-wettability To elucidate the influence of microstructure on the freezing behavior of microdroplets, a PTFE material with lower surface energy and higher stability is employed to construct hydrophobic surfaces via a multi-step spraying process. By varying the PTFE concentration during the secondary spraying, five surfaces with distinct microstructures are prepared and denoted as PTFE-0, PTFE-3, PTFE-5, PTFE-10, and PTFE-15. Scanning electron microscopy (SEM) reveals that the PTFE-0 surface exhibits a dense distribution of micro-nano strip-like features ~1.4 μm in length and 350 nm in width (Fig. 1a). Following secondary spraying, petal-like structures with the size of about 15 μm appear on the PTFE-3 surface (as shown in the lower left corner), which remains morphologically consistent across increasing concentrations, with structure sizes expanding to 22 μm in PTFE-5, 32 μm in PTFE-10, and 70 μm in PTFE-15.Fig. 1Morphology and composition analysis of different PTFE surfaces.a SEM morphology of 5 different PTFE surfaces. Each sample is independently observed three times to obtain representative morphologies. b three-dimensional contour images of 5 different PTFE surfaces. c XPS analysis of typical PTFE-0 surface (C1s and O1s spectra are shown by blue and orange arrows respectively). d EDS image of typical PTFE-0 surface. e FTIR curve of typical PTFE-0 surface (the illustration shows the schematic diagram of PTFE molecule structure). f Contact angle and sliding angle curves of different PTFE surfaces (droplets are 4 μL for each sample, the average value is taken after 5 tests and data are presented as mean values +/- standard deviation (SD)). Source data are provided as a Source Data file. Morphology and composition analysis of different PTFE surfaces. a SEM morphology of 5 different PTFE surfaces. Each sample is independently observed three times to obtain representative morphologies. b three-dimensional contour images of 5 different PTFE surfaces. c XPS analysis of typical PTFE-0 surface (C1s and O1s spectra are shown by blue and orange arrows respectively). d EDS image of typical PTFE-0 surface. e FTIR curve of typical PTFE-0 surface (the illustration shows the schematic diagram of PTFE molecule structure). f Contact angle and sliding angle curves of different PTFE surfaces (droplets are 4 μL for each sample, the average value is taken after 5 tests and data are presented as mean values +/- standard deviation (SD)). Source data are provided as a Source Data file. Three-dimensional contour confirms that the PTFE-0 surface exhibits an average roughness (Rq) of 0.1778 μm, characterizing it as nearly smooth (Fig. 1b). With the superposition of PTFE particles, the surface roughness increases from 0.9169 μm (PTFE-3) to 5.8426 μm (PTFE-15), and the maximum structure height concurrently rises from 29 μm to 92 μm, indicating a direct relationship between PTFE concentration and topographical development. Considering that the surface preparation process involves the use of non-ionic surfactants and high-temperature treatment, X-ray photoelectron spectroscopy (XPS) analysis is performed on the PTFE-0 surface to confirm the surface composition (Fig. 1c). The full spectrum contained only peaks corresponding to C and F without extraneous elemental signals. Peaks at 284.88 eV and 291.38 eV in the C1s spectrum corresponded to C-C and C-CF bonds, while peaks at 686.48 eV and 688.68 eV in the F1s spectrum are attributed to C-CF and CF3 bonds, respectively44. The subsequent Energy-Dispersive X-ray spectroscopy (EDS) confirms a uniform distribution of PTFE on the surface with a carbon-to-fluorine atomic ratio of 1:1.78, closely aligning with the theoretical ratio of (C2F4)n (Fig. 1d). Additionally, Fourier Transform Infrared (FTIR) spectroscopy further verifies that the asymmetric and symmetric stretching vibration peaks at 1214.04 cm−1 and 1156.58 cm_1 are F-C-F, and the bending and deformation vibration peaks at 633.96 cm−1 and 505.97 cm−1 are C-F, without evidence of contaminant species (Fig. 1e)45. Moreover, the wettability measurements demonstrate that PTFE-0 exhibits hydrophobic properties with a contact angle of 126.21° and a sliding angle of 22.4°. As surface roughness increases, the PTFE surface gradually changes from a hydrophobic state to a superhydrophobic state. PTFE-3 displays a contact angle of 155.63° and a sliding angle of 9.2°, while PTFE-15 achieves a contact angle of 166.45° and a sliding angle of 5.1°, confirming the correlation between surface microstructure and enhanced water-repellent behavior (Fig. 1f). Static icing behavior of microdroplets on PTFE surfaces Comparative experiments are conducted on hydrophobic (PTFE-0) and superhydrophobic (PTFE-10) surfaces to investigate the influence of surface properties on freezing behavior of microdroplets in a static environment. At −10 °C, dispersed microdroplets aggregate into larger droplets on PTFE-0 due to their non-wetting properties (see purple boxes), and then freeze into independent ice crystals after 39.67 s (see red boxes). As the freezing process progresses, continuous ice layer formation is achieved by 123.12 s (Fig. 2a). This process is influenced by the limited thermal conductivity of the PTFE substrate (only 9.45 W m-1·K−1) and the accumulation of macroscopic droplets. In contrast, the microstructure on PTFE-10 is more likely to contact with microdroplets, which can be served as preferential nucleation sites. Subsequently, the ice layer exhibits high porosity and a fragmented morphology due to the incomplete filling of microdroplets into ice voids. The freezing process at −5 °C is similar to that at −10 °C [Supplementary Fig. 1].Fig. 2Static freezing behavior of microdroplets on different hydrophobic surfaces.a The freezing evolution process on PTFE-0 surface and PTFE-10 surface at −10 °C. The droplet accumulation, ice nucleation and ice crystal growth are respectively indicated by the purple, red and blue boxes. b The freezing evolution process on PTFE-0 surface and PTFE-10 surface at −20 °C. The droplet accumulation, ice nucleation and ice crystal growth are respectively indicated by the purple, red and blue boxes. c The typical microscopic ice crystal morphology of ice layer on PTFE-0 surface and PTFE-10 surface after 300 s at −20 °C. d The initial formation time of ice nuclei on PTFE-0 surface and PTFE-10 surface under different temperatures. