Ice formation and accretion on solid surfaces have caused tremendous economic losses and dangerous conditions [1,2]. Passive anti-icing materials to reduce these problems are effective and consume no energy [3,4]. Liquid-repellent surfaces, either generally superhydrophobic (SH) or slippery-liquid-infused porous surfaces (SLIPS), are generally considered to be ideal passive anti-icing surfaces due to one or more of the following capabilities [[5], [6], [7]]: (1) rejection of incoming liquids that can potentially become ice, before the liquids freeze (2) low ice adhesion strength (S_IA) [8] (in general < 100 kPa [9]) so that natural forces (e.g., wind, vibration) can shed it, (3) delay of ice nucleation. [[10], [11], [12]]. Extremely low adhesion of water to the surface minimizes the contact time of the impinging drop (e.g., freezing rain) such that water can be shed before it freezes [7,13,14] or enables spontaneous jumping motion when condensate droplets merge. It decreases the speed of frost propagation and prevents the formation of a thick ice layer [4]. In addition, liquid-repellent surfaces generally delay ice nucleation time or lower the ice nucleation temperature by [5]: increasing the free energy barrier for ice nucleation [15], decreasing the water-solid contact area [16], and decreasing the viscosity of interfacial water [17]. However, icing is a complex process. The free-energy barrier can be reduced by many factors such as vibration or the presence of impurities, and variation in these factors can promote heterogeneous ice nucleation [18,19]. For these reasons, icing seems to be inevitable despite outstanding passive anti-icing by a liquid-repellent surface [1]. Therefore, passive anti-icing that exploits the properties of liquid-repellent surfaces has been unsatisfied, so elimination of ice accretion remains a challenge in the real world [20].

Liquid-repellent surfaces that also provide active heating could be more powerful to combat icing problems by combining passive and active anti-icing approaches synergistically [21]. With the liquid-repellence, the active heating ability of the surface can thaw the formed ice layer, and thereby recover the passive anti-icing property (i.e., liquid repellence) that was nullified by the formed ice covering the surface. Furthermore, during defrosting by heating, liquid-repellence allows melting water to form a slurry that easily slides off. Compared to non-liquid repellent surfaces, this process sheds water more quickly with less energy consumption, and leaves less melted water that can refreeze on the surface [22]. Recently, several hybrid passive-plus-active anti-icing surfaces that combine liquid repellence with active heating have been reported [20,23,24]. They usually involve electrically-conductive liquid-repellence, whereby the liquid repellent surface exploits an electrothermal or photothermal effect to generate heat that melts accreted ice [[23], [24], [25], [26]]. However, the reported surfaces have poor durability of liquid-repellence, and are fabricated by complex processes, which hinder their mass production and wide adoption in anti-icing applications [14,27].

Poor humidity tolerance of SH surfaces [28,29] or the drained lubricant in a SLIPS by icing/deicing cycles [30] or by volatilization [31,32,5] are the major impediments to their practical operation. These problems induce gradual loss of liquid-repellence locally or over the entire area. Several attempts to produce repairing/self-healing functionalities have made progress to extend the longevity of liquid repellence [31,[33], [34], [35], [36]]; examples from, Urata et al. [31], Zhu et al. [8], and Wang et al. [35] include organogels with regenerating liquid-repellent surfaces that can be replenished from inner matrices when damaged, and. Yan et al. [37] introduced a strategy for abrupt replenishment of lubricating layer onto locally depleted zone through well-designed internal pore structure in conjunction with surface structures. These strategies are desirable to extend the service lifetime. Other desirable attributes for practical anti-icing materials include multi-functionality, robustness against mechanical deformation, stability of liquid repellence against abrasion and scratching, and ease of fabrication with scaling availability [5,6].

In this paper, we report a highly durable liquid-repellent surface that has both passive anti-icing property and active de-icing capability. This surface is derived from a slippery liquid-repellent elastic conductor (SLEC) that has been previously pioneered by our group by mixing polydimethylsiloxane (PDMS) and oil with multi-walled carbon nanotubes (MWNTs) [38]. The developed SLEC is highly liquid-repellent (roll-off angle (ROA) < 5°) and can generate heat in response to electrical or photonic stimulation. These features of SLEC induce low S_IA, retardation of ice nucleation, removal of contacting liquids, and de-icing, which are necessary in ideal anti-icing material [Fig. 1]. Above all, this surface has highly durable liquid repellence because the surface lubricant is self-healable and its lubricant supply can be easily replenished, so it has long service time. The SLEC has additional advantages such as simple fabrication process with easy scaling, low cost, high flexibility and stretchability, and stability of liquid-repellence against external stress. These hybrid anti-icing functions with practical advantages of the SLEC give it great application potential.

