Thickness-invariant, durable and strongly adherent icephobic coatings via dynamic polyborosiloxane crosslinking Authors: Kun-Woo Nam, Sung-Hoon Park Published: 2026 ABSTRACT Ice accumulation poses severe hazards to transportation, energy systems, and infrastructure. Icephobic coatings, defined as surfaces with ice adhesion below 100 kPa, often suffer from poor durability, weak adhesion, or thickness dependence that limits scalability. Here, we harness the stress-dissipative nature of dynamic polyborosiloxane networks to overcome these limitations. However, their role in ice detachment through interfacial stress dissipation remains unclear. We design a robust, adherent coating that overcomes conventional thickness–adhesion scaling. The polyborosiloxane network, formed by reaction of boric acid with hydroxy-terminated polydimethylsiloxane, incorporates reversible B–O–Si bonds that reform under stress, dissipating energy while preserving structural integrity. Blending polyborosiloxane with polydimethylsiloxane and silicone oil yields a two-step ice release: interfacial sliding by chain mobility, then stress relaxation through bond dissociation. Consequently, the resulting coating exhibits low ice adhesion ( ~ 6.4 kPa), thickness-invariant performance down to 2 µm, and strong bonding ( ~ 9× higher lap-shear strength than polydimethylsiloxane), overcoming the durability–adhesion tradeoff. The coating retains performance under repeated icing/de-icing, abrasion, and extreme pH. This work establishes dynamic polyborosiloxane networks as a scalable design principle for durable icephobic surfaces. Introduction Ice accretion during cold weather presents substantial challenges in both industrial and everyday environments1–5. For instance, ice formation on wind turbine blades reduces power generation efficiency and induces structural damage6,7, while ice accumulation in cryogenic pipelines may cause malfunction or rupture8,9. In aerospace, ice on aircraft fuselages or wings increases weight and decreases lift and flight stability, potentially leading to catastrophic accidents10,11. In addition, due to the low temperature of hydrogen fuel, freezing occurs in the filling nozzle for hydrogen vehicles12,13. In particular, the adhesion strength for bare metal surfaces often exceeds several hundred kPa, and in some cases approaches 1000 kPa, requiring substantial shear force for effective ice removal14,15. Traditional strategies are generally categorized as anti-icing and de-icing. Anti-icing surfaces, such as superhydrophobic coatings, delay ice nucleation and prevent wetting by lowering surface energy and contact area4,16–18. However, these approaches often rely on fragile micro/nano structures and suffer from poor mechanical robustness, limiting large-area applications14,19,20. De-icing methods, including heating elements and electrothermal stimuli, can effectively remove ice but are energy-intensive and require additional infrastructure21,22. As an alternative, icephobic coatings have been developed to minimize ice adhesion and allow easy removal after full ice formation. The ice adhesion strength (τice) is typically defined as the force required to detach ice per unit area, and surfaces with τice below 100 kPa are generally categorized as icephobic14,23. Various designs have been explored to reduce ice adhesion, including soft elastomeric coatings, slippery liquid-infused porous surfaces (SLIPS), and low-surface-energy polymer coatings3,24–27. Elastomeric coatings exploit elastic mismatch with ice to promote interfacial cavitation, enabling partial detachment under shear; however, their performance is often strongly influenced by coating thickness28. SLIPS minimize interfacial friction by introducing a lubricating liquid layer, allowing smooth ice removal, although their long-term stability can be limited by lubricant depletion or environmental exposure26,29. Low-surface-energy materials, such as fluorinated polymers30–33 weaken ice–substrate interactions, but their effectiveness under repeated mechanical or environmental stress remains limited. From a mechanistic perspective, these strategies can be broadly categorized into two approaches: systems that rely on elastic deformation and systems that rely on interfacial lubrication. Elastomeric coatings reduce ice adhesion through elastic energy storage and release but exhibit strong thickness dependence, whereas lubricant-based systems rely on interfacial slippage yet suffer from stability issues associated with lubricant loss28,34–37. As a result, both approaches inherently involve a tradeoff between ice adhesion, durability, and scalability. Despite these advances, a key limitation remains the strong thickness dependence and the tradeoff between adhesion and durability28,34–37. Because ice detachment in these systems is governed by elastic energy storage or interfacial lubrication, thin coatings often fail due to insufficient stress dissipation or loss of interfacial functionality, making thicker coatings necessary for stable performance. Such thickness dependence fundamentally constrains scalability and represents a critical limitation for real-world applications. Here we introduce dynamic polyborosiloxane (PBS) crosslinking into icephobic design and demonstrate that stress dissipation through reversible borosiloxane linkages effectively overcomes the durability–adhesion trade-off. This strategy leverages reversible B–O–Si bonds that dissociate and reform under stress, dissipating interfacial energy while preserving structural integrity38–44. Although PBS-based systems have recently been explored in icephobic and related applications, their role has largely been interpreted in terms of self-healing or lubricant stabilization45–47. As a result, the fundamental contribution of reversible B–O–Si linkages to ice detachment, particularly in terms of stress dissipation at the interface, has not been clearly established. By dissipating stress through reversible B–O–Si linkages, PBS mitigates the durability–adhesion trade-off and enables low ice adhesion even at micrometer-scale thicknesses ( ~ 2 μm), while maintaining mechanical robustness. In addition, hydroxyl groups along PBS chains enable hydrogen bonding with hydroxylated substrates, providing substantially stronger interfacial adhesion than conventional PDMS-based systems48. The icephobic behavior arises from a cooperative two-step mechanism: initial interfacial slippage that reduces shear stress at the coating interface, followed by dynamic bond dissociation that relaxes residual stresses within the network. Collectively, these processes overcome the durability–adhesion tradeoff and establish a scalable route toward practical icephobic surfaces. Results and discussion Design and mechanism of the icephobic coating To address the limitations of conventional icephobic coatings, including thickness dependence and lubricant instability, we designed the P-PBS coating by introducing dynamic polyborosiloxane