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. e The expansion time of ice layer on PTFE-0 surface and PTFE-10 surfaces under different temperatures. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. f The static ice nucleation mechanism on the PTFE-0 surface and PTFE-10 surface. Source data are provided as a Source Data file. Static freezing behavior of microdroplets on different hydrophobic surfaces. a The freezing evolution process on PTFE-0 surface and PTFE-10 surface at −10 °C. The droplet accumulation, ice nucleation and ice crystal growth are respectively indicated by the purple, red and blue boxes. b The freezing evolution process on PTFE-0 surface and PTFE-10 surface at −20 °C. The droplet accumulation, ice nucleation and ice crystal growth are respectively indicated by the purple, red and blue boxes. c The typical microscopic ice crystal morphology of ice layer on PTFE-0 surface and PTFE-10 surface after 300 s at −20 °C. d The initial formation time of ice nuclei on PTFE-0 surface and PTFE-10 surface under different temperatures. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. e The expansion time of ice layer on PTFE-0 surface and PTFE-10 surfaces under different temperatures. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. f The static ice nucleation mechanism on the PTFE-0 surface and PTFE-10 surface. Source data are provided as a Source Data file. At −20 °C, the increased supercooling enhances nucleation efficiency, leading to a growth advantage of the initial ice crystal. The initial freezing is observed at 48.42 s on PTFE-0 with smaller droplet sizes (Fig. 2b). However, although nucleation identically occurs within superhydrophobic microstructures, the initial crystal blocks further droplet infiltration into ice gaps, significantly increasing the porosity of ice layer. Freezing behavior at −15 °C is highly consistent with that at −20 °C [Supplementary Fig. 2]. Further analysis of ice crystal evolution reveals that converged droplets atop initial crystals tend to flow downward and freeze near the PTFE-0 substrate, leading to the interconnection of adjacent ice nuclei to form a ridge-like structure (Fig. 2c). Subsequently, the effect of temperature on nucleation time is quantified (Fig. 2d). Although nucleation times descend for both surfaces as temperature declines from −5 °C to −20 °C due to the increased driving force of water phase change, the time required on PTFE-10 remains 1.74 times longer than that on PTFE-0. This phenomenon is attributed to the thermal insulation induced by entrapped air within the microstructural surface, which also promotes droplet coalescence and postpones phase change at low-temperature. Statistical analysis indicates that the expansion time of ice layer is rapidly shortening on PTFE-0 with decreasing temperature, reaching 42.33 s at −20 °C (Fig. 2e). In contrast, ice expansion on PTFE-10 remains slow and relatively insensitive to temperature, with complete coverage achieving only after 218.86 s at −20 °C, which is owing to the obstruction effect of the microstructure on droplet merging. This perspective is further supported by statistical data on the ice layer porosity of PTFE-0 and PTFE-10 surfaces under varying temperatures and times, as illustrated in Supplementary Figs. 3 and 4. The corresponding mechanism of ice formation is descripted in Fig. 2f. Icing characteristic of microdroplets under flow condition In realistic flight scenarios, surface icing primarily results from the high-speed impact of microdroplets during flight. Unlike the static environment, dynamic airflow parameters such as wind velocity and liquid water content (LWC) exert significant influence on icing behavior. Considering that the LWC at cruising altitudes typically ~0.5 g cm−3 (with droplet diameters around 20 μm), the dynamic icing response of microdroplets on hydrophobic surfaces is investigated under variable velocities and temperatures. The macroscopic morphology of the ice layers on PTFE-0 surfaces reveals that as the temperature decreases from −5 °C to −10 °C, the ice on the wind-facing surface evolves from transparent and dense to opaque and scaly (Fig. 3a). When the temperature decreases to −20 °C, the scaly morphology remains unchanged while the size gradually decreases and the quantity increases sharply. it can be concluded that the ice formed at −5 °C is clear ice, the rime is predominantly generated at −20 °C, while a mixed ice composed of clear ice and rime appears at −10 °C and −15 °C, which is also verified by the microscopic morphologies and density analysis of ice [Supplementary Figs. 5, 6, and 7].Fig. 3Ice formation behavior of microdroplets on hydrophobic surfaces under flow condition.a Evolution process of ice structures on PTFE-0 surface under different temperature and flow velocities. b Typical cross-sectional morphology of clear ice, rime and mixed ice under an icing time of 60 s. Each ice condition is independently repeated and verified three times to obtain representative ice morphologies. c Statistics of ice types on PTFE-0 surface under different flow environment parameters. The clear ice, mixed-ice and rime are marked in yellow, green and blue areas respectively. d High-speed images of the typical ice formation process of mixed ice on PTFE-0 surface under −20 °C & 40 m s−1, where the yellow box indicates the typical morphology evolution of droplets. e The influence of different temperatures on the start freezing time and complete freezing time on PTFE-0 surface under 40 m s-1. f The influence of different flow velocities on the start freezing time and complete freezing time on PTFE-0 surface at −20 °C. g The influence of flow velocity on the droplet freezing size at the solid-ice interface under −20 °C, and the influence of different temperatures on the droplet freezing size under 40 m s−1. The repetitive experiments are conducted for 10 times and data are presented as mean values +/- SD. Source data are provided as a Source Data file. Ice formation behavior of microdroplets on hydrophobic surfaces under flow condition. a Evolution process of ice structures on PTFE-0 surface under different temperature and flow velocities. b Typical cross-sectional morphology of clear ice, rime and mixed ice under an icing time of 60 s. Each ice condition is independently repeated