Strong yet highly conductive liquid metal composite film constructed via friction induced in-situ synthesis

Abstract

Contact resistance is a crucial bottleneck for various applications. Liquid metal and its composites can reduce contact resistance of interfaces due to their high conductivity and deformability, yet their mechanical strength is insufficient under external forces. Here, we prepare a deformable liquid-metal-bridged nanogranular composite via friction induced in-situ synthesis, realizing low contact resistance alongside high current density of 1.5 GA/m² and high hardness of 2.6 GPa. This remarkable performance originates from the unique structure: a high fraction ( ~ 66 vol%) of nano-sized conductive ceramic particles bridged by high-surface-tension liquid metal through metallic bonding, which is formed in situ under mechanochemical stimulation during friction. This strategy provides a route to fabricating functional liquid metal composites, shedding light into simultaneous improvement of the mechanical and electrical properties.

Contact resistance limits device performance. Here, the authors fabricate a strong yet highly conductive liquid-metal-bridged nanogranular composite via friction induced in-situ synthesis, achieving low contact resistivity and high current density.

Introduction

Contact resistance is a ubiquitous interfacial phenomenon, which has long constrained the performance improvement of many critical devices, such as solid-state batteries¹, field-effect transistors², nanoelectromechanical switches³, and electrical rotary joints⁴. Specifically, high contact resistance causes poor thermal conduction⁵ or electrical transmission⁶, thereby failing to meet the growing demands for energy-intensive systems^7,8. The main source of contact resistance is the insufficient contact at the rough interface. It is generally considered that the actual contact area constitutes only a tiny fraction of the apparent contact area^9,10, and it may further diminish after inevitable surface wear¹¹, leading to the conductivity of interfaces far inferior to that of bulk phases.

Liquid metal (LM) shows great potential in achieving intimate contact and reducing contact resistance owing to its intrinsic deformability and high conductivity^12–16. Compared to solid contact-resistant materials such as copper and silver, LM can significantly reduce contact voltage drop and avoid arc erosion^17,18. Yet its fluidity introduces drawbacks such as leakage¹⁹ and consequent short circuit²⁰ or structural corrosion^21,22. In recent years, LM composites are prepared and have made great progresses in solving the problems, while also introducing many interesting properties through the addition of various fillers^20,23,24. For example, LM-polymer composite can serve as conductive solder to produce robust electrical contact interface²⁵. However, formation of the composites typically leads to a 40~95% decrease in conductivity because LM/additive interfaces hinder electron transport^25–28. Moreover, most LM composites exhibit low hardness and modulus^29–32, making them susceptible to extrusion from the contact zone or wear under external forces. Although increasing the fraction of solid fillers can enhance the mechanical properties to a certain extent, it is difficult to prepare uniform and integrating LM composites with high filler content (>60 vol%) by the conventional direct-mixing method, which instead often produces loose powders³³. This significantly limits their durability in mechanical-electrical systems, such as flexible electronics, switches, and joints^16,34–36. Consequently, how to obtain LM composites combining low contact resistance with high mechanical strength remains a major challenge.

Here, we fabricated a strong and conductive film, i.e., LM-bridged nanogranular composite (LMNC), via friction induced in-situ synthesis (FIIS). Surprisingly, we found that even after friction under GPa-level contact pressure, the LM composite experienced no wear, but instead, a tribofilm grew on its surface. More intriguingly, this film possesses superior conductivity of ~4 × 10⁶S/m and high mechanical strength with a hardness of 2.6 GPa. The exceptional performance of the film originates from the simultaneous processes of LM release, conductive ceramic nanocrystallization, and mixing under GPa-level contact pressure, which facilitate the formation of metallic bonds across the interfaces, thus yielding a film composed of ~66 vol% nano-sized particles bridged by LM. This work presents a manufacturing method for the LM composite with a unique structure, and offers a promising pathway to address the widely existing challenge of reducing contact resistance.