crosslinking. A comparative summary of representative strategies and their limitations is provided in Supplementary Table 1. Figure 1a compares ice detachment on bare metal, neat PDMS, P-Oil, and the P-PBS coating. Bare metal exhibits τice up to ~1000 kPa, requiring substantial shear for removal. Replacing the surface with soft PDMS lowers the elastic mismatch with ice, enabling interfacial cavitation under shear stress and thereby facilitating partial detachment, but τice still exceeds 100 kPa14,49–51. However, the τice on neat PDMS typically exceeds 100 kPa, which limits its practical utility. Incorporating silicone oil into the PDMS matrix (P-Oil) increases free volume between polymer chains, promoting chain mobility and enabling interfacial slippage14,49,52. While this reduces ice adhesion to 20–30 kPa, P-Oil coatings remain vulnerable to mechanical damage and suffer from weak bonding with underlying substrates.Fig. 1Schematic illustration of icephobic mechanism.a Representative ice removal mechanisms on conventional surfaces. Bare metal exhibits increasing interfacial stress during ice detachment, whereas elastomeric coatings and oil-infused coatings release ice through cavitation and interfacial slippage, respectively. Representative ice adhesion strengths of approximately 100, 30, and 6 kPa are included to illustrate the relative performance of bare metal, oil-infused, and P-PBS coatings, respectively. The stress–time curves are schematic illustrations and do not represent experimental data, and the time axis is qualitative and not to scale. b Proposed ice detachment mechanism of the P-PBS coating. Reversible Si–O–B linkages within the polyborosiloxane (PBS) network undergo dissociation and reformation under stress, enabling stress relaxation and redistribution. The highlighted features summarize the key characteristics of the P-PBS coating, including outstanding icephobicity, durability, robust substrate adhesion, and thickness-invariant performance. c Schematic illustration of the two-step icephobic mechanism of the P-PBS coating. Silicone oil promotes interfacial slippage, while the PBS network dissipates stress through reversible bond dissociation. PBS polyborosiloxane, P-PBS PDMS/PBS/silicone oil coating. Schematic illustration of icephobic mechanism. a Representative ice removal mechanisms on conventional surfaces. Bare metal exhibits increasing interfacial stress during ice detachment, whereas elastomeric coatings and oil-infused coatings release ice through cavitation and interfacial slippage, respectively. Representative ice adhesion strengths of approximately 100, 30, and 6 kPa are included to illustrate the relative performance of bare metal, oil-infused, and P-PBS coatings, respectively. The stress–time curves are schematic illustrations and do not represent experimental data, and the time axis is qualitative and not to scale. b Proposed ice detachment mechanism of the P-PBS coating. Reversible Si–O–B linkages within the polyborosiloxane (PBS) network undergo dissociation and reformation under stress, enabling stress relaxation and redistribution. The highlighted features summarize the key characteristics of the P-PBS coating, including outstanding icephobicity, durability, robust substrate adhesion, and thickness-invariant performance. c Schematic illustration of the two-step icephobic mechanism of the P-PBS coating. Silicone oil promotes interfacial slippage, while the PBS network dissipates stress through reversible bond dissociation. PBS polyborosiloxane, P-PBS PDMS/PBS/silicone oil coating. To overcome these shortcomings, PBS was incorporated into the P-Oil system, resulting in the formation of the P-PBS coating (Fig. 1b). The P-PBS coating combines interfacial slippage induced by silicone oil with the dynamic bonding capability of PBS, enabling low ice adhesion, strong substrate bonding, and durability that remains effective even at reduced thicknesses. In particular, the reversible B–O–Si linkages in PBS can dissociate under external stress to dissipate localized energy and subsequently reform once the stress is removed, enabling structural recovery and maintaining coating integrity. The corresponding two-step ice detachment mechanism is illustrated in Fig. 1c. In the first stage, interfacial mobility is enhanced by silicone oil and PDMS chain flexibility, allowing lateral sliding of ice. As the stress continues, dynamic B–O–Si linkages between PBS and PDMS dissociate, relieving localized stress and enabling complete detachment. Li et al. reported that reducing interfacial stress during detachment can effectively lower ice adhesion53. The network subsequently recovers its structure through bond reformation after unloading. This process reflects the combined role of reversible bonding and polymer mobility in dissipating stress at the ice-coating interface. The two-step mechanism of the P-PBS coating effectively dissipates internal stresses, thereby enhancing durability. Furthermore, unlike conventional icephobic coatings whose performance strongly depends on thickness28,34–37, the P-PBS coating maintains stable performance irrespective of thickness. In addition, hydrogen bonding between PBS and hydroxylated substrates enhances interfacial adhesion. Therefore, the P-PBS coating achieves low ice adhesion together with improved durability under repeated icing/de-icing and mechanical loading. To further clarify the role of PBS in governing icephobic and adhesive performance, we investigated how PBS viscosity influences icephobic behavior and used this to guide the design of an optimal P-PBS coating. Two types of PBS were synthesized from hydroxy-terminated PDMS precursors with viscosities of 25 and 750 cSt, designated as PBS1 and PBS2, respectively. Confirmation of reversible B–O–Si network formation As shown in Supplementary Fig. 1, PBS formation involves multiple bonding mechanisms. Boric acid reacts with hydroxy-terminated PDMS through both hydrogen bonding and condensation. In addition, the boron atom in boric acid acts as a Lewis acid and may form reversible Lewis acid–base coordination interactions with siloxane oxygens in PDMS chains39,54. These interactions support the formation of a dynamic PBS–PDMS network39,41,42. The surface wettability results (Fig. 2a–c) provided preliminary evidence for network formation. The neat PDMS coating exhibited a static water contact angle of 104°, whereas the P-Oil coating showed a higher angle (113°), attributed to silicone oil exposed at the surface. In contrast, P-PBS1 (5 wt%) showed a reduced contact angle (93°) compared to PDMS and P-Oil, which can be ascribed to residual and/or surface-reoriented hydroxyl groups originating from incomplete PBS formation. The decrease indicates the presence of polar groups, indirectly suggesting hydroxyl participation in PBS–PDMS crosslinking.Fig. 2Characterization of PBS–PDMS network formation.a–c Static water contact angles of PDMS, P-Oil, and P-PBS1 coatings, respectively. Scale