and verified three times to obtain representative ice morphologies. c Statistics of ice types on PTFE-0 surface under different flow environment parameters. The clear ice, mixed-ice and rime are marked in yellow, green and blue areas respectively. d High-speed images of the typical ice formation process of mixed ice on PTFE-0 surface under −20 °C & 40 m s−1, where the yellow box indicates the typical morphology evolution of droplets. e The influence of different temperatures on the start freezing time and complete freezing time on PTFE-0 surface under 40 m s-1. f The influence of different flow velocities on the start freezing time and complete freezing time on PTFE-0 surface at −20 °C. g The influence of flow velocity on the droplet freezing size at the solid-ice interface under −20 °C, and the influence of different temperatures on the droplet freezing size under 40 m s−1. The repetitive experiments are conducted for 10 times and data are presented as mean values +/- SD. Source data are provided as a Source Data file. Subsequently, the influence of flow velocity on icing behavior at PTFE-0 surfaces is investigate at a constant −20 °C within 180 s. Microscopic observations reveal that lower-velocity (<30 m s−1) flow promotes density increases, while higher-velocity flow drives a morphological transformation from opaque rime to transparent clear ice, resulting in mixed-ice structures. This transition is corroborated by macroscopic morphology and density evolution of ice [Supplementary Figs. 8 and 9]. It is inferred that ice thickens and extends the heat transfer path between the surface and incoming microdroplets as freezing proceeds, leading to the hindrance of latent heat release for the microdroplets. Consequently, partial microdroplets that cannot freeze immediately infiltrate the rime layer, causing the ice to become denser. Meanwhile, the remaining microdroplets accumulate on the ice surface to form a water film and finally convert to clear ice. The cross-sectional morphologies of three typical ice types (clear ice, rime, and mixed ice) are characterized to further elucidate the mechanisms governing ice morphology under flow environments. The voids within the ice layer are completely filled with water under a velocity of 30 m s−1 at −5 °C, presenting a highly transparent form of clear ice, which arose from reduced thermal gradients that facilitate continuous water film formation prior to freezing [Fig. 3b, Supplementary Figs.10 and 11]. When the temperature drops to −20°C, the elevated temperature gradient promotes droplet solidification upon impact immediately, producing a rime layer composed of discrete ice crystals [Supplementary Figs. 12 and 13]. However, a distinct layered structure consists of the rime in contact with the substrate (interfacial-rime) and the clear ice away from the solid-liquid interface emerges and is defined as mixed ice when the flow velocity increases to 40 m s-1. It is confirmed that the ice formation behavior of microdroplets is governed by the competition between phase transformation rate and accumulation rate of water film. Meanwhile, different types of interfacial-ice are also observed on representative superhydrophobic materials (Polyvinylidene fluoride (PVDF), Fluorinated ethylene propylene (FEP), Cerium Stearate, Heptadecafluoro-1,1,2,2-tetradecyltrimethoxysilane (FAS-17@Si) and F-SiO2@PVDF) with various surface topographies under flow conditions, indicating a universal existence of interfacial-ice [Supplementary Figs. 14, 15, 16, 17 and 18]. Notably, this phenomenon is not solely attributable to surface energy, as interfacial ice layers can even form on hydrophilic surfaces [Supplementary Figs. 19 and 20]. A statistical analysis of ice types across temperature and flow velocity conditions reveals that clear ice predominates at −5°C and −10 °C, while rime is absent at −5°C (Fig. 3c). As temperature decreases to −10°C and below, the environmental range where rime form has expanded while the occurrence of clear ice is diminished. The flow velocity required for the transformation from rime to mixed ice also ascends from 20 m s-1 at −10 °C to 40 m s-1 at −20 °C. Notably, interfacial rime is consistently formed during initial microdroplet contact regardless of flow velocity at low-temperature (<−10 °C). Considering that mixed ice predominates in the ice distribution under flow condition, a high-speed camera system is adopted to capture the dynamic evolution of mixed ice (Fig. 3d). It is observed that the formation of mixed ice can be delineated into four distinct stages: (1) droplet coalescence (0.4494 s), wherein non-wetting characteristic induces rapid microdroplet aggregation, initiating the incubation period of freezing; (2) discrete droplet freezing (0.9402 s), discrete droplets freeze before forming a water film without mutual connection; (3) interfacial rime formation (1.7974 s), subsequent microdroplets freeze on contact with the protruding ice crystals, creating a porous substructure due to the blocking effect. (4) liquid film accumulation (after 13.826 s), wherein reduced interfacial heat transfer caused by ice thickening promotes surface water retention, resulting in the formation of clear ice [Supplementary Movie 1]. Given the key role of initial freezing dynamics in defining interfacial ice morphology, the influence of temperature on droplet freezing time is investigated at a representative flow velocity (40 m s-1). Since microdroplets start to freeze in the form of water films at -5 °C, the temperature range below −10 °C is chosen for analysis. At −10 °C, microdroplets begin to freeze after a pre-cooling period of 0.9513 s (Fig. 3e). As temperature decreases to −20 °C, this delay contracts significantly to 0.4437 s, with complete freezing achieved by 0.9385 s. Interestingly, further temperature reductions have few effects on icing time, indicating the limitation in phase change efficiency driven by temperature [Supplementary Fig. 21]. Meanwhile, flow velocity is also found to significantly influence freezing kinetics. The droplet aggregation stage shortens sharply with increasing velocity at −20 °C (Fig. 3f). At 60 m s-1, the complete freezing time of the droplets is even only 21.98% of that at 20 m s-1, indicating a significant promotion of the freezing process. This acceleration is attributed to higher droplet