Silicone oil was purchased from Sigma Aldrich (viscosity μ =500 cSt) or SFX KOREA (μ =1000 cSt). MWNTs (outer diameter 10-30 nm, inner diameter 5-10 nm, length 10-30 μm) were purchased from Nanostructured and amorphous materials Inc. PDMS (prepolymer and curing agent) (Sylgard 184) was purchased from K1 solution. Calcium sulfate powder (Drierite) and stainless steel 304 were obtained from Han Lab. And, Plastikote spray was provided by Korea Atomic Energy Research Institute.

The SLECs are composites composed of PDMS, MWNT and silicone oil, and were obtained by mixing 1 g of PDMS prepolymer, 0.1 g of curing agent, silicone oil, and MWNTs in a planetary mixer (Thinky ARE-310) for 8 min at 2000 rpm. The resulting homogeneously-dispersed paste was placed onto glass and degassed in a desiccator for 20 min. The degassed paste was pressed manually between two glass sheets that were separated by 500 μm using a spacer to achieve SLEC layer that had that thickness. Then a pressure of ~20 N was applied using a calibration weight. The glass-paste-glass combination was cured at 90 ℃ on a hot plate for 3 h. Subsequently, the cured SLECs were baked in a convection oven at 70 ℃ to induce self-lubrication (syneresis).

Composites were fabricated with different mixing ratios; samples were named to express the weight percent of each component (e.g., M7L90 refers to composite prepared by mixing 7 wt% MWNTs (M) and 90 wt% lubricant (L) of PDMS weight. SLECs with names that start with M11,M14, and M0 were made with silicone oil that had μ =500 cSt, and SLECS with names that start with M12 and M15 were made silicone oil that had μ =1000 cSt. All anti-icing experiments used the M11 and M14 series. The self-lubricating organogels (SLUGs) (i.e., M0L90, M0L60) are composites of PDMS/silicone oil [31].

For the comparison to demonstrate the advantages of the SLEC, two SH surfaces were prepared. A black SH surface (SHPo) was achieved by spray-coating Plastikote onto a Si wafer, then drying in air for 60 min, then heating in an oven at 340 ℃. The resulting film was 12 μm thick. Another SH surface (named SS) was achieved on stainless steel by electrochemical etching [39]. Detailed wetting information (i.e., syneresis behavior, contact angle) for samples tested here are presented in Section 1 of the supplementary information (SI).

2.3.1. Experiment to confirm the retention of liquid repellence

SLUG, SLEC, and SHPo liquid-repellent surfaces were prepared for use in two tests to prove the retention of liquid repellence. To test under a condition in which condensation occurred, SLEC with M11L90, SLUG (M0L90), and SHPo were put on a Peltier plate for 10 min at temperature T_P = 0 ℃ in ambient temperature T_A = 23 ℃ and relative humidity (RH) ~ 40 %. After condensation for 10 min., 10- μL water droplets were deposited to check whether the surface had retained liquid repellence.

For another test after defrosting, SLEC with M14L90, and SHPo surfaces were used. They were frosted on the Peltier plate at T_P = -11 ℃ for ~ 25 min, at T_A = 23 ℃ and RH ~ 60 %. After a frost layer had covered the entire surface, the Peltier plate was turned off, so the frost layer melted. Then 10-μL water droplets were deposited to test the liquid repellence.

2.3.2. Experiments for delay of ice nucleation

Tests of ice-nucleation effect on onset of freezing and frost propagation were conducted on the SLECs with M11L90, M14L90 and SHPo. Samples with water droplet of 10 u L was placed on a Peltier plate with a setting temperature of -5 ℃. The entire process was video-recorded to allow observation of the onset of freezing, and the frost propagation process.