Results

Friction-induced composite film growth

We developed the FIIS method for preparing the LM composite film. We first stored some LM in the tungsten carbide (WC) matrix (Fig. 1), which is a ceramic material with high conductivity and mechanical strength, and unlike most metals, does not react with LM. This tungsten carbide-liquid metal (WC-LM) composite was then used as the substrate for film growth. The LM selected in this work is eutectic gallium-indium alloy (EGaIn), with a melting point of 15.7 °C, which enables the convenient mixing of WC and LM by mechanical stirring at room temperature (Fig. 1a and Supplementary Fig. 1). Consequently, the storage of LM was successfully achieved through fast-hot-pressed sintering this mixture at 1350 °C. The cross-section morphology of the sintered WC-LM substrate (Fig. 1c) confirms the presence of both solid and liquid phases. Characterization of the internal element content suggests that LM is evenly distributed in the WC matrix (Supplementary Fig. 2), which ensures that LM can be continuously released to the surface during the subsequent friction process. The high magnification image of the internal morphology and the corresponding crystal structures of the WC-LM specimen (Fig. 1d and Supplementary Fig. 3) were obtained after ion beam thinning. The oxidation of LM may mainly occur during the thinning process, and LM has almost no oxidation in the original sample due to the absence of Ga₂O₃ characteristic peaks³³ in the X-ray diffraction spectrum (Supplementary Fig. 4). In addition, the diffraction peak width decreases and there is no new peak appearing after sintering, which indicates that the WC grains grow and do not react with LM. The chemical state of elements at the surface of the composite (Supplementary Fig. 5) further demonstrates that the Ga and In elements exist in metallic states. The density, hardness, and conductivity of the WC-LM composite were also tested, and the pure WC was used for reference (Supplementary Table 1).

Fig. 1. Preparation of the tungsten carbide-liquid metal (WC-LM) substrate.

a Schematic diagram of the preparation process, indicating how to storage LM in the WC matrix. b Photo of the sintered WC-LM composite. c Morphology of the composite cross section, showing the coexistence of liquid phase LM and solid phase WC. The preparation process was repeated independently at least five times and similar results were obtained. d Microstructure and corresponding crystal structure images of the composite.

Next, we fabricated a composite film using the WC-LM substrate. This fabrication process originated from an unexpected finding. We slid a WC-Co (YG6) ball with a diameter of 8 mm on the WC-LM composite disk under a normal load of 1.5 N (Supplementary Fig. 6), corresponding to an initial contact pressure of ~1 GPa. Meanwhile, an electric current of 2 A was applied during the sliding process. We recorded the variations in coefficient of friction (COF) and electrical contact resistance (ECR) throughout the sliding process. During the initial stage of dozens frictional cycles, the ECR rapidly decreased from 40 mΩ to 5 mΩ (Fig. 2a inset). More intriguingly, we observed no wear of the WC-LM substrate after sliding under such high contact pressure. Instead, a significant protrusion appeared at the contact zone (Fig. 2b, c), suggesting that a highly conductive film grew on the WC-LM substrate. As a comparison, we replaced the disk with pure WC and CuCrZr, both of which exhibited increasing ECR and normal wear behavior without film growth (Supplementary Figs. 7, 8). Therefore, in contrast to the conventional understanding that friction typically leads to interfacial material removal, we discovered a friction induced in-situ synthesis (FIIS) phenomenon involving LM participation, as illustrated in Fig. 2d–f.

Fig. 2. Fabricating liquid metal (LM) composite film via friction induced in-situ synthesis (FIIS).

a Coefficient of friction (COF) and electrical contact resistance (R_c) evolution of the contact pair under 2 A electric current, 10 mm/s sliding velocity and 1.5 N normal load. Inset is the variation of COF and R_c during the first 192 frictional cycles. b 3D morphology and c cross-section profile along the dotted line in b, indicating that a thin film has grown on the contact zone. d Schematic diagram of the FIIS process. Comparison of the contact interface e before and f after sliding. The actual contact area of the rough interface before sliding is small, while after sliding, the LM composite film fills the gaps between the asperity peaks, thereby increasing the contact area. Source data are provided as a Source Data file.

Structure and properties of LMNC

Further tests reveal that the LM composite film fabricated via FIIS exhibits exceptional mechanical properties and electrical conductivity. The nano-indentation test (Fig. 3a and Supplementary Fig. 9) reflects that this film has a hardness of up to 2.6 GPa and an elastic modulus of 65 GPa, significantly higher than that of existing LM composites^{21,26,27,29,33,34,37} (Fig. 3c and Supplementary Table 2). This illustrates the solid-like characteristic and load-bearing capacity of our film. Besides, this film experiences deformation under a press-in depth of 1 μm without collapse and fracture (Supplementary Fig. 10), suggesting its high deformability. Both the WC-LM substrate and the film possess a high electrical conductivity about 4 × 10⁶S/m, higher than that of the pure LM and other types of LM composites^{30,34,38–40} (Fig. 3d and Supplementary Table 3). The film reduces the interfacial contact resistance to 1/7 of its initial value (Fig. 3b). Besides, the ECR and its variation with load decrease after sliding (Supplementary Fig. 11), which provides high-quality and load-insensitivity electrical contact.