kinetic energy and enhanced convective heat transfer induced by flow, which expedites nucleation and suppresses droplet aggregation [verified by Supplementary Fig. 22]. Additionally, quantification of ice crystal dimensions at the solid-ice interface indicates that crystal size decreases from 232.47 μm at −10 °C to 63.83 μm at −20 °C under a constant velocity of 40 m s−1 (Fig. 3g). Similarly, the ice crystal size is reduced from 121.99 μm to 20.23 μm when the flow velocity progresses from 20 m s−1 to 60 m s−1 at −20 °C. Specifically, the ice crystal sizes approach the scale of individual microdroplet under the velocity of 60 m s−1, implying instantaneous freezing upon impact. It is affirmed that the dominance of early-stage freezing driven by intense convective heat transfer leads to a suppressed aggregation and finer crystal structures although enlarged droplet flux can theoretically favor coalescence. Transformation of interfacial-ice induced by microstructure The thickness of the interfacial-rime on PTFE-10 and PTFE-0 surfaces is compared within 60 s at varying temperatures under a constant flow velocity of 40 m s-1 to elucidate the effect of microstructure on the formation process of interface-rime (Fig. 4a). At −5 °C, both surfaces exhibit clear ice with no observable interfacial-rime. The obvious disparity in the interfacial-rime thickness occurs at −10 °C, with the interfacial-rime thickness on the PTFE-10 surface being 0.67 mm, which is 294.12% higher than that on the PTFE-0 surface. As the temperature drops, the dominant influence of droplet freezing rate weakens the effect of microstructure on microdroplet accumulation, nevertheless, the PTFE-10 surface still presents an incrassated interfacial-rime layer, reaching a thickness of 2.43 mm at −20°C (right side of Fig. 4a).Fig. 4Icing behavior on various superhydrophobic surfaces under flow condition.a Thickness of interfacial-rime on PTFE-0 and PTFE-10 surfaces under different temperatures at a constant velocity of 40 m s−1 within 60 s, the evolution process of interfacial-rime on PTFE-10 surface with different temperatures is illustrated on the right side. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. b Droplet freezing process on PTFE-10 surface at −20 °C and 40 m s−1, the green box indicates the spherical cap-shaped ice crystals. c Initial freezing time and complete freezing time of droplets on PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces at -20 °C and 40 m s−1. All repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. d Initial ice crystal size and transformation time of ice type on PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces at −20 °C and 40 m s−1. e High-speed image of droplet splitting process on PTFE-10 surface. f Schematic diagram of the droplet rupture model underflow condition. g Ice type statistics of PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces under different flow conditions. The clear ice, mixed ice and rime are marked in purple red, blue and green respectively. Source data are provided as a Source Data file. Icing behavior on various superhydrophobic surfaces under flow condition. a Thickness of interfacial-rime on PTFE-0 and PTFE-10 surfaces under different temperatures at a constant velocity of 40 m s−1 within 60 s, the evolution process of interfacial-rime on PTFE-10 surface with different temperatures is illustrated on the right side. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. b Droplet freezing process on PTFE-10 surface at −20 °C and 40 m s−1, the green box indicates the spherical cap-shaped ice crystals. c Initial freezing time and complete freezing time of droplets on PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces at -20 °C and 40 m s−1. All repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. d Initial ice crystal size and transformation time of ice type on PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces at −20 °C and 40 m s−1. e High-speed image of droplet splitting process on PTFE-10 surface. f Schematic diagram of the droplet rupture model underflow condition. g Ice type statistics of PTFE-3, PTFE-5, PTFE-10 and PTFE-15 surfaces under different flow conditions. The clear ice, mixed ice and rime are marked in purple red, blue and green respectively. Source data are provided as a Source Data file. Subsequently, high-speed images indicate that microdroplets coalesce into larger spherical-cap-shaped aggregates (marked by arrows) by 0.737 s on PTFE-10 at −20°C and 40 m s−1 (Fig. 4b). After 0.747 s, several larger droplets come closer to each other during the freezing process, inhibiting the subsequent filling of the gaps between the ice crystals. Complete surface freezing is observed by 2.384 s, producing numerous isolated ice crystals (marked by box). This process parallels that observed on PTFE-0 [Supplementary Movie 2]. Therefore, further analysis is conducted on the influence of microstructure size on the droplet size and freezing time to clarify the promoting effect of the microstructure on interfacial-rime under the same environment conditions in Fig. 4b. The PTFE-10 surface exhibits a prolonged initial freezing time of 0.74 s, which is attributed to its appropriately scaled microstructure that effectively suppresses water molecule nucleation by elevating the nucleation energy barrier and disrupting the ordered arrangement of water molecules [Fig. 4c and Supplementary Fig. 23]. However, the complete freezing time of droplets on PTFE-10 surface is generally 10.19% to 26.02% earlier than that on the PTFE-0 surface, proving that the microstructure can actually promote the freezing process of coalesced droplets, leading to the rapid formation of the initial ice crystal. Moreover, the transition time from interfacial-rime to clear ice displays a 72.95% delay on the PTFE-10 relative to PTFE-0 surface, leading to an enlargement in the thickness of interfacial-rime (Fig. 4d and Supplementary Fig. 24). Since both bare droplet rebound phenomenon is observed and droplet contact rates are comparable on either surface, this effect cannot be attributed to differences in droplet accumulation. Rather, the hindered infiltration of microdroplets into rime voids is linked to smaller initial ice crystals. Meanwhile, the initial crystal size on PTFE-10 surface is 29.74 μm, which is only 45.6% of that on PTFE-0 