Introduction

Unmanned aerial vehicles (UAVs) are intelligent flying robots that are widely used in various fields, including aerial photography, remote sensing mapping, logistics transportation, and disaster relief1,2,3. However, in-flight icing, which is a leading cause of drone accidents, hinders their use, especially in extreme weather4,5,6,7,8. This is because UAVs are usually small and sensitive to weight1,9,10,11,12,13. Moreover, due to their limited payload and scant excess power, traditional anti-icing and deicing methods, such as mechanical or electrothermal deicing, are not very suitable or energy-saving for UAVs11,14,15,16,17. Electrothermal systems for UAVs would be carefully designed to minimize the required heat to operate the system. Therefore, effective anti-icing coating would be an ideal solution to prevent icing on the smart flying robots18,19,20,21,22,23,24,25,26.

In addition to anti-icing, continuous real-time monitoring of in-flight icing is urgently desired for intelligent control in time to avoid UAV accidents. This requires a UAV skin that enables both anti-icing and real-time icing monitoring to improve the intelligence of smart robots, rather than only relying on the traditional anti-icing coatings. However, to the best of our knowledge, there is no investigation towards an intelligent skin of UAVs for ice accretion prevention and monitoring. That may be because, first of all, the integration of anti-icing and sensing performance into an ultrathin and skin sensor device is challenging27,28,29. At freezing temperatures, the substrate/electronic materials that are comprised of skin sensors, such as hydrogels or liquid metals, commonly suffer from increased rigidity or decreased conductivity30,31,32,33,34. In addition, the reduced molecular fluidity at freezing temperatures further hinders the development of self-healing skins or coatings that can withstand surface injuries. Second, the commonly-used microstructure-dependent anti-icing coatings on in-flight UAVs are inevitably prone to mechanical damage from sandstorms or impacts, especially under extreme weather conditions, resulting in the loss of sensing or anti-icing functions35.

Herein, a freezing-tolerant self-healing skin was developed for UAVs with both anti-icing/deicing and real-time icing monitoring abilities (Fig. 1). The flexible skin consists of five layers comprising of overall self-healing elastomers, electrodes, and an anti-icing encapsulation layer. The encapsulation layer provides the skin with both anti-icing and icephobic properties, lowering the ice nucleation temperature to −28.4 °C and reducing the ice adhesion strength to below 33.0 kPa. Owing to its overall self-healing sensing construction design, the developed skin enables continuous real-time monitoring of in-flight frost and icing on a UAV, with its sensory function completely self-recovering after damage, even at temperatures as low as −20 °C. This work contributed to the concept of intelligent anti-icing skins for flying mini robots, enabling their operation under extreme weather.

Fig. 1: Schematic and properties of self-healing intelligent UAV skin.
figure 1

a Self-healing intelligent UAV skin for the detection and prevention of in-flight icing, b developed five-layer UAV skin and the mechanism of in-flight icing monitoring, and c anti-icing, icephobic, and self-healing properties of the developed UAV skin.

Full size image

Results

Design and characterization of the self-healing layers

To endow the proposed UAV skin with anti-icing and self-healing properties, a universally autonomous self-healing elastomer (SE) was prepared based on synergistic interactions of multiple types of dynamic bonds within a supramolecular polydimethylsiloxane (PDMS) framework, serving as the substrate and dielectric layers (Supplementary Figs. 1, 2). Based on this low surface energy of silicone-based SE, PDMS-based fluoropolymer (PFS) was designed and integrated into the SE as an anti-icing encapsulation layer for UAV skin (Supplementary Figs. 3, 4). The fluorine atoms in PFS have a strong affinity to electron clouds, and the C–F bond has high bond energy, making it difficult to polarize. This results in lower surface energy and good ice-phobic properties. Therefore, the introduction of PFS into the SE substrate has the potential to cover the polar groups (–C=O and –NH) on the SE surface to inhibit the interaction between dynamic hydrogen bonds and water molecules and improve the anti-icing and deicing performance. Furthermore, the interaction forces between water molecules and fluorine differ from those between water molecules and Si–O, disrupting the orderly arrangement of the liquid-like layer between ice and the skin encapsulation layer surface, thereby reducing ice adhesion strength36,37. Meanwhile, because of its PDMS segments, PFS exhibits compatibility with the SE substrate, ensuring its dispersibility. To investigate the influence of PFS on the anti-icing/deicing and self-healing properties, we prepared encapsulation layers with varying weight ratios of PFS at 5, 15, 25, and 35 wt%, termed as SEF5, SEF15, SEF25, and SEF35, respectively, compared to the SE.