2.3.3. Experiment for ice adhesion strength

The test to measure S_IA used a generally used method [11,26]. The experimental setup was composed of a lab-built cooling system equipped with a Peltier plate, a load cell (1 kg, Bongshinloadcell) clamped onto an XY motorized stage. The sample with a plastic cuvette (1 cm × 1 cm cross section) filled with water was put on the Peltier plate. A cuvette filled with deionized water (1 cm × 1 cm × 2 cm) was placed on the sample and cooled on the Peltier plate at T_P = -15 °C for > 2 h, after which the water in the cuvette had fully frozen. After the water froze, probe connected to the motorized stage was moved toward the bottom of the cuvette column at ~83 μm/s; the speed was controlled by Labview software. When the probe contacted the cuvette, the force that the probe exerted increased until the interface between substrate and the ice fractured; at that point, the force suddenly dropped. S_IA was obtained from the peak force. This test was repeated > 3 times. The measured values were normalized by the cross-sectional area of the cuvette to yield S_IA.

2.4.1. General heat generation capability

Temperature profiles during heat generation in response to electronic or photonic stimulation were obtained using a thermal (FLIR) camera (Sections 5 and 6 in SI).

2.4.2. Defrosting by electro-thermal effect

To demonstrate the effect of thermogenesis under electronic stimulation, two experiments were performed: (1) application of voltage before ice formation (2) application of voltage after ice formation. Both experiments used the M14L90 SLEC and the M0L90 SLUG. For experiment (1), the SLEC and SLUG were placed on a Peltier plate at T_P = -7 ℃, T_A = 23 ℃ and RH ~ 58 %. Voltage (25 V) was applied from the onset. The samples were monitored for 30 min, then 10-μL water droplets were dropped onto the SLEC to confirm its liquid repellence.

For experiment (2), the samples were frosted for ~ 20 min by being placed on a Peltier plate at T_P = -11 ℃, T_A ~23 ℃ and RH ~ 60 %. When the entire surfaces were covered by ice, voltage (25 V) was applied to test the de-icing function.

2.4.3. Defrosting by photo-thermal effect

To demonstrate the effect of thermogenesis under photonic stimulation, a laser (532 nm, 500 mW/cm²) was used to illuminate the ice layers. SLEC (M14L60) and SLUG (M0L60) were held on a Peltier plate at T_P = -3 °C at T_A =25 °C and RH ~ 64 %. After the ice layer had formed for > 40 min, they were irradiated at one spot for 30 s. Optical data were captured using an ordinary camera; temperature information was monitored using an IR camera.

2.5.1. General heat-generation capability

To quantify self-healable liquid repellence, syneresis behavior was induced by electric or photonic stimulation, and the efficacy of the replenishment process was demonstrated. These tests used the same electrical and photonic systems that had been used to quantify active deicing. The lubricant was removed from the surface, then electrical stimulation was given by applying voltages of 13 or 20 V 500 min, or photonic stimulation was applied using the laser (532 nm, 500 mW/cm²) for 1 h.

2.5.2. Replenishment

Replenishment was achieved simply by dropping lubricant (oil) onto the SLECs for 20 h. Then syneresis was induced at 70 °C for 24 h.

To confirm the self-cleaning ability, calcium sulfate was spread on SLEC of M15L90 and cleaning by rolling water was observed. For the durability of liquid-repellence against abrasion, a 70-cm abrasion was applied to an M11L90 SLEC by dragging it on 2000-mesh sandpaper. ROA was measured every 10 cm along the abrasion (Smartdrop, Femtobiomed). After abrasion over 70 cm, it was stored in 70 °C for syneresis for > 24 h. Lastly, Durability confirmation of SLECs under cyclic loading of stretching (ε = 35 %) and bending (ε = 60 %) were conducted. The resistance change of M15L50 (1 cm x 1 cm x500 μm) during cycles was measured by a digital multimeter (34401A, Agilent technologies Inc.). Both ends of sample were clamped on a motorized XY stage with a gap of 2 cm where one end of sample was moved into the other end for bending, or was moved away from the other end for stretching up to given strain.

The SLEC must retain their liquid repellence in harsh conditions. This ability is important because after anti-icing surfaces lose their liquid repellence, water that is deposited on them by condensation or rain can freeze, and the resulting ice can be difficult to remove. These changes nullify all of the advantages of the icephobic properties of liquid-repellent surfaces. Liquid-repellent surfaces must overcome these complications to increase anti-icing activity.