surface, indicating a trend that the ice crystal size descends as the microstructural size augments. Unfortunately, a further increase in microstructural size generates a larger initial crystal size (34.99 μm on PTFE-15 surface). Supplementary Fig. 25 demonstrates that the size of the pre-freezing droplet on PTFE-10 surface is markedly smaller than that on other surfaces. Larger droplets always lead to a reduction in ice crystal gaps and make it easier for subsequent microdroplets to fill them, giving rise to a shortened transformation process of ice type. Considering that the growth of droplets mainly depends on the merging of microdroplets, it can be inferred that the reduced ice crystal size on PTFE-10 surface likely results from inhibited droplet aggregation. Similarly, the structural dimensions also have a significant impact on the interfacial-rime for various superhydrophobic materials, revealing a common scheme for regulating interfacial-ice [Supplementary Fig. 26]. Notably, a unique droplet separation phenomenon is observed exclusively on PTFE-10 surface, wherein merged droplets spontaneously divide into two smaller droplets within 36.3 ms and rapidly freeze at 38.7 ms (Fig. 4e and Supplementary Movie 3). This atypical fragmentation which is contrary to the conventional behavior on hydrophobic surfaces significantly reduces the final ice crystal size. Hence, theoretical analysis of this separation process is performed to further explore the relationship between critical droplet size that can exhibit splitting behavior and the microstructure size. Considering that the microstructural heights are always 3 ~ 4 times the radius on various surfaces, the microstructure is simplified to a cylinder, and the initial droplet is modeled as a spherical cap (Supplementary Fig. 27). According to volume conservation, the critical condition of droplet rupture can be expressed as:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{init}={V}_{top}+{V}_{side}$$\end{document}Vinit=Vtop+Vsidewhere Vinit is the volume of the droplet before it ruptures, Vtop and Vside represent the volumes of droplets that adhere to the top and sides of the microstructure after the droplets rupture, respectively. For the convenience of analysis, the cross section of the Vinit and Vside is simplified as the circular segment, as shown in Fig. 4f. Considering the liquid on the top as a thin film with a close-to-zero thickness at the moment of droplet rupture, the volume of the droplet on the top is negligible, while only the surface area is considered when calculating the energy. in which2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{init}=\frac{\pi }{3}{R}_{0}^{3}\left[4-{(1+\,\cos \theta )}^{2}(2-\,\cos \theta )\right]$$\end{document}Vinit=π3R034−(1+cosθ)2(2−cosθ)3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{top}=0$$\end{document}Vtop=04\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{side}= \int\limits_{-\pi+\theta }^{\pi -\theta }[\pi {R}_{d}{({R}_{d}\,\cos \varphi+{R}_{s}-{R}_{d}\,\cos \theta )}^{2}-\pi {R}_{s}^{2}]\cos \varphi d\varphi \\ +4{R}_{d}^{4}(\sin \theta -\frac{1}{3}{\sin }^{3}\theta )\int\limits_{\pi -\theta }^{\frac{\pi }{2}}[\pi {R}_{d}{({R}_{d}\,\cos \varphi+{R}_{s}-{R}_{d}\,\cos \theta )}^{2}-\pi {R}_{s}^{2}]\cos \varphi d\varphi \\ +2\pi {R}_{d}^{3}[\sin \theta \,\cos \theta ({R}_{d}\,\cos \theta -2{R}_{s})-(\frac{\sin 2\theta }{2}+\theta )({{R}_{d}}\, \cos \theta - {{R}_{s}})]$$\end{document}Vside=∫−π+θπ−θ[πRd(Rdcosφ+Rs−Rdcosθ)2−πRs2]cosφdφ+4Rd4(sinθ−13sin3θ)∫π−θπ2[πRd(Rdcosφ+Rs−Rdcosθ)2−πRs2]cosφdφ+2πRd3[sinθcosθ(Rdcosθ−2Rs)−(sin2θ2+θ)(Rdcosθ−Rs)]where R0 represents the initial droplet radius, θ is the droplet contact angle, Rd is the radius of the separated droplet, and Rs is the radius of microstructure. According to the energy conservation principle before and after the rupture, it can be known46:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{k0}+{E}_{g0}+{E}_{s0}+{E}_{f0}={E}_{k}+{E}_{g}+{E}_{s}+{E}_{f}+W$$\end{document}Ek0+Eg0+Es0+Ef0=Ek+Eg+Es+Ef+WWhere Ek0, Eg0, Es0 and Ef0 denote the kinetic energy, potential energy, surface energy and flow impact energy at the initial stage. Following droplet breakup, these energies are represented as Ek, Eg, Es and Ef, respectively. W is the viscous dissipation of the droplet. Taking into account that the kinetic energy of droplet does not exceed 5% of the initial total energy at the moment of rupture46,47:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{k0}={E}_{k}\approx 0$$\end{document}Ek0=Ek≈07\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{g0}-{E}_{g}=\frac{\pi {R}_{0}^{4}\rho g\,\cos \theta }{3}[{(1+\,\cos \theta )}^{2}(2-\,\cos \theta )-4]$$\end{document}Eg0−Eg=πR04ρgcosθ3[(1+cosθ)2(2−cosθ)−4]8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{s0}-{E}_{s}=\, {\sigma }_{g-l}\bigg[\pi {R}_{0}^{2}(2-2\,\cos \theta -{\cos }^{2}\theta )-\pi {R}_{s}^{2} \\ -4\pi {R}_{d}({R}_{d}\theta \,\cos \theta -{R}_{s}\theta+{R}_{s}\,\sin \theta -{R}_{d}\,\sin \theta )\bigg]$$\end{document}Es0−Es=σg−lπR02(2−2cosθ−cos2θ)−πRs2−4πRd(Rdθcosθ−Rsθ+Rssinθ−Rdsinθ)9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{f0}-{E}_{f}= \frac{1}{2}\rho {V}_{0}^{3}\left[\pi {R}_{0}^{2}-2{R}_{d}^{2}(\theta -\frac{1}{2}+{\sin }^{2}\theta ) \right. \\ \left.+{R}_{s}^{2}(\pi -2\arcsin \frac{{R}_{d}\,\sin \theta }{{R}_{s}})-2{R}_{d}\,\sin \theta \sqrt{{R}_{s}^{2}-{R}_{d}^{2}{\sin }^{2}\theta }\,\right]$$\end{document}Ef0−Ef=12ρV03πR02−2Rd2(θ−12+sin2θ)+Rs2(π−2arcsinRdsinθRs)−2RdsinθRs2−Rd2sin2θ10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$W=\frac{4}{3}\pi \mu \sqrt{\frac{2{\pi }^{2}{{V}_{0}}^{3}{R}_{0}^{5}{R}_{e}}{3}\left[4-{(1+\,\cos \theta )}^{2}(2-\,\cos \theta )\right]}\,({R}_{S}^{2}+4{R}_{S}{R}_{d}\,\sin \theta )$$\end{document}W=43πμ2π2V03R05Re34−(1+cosθ)2(2−cosθ)(RS2+4RSRdsinθ)Where σg-l represents the tension coefficient of gas-interface, V0 is the flow velocity, S0 represents the initial windward area of droplet, and S represents the windward area of the dissociated droplet. Combining Eq. (1) ~ (10), and eliminating the intermediate variable Rd, a quantitative relationship between flow velocity (V0) and microstructure size (Rs) is established. That is, microstructures that satisfy this dimensional criterion can effectively facilitate the fragmentation of accumulated droplets on the surface under a specific flow