Fig. 2a and Supplementary Fig. 5 show the surface topography of all samples using atomic force microscopy. The SEF5, 15, 25 layers exhibited the smooth and flat surfaces with negligble phase separation, similar with SE. While at a high PFS concentration of 35%, the phase separation was pronounced and the agglomerates increased the surface roughness, so an optimal FPS ratio of 25% for SE substrate was used for the encapsulation layer. Fig. 2b–c and Supplementary Fig. 6 show the energy dispersive spectroscopy elemental analysis (EDS) of SE and SEFs, proving the increased amounts of fluorine elements on the material surfaces and the decrease of their surface energy. Fig. 2d presents the chemical structures of all of the samples detected by attenuated total reflection–fourier transform infrared spectroscopy, to evaluate the surface chemical compositions of the SEF samples. The SEFs showed the vibration peaks of CF2 and CF3 at 1143 and 1240 cm−1, respectively, indicating the successful integration of the PFS polymers into the SE matrix38,39. The intensity of these two characteristic peaks was directly proportional to PFS concentrations (Fig. 2e). The intensities of the C=O stretching vibration peak at 1709 cm−1 and N–H bending vibration peak at 1537 cm−1, which contribute to hydrogen bond formation as polar groups, were inversely proportional to the PFS content (Fig. 2f, g). This suggests that the addition of PFS exposes C–F groups on the substrate surface instead of polar groups, thereby decreasing ice adhesion strength.

Fig. 2: Characterization of the self-healing layers.
figure 2

a Atomic force microscopy images of the SE and SEFs at 5 μm × 5 μm scan sizes. Energy dispersive spectroscopy elemental analysis of b SE and c SEF25 on the surface. The attenuated total reflection–Fourier transform infrared spectra of SE and SEFs in the ranges of d 500–4000 cm−1, e 1125–1255 cm−1, f 1660–1760 cm−1, and g 1450–1620 cm−1.

Full size image

As shown in Supplementary Fig. S7, a scratch on the SE surface healed to become almost invisible at −20 °C and −40 °C, as well as under saltwater at −10 °C for 12–24 h, indicating the low-temperature-tolerant and water-resistant self-healing performance of SE. Because in the dynamic supramolecular polymer network, the viscous and fluid-like property of SE with a low Tg (<−110 °C) prevented it from freezing state and enhanced its self-healing efficacy even in a low temperature environment. In addition, the sufficient multiple dynamic bonds could synergistically provide more reconfiguration opportunities when polymer chains were re-entangled at the damaged surface. Thereinto, the disulfide bonds with a low bond energy (60 kcal mol−1) mainly contributed to the self-healing property40. Moreover, the mechanical self-healing (Supplementary Fig. 8), rheological (Supplementary Fig. 9a–d), and cyclic stress–strain tests (Supplementary Fig. 9e, f) showed that both SE and SEF films exhibited the outstanding self-healing properties, stable elastic properties, and mechanical durability at room and low temperatures. The results also demonstrated that the outstanding mechanical and self-healing properties were well maintained even FPS copolymer addition into SE dynamic network.