bars: 0.5 mm. d ATR–FTIR spectra of PDMS and P-PBS1 coatings. The appearance of characteristic borosiloxane-related vibrations indicates the formation of Si–O–B linkages in the PBS network. e Magnified ATR–FTIR spectra highlighting the B–O–H and Si–O–B absorption regions from (d). f XPS survey spectrum of the P-PBS coating, confirming the presence of boron together with carbon, oxygen, and silicon. g High-resolution B 1 s XPS spectrum of the P-PBS coating. The peak centered at 191.5 eV is attributed to borosiloxane bonding. h Thickness profile of the P-PBS1 coating measured using a stylus profilometer. i Digital image of uncoated and P-PBS-coated copper substrates, showing minimal visual change after coating. Scale bars: 10 mm. Source data are provided as a Source Data file. Characterization of PBS–PDMS network formation. a–c Static water contact angles of PDMS, P-Oil, and P-PBS1 coatings, respectively. Scale bars: 0.5 mm. d ATR–FTIR spectra of PDMS and P-PBS1 coatings. The appearance of characteristic borosiloxane-related vibrations indicates the formation of Si–O–B linkages in the PBS network. e Magnified ATR–FTIR spectra highlighting the B–O–H and Si–O–B absorption regions from (d). f XPS survey spectrum of the P-PBS coating, confirming the presence of boron together with carbon, oxygen, and silicon. g High-resolution B 1 s XPS spectrum of the P-PBS coating. The peak centered at 191.5 eV is attributed to borosiloxane bonding. h Thickness profile of the P-PBS1 coating measured using a stylus profilometer. i Digital image of uncoated and P-PBS-coated copper substrates, showing minimal visual change after coating. Scale bars: 10 mm. Source data are provided as a Source Data file. The ATR–FTIR spectra of PDMS and P-PBS1 coatings (Fig. 2d, e) confirmed the successful formation of PBS and its incorporation into the PDMS matrix. Characteristic B–O–H stretching bands were observed at approximately 660 and 886 cm⁻¹, and a B–O–Si stretching band appeared at 1340 cm⁻¹, indicating that PBS was chemically bonded to the PDMS chains55–58. Although surface hydroxyl groups significantly influenced the wettability of the coatings, they were not detected in FTIR due to their low surface concentration and the overwhelming intensity of dominant Si–O–Si and Si–CH3 bands44,58. X-ray photoelectron spectroscopy (XPS) further confirmed PBS incorporation through the B 1 s peak observed in P-PBS1 (Fig. 2f, g), which was absent in neat PDMS (Supplementary Fig. 2). Cross-sectional SEM (Supplementary Fig. 3) revealed a uniform coating morphology with an average thickness of ~2.0 µm, thinner than typical icephobic coatings49,59–61. This was verified by α-step profilometry (Fig. 2h). The coated copper substrate also exhibited high optical transparency, appearing visually indistinguishable from the uncoated substrate (Fig. 2i). Mechanical analysis further supported PBS–PDMS bonding (Fig. 3). Coating mixtures were cast into Petri dishes, solvent removed, and cured under the same conditions as the coatings. Crosslink density, measured by swelling, was highest for neat PDMS (577 mol/m³), consistent with a fully crosslinked siloxane network (Fig. 3a). Adding silicone oil reduced density to 366 mol/m³ in P-Oil, as oil penetrated the matrix, increasing chain mobility and acting as a plasticizer. In P-PBS1 (5 wt%), density dropped further. Although PBS chains form dynamic B–O–Si bonds with PDMS, these linkages dissociate in solvents, making PBS behave as a plasticizer-like, stress-dissipating but non-reinforcing component62,63. P-PBS2 showed the lowest density, due to its use of higher-viscosity PDMS precursors that generated longer PBS chains and a more compliant network. Increasing PBS to 30 wt% caused further density reduction, especially in P-PBS2. Tensile tests provided indirect evidence of PBS–PDMS bonding (Fig. 3b, c). Neat PDMS coating exhibited the highest modulus (534 kPa) and elongation (312.6%), attributed to its fully crosslinked network. In contrast, P-Oil showed a reduced modulus (298 kPa) due to the plasticizing effect of silicone oil, which lowered the crosslink density (Fig. 3a). Silicone oil also hindered efficient stress dissipation during tensile deformation, resulting in a lower elongation (187%) compared to PDMS.Fig. 3Mechanical properties of the coatings.a Crosslinking density determined from equilibrium swelling measurements. b Young’s modulus obtained from the initial linear region (0–20% strain) of the tensile stress–strain curves. c Elongation at break of the coatings. The incorporation of silicone oil and PBS reduced the crosslinking density and modulus of the PDMS network while maintaining high extensibility. Bars represent mean values, and error bars represent standard deviation. Individual data points are shown. The number of measurements is indicated by the plotted data points (n = 5). Measurements were performed on independently prepared samples. Source data are provided as a Source Data file. Mechanical properties of the coatings. a Crosslinking density determined from equilibrium swelling measurements. b Young’s modulus obtained from the initial linear region (0–20% strain) of the tensile stress–strain curves. c Elongation at break of the coatings. The incorporation of silicone oil and PBS reduced the crosslinking density and modulus of the PDMS network while maintaining high extensibility. Bars represent mean values, and error bars represent standard deviation. Individual data points are shown. The number of measurements is indicated by the plotted data points (n = 5). Measurements were performed on independently prepared samples. Source data are provided as a Source Data file. P-PBS1 (5 wt%) had a slightly higher modulus (310.3 kPa) than P-Oil, attributed to short PBS chains dispersed in PDMS forming transient B–O–Si crosslinks. These bonds acted as reversible reinforcements during initial deformation. However, as strain increased, bond dissociation induced localized stresses, reducing elongation compared to P-Oil. At 30 wt% PBS1, both modulus and elongation decreased further due to reduced crosslink density, indicating a soft, sparsely crosslinked PDMS network dominated performance. P-PBS2 coatings had lower modulus and elongation than P-PBS1, stemming from longer PBS chains formed from high-viscosity precursors. These chains dispersed less efficiently and provided limited reinforcement. Moreover, to maintain a 1:1 hydroxyl-to-boric acid ratio, less boric acid was used, resulting in fewer B–O–Si linkages and weaker PBS–PDMS connectivity. With fewer dynamic crosslinks, stress transfer efficiency decreased, leading to reduced stiffness. Although dynamic bonds offered local reinforcement, the overall compliant network caused stress concentration and impaired performance. At 30 wt% PBS2, modulus declined further, but elongation increased due to phase separation that produced an oil-rich surface layer capable of dissipating strain energy. Optical microscopy confirmed this lubricating