velocity, which contributes to a reduction in the initial ice crystal dimensions. The full derivation is provided in Supplementary Section 1.5, and the reliability of theoretical predictions have been verified in Supplementary Fig. 28. Furthermore, ice pattern distributions are statistically analyzed for four superhydrophobic surfaces to evaluate the influence of microstructure on the ice types (Fig. 4g). The ice types on the PTFE-3 surface are similar to those on the PTFE-0 surface, suggesting that fine microstructures exert minimal influence on initial crystal formation. As microstructure size aggrandizes, the environmental envelope supporting rime formation expands, whereas the range supporting clear ice formation is shrunk. This confirms that appropriately sized microstructures not only thicken48,49 the interfacial-rime but also broaden the environmental parameter to form rime and mixed ice under flow conditions. Interestingly, compared to PTFE-10, PTFE-15 facilitates the formation of clear ice across a broader range of environmental conditions while suppressing the appearance of rime, which is attributed to the larger microstructure promoting droplet roll-off rather than fragmentation, facilitating the pores between initial ice crystals easily filled by subsequent microdroplets. These results suggest that contrary to conventional beliefs, the ice types on hydrophobic materials are not governed solely by environmental conditions. Rather, microstructure size plays a pivotal role in dictating ice type. Furthermore, the types of interface-ice on various superhydrophobic surfaces all show a variation with the change in structural size, indicating that the influence of non-wetting microstructural parameters on the interface-ice type is beyond the limitations of material category and structural morphology (Supplementary Figs. 29, 30, 31, 32 and 33). Solid-ice interface evolution during melting process A comparative assessment is conducted on PTFE-0 surfaces subjected to varied icing conditions to explore the influence of interfacial ice on the melting and de-icing processes (detailed icing parameters as shown in Fig. 3b). At −20 °C, the melting times across clear ice, frost ice, and mixed ice at the interface are similar while the complete detachment times are 223 s, 54 s, and 72 s respectively (Fig. 5a). Macroscopic images reveal that the thickness of clear ice gradually decreases until it completely melts, leaving residual water on the surface. In contrast, mixed ice exhibits a limited sliding behavior with a relatively constant thickness while rime is abruptly detached from the surface, and little residual water can be observed on the two surfaces (Supplementary Movie 4).Fig. 5Melting behavior of different ice types on different hydrophobic surfaces.a The start melting time and complete detachment time of clear ice, rime and mixed ice on the PTFE-0 surface at -20 °C, the macroscopic detachment process shown in the upper inset (detailed freezing parameters: clear ice ~ -5 °C@30 m s-1, frost ice ~ -20 °C@30 m s-1 and mixed ice ~ -20 °C@40 m s-1). The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. b The microscopic evolution process of the melting processes of the above three typical ice types, where the yellow boxes indicate the densified areas and interface defect areas of rime, and the blue boxes indicate the densified areas of mixed ice. c The start melting time and complete detachment time of clear ice, rime and mixed ice on the PTFE-10 surface, the blue dotted line representing the reference value of the complete detachment time on PTFE-0 (detailed freezing parameters: clear ice ~ -5 °C@50 m s-1, frost ice ~ −20 °C@30 m s-1 and mixed ice ~ −20 °C@40 m s−1). The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. d The microscopic evolution process of the melting process of rime and mixed ice in Fig. 5c, where the yellow box indicates the densified area of rime. e The complete detachment times of the clear ice, rime and mixed ice on different superhydrophobic PTFE surfaces during the melting process. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. f The mechanism diagram for controlling the interfacial-ice through microstructure to promote the deicing process. Source data are provided as a Source Data file. Melting behavior of different ice types on different hydrophobic surfaces. a The start melting time and complete detachment time of clear ice, rime and mixed ice on the PTFE-0 surface at -20 °C, the macroscopic detachment process shown in the upper inset (detailed freezing parameters: clear ice ~ -5 °C@30 m s-1, frost ice ~ -20 °C@30 m s-1 and mixed ice ~ -20 °C@40 m s-1). The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. b The microscopic evolution process of the melting processes of the above three typical ice types, where the yellow boxes indicate the densified areas and interface defect areas of rime, and the blue boxes indicate the densified areas of mixed ice. c The start melting time and complete detachment time of clear ice, rime and mixed ice on the PTFE-10 surface, the blue dotted line representing the reference value of the complete detachment time on PTFE-0 (detailed freezing parameters: clear ice ~ -5 °C@50 m s-1, frost ice ~ −20 °C@30 m s-1 and mixed ice ~ −20 °C@40 m s−1). The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. d The microscopic evolution process of the melting process of rime and mixed ice in Fig. 5c, where the yellow box indicates the densified area of rime. e The complete detachment times of the clear ice, rime and mixed ice on different superhydrophobic PTFE surfaces during the melting process. The repetitive experiments are conducted for 5 times and data are presented as mean values +/- SD. f The mechanism diagram for controlling the interfacial-ice through microstructure to promote the deicing process. Source data are provided as a Source Data file. Afterward, the evolution process of the solid-ice interface indicates that clear ice produces a liquid water film at interface by 20 s during the melting process, leading to a strong adsorption effect attributed to external atmospheric pressure and surface tension (Fig. 5b). As the melting time extends from 20 s to 51 s, the ice layer thickness continuously reduces from 2.79 mm to 1.47 mm (as shown by dotted line). Additionally, the rime shows a densification