Anti-icing/deicing performance of SEFs

The integration of PFS into SE reduced its surface wettability, as indicated by the static, advancing, and receding water contact angle measurements. With increasing PFS content, the static water contact angle of the material surfaces increased from 88.0° (SE) to 114.2° (SEF35), indicating a transition of the surface properties from hydrophilicity to hydrophobicity. Thus, the interaction between water molecules and the substrate surface was reduced (Fig. 3b). According to the Owens model, a low surface energy of SEF25 material was calculated as a result of 21.82 mN/m (Supplementary Fig. 10). The advancing contact angle and a receding contact angle of SEF25 were 119° and 70°, respectively, confirming the hydrophobicity and low surface energy of the SEF25 (Supplementary Fig. 11). In addition, the anti-icing properties of PFS are reflected in the coverage of the polar groups (–C=O and –NH) on the substrate surface, reducing the dynamic hydrogen bonds and the ice adhesion strength (Fig. 3a). We further investigated the inhibition of ice nucleation on the SEF surfaces, considering both the freezing delay time (Td) and ice nucleation temperature (TIN). Fig. 3c and Supplementary Fig. 12 show the freezing process of water droplets on the material surfaces and Td. The Td of the SEF25 surface was 243 s (Supplementary Movie 1), which is 35 times longer than that of the steel sheet (7 s). After 6 icing-deicing cycles, a negligible decrease was observed in Td on SEF25, indicating the stability of its icephobic properties (Supplementary Fig. 13). Supplementary Fig. 14 indicates that the Td of the injured SEF25 was decreased from 241 to 93 s and recovered to 231 s after self-healing at room temperature after 2 h. This result demonstrated the recovered anti-icing performance of the self-healed SEF25. The dynamic anti-icing properties were carried out by dropping supercooled water onto the 30° tilted SEF25 at −20 °C and 72% RH environment. The time of icing on the SEF25 surface was >540 s, significantly longer than that on the bare steel (~24 s), indicating a good dynamic anti-icing performance of SEF25 (Supplementary Fig. 15). TIN is the temperature at which ice nucleation first occurs41,42. As shown in Fig. 3d, SEF25 exhibited the lowest TIN (−28.4 °C) which is 8.1 °C lower than that of the blank control sample (steel sheet). The TIN of SEF35 (−27.0 °C) was slightly increased probably due to its uneven surface topography, which can reduce the ice nucleation energy barrier. The PFS addition reduced the TIN and prolonged the Td of SE, indicating its anti-icing ability in inhibiting ice nucleation. The deicing performance of the SEF samples was evaluated using a manmade ice shear force device (Supplementary Fig. 16). Owing to its smooth surface and low surface energy, SEF25 exhibited the lowest ice adhesion strength (33.0 ± 1.214 kPa) at −15 °C (Fig. 3e) (Supplementary Movie 2), lower than that of general anti-icing/ice-phobic surfaces (<100 kPa)43. The ice adhesion strengths on surfaces of bare steel (880 kPa), copper (304 kPa), aluminum (235 kPa), and glass (717 kPa) were all >7 times that of SEF25, while the commercial icephobic coating surface (77 kPa) was twice of SEF25. In addition, during the cyclic icing–deicing test, a negligible increase in the ice adhesion strength on SEF25 was observed under 50 icing–deicing cycles, indicating the stability of its icephobic properties (Fig. 3f). As shown in Supplementary Figs. 17, 18, in cyclic damage/healing and abrasion tests, the ice adhesion strengths on SEF25 presented a negligble increase, indicating its stable anti-icing performance (<100 kPa). As temperature decreased to −45 °C, the ice adhesion strength on SEF25 was also constant (Fig. 3g). It indicates the excellent deicing performance of SEF25 across a wide temperature range, making it promising for application in UAVs. Furthermore, the SEF25 also exhibited a good anti-frosting performance. The delay time of frosting on SEF25 (865 s) was 11.5 times than that of bare glass sheet (75 s) during the cooling process from 10 °C to −10 °C (Supplementary Fig. 19).

Fig. 3: Anti-icing and deicing performance of the encapsulation layer.
figure 3

a Proposed anti-icing and deicing mechanisms of the SEFs. b The static water contact angel of steel, SE, and SEF material surfaces. c Freezing delay process at −15 °C (the individual water droplets are 5 µL), and d TIN on the blank steel, SE, and SEF material surfaces (the individual water droplets are 1 µL). e The ice adhesion strength on bare steel, copper, aluminum, glass, commercial icephobic coatings, SE, and SEF material with different PFS contents. Statistical significance of ice nucleation temperature and adhesion strength were assessed using t tests, with a predefined level of significance (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001) (The control group in t test of b, d, e was steel). f During 50 icing/deicing cycles at −15 °C and g different freezing temperatures. The average thickness of SE and SEFs is 329 ± 13 µm.