layer, which was removed after gentle rubbing (Supplementary Fig. 4). Dynamic mechanical analysis (DMA) further examined viscoelastic response (Supplementary Fig. 5). Measurements were performed at 0.1% strain over 0.01–10 Hz. The viscoelastic response was not apparent in P-PBS1 (5 wt%), likely masked by the dominant elastic response of the PDMS network. In contrast, neat PBS (100 wt%) exhibited clear viscoelastic relaxation. This indicates dynamic B–O–Si bonding exists in the hybrid system, but its contribution is overshadowed at low PBS concentrations by the percolated PDMS network. Adhesion enhanced via dynamic B–O–Si interfacial interaction Figure 4 presents the adhesion strength of the coatings to the substrate. Strong interfacial adhesion is critical for the long-term durability of icephobic coatings, as weak bonding can cause delamination or failure under repeated icing/de-icing cycles. The adhesion mechanisms of neat PDMS and P-PBS are schematically illustrated in Fig. 4a. For neat PDMS, adhesion arises mainly from weak van der Waals forces due to the absence of polar groups capable of strong interfacial bonding64,65. In contrast, PBS chains in P-PBS introduce hydroxyl groups (Fig. 2) that form hydrogen bonds with hydroxylated substrates48,66,67. PBS forms B–O–Si linkages with PDMS chains that act as molecular bridges between the PDMS network and the substrate. As shown in Fig. 4a, PBS chains (highlighted in purple) can interact with the substrate, PDMS, and neighboring PBS chains, which strengthens interfacial adhesion.Fig. 4Adhesion performance of coatings.a Schematic illustration of the adhesion mechanisms of neat PDMS and P-PBS coatings. The incorporation of PBS introduces hydroxyl-containing borosiloxane structures capable of forming interfacial hydrogen-bonding interactions with the substrate. b Photograph of the lap shear test configuration used to evaluate coating adhesion according to ASTM D1002. Scale bar: 2.5 cm. c Lap shear adhesion strength as a function of PBS content. d Effect of UV–ozone treatment on the lap shear adhesion strength of the coating. e Lap shear adhesion strength of the P-PBS coating on different substrates, including Cu, Al, Fe, and PMMA. Bars represent mean values, error bars indicate standard deviation, and individual data points are shown (n = 5). Measurements were performed on independently prepared specimens. PBS polyborosiloxane, PMMA poly(methyl methacrylate). Source data are provided as a Source Data file. Adhesion performance of coatings. a Schematic illustration of the adhesion mechanisms of neat PDMS and P-PBS coatings. The incorporation of PBS introduces hydroxyl-containing borosiloxane structures capable of forming interfacial hydrogen-bonding interactions with the substrate. b Photograph of the lap shear test configuration used to evaluate coating adhesion according to ASTM D1002. Scale bar: 2.5 cm. c Lap shear adhesion strength as a function of PBS content. d Effect of UV–ozone treatment on the lap shear adhesion strength of the coating. e Lap shear adhesion strength of the P-PBS coating on different substrates, including Cu, Al, Fe, and PMMA. Bars represent mean values, error bars indicate standard deviation, and individual data points are shown (n = 5). Measurements were performed on independently prepared specimens. PBS polyborosiloxane, PMMA poly(methyl methacrylate). Source data are provided as a Source Data file. Adhesion was quantified by lap shear testing (ASTM D1002, Fig. 4b). P-PBS1 (5 wt%) achieved the highest adhesion ( ~ 2620 kPa), representing an 8.7-fold increase over neat PDMS (300 kPa) and a 13.8-fold increase over P-Oil (190 kPa). Notably, this result deviates from the general expectation that a decrease in crosslinking density leads to reduced adhesion strength due to lowered cohesive integrity of the coating, as shown in Fig. 3. Despite the reduced crosslinking density, the P-PBS coating exhibits significantly enhanced adhesion, suggesting that interfacial interactions play a dominant role in this system. This behavior can be understood by considering the dominant failure mode during mechanical loading. In the case of neat PDMS, irregular coating residues were observed on the substrate after lap shear testing (Supplementary Fig. 6), indicating that failure occurred at the coating–substrate interface (adhesion failure). In contrast, the P-PBS coating remained uniformly attached to the substrate after fracture, demonstrating that failure predominantly occurred within the coating layer (cohesive failure). This transition in failure mode indicates that, although the internal cohesion of the coating is reduced due to the lower crosslinking density, enhanced interfacial interactions arising from hydroxyl-functional PBS contribute to stronger overall adhesion. Increasing PBS content beyond 5 wt% led to reduced adhesion (Fig. 4c), likely due to PBS chain aggregation and decreased PDMS crosslink density, which together weakened coating cohesion48. P-PBS2 coatings consistently showed lower adhesion than P-PBS1, mainly because the longer hydroxy-terminated PDMS precursors used in PBS2 synthesis required less boric acid to maintain the 1:1 OH-to-boron ratio. As a result, fewer boron atoms were available to form B–O–Si linkages with PDMS chains, weakening interfacial bonding. By contrast, P-Oil exhibited even weaker adhesion than PDMS because surface and bulk silicone oil disrupted interfacial bonding. Incorporation of PBS restored adhesion despite the presence of oil. Figure 4d, e compares the lap shear strength of P-PBS1 (5 wt%) before and after substrate hydroxylation and evaluates adhesion on various substrates. The effectiveness of hydroxyl group introduction via UV–ozone treatment was evaluated by comparing the water contact angles of treated and untreated surfaces (Supplementary Fig. 7). UV–ozone treatment on metallic substrates such as copper induces the formation of surface oxides like CuO and Cu(OH)2, which introduce hydroxyl groups that enhance surface wettability and consequently reduce the contact angle68. Even without surface functionalization, the inclusion of PBS increased adhesion to 587.7 kPa, slightly higher than that of neat PDMS due to hydrogen bonding from PBS hydroxyl groups and its intrinsic tackiness. However, this value was significantly lower than that achieved after UV–ozone treatment, where hydrogen bonding with the hydroxylated substrate strongly enhanced interfacial adhesion. P-PBS1 coatings exhibited strong adhesion not only to copper substrates but also to aluminum, steel, and non-metallic surfaces such as PMMA. Water contact angle measurements for different substrates following identical UV–ozone treatment are summarized in Supplementary Table 3. Aluminum showed the lowest contact angle (16°), followed by steel (34.5°), copper (41°), and PMMA (74.2°). The trend indicates that the extent of hydroxylation depends on the substrate, which in turn affects interfacial adhesion. Across all substrates