phenomenon at interface after being heated for 25 s (from white to transparent), indicating that the liquid water formed by the melting process can penetrate into the voids of rime. Subsequently, obvious interface defects are generated since the liquid water is absorbed by interfacial-rime, resulting in an abrupt detachment of ice after 54 s. Diversely, mixed ice that contains interfacial-rime with limited thickness is insufficient to accommodate all the meltwater. The rime crystals near the melting area effectively promote the contact and coalescence between droplets, forming a continuous thin water film50. Particularly, this water film acting as a lubricating medium facilitates the slippage of ice layer due to the discontinuous contact between the interfacial-rime and the substrate. However, the detachment times for all ice types on PTFE-10 surface are significantly shorter than those on PTFE-0 surface (Fig. 5c). Notably, the separation time of rime and mixed ice are only 24 s and 32 s, respectively, which are 55.56% and 56.94% lower than that on PTFE-0 surface. Surprisingly, clear ice and rime follow similar melting regulation as on PTFE-0, while mixed ice displays an abnormal detachment model similar to rime [Supplementary Fig. 34 and Movie 5]. The microstructural analysis clarifies that a limited densification phenomenon is observed within rime on the PTFE-10 surface during heating up to 16 s (indicated by orange box), which is due to the elevated ice porosity and reduced initial crystal size induced by microstructure. Meanwhile, cracks have already emerged and propagated across the solid-ice interface within 2 s. It is deduced that fewer initial solid-ice contact interfaces arose from low-energy microstructure, reducing the burden of de-icing and promoting the ice to detach within 5 s. Particularly, different superhydrophobic surfaces also confirms that the presence of interfacial-rime can significantly reduce the deicing time, demonstrating that regulating interfacial-ice by optimizing the geometric parameters of non-wetting microstructure has become a universal design criterion for thermal de-icing process [Supplementary Figs. 35, 36, 37, 38 and 39]. It is worth noting that the mixed ice exhibits a localized densification, defects initiation and interface separation process during heating, which is similar to that of rime (Fig. 5d). This means that the thickened interfacial-rime induced by microstructure inhibits the formation of continuous water film, converting the interfacial mode from solid-liquid-ice to solid-ice contact through avoiding the additional adsorption effects brought by liquid tension and external atmospheric pressure. Similarly, the less initial solid-ice contact is also conducive to the initiation and expansion of interface cracks. It is confirmed that the interfacial-rime layer with thickness exceeding the threshold value can be peeled off from the substrate, which analogous to the melting behavior of rime under static conditions51. The estimation process of the threshold thickness is described in Supplementary section 1.6. Moreover, although the clear ice on the superhydrophobic surface still needs to be enormously melted for detachment, the enhanced thermal conductance at the solid-liquid interface triggered by microstructure also improves melting efficiency (Supplementary Fig. 40). Two typical melting processes of interfacial-clear ice and interfacial-rime on the PTFE−10 surface are shown in Supplementary Movie 6. Subsequently, detachment times for different ice types on various superhydrophobic surfaces are systematically evaluated to assess the effect of microstructure scale on de-icing performance (Fig. 5e). It is obvious that the detachment time of clear ice exhibits a progressive decrease with increasing microstructural size, whereas that of rime always remains at ~25 s. Specifically, mixed ice exhibits a minimum detachment time of 32 s on PTFE-10, followed by an increase to 36 s on PTFE-15. This means that the microstructure not only governs the melting dynamics of mixed ice but also narrows the time difference between mixed ice and rime, expanding the applicable range of the de-icing process promoted by interfacial-rime. Even in cases with minimal interfacial-rime thickness (e.g., 1 mm on PTFE-3 at −15 °C and 30 m s-1; 0.42 mm on PTFE-5 at −10°C and 40 m s-1), the microstructure can still suppress the formation of water film and the generation of interface adsorption effects, accelerating the de-icing process [Supplementary Fig. 41]. Generally, the size of microstructures can not only regulate the organization and thickness of interfacial-ice induced by the droplet separation effect, but also govern the solid-ice contact behavior during melting, enabling a significant reduction in the energy consumption for thermal de-icing under diverse icing conditions (Fig. 5f). Discussion The results obtained in the static environment confirm that hydrophobic surfaces can facilitate the coalescence of dispersed microdroplets into larger isolated droplets prior to freezing due to their intrinsic non-wetting characteristics. Conversely, microstructures on superhydrophobic surfaces serve dual roles by impeding droplet coalescence and providing preferential nucleation sites for ice formation, promoting initial crystal growth. Correspondingly, the freezing behavior of microdroplets is predominantly governed by the competitive interplay between droplet phase-change rates and the accumulation efficiency of a water film on the hydrophobic surface under flow conditions. The intense convective heat transfer rapidly elicits the freezing of microdroplets upon initial contact with the surface at low temperatures (≤ -10 °C), always forming an immediate rime layer. As freezing progresses, the elongation of the thermal pathway between the droplets and the PTFE substrate inhibits latent heat transfer, leading to a deferred freezing of subsequent microdroplets. These microdroplets that cannot freeze rapidly on the ice penetrate the rime layer, generating a stratified ice morphology characterized by distinct rime and clear ice. However, microstructures can actually facilitate the freezing process of converging droplets with the assistance of the flow field, promoting the rapid formation of the initial ice crystal. Subsequently, these initial crystals facilitate the