Full size image

Self-healing and sensitive intelligent skin sensor

Fig. 4a shows the structure of a five-layer self-healing intelligent skin (SHIS) sensor. The structure mainly comprises three parts: an anti-icing encapsulation layer (SEF25), top and bottom electrode layers, and dielectric and substrate layers (SE). The SHIS sensor was fabricated using a layer-by-layer strategy with a flexible electronics printer. These layers were all based on the SE that endows them with fast self-healing properties. Meanwhile, SE also ensured the interface compatibility of multilayer structure sensor. The obtained sensor was ultrathin (1.25 mm thick, Supplementary Fig. 20). It is attributed to the supramolecule polymeric SE serving as a 3D printable ink and controlling the sensor structure via microelectronic printing. As shown in Fig. 4b, a thin substrate SE layer was initially obtained by glue dispensation, and its thickness was varied according to the pinhead sizes and ink concentrations. We prepared an SE-based electrode, a manmade electronic ink. For anti-icing skin on UAV applications, an ideal electrode should exhibit good conductivity and self-healing properties at temperatures as low as −20 °C. Therefore, the e-ink was designed to integrate SE elastomer, liquid EGaIn, and solid silver flake powers (SFP), where EGaIn serves as an electrical anchor between Ag flakes to provide conducting paths44, and SE serves as the matrix to shape the solid and self-healing conductive electrode. During the healing process, liquid EGaIn aggregated and fused quickly to recover the electrical conductivity. Meanwhile, the dynamic bonds in the SE matrix facilitated the healing process. The synergistic effect of the conductive fillers and matrix ensures the stable recovery of electrical conductivity. After systemic investigation towards conductivity, bending stability, and tensile property of the different ratios of electrode, the optimal ratio of components was optained as SE:EGaIn:SFP = 1:2:2, as shown in Supplementary Figs. 21, 22. The self-healing electrode exhibited a direct ink-writing performance, and thus could be directly printed onto the substrate and dielectric layer (Fig. 4b, c). The electrode exhibited conductivity of up to ∼7736 S/m (Fig. 4d) and an electrical self-healing efficiency of 98.0%, and ~88% was retained after cyclic damage–healing tests (Fig. 4e).

Fig. 4: Layer-by-layer fabrication of SHIS sensor and its sensing and self-healing performance.
figure 4

a Schematic of the five-layer sensor and the composition of each layer. b Layer-by-layer fabrication procedure for the sensor. c Photographs of the three-dimensional printable electrode conductor (SE:EGaIn:SFP = 1:2:2) in various shapes. Scale bar: 0.5 cm. d Conductivity of the electrodes self-healing at −20 °C. The electrodes were subjected to damage and self-healing at −20 °C for five times. Statistical significance of ice nucleation temperature and adhesion strength were assessed using t tests, with a predefined level of significance (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001) (The control group in t test was the original conductivity). e Electrical healing efficiency of the electrode under five cycles of damage–healing process at −20 °C. Error bars: standard deviation (n = 3). f Stress response of the SHIS sensor. g Photographs of SHIS damaged and healed at room temperature for 24 h. h Sensitivity of SHIS before and after self-healing at room temperature. i Sensitivity of SHIS that were self-healed or placed at −20 °C for 10 days.

Full size image

Based on the above printable five layers, a capacitive sensor (SHIS) was prepared. The capacitance of SHIS increased with increasing extra stress (Fig. 4f),attributed to the decrease in the distance between the two electrodes. This pressure-capacitance relationship served as the basis for the response of the SHIS sensor during the frosting or icing process onto the surfaces of UAVs. The pressure induced by frost and ice accumulation could reduce the electrode spacing to change capacitance of the SHIS sensor. The signal acquired by SHIS sensor experienced less interference during transmission (the signal-to-noise ratio was equal to 29 dB at 9.8 kPa), resulting in a high-quality data transmission (Supplementary Fig. 23). The sensitivity of SHIS also exhibited appreciable stability in cyclic tests under 1.0 kPa. After being placed at ultra-low temperature (−40 °C, 71% RH) for 7 days, it also exhibited a good stability. (Supplementary Fig. 24).