tested, the P-PBS coating maintained substantially higher adhesion strength than neat PDMS. This consistency highlights its applicability to both metallic and polymeric surfaces. Icephobic adhesion performance Figure 5a illustrates the two-step ice detachment mechanism of the P-PBS1 coating. Initially, silicone oil dispersed within the PDMS matrix enhances chain mobility, enabling interfacial slippage between ice and the coating69. This mobility is effective only when the oil is well integrated into the polymer network, avoiding the formation of a discrete oil layer14,49,52. With continued shear, reversible Si–O–B borosiloxane linkages within the PBS network undergo reversible dissociation, locally relaxing stress and allowing progressive sliding of ice under reduced friction. Upon unloading, the broken bonds re-form, restoring network integrity. As depicted in Figs. 1b and 5a, the cooperative action of oil-induced slippage and reversible bond dissociation enables ice removal at forces far lower than those required for PDMS or P-Oil coatings. To quantify ice adhesion, shear tests were conducted using a custom-built apparatus (Supplementary Fig. 8). As shown in Fig. 5b–d, the bare copper substrate exhibited very high τice (~454 kPa), where interfacial fracture of ice often preceded detachment, evidencing strong ice–substrate bonding. Neat PDMS reduced τice to ~103 kPa due to elastic mismatch and interfacial cavitation under shear stress; in elastomeric coatings, such cavitation destabilizes the interface and leads to abrupt force drops once cavities coalesce. By contrast, incorporating silicone oil (P-Oil) lowered τice to 31 kPa, attributed to the chemical similarity (Si–O backbone) and high miscibility between PDMS and silicone oil, which reduces crosslink density and enhances chain mobility14,49. Under shear, this mobility facilitates interfacial slippage, producing gradual ice sliding even beyond the peak stress, as evidenced by the smooth decay in shear force in Fig. 5d. However, excessive oil causes phase separation, forming a discrete surface oil layer70,71 that undermines coating integrity and durability, as confirmed by surface imaging showing increased adhesion at higher oil contents (Supplementary Fig. 9).Fig. 5Icephobic performance of coatings.a Schematic illustration of the proposed two-step ice detachment mechanism in the P-PBS coating. Silicone oil enhances PDMS chain mobility and promotes interfacial slippage under shear loading. As stress accumulates, reversible Si–O–B linkages within the PBS network undergo dissociation, dissipating localized stress and facilitating ice release. Following detachment, the dynamic network recovers through bond reformation. b Ice adhesion strength (τice) of various coating systems. c Effect of PBS content on ice adhesion strength (τice). d Shear force–time curves recorded during ice detachment. The inset highlights the low-force detachment behavior of PBS-containing coatings. e Ice adhesion strength as a function of coating thickness. The two thinnest coatings exhibited thicknesses of 0.378 and 0.91 μm, respectively. f, g Time-lapse images of ice detachment from an uncoated Cu substrate (f) and a P-PBS1 (5 wt%) coating (g) under applied shear force. Scale bars: 1 cm. Bars represent mean values, error bars indicate standard deviation, and individual data points are shown. Measurements were performed on independently prepared samples. Sample sizes were n = 5 or 6, as indicated by the plotted data points. PBS polyborosiloxane, PDMS polydimethylsiloxane. Source data are provided as a Source Data file. Icephobic performance of coatings. a Schematic illustration of the proposed two-step ice detachment mechanism in the P-PBS coating. Silicone oil enhances PDMS chain mobility and promotes interfacial slippage under shear loading. As stress accumulates, reversible Si–O–B linkages within the PBS network undergo dissociation, dissipating localized stress and facilitating ice release. Following detachment, the dynamic network recovers through bond reformation. b Ice adhesion strength (τice) of various coating systems. c Effect of PBS content on ice adhesion strength (τice). d Shear force–time curves recorded during ice detachment. The inset highlights the low-force detachment behavior of PBS-containing coatings. e Ice adhesion strength as a function of coating thickness. The two thinnest coatings exhibited thicknesses of 0.378 and 0.91 μm, respectively. f, g Time-lapse images of ice detachment from an uncoated Cu substrate (f) and a P-PBS1 (5 wt%) coating (g) under applied shear force. Scale bars: 1 cm. Bars represent mean values, error bars indicate standard deviation, and individual data points are shown. Measurements were performed on independently prepared samples. Sample sizes were n = 5 or 6, as indicated by the plotted data points. PBS polyborosiloxane, PDMS polydimethylsiloxane. Source data are provided as a Source Data file. Notably, P-PBS1 (5 wt%) exhibited low τice of 6.4 kPa, the lowest among all tested coatings, while maintaining strong substrate adhesion. During ice removal, as illustrated in Fig. 5a, silicone oil facilitated interfacial stress distribution, whereas reversible dissociation of PBS–PDMS bonds relaxed localized stresses, resulting in clean and smooth detachment. After the ice was removed, the dynamic network re-associated, preserving the coating’s structural integrity and durability. The enhanced icephobic performance of the P-PBS coating is not attributed to the presence of a lubricating layer. After toluene washing to remove any extractable oligomeric species72, the ice adhesion strength remained unchanged, and no measurable mass loss was detected, indicating that the coating performance is not governed by a free lubricating phase (Supplementary Fig. 10). To further verify this stress-relief mechanism, P-PBS coatings prepared without silicone oil were evaluated (Supplementary Figs. 11 and 12). Even in the absence of oil, the coating containing 5 wt% of PBS1 exhibited a τice of 15.2 kPa, which was lower than that of the P-Oil coating, indicating that dynamic PBS–PDMS bond dissociation alone contributes to interfacial stress relaxation during ice removal. However, when oil was absent, interfacial slippage was suppressed, and detachment occurred simultaneously with bond breakage, resulting in a cavitation-like force–time response (Supplementary Fig. 12). Increasing the PBS1 content to 30 wt% increased the τice to 38 kPa, attributed to chain aggregation and restricted PDMS mobility (Fig. 5d, inset). PBS viscosity also influenced the icephobic performance. The P-PBS2 coating containing 5 wt% of PBS2 exhibited higher adhesion than P-PBS1 because (i) the longer PBS2 chains were poorly dispersed within the PDMS matrix, limiting molecular rearrangement, and (ii) less boric acid was required during synthesis to maintain the stoichiometric OH ratio, resulting in fewer boron atoms and a lower density of dynamic B–O–Si linkages. Consequently, reversible