We performed tests under condensation conditions and de-icing conditions, which are plausible situations that cause ice accretion that has caused trouble for many liquid-repellent surfaces. To quantify the retention of liquid repellence under condensation, SLUG, SLEC, and SHPo liquid-repellent surfaces were held below the dew point (Wetting information at room temperature is in SI). While the surfaces went through a temperature drop to 0 ℃ for 10 min, condensed liquids covered them entirely (Supplementary video S1). Impacting droplets slid off the SLUG and SLEC, but stuck to SHPo (Fig. 2a, and Supplementary video S1). This result for SHPo occurs because it loses its liquid repellence when condensed liquids form within its surface structures [4,40].

A similar tendency was observed in another de-icing test (Fig. 2b-e). After the ice layer was melted by a de-icing process, the resulting water droplets rolled of the SLEC, but remained on the SHPo. Subsequently, water droplets that struck the surfaces could only slide off the SLEC. They did not slide off the SHPo, because water that permeates its structures can pin them.

These results demonstrate that SLEC surfaces have better anti-icing property than SH surfaces. We believe that use of SLIPS series (including SLECs), which are relatively insensitive to condensed water, can be an effective passive anti-icing strategy. These experiments also demonstrated that the SLEC can retain its liquid repellence.

To further explore the passive anti-icing property of SLEC, the ice nucleation effectiveness was characterized (Fig. 2f-h). For the test, time of onset of freezing and frost coverage were measured. Water droplets of 10 μL on M11L90, M14L90 and SHPo were cooled down to -5 °C. Droplets froze at 115 s on the SHPo first, at 318 s on M14L90, and at 413 s on M11L90. From the perspective of icephobicity, SLIPS benefits from a reduction in the potential nucleation sites because it provides chemically homogeneous interface and can presumably get rid of potential nucleation sites [1,41,42]. Apart from this, SLIPS is less vulnerable to the humidity. In contrast, SHP surfaces are generally prone to localized coating peeling, cracks, scratches by external stress according to the fabrication process or materials [30]. Such defects provide initiation sites for ice nucleation. We believe these characteristics explain the longer icing delay time on SLECs than on SHPo. The prepared SHPo was made on the silicon wafer by spray silicone based material coating which can induce the delamination at interface due to modulus mismatch by external stress [43] or low robustness such as low hardness. These might play the role of initiation sites. Apart from the comparison between SHPo and SLEC, M11L90 shows longer onset of freezing time than M14L90. It might take advantage of the thicker lubricant film and lower thermal conductivity of M11L90 than M14L90. Note that water droplet floats on the lubricant (Section 3 in SI).

Similarly, a test for frost propagation shows benefits of lubricant film on anti-icing. After onset of frost, the coverage percentage in frost area was measured over time. Frost covered the surface of M5.8L0 for 61 s, of M14L90 for 460 min, and of M11L90 for 774 min. Frost nucleates at defects or edges, and proceeds to propagate by forming an inter-bridge (e.g., ice dendrites) among peripheral condensates [4]. The frost is restrained to grow when condensate droplets are cloaked in lubricant [30] so, we presume that condensate droplets cloaked in thin lubricant film suppress the growth of ice dendrite that was caused by vapor of neighboring condensate droplets. Also, lubricants provide low thermal conductivity and suppress the nucleation chances by removing possible nucleation sites. The SLEC forms cloaking droplets (Section 3 in SI). The effect of lubricant is also observed in the comparison between SLECs. On the oil film, frost propagates along the thin oil layer first, then the thick oil. So, frost spreads more slowly on M11L90 than M14L90. Lower speed of frost propagation on M11L90 also might be attributed to lower thermal conductivity, suppression of the nucleation sites, and thickness of the oil.

Additionally, ease of tuning in SLEC’s property offers a way to increase icing delay time and reduce frost propagation speed. The surface structuring, lower ratio of MWNTs and higher ratio of oil could be the examples. However, despite the better ice nucleation, these efforts could adversely affect the active anti-icing functions (e.g., Joule heating for defrosting). Therefore, materials should be selected to match applications.