Furthermore, the self-healing performance of the sensor was systemically investigated. A scratch on the SHIS surface was almost invisible after 2 h healing at room temperature and 24 h healing at both −20 °C and −40 °C (Supplementary Fig. 25), and the broken SHIS could withstand a certain degree of tension after 24 h of healing, indicating the outstanding self-healing performance of the skin at room temperature (Fig. 4g). After healing, the sensitivity and sensing performance of the sensor were almost completely recovered, as shown in Fig. 4h and Supplementary Fig. 26. Notably, the sensitivity and self-healing properties of SHIS at −20 °C were evaluated. The average sensitivity of SHIS under tiny pressure (0–2.0 kPa) was 0.029 kPa−1, and that in the pressure range of 2.0–10.0 kPa was 0.019 kPa−1 (Fig. 4i). There was no significant difference between the sensitivity of SHIS at −20 °C for 10 days and room temperature, indicating the stability of the sensitive sensing property even at low temperatures (Supplementary Fig. 27). After 6 h of healing of the damaged sensor at −20 °C, its sensing performance was almost completely recovered. After 24 h of healing, the whole skin sensor exhibited the recovered stretchability, indicating its low-temperature-tolerant self-healing properties (Supplementary Fig. 28) (Supplementary Movie 3).

Frosting and icing monitoring of SHIS on a UAV

Attaching the SHIS to the drone provides an intelligent icing monitoring system that can adjust the flight status of the drone in time (Fig. 5a, b). Considering icing or frosting on UAVs in low-temperature and high-humidity environments, we evaluated the real-time monitoring ability of the SHIS sensor in a model experiment and in-flight UAV setting investigation (Supplementary Figs. 29, 30). As shown in Supplementary Fig. 31, SHIS exhibited a rapid response time (~1.4 s) and sensing stability for a certain weight of ice blocks in the model experiment (Supplementary Fig. 32). The sensing performance of the frosting on the SHIS under 90% humidity was then evaluated (Fig. 5c). A sharp decrease in the capacitance of the sensor was observed during the cooling process (sharp decrease in temperature as represented by purple area), which can be attributed to electromagnetic interference. When the temperature was reduced to maintain a stability of −15 °C, with the extension of the freezing time, a frost layer accumulated on the SHIS surface. It induces a gradual increase in the capacitance signal, indicating the monitoring of the frosting. As shown in Fig. 5d, the capacitance increased as water dropped onto the SHIS surface. When the sensor was cooled to a stable low temperature, its capacitance was considerably higher than the initial value because of ice formation and accretion. Furthermore, when the ice was removed from the SHIS surface, the capacitance sharply decreased. These results demonstrate the effective real-time monitoring of the icing–deicing process, ensuring the safety of small flying robots. The developed SHIS was applied on a mini UAV for outdoor in-flight icing tests (Supplementary Movie 4). SHIS with a thin thickness of 1.25 mm and a light weight of ~1.16 g was tightly attached to the UAV to serve as sensing skin. According to ASTM D3359, the SE layer presented a high adhesion level (5B) on the UAV fuselage even after 5 days at −20 °C, 63% RH, probably due to the strong interaction (hydrogen and imine bonds) between SE layer and the substrate (Supplementary Fig. 33). The high adhesion level ensured the stable sensing performance of SHIS during drone flight. Whether in warm (28 °C), breezy, cold (−12 °C), or windy (~15 km/h) environments, the sensor exhibited good performance in monitoring rain or ice accretion (Fig. 5e, f). Compared to the recently-reported anti-icing/icephobic coatings, this skin sensor exhibited the superior self-healing and real-time monitoring properties at a freezing temperature (−20 °C), showing a huge potential use on the smart UAV surface (Supplementary Fig. 34).

Fig. 5: Real-time monitoring of frosting and icing on a UAV.
figure 5

a The real-time intelligent monitoring system. b Photographs of SHIS on an in-flight UAV (DJI MINI 2 SE). Capacitance signal response of SHIS during c frost accumulation, and d icing and deicing progress (at −15 °C). The orange section of the figure represented room temperature. The green section of the figure represented the start of freezing. The purple section of the figure represented a temperature of −15 °C. The SHIS’s monitoring of e simulated rain and f icing in cold (−12 °C) and windy (~15 km/h) environments.