dissociation was less effective in relieving stress. At higher PBS2 contents, oil-rich phase separation occurred (Supplementary Fig. 4), which temporarily reduced ice adhesion through surface lubrication but compromised long-term durability during repeated icing and de-icing cycles. Unlike conventional soft polymeric coatings, the P-PBS system exhibits icephobic behavior that is effectively independent of coating thickness, owing to its stress-relief mechanism. For soft materials, τice typically follows the relationship:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tau }_{{ice}}\propto {\left(\frac{2{w}_{a}K}{t}\right)}^{1/2}$$\end{document}τice∝2waKt1/2where wa is the work of adhesion, K is the bulk modulus, and t is the coating thickness28,36. According to Eq. (1), thicker coatings are generally expected to show lower ice adhesion23,28,36,37. This model is based on an elastic energy balance framework, where ice detachment occurs when the elastic strain energy stored in the coating exceeds the interfacial fracture energy. In this framework, ice removal is governed by the competition between the energy required to separate the ice-coating interface (work of adhesion) and the elastic energy that can be stored and released by the deformable coating. As the coating thickness increases, a larger volume of material can deform and store elastic energy before instability, facilitating interfacial cavitation and reducing the apparent ice adhesion strength. In conventional elastomeric coatings, the energy-driven detachment mechanism leads to the characteristic thickness dependence described by Eq. (1), where thicker layers can store more elastic energy before instability, thereby facilitating interfacial cavitation and reducing the apparent ice adhesion strength. The dependence of ice adhesion on coating thickness was subsequently investigated by controlling the viscosity of the coating solution (Fig. 5e). Conventional PDMS coatings showed a decrease in τice from approximately 103 kPa at 2 μm to about 50 kPa at 500 μm, consistent with classical scaling models (Eq. (1)) in which thicker elastomeric layers reduce interfacial stress28,36. The P-Oil coating followed a similar trend, reaching about 7 kPa at a thickness of 500 μm, although optical microscopy revealed surface oil accumulation (Supplementary Fig. 13). These surface oil layers gradually diminished after repeated icing and de-icing cycles (Supplementary Fig. 14), leading to a progressive increase in ice adhesion. In contrast, the P-PBS1 (5 wt%) coating maintained consistently low ice adhesion (~ 6–8 kPa) over a wide thickness range, and notably preserved its performance even in the sub-micrometer regime, without evidence of oil phase separation. Time-lapse images (Fig. 5f, g) further highlight the improved icephobic performance. On the bare copper substrate, ice remained immobile even after 60 s under a constant shear load (Supplementary Video 1). In contrast, on the P-PBS1 (5 wt%) coating, ice began sliding within 0.5 s and fully detached within 6 s (Supplementary Video 2). Significantly, the ice block on the P-PBS1 coating could even slide off solely under gravity (Supplementary Video 3), providing direct evidence of the low adhesion enabled by the dynamic PBS–PDMS network. To evaluate the scalability of the icephobic coating, it was applied to a large-area aluminum substrate. The coating was deposited using a spray-coating method, and detailed processing conditions are provided in the Supplementary Information. After ice formation at –20 °C, the substrate was tilted to assess ice removal behavior. Ice formed on the coated region readily detached at a tilt angle of ~20° (Supplementary Video 4). In contrast, ice on the uncoated region remained strongly adhered, requiring a substantial external force for removal. These results demonstrate that the coating can be uniformly applied over large areas while maintaining effective icephobic performance under practical conditions. Durability in various environments Figure 6 summarizes the durability of the icephobic coatings under diverse environmental stresses. For real-world applications, icephobic coatings must sustain their performance under repeated icing/de-icing, mechanical abrasion, tape peeling, and chemical attack. Despite their thin structure (1.5–2.5 μm), which typically increases vulnerability to damage, the P-PBS1 (5 wt%) coating exhibited robustness across all tests.Fig. 6Durability of icephobic coatings under various environmental conditions.a Ice adhesion strength (τice) of P-Oil and P-PBS1 (5 wt%) coatings during repeated icing/de-icing cycles. b Optical microscopy images of P-Oil and c P-PBS1 (5 wt%) coatings after 30 icing/de-icing cycles. The red dashed regions in b indicate fractured and delaminated coating areas after repeated ice removal, whereas the P-PBS1 coating remains intact. Scale bars: 20 μm. d τice after abrasion tests using 1000-grit sandpaper under a 250 g normal load. The inset schematically illustrates the abrasion test configuration. e τice after repeated tape peeling tests. The green dashed line indicates the average ice adhesion strength calculated from all measured data points in this panel after repeated tape adhesion cycles. f τice after immersion in aqueous solutions adjusted to the indicated pH values using H₂SO₄ or NaOH, followed by drying at 100 °C for 1 h. Symbols represent mean values, and error bars indicate standard deviation. Measurements were performed on independently prepared samples. Sample sizes were n = 5 for all conditions, except the first cycle of P-PBS1 in panel a and the 0-cycle abrasion condition of P-PBS1 in panel d (n = 6). P-Oil PDMS/silicone oil coating, P-PBS1 PDMS/PBS/silicone oil coating containing PBS1. Source data are provided as a Source Data file. Durability of icephobic coatings under various environmental conditions. a Ice adhesion strength (τice) of P-Oil and P-PBS1 (5 wt%) coatings during repeated icing/de-icing cycles. b Optical microscopy images of P-Oil and c P-PBS1 (5 wt%) coatings after 30 icing/de-icing cycles. The red dashed regions in b indicate fractured and delaminated coating areas after repeated ice removal, whereas the P-PBS1 coating remains intact. Scale bars: 20 μm. d τice after abrasion tests using 1000-grit sandpaper under a 250 g normal load. The inset schematically illustrates the abrasion test configuration. e τice after repeated tape peeling tests. The green dashed line indicates the average ice adhesion strength calculated from all measured data points in this panel after repeated tape adhesion cycles. f τice after immersion in aqueous solutions adjusted to the indicated pH values using H₂SO₄ or NaOH, followed by drying at 100 °C for 1 h. Symbols represent mean values, and error bars indicate standard deviation. Measurements were performed on independently prepared samples. Sample sizes were n = 5 for all conditions, except the first cycle of P-PBS1 in panel a and the 0-cycle