growth of the interfacial-rime through the structural blocking effect. Notably, pre-frozen droplets that satisfy a corresponding mathematical relationship can be effectively fragmented by specific-sized microstructure under flow condition, significantly reducing the initial ice crystal size and substantially increasing the thickness of interfacial-rime. Furthermore, contrary to conventional beliefs, microstructures also have a decisive impact on ice type by regulating the freezing behavior of droplets, expanding the range of environmental parameters conducive to interfacial-rime formation. Remarkably, a superhydrophobic surface can transform the traditional contact form of solid-liquid-ice to a direct mode of solid-ice for mixed ice through controlling the surface microstructure size during the melting process. This approach avoids the additional adsorption effects brought by liquid tension and external atmospheric pressure, significantly narrowing the difference in de-icing time between mixed ice and rime. This strategy enables a substantial reduction of 55.56% to 56.94% in de-icing times for conventional hydrophobic materials across a broader spectrum of icing conditions without additional energy consumption. Methods Preparation and characterization of hydrophobic surfaces A 2cm × 2cm × 2mm specimen of 2024 aluminum alloy was mechanically polished and buffed, followed by sequential ultrasonic cleaning in acetone, anhydrous ethanol, and deionized water for 15 min. After drying, 0.08 mL of a polytetrafluoroethylene (PTFE) nanoparticle dispersion (60 wt%), containing a non-ionic surfactant, was sprayed onto the aluminum substrate at a pressure of 0.4 MPa from a distance of 80cm to yield the PTFE-0 sample. Additionally, when the as -prepared PTFE-0 sample was naturally air-dried, additional PTFE dispersions of 3 mL, 5 mL, 10 mL, and 15 mL were sprayed onto its surface again using the same parameters to generate the PTFE-3, PTFE-5, PTFE-10, and PTFE-15 samples, respectively. All these samples were subsequently annealed in a muffle furnace by heating at a rate of 5 °C min−1to 360 °C, held isothermally for 1h, and then water-quenched to obtain the final samples. Surface morphology and roughness were characterized using field-emission scanning electron microscopy (FE-SEM) and optical profilometry. Chemical composition was determined via Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), and energy-dispersive X-ray spectroscopy (EDS). Surface wettability was assessed by contact angle measurements. Detailed procedures and materials are provided in Supplementary Section 2.1. Evaluation of icing behavior in static environments Static icing experiments were performed in a self-made low-temperature environment chamber. To ensure sufficient thermal exchange, the test samples were placed atop a DC-6506 low-temperature constant-temperature platform, with the stage temperature matched to the ambient chamber conditions and a humidity of about 95%. Subsequently, the microdroplets generated by a precision droplet emitter were pre-cooled and transported via airflow to uniformly deposit on the sample surfaces to form ice. The ice formation process on the material surface was observed using a stereo microscope (ASV0870), and the corresponding icing time and ice formation morphology were systematically recorded. The schematic diagram of the test equipment is shown in Supplementary Fig. 42. A dedicated laser-based diagnostic system was employed to quantify the size of microdroplets. Based on the laser reflection intensity, the diameter distribution of the microdroplets was determined to be 6.14 μm, as demonstrated in Supplementary Fig. 43. The expansion time of ice layer was defined as the time elapsed from the moment the initial ice crystal was detected within the observation field until the ice crystals completely covered the observed surface. Icing process analysis of microdroplets in flow environment The experiments were conducted in a laboratory-designed icing wind tunnel to simulate dynamic flow conditions (Supplementary Fig. 44). The diameter of microdroplets was regulated to ~20 μm by adjusting the air and water pressures within the droplet generation system. The grid freezing test confirmed a relatively uniform microdroplet distribution across the flow field [Supplementary Fig. 45]. The temperatures of the icing wind tunnel were set at −5 °C, −10 °C, −15 °C, and −20 °C, with flow velocities ranging from 0 m s−1 to 70 m s−1. The ice morphology on the windward side and the lateral side was visualized using a confocal microscope, while the macroscopic freezing process was captured and recorded by a CCD camera. Additionally, a high-speed camera with 200,000 frames s-1 was used to precise capture and record the initial freezing time, complete freezing time and the size of the initial ice crystal under varying flow conditions. Particularly, the ice layer was sectioned with a cryogenic blade matched to ambient conditions to accurately identify the interfacial ice morphology, and the interface microscopic images were imaged and analyzed. Implementation of thermal de-icing process A commercial polyimide electric heating film was affixed to the rear surface of the sample to implement the de-icing process. The sample was initially placed in the icing wind tunnel at various temperatures and velocities for 60 s to induce surface icing. Upon an ice layer formed on the sample, the heating film was activated at a constant voltage of 28 V, corresponding to a power density of 1 W cm-2. The variation of microscopic morphology at the solid-ice interface was observed through a microscope during heating, while the macroscopic processes of ice detachment behavior driven by gravitational forces were captured and recorded by a CCD imaging system. All data analyses were conducted using the Origin (Learning Edition) software. Supplementary information Supplementary Information Description of Additional Supplementary File Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Supplementary Movie 5 Supplementary Movie 6 Reporting-summary Transparent Peer Review file Supplementary Information Description of Additional Supplementary File Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Supplementary Movie 5 Supplementary Movie 6 Reporting-summary Transparent Peer Review file Source data Source Data Source Data