Full size image

Discussion

In summary, a low-temperature-tolerant self-healing UAV skin was developed to achieve anti-icing, deicing and real-time in-flight icing monitoring characteristics. The skin was designed as a self-healing five-layer capacitive sensor comprising an anti-icing encapsulation layer, flexible electrode layers, and SE. All of above layers are based on multiple dynamic bond-based supramolecular polymers. The obtained skin effectively reduced the ice nucleation temperature to approximately −28.4 °C and ice adhesion strength to ~33.0 kPa, while monitoring the frosting and icing on an in-flight UAV surface. In case of suffering from mechanical injury, this skin can self-heal to fully recover its sensing performance. This work integrates anti-icing properties with intelligent monitoring capabilities into a thin skin sensor, thereby benefiting mini robots that are sensitive to weight increase.

Methods

Materials

Dihydroxyl-terminated PDMS (OH-PDMS-OH, Mw = 5600 g mol−1) was purchased from Sigma-Aldrich. Isophorone diisocyanate (IPDI) was provided by Shanghai Meryer Chemical Technology Co., Ltd. Dibutyltin dilaurate (DBTDL) was obtained from Tianjin Yuanli Chemical Engineering Co., Ltd. 4,4′-dithiodianiline (SS) was purchased from Energy Chemical. 4,4′-bis(hydroxymethyl)-2,2’-bipyridine (BNB) was purchased from Shanghai UCHEM Biological Technology Co., Ltd. Vinylmethylsiloxane-dimethylsiloxane copolymers P-(VMS-DMS) (800–1200 cSt) was purchased from Gelest, Inc. (FS) was purchased from J&K Scientific. 2,2-bimethoxy-2-phenylacetophenone (DMPA) and EGaIn were purchased from HEOWNS Biochem Technologies LLC, Tianjin, China. Silver flake was purchased from Macklin Biochemical Technology Co., Ltd. N, N′-dimethylacetamide (DMAc) was obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. THF was purchased from Tianjin Kermel Chemical Reagent Co., Ltd. All reagents were used as received without further purification.

Synthesis of self-healing polymer (SE)

The synthesis procedure has been published in our previous work. Firstly, OH-PDMS-OH (11.2 g, 2 mmol) was added in a 50 mL flask and then stirred at 100 °C under vacuum for 1 h. This step was used to remove the moisture. After cooling to 70 °C, the hybrid solution of IPDI (0.9782 g, 4 mmol) and DBTDL (0.05 g) in DMAc (8 mL) was subsequently added into above flask and then stirred for 3 h under the nitrogen atmosphere. Afterwards, the hybrid solution of SS (0.2484 g, 1 mmol) and BNB (0.2162 g, 1 mmol) in DMAc (2.4 mL) was added into the flask and stirred for another 3 h. After the reaction, the product was put in a vacuum oven at 90 °C for 24 h to remove the residual solvent.

Synthesis of fluoropolymer (PFS)

P-(VMS-DMS) (5 g, 0.18 mmol) and DMPA (0.33 g, 1.28 mmol) were dissolved in 15 mL THF in a 25 mL flask, which was equipped with a magnetic stirrer under the nitrogen atmosphere. Then, FS (3.078 g, 8.09 mmol) was added to the above solution, and the resultant solution was kept stirring under ultraviolet light for 6 h to obtain the P-(VMS-DMS)-FS (PFS). The synthesis of PFS was carried out at room temperature.

Synthesis of self-healing anti-icing encapsulation layer

First, PFS (0.5 g) was dissolved in THF (3 mL). After stirring for 20 min, SE (2 g) was added to the solution and the hybrid solution was stirred for 3 h. Then, the obtained mixture was coated onto the polished steel and then cured at room temperature for 24 h. Four materials were obtained by the same procedure, named SEF5, SEF15, SEF25, and SEF35, where the number represented the weight ratio (%) of FS to SE.

General characterization

The chemical structure of SE and PFS was characterized via various measurements, including H NMR,19F NMR (Bruker, Avance 400 MHz), and FTIR spectra (Bruker, Optics). The morphologies of SE and SEF were characterized by SEM (FEI, Apreo S LoVac), which is also equipped with the EDS for elemental analysis and AFM (Bruker, Dimension icon).