abrasion condition of P-PBS1 in panel d (n = 6). P-Oil PDMS/silicone oil coating, P-PBS1 PDMS/PBS/silicone oil coating containing PBS1. Source data are provided as a Source Data file. Repeated icing/de-icing performance (Fig. 6a–c) confirmed the long-term stability of the P-PBS1 coating. Ice was repeatedly frozen and removed at –15 °C, with τice measured after each cycle. The P-Oil coating initially maintained low τice (20~40 kPa) for ten cycles but began to degrade after fifteen cycles, showing visible delamination after twenty cycles and a sharp increase in τice to >120 kPa. After 30 cycles, τice exceeded 140 kPa (Fig. 6b). In contrast, the P-PBS1 coating retained a low τice of ~6 kPa even after 100 cycles without observable surface defects (Fig. 6c). To verify that the coating performance is not governed by oil depletion, the mass of the coating was also measured over repeated icing/de-icing cycles. After each cycle, the total mass of the substrate and coating was recorded using a balance with a resolution of 0.0001 g. The results showed that the mass change remained within 0.005% even after 50 cycles (Supplementary Fig. 15). This cyclic stability arises from the dual effect of PBS chains, which (i) strengthen coating–substrate adhesion through interfacial bonding and (ii) dissipate localized stress during detachment via dynamic B–O–Si bond exchange. Mechanical abrasion resistance further highlighted the coating’s robustness. Under a 250 g normal load using 1000-grit sandpaper, the P-Oil surface exhibited partial delamination after five cycles and completely failed by 100 cycles (Supplementary Fig. 16), accompanied by τice > 100 kPa. Conversely, the P-PBS1 (5 wt%) coating preserved its structure and icephobicity even after 500 abrasion cycles (Fig. 6d), with negligible changes in τice. Optical microscopy confirmed that the reversible dissociation of PBS–PDMS bonds efficiently redistributed shear stress, preventing crack propagation and material loss, whereas the P-Oil coating was almost completely removed even after only 60 abrasion cycles (Supplementary Fig. 16). Adhesion durability was evaluated by repeated tape peeling and cross-cut adhesion tests (Fig. 6e and Supplementary Fig. 17). In tape tests (ASTM D3359), the P-PBS1 coating maintained τice < 10 kPa even after sixty peel cycles, demonstrating robust surface bonding. In cross-cut adhesion tests, the coating retained a perfect 5B rating with no observable delamination or cracking, confirming strong interfacial integrity. Chemical stability was assessed under a wide pH range (Fig. 6f). The P-PBS1 coating was immersed in acidic (pH 1–5) and basic (pH 9–12) environments for up to 12 h, then dried at 100 °C for 1 h before testing. The coating preserved its icephobic performance across all pH conditions, maintaining τice < 10 kPa. Under strong acid (pH 1), slight increases in τice (~15 kPa) were attributed to partial hydrolysis of Si–O or B–O bonds, generating surface hydroxyls that marginally increased wettability54,73. Nevertheless, τice remained far below that of P-Oil coatings, indicating chemical durability and environmental resilience of the P-PBS system. Collectively, these results demonstrate that dynamic PBS–PDMS crosslinking provides an effective self-protective mechanism against physical and chemical degradation, enabling long-term icephobic performance even under harsh conditions. To evaluate stability under rain-like conditions, the coating was immersed in a mildly acidic aqueous solution74 (pH ≈5.2). The ice adhesion strength remained unchanged after prolonged exposure, indicating that the coating is not significantly affected by weakly acidic environments typically encountered in natural rainfall (Supplementary Fig. 18a). The durability of the coating was further examined under high-humidity conditions (99% relative humidity at 2 °C for 24 h) using a humidity chamber (Supplementary Fig. 18b). No noticeable change in τice was observed after exposure, demonstrating that the coating maintains its icephobic performance even in moisture-rich environments associated with frost formation (Supplementary Fig. 18c). A clear difference between coated and uncoated regions was observed after the humidity test (Supplementary Fig. 18d). The uncoated substrate exhibited evident surface corrosion, whereas the coated region remained intact without visible degradation. This result indicates that the P-PBS coating not only maintains icephobic performance but also provides effective protection against moisture-induced surface degradation. The stability of the coating under ultraviolet (UV) exposure was also evaluated. The P-PBS coating was subjected to UVA irradiation (365 nm) for up to 168 h, and the cumulative exposure dose was calculated based on the irradiation time. This duration corresponds to a period comparable to several weeks of outdoor exposure. No noticeable change in τice was observed after irradiation (Supplementary Fig. 19), indicating that the coating maintains its icephobic performance under prolonged UV exposure. In summary, we developed a durable, strongly adherent, and thickness-invariant icephobic coating by integrating dynamic polyborosiloxane (PBS) networks into a PDMS-based matrix. The PBS, synthesized from boric acid and hydroxy-terminated PDMS, introduces reversible B–O–Si linkages and hydroxyl groups that form multiple interfacial bonding pathways. These dynamic linkages not only reinforce adhesion, achieving over ninefold higher lap shear strength compared to neat PDMS, but also enable self-adaptive stress dissipation within the coating network. The icephobic behavior of the P-PBS coating arises from a two-step cooperative mechanism: (1) interfacial slippage facilitated by silicone oil enhances chain mobility and reduces shear stress, and (2) reversible dissociation of Si–O–B borosiloxane linkages relaxes localized stresses within the network, resulting in smooth and complete ice detachment. This synergistic mechanism achieves a low ice adhesion strength of 6.4 kPa, approximately sixteen times lower than that of PDMS and maintains its performance even at sub-2 μm thickness, thus effectively overcoming the conventional thickness–adhesion scaling that constrains existing icephobic coatings. Comprehensive durability evaluations, including repeated icing/de-icing, mechanical abrasion, tape peeling, and chemical immersion across a broad pH range, demonstrated that the P-PBS coating sustains its icephobicity and mechanical integrity over prolonged use. This stability stems from the self-recoverable nature of dynamic B–O–Si bonding, which redistributes applied stress and mitigates interfacial damage. Furthermore, the dynamic crosslinking concept is broadly applicable to diverse elastomeric matrices and is compatible with scalable fabrication routes such as dip- or spray-coating, ensuring industrial feasibility for large-area or complex-shaped substrates. Overall, this study establishes dynamic covalent bonding as a