S_IA is another important measure of passive anti-icing ability. S_IA as low as ~20 kPa can allow removal of ice by natural forces like wind and vibration [1,32]. Low S_IA can be achieved by non-interlocking of ice with a surface structure, low shear modulus, and low chemical or physical interaction between ice and substrate [[44], [45], [46]]. S_IA (Fig. 3a) were obtained for as-prepared SLECs with different MWNTs and oil contents. As-prepared SLECs had S_IA ~ 20 kPa, which is similar to a value reported for other SLIPS [33]. Compared to SHPo and SS SH surfaces (Fig. S3 in SI) and M09L0 that have high S_IA > 50 kPa, we confirm that highly-mobile lubricant guarantees weakening of the interaction between ice and the substrates, and yields promising anti-icing materials. Reported S_IA of SH surfaces are typically 50-100 kPa [1,47]; i.e., SLIPS achieve superior reduction in ice adhesion.

Among SLECs with different content of components, M11L90 had the lowest S_IA ~14 kPa, and M14L60 had the highest S_IA ~25 kPa. These results are easily predictable, because an excess of oil films greatly reduces ice adhesion (i.e., cohesive failure) [46,48]. Under the same MWNTs amounts, the higher content of oil leads to lower S_IA as a result of the lower shear modulus. Incorporation of oil reduced S_IA by ~2/3 compared to M09L0. Nevertheless, S_IA of M09L0 is still quite small, possibly as a result of the large difference in elastic modulus E_Y between ice and SLEC [49].

The result of the ice adhesion cycle did not show significant increase in S_IA even though no oil layer could be seen (Fig. 3b). This stable S_IA may be a result of weak interaction and the viscoelastic property of the SLEC. SLEC consists of PDMS and oil, which infiltrates the crosslinked network of the PDMS. At the outermost surface of the SLEC, oil is exposed to the environment although oil is not visible. When water freezes on the SLEC, the resulting ice contacts the oil beneath it. The low E_Y of the SLEC may also explain the consistently low S_IA. SLEC has a lower E_Y (i.e., is softer) than unadulterated PDMS. The large difference in E_Y induces mismatch in strain when force is applied to remove the ice. The stress induced by applied force to remove the ice is not evenly distributed in the SLEC, but builds up at the interface which fractures [49]. As a result, the low interaction between two results in low S_IA. This viscoelastic property and the oil components of the SLEC are also beneficial in outdoor applications even when oil films are lost. These results demonstrate that the SLEC has desirably low S_IA.

At temperature < 0 °C, water could eventually freeze on a surface, even if it has outstanding passive anti-icing properties. Once ice has accreted on a surface, the phase change of subsequent water to ice accelerates as a result of heterogeneous nucleation, and this acceleration aggravates the icing problem. Therefore, a promising anti-icing surfaces must remove the accreted ice on the surface as well recover passive anti-icing capability quickly and reliably.

The SLEC generates heat by electro-thermogenesis when it is stimulated electrically. The MWNTs in the SLEC form conductive pathways that undergo Joule heating when an electrical current passes through them. The maximum temperature that can be reached can be tuned by adjusting the amount of each component in the SLEC, and by adjusting the applied voltage (ref [38]; sections 5 and 6 in SI). To combat icing, electro-thermogenesis could be exploited to prevent ice formation, and to remove an ice layer after it has formed.

To quantify the ability of electro-thermogenesis to prevent water from freezing, SLUG and SLEC were placed on a Peltier plate at -7 ℃, then voltage of 25 V applied to them until their temperature reached 17 ℃. The SLUG does not contain MWNTs, so it did not generate heat; as a result, ice accreted in 20 min (Fig. 4a and b, and Supplementary video 2). The SLEC was warmed by Joule heating so ice did not form; only condensate droplets formed (Fig. 4a and b, and Supplementary video S2). Moreover, droplets could still slide when they fell onto a surface that was covered by condensate droplets (Fig. 4c). This sustainable liquid-repellence of SLEC with icing/deicing cycle and condensation is an advantageous anti-icing property for outdoor applications. Many SH surfaces lose liquid repellence when liquids condense on them (Fig. 2). These results demonstrate that the electro-thermogenesis capability of the SLEC in response to voltage effectively prevents ice formation while maintaining liquid repellence.

After ice has accreted, it should be removed to avoid exacerbation of the icing problem. In this situation, electro-thermogenesis can stop worsening of the problem. After voltage was applied, ice remained on the SLUG surface (Fig. 4d), but the entire ice surface melted from the SLEC as a result of Joule heating (Fig. 4d). The melting ice can slide off the surface. This electro-themogenesis can act synergistically with the liquid-repellent property, because without complete melting, the ice-slurry can easily slide off the surface due to its low adhesion to water [22]. These results demonstrate that electro-thermogenesis boosted by liquid repellence can amplify the anti-icing effect in real environments.