Mechanical properties

Adhesion strength test

According to ASTM D3359, a knife was used to create regular grids with an interval of 1 mm on the SE. Then, the tape was tightly attached to the coated surface and then peeled off. The intactness of the SE is observed.

Sandpaper abrasion

The 1000-mesh sandpaper was placed horizontally on the table and the skin surface was contacted with the sandpaper. A 50 g weight was placed on the sample (2.25 cm2 area, 4.4 kPa pressure), to which a horizontal force was applied, resulting in the movement at a rate of 1 cm/s. The corresponding ice adhesion strength were measured for each 20 cm grinding distance, while the total grinding distance was 1 m.

Mechanical self-healing properties

Uniaxial tensile - stress tests were carried out by employing a Shenzhen Suns Electronic Universal Testing Machine furnished with a 500 N load cell. The notched specimens were secured in two clamps with a pre-determined distance of ~20 mm. Subsequently, they were stretched at a deformation rate of 20 mm/min until a complete fracture occurred at the notch.

The tensile fatigue properties

The cyclic stress–strain tests were performed at a tensile rate of 20 mm/min and a recovery rate of 20 mm/min, with a fixed strain level of 30%.

The rheological properties

The rheological behavior was measured on a DHR-2 rheometer. Frequency and temperature sweeps were performed on circular samples with a diameter of 2 cm using 2 mm parallel plates. Frequency sweeps in the range of 0.1–100 Hz were measured at a strain of 0.1% at room temperature (25 °C). Temperature sweeps were performed from −20 °C to 80 °C at a frequency of 1 Hz.

Anti-icing performance

Static anti-icing test

The water droplet (5 µL) was placed atop the skin surface (−15 °C) in a chamber with nitrogen gas. The time from the beginning of contact with the SEF surface to the emergence of the small sharp tips of the water droplet was defined as Td, which was obtained on the cooling stage coupled with a high-speed CCD camera. The experiment was repeated three times for each type of skin to obtain an average value.

Dynamic anti-icing test

Cold water droplets were continuously dropped from 10 cm height onto the 30° tilted SEF25 in the −20 °C and 72% RH environment. The processes were recorded by a phone to measure the water freezing time.

heterogeneous nucleation temperature test

The SEF25 was fixed on the cryo-stage and a water droplet (1 µL) was placed atop the SEF25 surface. The cryo-stage was a sealed cell and the cooling rate was set to 5 °C min−1. The ambient relative humidity was ~44%. The temperature at which the transparency of water droplets suddenly changed was defined as TIN, which was measured by an optical microscope equipped with a cryo-stage. Notably, the freezing had not started when the light spots appeared on the water droplets. The ice nucleation temperature was tested three times at different positions to obtain an average value.

Anti-frosting performance

The samples in 2 × 2 cm2 were placed on a cooling stage that was controlled at −10 °C, relative humidity of 69%. The time was recorded during the stage temperature cooled to −10 °C from 10 °C.

Deicing performance

The ice adhesion strength was measured by a homemade setup, which consists of a cooling stage, an XY motion stage, and a force transducer (Supplementary Fig. 13). Firstly, the skins were fixed on the cooling stage and the bottomless cuvettes (1 cm in diameter, 450 µL water) were placed on the skins. To ensure the total ice freezing, the system was maintained at the testing temperature for 4 h. Meanwhile, the nitrogen gas was purged into the glass box to cover the cooling stage to minimize the effect of frost formation, and the relative humidity was 28%. After 4 h freezing, the force transducer was driven toward the cuvette with a speed of 0.1 mm s−1. The distance between the force probe and skin surface was set as close as possible (~1 mm) to avoid additional torque. The ice adhesion strength was calculated according to the following equation.

$$\tau =\frac{{F}_{\max }}{S}$$
(1)

where F is the maximum force, and S is the contact area between the ice and the skin surface. Three samples were investigated to obtain an average value.

Electrical properties testing of electrodes

The electrical resistances of the original and self-healed electrodes were tested using a LCR meter (Tonghui Electronics Co., Changzhou) at −20 °C.

The conductivity (σ, S/m) was calculated by the following equations:

$$\sigma =L/(R\times A)$$
(2)

where L (m) is the electrode length; R (Ω) is the electrical resistance of the electrode; A (m2) is the sectional area of the electrode.