general design strategy for durable icephobic surfaces. By enabling simultaneous durability, strong adhesion, and thickness-invariant performance, the P-PBS system provides a scalable and material-agnostic approach for anti-icing applications in energy, transportation, and infrastructure systems. Methods Materials Sylgard 184 (polydimethylsiloxane, PDMS) was purchased from Dow Corning and used by mixing the base and curing agent at a 10:1 weight ratio. Hydroxy-terminated PDMS (25 cSt and 750 cSt) and boric acid were obtained from Sigma-Aldrich. Before use, boric acid was finely ground with a mortar and dried in a convection oven at 100 °C for 12 h. Silicone oil (100 cSt) was purchased from Shin-Etsu Silicone. n-Hexane (99.5%), toluene (95%), and isopropyl alcohol (IPA, 99.9%) were obtained from Duksan Pure Chemicals (Korea). Preparation of substrates Metal substrates (copper, steel, and aluminum) were first polished using 600-grit sandpaper, followed by 1500-grit sandpaper to achieve a smooth surface. The polished substrates were then ultrasonically cleaned twice in isopropyl alcohol (IPA) and dried in an oven at 80 °C for 1 h. Surface oxidation was performed using a UV–ozone cleaner (JSE Co.) for 900 s to introduce hydroxyl groups onto the substrate surface. The presence of hydroxyl groups was confirmed by measuring the static water contact angle. Preparation of icephobic coating Polyborosiloxane (PBS) was synthesized by mixing hydroxy-terminated PDMS (25 and 750 cSt, respectively) and boric acid at a stoichiometric molar ratio of 1:1 (based on hydroxyl groups) in a beaker. Subsequently, PDMS prepolymer (Sylgard 184) was added to the mixture, and the solution was magnetically stirred at 120 °C for 12 h to facilitate coordination between the PBS chains and PDMS matrix. Depending on the viscosity of the hydroxy-terminated PDMS used, the samples were designated as PBS1 (prepared from 25 cSt hydroxy-terminated PDMS) and PBS2 (prepared from 750 cSt hydroxy-terminated PDMS). After cooling the mixture to room temperature, a curing agent was added at a weight ratio of 10:1 (prepolymer: curing agent). Silicone oil (fixed at 25 wt% relative to the PDMS prepolymer, irrespective of PBS content) and n-hexane (200 mg/mL) were then introduced, followed by stirring at room temperature for 30 min. Detailed compositions are summarized in Supplementary Tables 4 and 5. The viscosity of the icephobic coating solution was 7.73 mPa·s, as measured by a rheometer. PDMS and P-Oil coating solutions were prepared using the same procedure, excluding PBS. Cleaned substrates were dip-coated in the prepared coating solutions for 30 min, slowly withdrawn, and then cured in an oven at 80 °C for 2 h. To evaluate the mechanical properties of the coatings, the same coating solutions were poured into Petri dishes and dried in a vacuum oven at 80 °C for 2 h to obtain uniform films. Characterization The contact angle of the substrate and coating was measured using a contact angle analyzer (SmartDrop, Femtofab, Korea) to analyze the effect of UV ozone cleaning and hydroxyl groups of the P-PBS coating. The reaction of polyborosiloxane was confirmed by Fourier-transform infrared spectroscopy (FTIR, Frontier, PerkinElmer, USA) and X-ray photoelectron spectroscopy (XPS, Axis Supra, Kratos, UK). The thickness of the coating was measured using a surface profiler with a length of 2 mm (Alpha-Step D-300, KLA, USA) and scanning electron microscopy (SEM, ZEISS, Germany). The crosslinking density of the icephobic coating was determined using the swelling method. Films were prepared by solvent evaporation and cut into 1 cm × 1 cm samples. The samples were immersed in toluene (95%) until equilibrium swelling was reached, while replacing the solvent every 24 h. The swollen weight of the samples was measured, followed by drying in a vacuum oven at 80 °C for 24 h, after which the dried weight was recorded. The crosslink density was calculated using the Flory–Rehner equation75,762\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${v}_{e}=-\frac{{{\mathrm{ln}}}\left(1-{V}_{2}\right)+{V}_{2}+\chi {V}_{2}^{2}}{{V}_{s}\left({V}_{2}^{1/3}-\frac{{V}_{2}}{x}\right)}$$\end{document}ve=−ln1−V2+V2+χV22VsV21/3−V2xwhere ve is the crosslinking density of the coating, Vs is the molar volume of the toluene, V2 is the volume fraction of coating in the swollen mass, χχ is the toluene-coating interaction term. The mechanical performance of the coatings was evaluated using a universal testing machine (UTM, DRTEC, Korea) to obtain stress–strain (SS) curves and a dynamic mechanical analyzer (DMA 8000, PerkinElmer, USA) to analyze viscoelastic behavior. The adhesion strength to the substrate was measured using the UTM according to ASTM D1002 at a loading rate of 1.13 mm min⁻¹. The icephobic performance was evaluated by measuring the τice using a custom-built apparatus (Ice Adhesion Measurement Device, SewonLabs, Korea). Ice blocks (1 cm³) were prepared using a 3D-printed mold and frozen for over 30 min on a Peltier plate maintained at −15 °C. The ice blocks were displaced at a constant speed of 0.25 mm s⁻¹ while the debonding force was recorded in real time. For cyclic icing/de-icing tests, ice was frozen for 30 min and subsequently removed. The τice was calculated by dividing the maximum shear force by the contact area of the ice. The peak value of the force vs. time graph is defined as the force required to remove the ice from the surface. For large-area spray coating experiments, a custom-built spray coating system was used. The substrate (SUS) was treated with UV–ozone for 900 s before coating, and the solution was sprayed at a pressure of 2 bar from a distance of 30 cm. The durability against abrasion was tested by attaching sandpaper (grit size 1000) to a 250 g weight and sliding it across the coating surface at a speed of 10 mm s⁻¹. After abrasion, the τice was re-evaluated. The surface morphology of the coating after repeated icing/de-icing and abrasion tests was observed using optical microscopy (OM, Olympus, Japan). Repeated tape tests and cross-cut adhesion tests were conducted using adhesive tape (adhesion strength: 6 N cm⁻¹). After each cycle of attaching and detaching the tape, the τice was measured. For pH stability tests, the coatings were immersed in sulfuric acid or sodium hydroxide aqueous solutions with adjusted pH values, then dried at 100 °C for 1 h, followed by evaluation of the τice. Statistics and reproducibility No statistical method was used to predetermine sample size. No data were excluded from the analyses. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment. Quantitative data are presented as mean ± standard deviation (s.d.). The exact number of measurements (n) is indicated in the corresponding figure legends. Data processing and statistical analyses were performed using OriginPro 2021 (OriginLab Corporation). Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Transparent Peer Review file Source data Source Data Source Data