The SLEC can also undergo photo-thermogenesis, which is heat generation in response to photonic stimulation. This heat generation is also induced by the MWNTs, so the power of the heat-generation can be tuned by the adjusting the content of MWNTs in the material, and by selecting an appropriate laser wavelength.

Photo-thermogenesis of the SLEC offers a remarkable alternative when electro-thermogenesis is not available to melt the ice. This substitutional choice provides flexibility for application. To confirm the capability of defrosting by the photo-thermal effect, a laser with wavelength of 532 nm was beamed at ice layer on the SLEC surface. The characteristics of thermogenesis in response to photonic energy in this system were investigated (Fig. S5 in SI). Beams with power density of 500 mW/cm² irradiated the ice-layered samples. The SLUG does not have a photo-thermogenesis effect, so the ice was not affected. In contrast, on the SLEC the ice melted in the spots that were irradiated by the laser (Fig. 5, and Supplementary video S3), because the light is absorbed by MWNTs and increases their temperature. The temperature of the SLEC increased from -3 °C to 25 °C after irradiation for 30 s (Fig. 5). Photo-thermogenesis by the SLEC can offer the significant benefit of eliminating ice buildup.

The self-healing property of SLIPS ensures more stable liquid repellence than SH surfaces can achieve [50], but SLIPS still lack durability for outdoor applications [32]. The poor durability is mainly a result of depletion of the lubricant by processes such as volatilization, leakage, and migration. Especially, cloaking of the droplet in oil accelerates the depletion of lubricant (Section 3, 4 in SI). Water is cloaked on the SLEC. For these reasons, to attain long-lasting liquid repellence, the SLEC should have self-sustainable liquid repellence with minimized dependence on the environmental conditions. Self-lubricating organogels can show a special thermo-responsive syneresis behavior that can recover their liquid repellence [31,38]. Nevertheless, in reality, an environment that can allow the system to control the surrounding temperature is rare. In addition, although liquid-repellence can be recovered by syneresis, the amount of oil entrapped in the crosslinked PDMS network is finite, so the syneresis rate gradually decreases, and eventually fails to maintain sufficient liquid repellence. Therefore, an ice-repellent surface must have sustainable syneresis ability.

From these perspectives, our SLEC has superior advantages for better durability than SLIPS without self-lubrication or refill function. The thermogenesis in the SLEC in response to stimulation by electricity or light allows for syneresis because of temperature dependence of the SLEC’s syneresis. Additionally, by a replenishment process, the amount of oil entrapped in the crosslinked PDMS network can be maintained above a level that ensures sustainable syneresis.

On the SLEC, syneresis occurred under electrical and photonic stimulation (Fig. 6). Syneresis by external stimulation has rarely been observed in other materials. Thermogenesis in the SLEC in response to electrical or photonic energy can make syneresis available as syneresis of the SLEC is presumably driven by the diffusion. Electronic energy with voltage applications of 13 or 20 V for 500 min imparted lubrication (Fig. 6a). The syneresis rate increased as the temperature of the SLEC increased (Fig. 6a). Syneresis reduced the ROA from ~20° to < 5° after electrical stimulation for 4 h, so liquid easily rolled off the surfaces (Fig. S6b).

The SLEC also shows photo-responsive syneresis behavior (Fig. 6b). The photonic system is useful as an alternative way to induce self-lubrication in environments where Joule heating is not feasible. This photo-responsive thermogenesis induces syneresis (Fig. 6b). M12L90 that was lubricant-wiped was irradiated for 1 h; the irradiated spot released lubricant. This combination of slipperiness and heat-generating ability under voltage or laser illumination may have practical advantages in outdoor applications.

The self-healing capability of the SLEC induced by syneresis is necessary for lifetime-long liquid repellence. The reduction of syneresis rate may accelerate when an SLEC is used outdoors, and this loss of lubrication can be fatal to anti-icing applications. However, the SLECs can be replenished by bringing them into contact with the silicone oil that is used as the lubricant (Fig. 6c). The silicone oil wicks into the crosslinked PDMS network, so the PDMS swells; the silicone oil may dissolve into the network [51]. We prepared an SLEC from which most of the oil had been consumed, and that did not release sufficient oil. Then we compared this sample to one in which the oil had been replenished by contact with a reservoir of the lubricant. The quantity of released oil differed greatly between the two samples (Fig. 6d). The sample that had been replenished released as much oil as the as-prepared sample; this result confirms that contact with the lubricant can restore the liquid repellence. The durability of liquid repellence is the most important requirement for practical anti-icing applications, so this replenishment approach is useful.