Preparation of SHIS by flexible electronics printer

Ink preparation

First, SE (1 g), EGaIn (2 g), and silver flake (2 g) were dissolved in THF (1.5 mL). The product was stirred at room temperature for 12 h and the e-ink was obtained. Then, SE (2 g) or SEF (2 g) were dissolved in THF (4 mL), separately. After stirring at room temperature for 30 min, the self-healing ink for the substrate layer and dielectric layer or encapsulation layer was obtained.

Fabrication of the SHIS

A flexible electronics printer was used to fabricate the SHIS.

The printing mode was selected as the dispensing mode; the needle size was 250 μm; the extrusion pressure of the ink was generally set to 70–100 kPa; and the printing speed was 1 mm/s. First, the substrate SE was printed on the PDMS-treated glass sheet and dried at room temperature for 12 h. Then the down electrode was printed on the substrate layer and dried at room temperature for 6 h. The SE dielectric layer was then printed and dried at room temperature for 12 h. The up electrode was printed on the dielectric layer and dried at room temperature for 6 h. Finally, the SEF encapsulation layer is printed and dried at room temperature for 12 h to obtain SHIS.

Self-healing performance

The self-healing performance of the SHIS encapsulation layer (SEF25) was first tested by applying scratches on the surface of SEF25 with a blade and placing it at room temperature for 2 h, and then the optical morphology before and after healing was recorded by a microscope. Then, the self-healing performance of the entire SHIS device was tested. The SHIS device was cut into two pieces and put together to heal at room temperature for 24 h.

Sensing performance of the SHIS

An LCR meter (TH2830) was used to measure the capacitance change of the SHIS. The sensitivity, expressed as \(\frac{\Delta C/{C}_{0}}{\Delta p}\), is usually used to evaluate the performance of the pressure sensor, where C0 and C are the initial capacitance and capacitance under applied pressure, respectively.

The signal-to-ratio (SNR) values were estimated via the following equation1,2:

$${SNR}\,\left({dB}\right)=10\times \log \frac{{C}_{{rmsl}}}{{c}_{{rmsn}}}$$

where \({C}_{{rmsl}}\) was the root-mean-square value of the filtered signal, and \({c}_{{rmsn}}\) was the root-mean-square value of the noisy signal.

Frost and ice sensing performance of the SHIS

The ice sensing test

1 mL of deionized water was injected into an acrylic cylinder with an inner diameter of 1 cm. After a cooling process, water freezes into icicles. Then the SHIS was placed on the cryostage to maintain the temperature of the SHIS surface at about −15 °C. The electrode of the SHIS was connected to the LCR meter. After the capacitance signal was stabilized, the icicle was placed on the SHIS surface to observe the capacitance change.

The frost sensing test

We designed a small sample chamber with ambient humidity controlled at 90%. At the bottom of the sample chamber is the cryostage, which can control the temperature of the chamber. The SHIS was placed on the cryostage and connected to the LCR meter. The thermocouple probe was used to record the temperature change of the SHIS surface, so it was intimately attached to the surface of the SHIS. During the test, SHIS was stabilized at room temperature for 10 min and began to cool down. When the surface temperature of SHIS dropped to about −20 °C, it was kept for about 90 min, waiting for the accumulation of frost layer, and the signal response during the whole test process was recorded by LCR meter.

The water icing process sensing

The SHIS was placed on the cryostage and connected to the LCR meter. The thermocouple probe was intimately attached to the surface of the SHIS. During the test, SHIS was stabilized at room temperature for 10 min. Then, an acrylic cylinder with an inner diameter of 1 cm was placed on the SHIS surface. 1 mL of deionized water was injected into an acrylic cylinder quickly. After the capacitance signal was stabilized, it began to cool down until the SHIS surface temperature was maintained at about −15 °C. The water in the acrylic cylinder was slowly frozen at −15 °C. After about 70 min, the icicle on the surface of SHIS was removed by external force. The capacitance change in the whole process is recorded by an LCR meter.

The outdoor in-flight icing tests

The UAV (DJI MINI 2 SE) was purchased from SZ DJI Technology Co., Ltd. The SHIS was mounted onto the fuselage of the UAV.