SHPo and SLIPS have been proven to be effective anti-icing surfaces, so their practicality such as ease of fabrication or durability has been considered important [5,52]. To meet these requirements, the SLEC was developed with the intention of achieving a strong practical coating strategy. First, SLEC is a PDMS-based material made by standard physical mixing process, thus the process is scalable, inexpensive, and applicable to diverse shapes (Fig. 7). Furthermore, this method also allows the material’s properties to be easily tuned simply by changing the components’ ratio, which offers versatility for a wide range of applications. We strongly believe that such simplicity and effectiveness of fabrication process for such multi-functionality has value in fields of engineering and surface modification, to name a few. Secondly, SLEC possesses promising stability against external stress such as mechanical and environmental damages. In practical applications, this stability is another important consideration to maintain the surface’s original property against external stress and to achieve longevity.

Many factors can impede the maintenance of functionality in real circumstances. Accordingly, we tested the stability of SLEC under four conditions: impurity, mechanical stress (i.e., abrasion, scratch, stretching and bending). These stresses are commonly encountered in many applications. Impurities such as dirt can contaminate the surface and reduce its functionality. Although SHP surfaces or SLIPS have self-cleaning function, it does not always remove the dirt [53]. Furthermore, self-cleaning by a rolling motion is more efficient than by a sliding motion [54,55]. An SLEC that is composed of an appropriate material can remove powdered calcium sulfate dirt by a rolling motion (Fig. 7b, Supplementary video). This result demonstrates a reliable self-cleaning ability.

Next, we confirmed stability against abrasion and scratching. Although the depletion of the lubricant due to abrasion with sandpaper increased the wettability (i.e., increased ROA), it remains low (ROA < 15°). More importantly, after self-lubrication, the surface recovered its repellence (ROA ~2°) (Fig. 7c). Similarly, a water droplet smoothly passed a line of damage that was imposed by cutting using a blade (Fig. 7d). All results demonstrate that the outer liquid layer yields high liquid repellence that is resilient to physical damage.

Further tests of stability against mechanical stress were performed under cyclic stretching and bending deformation (Fig. 7e). The SLEC composed of M15L50 shows good mechanical compliance under repetitive stretching (35 % strain) and bending deformation (60 % strain with displacement of 1.2 cm from an initial gap of 2 cm). This elastomer/nanofiller composite provides a method to fabricate materials with superior mechanical compliance comparing to the material with low strain to break ratio [56]. We believe that the simple fabrication process, durability of liquid repellence and mechanical compliance increase the practicality and reliability of the material in a wide range of real applications.

We have shown a superior anti-icing strategy (SLEC) that is capable of both passive anti-icing and active de-icing. The SLEC has a slippery surface that imparts liquid repellence, and undergoes thermogenesis in response to electric or photonic stimulation. The liquid repellence of the SLEC enables water droplets to slide off a wet surface that frequently encounters conditions that cause condensation and defrosting. The lubricant can retard ice nucleation. The slippery oil of the material and undermatching in modulus between SLEC and PDMS induce low ice adhesion. The electrical or photonic stimulation cause the SLEC to heat up, and this process can help prevent ice accretion or melt an accreted ice layer on the SLEC. The liquid repellence of the SLEC is sustainable by self-lubrication behaviour, which is boosted by thermogenesis with the help of an electronic/photonic system. Oil that is lost during service can be replenished by simple immersion. As a result, SLEC shows exceptional durability of the anti-icing property and extended service time. SLEC has many practical advantages such as stability against mechanical stress, and a simple and scalable fabrication process. These passive and active anti-icing functionalities in conjunction with practical advantages such as durable liquid repellence offer strong potential as an anti-icing tool in a general icing environment.

The authors declare no competing financial interests.

We thank Dr. Junghwan Byun, Dr. Hyejeong Kim, Dr. Joonphil Choi, Dr. Changyong Yim and MNT for valuable comments, supports, and discussion.

Supplementary material related to this article can be found, inthe online version, at doi:https://doi.org/10.1016/j.jmst.2020.02.022.