Fig. 3. Performance and structure of liquid-metal-bridged nanogranular composite (LMNC).

a Load-depth curves of the LMNC film and the tungsten carbide-liquid metal (WC-LM) substrate, respectively. The nano-hardness (H) and elastic modulus (E) obtained from the corresponding tests are marked next to the curves. Inset is the electron localization function (ELF) of the ideal WC/LM interface. b I-V curves of the film and substrate, respectively. Inset is the electrostatic potential at the ideal WC/LM interface, and the red dotted line indicates the Fermi level. c Modulus of the LMNC and other LM composites. d Conductivity and solid fraction of the LMNC, WC-LM substrate and other types of LM composites. e Comparison of the contact resistivity and interfacial current density (J_c) of different types of contact pairs. f Comparison of the contact quality (R_{c, ideal}/R_{c, real}) of different contact pairs. g Morphology and the corresponding elemental distribution at the contact area after sliding. h Structure of the LMNC film and the substrate. i Cross-sectional topography of the contact zone, showing the significant difference in the morphology between the protruded film and the substrate. The fabricating of LMNC film was repeated independently at least three times and similar topographies were observed. j The microstructure and k the corresponding electron diffraction pattern of LMNC. Source data are provided as a Source Data file.

Moreover, the composite film can still maintain excellent performance under a large current of 10 A and 0.4 GPa contact pressure. This is equivalent to a high interfacial current density above 1.5 GA/m² (Supplementary Fig. 12). After a prolonged sliding test lasting 50,000 cycles (i.e., 17.4 h), the ECR maintains at a low level about 1.0 mΩ (Fig. 2a), and the film still has not been extruded or damaged (Fig. 2b, c). The contact resistivity and interfacial current density of different contact pairs reported in literature^41–57 are compared with those of our contact pair (Fig. 3e and Supplementary Table 4), showing that the film prepared via FIIS exhibits a low contact resistivity and the highest interfacial current density. Notably, Fig. 3e only compares electrical contact systems with load-bearing capacity and does not include other conductors as well as tribofilms with other excellent properties but without reported interfacial conductivity. In addition, the ratio of ideal ECR (R_{c, ideal}) and real ECR (R_{c, real}) is calculated to evaluate the contact quality. The R_{c, ideal} is calculated using the Holm formula, representing the contact resistance between two materials under ideal conditions, that is, with perfectly smooth interfaces and no interfacial contaminants. Although the existing metal-based electrical contact materials have high bulk conductivity, their interfacial conductivity in the actual system is always inferior to that in the ideal situation^42,43,46,48. In comparison, the contact pair proposed in this work can achieve a high R_{c, ideal}/R_{c, real} value of 0.99 (Fig. 3f and Supplementary Table 5). This implies that the LM composite film possesses deformability, allowing it to fill almost all the gaps at the rough contact interface (Fig. 2f), thereby achieving a contact quality close to the theoretical limit.

To elucidate the origin of the exceptional performance, we investigated the composition and structure of the LM composite film. First, elemental analysis indicates the distinct enrichment of Ga element inside the film (Fig. 3g). Besides, the LM is almost concentrated in the contact area, without spreading around, which avoids the risk of LM leakage and promotes the formation of the protuberant film. The slight wear of the grinding ball (Supplementary Fig. 13) implies that material transfer is not the main factor for the film growth. In addition, Fig. 3g shows that the film is a uniform composite containing both W and Ga element.

In order to analyze the internal structure of the composite film and reduce the damage on LM during the milling for specimen characterization⁵⁸, we used the cryo-plasma focused ion beam (Cryo-PFIB) technology to expose the cross-sectional structure. There is a significant structural difference between the film and the substrate (Fig. 3h, i; Supplementary Fig. 14). The microstructure image (Fig. 3j) and the corresponding electron diffraction pattern (Fig. 3k) show that some WC nanocrystals and gallium oxide are embedded in the film, encapsulated by amorphous Ga. From Fig. 3j, there are no distinct boundaries between WC, Ga₂O₃, and Ga, indicating a very tight bonding. Based on these observations, we can conclude that the frictional shear under high contact stress transforms the WC particles into nanocrystals, and homogenizes them with LM. Consequently, the generated film is a type of LMNC, i.e., a dense and uniform solid-liquid composite mixed with WC/Ga₂O₃ nanocrystals bridged by high-surface-tension LM.

To explain why the above structure exhibits high mechanical properties and electrical conductivity, on the one hand, we calculated the interfacial electronic characteristics of WC and LM using density functional theory. First, the electron localization function (ELF) values between Ga/W, In/W, and C/W atoms are all close to 0.5, indicating that WC and LM in the film are bonded through metallic bonds (Fig. 3a, inset, Supplementary Data 1). As a result, these uniformly dispersed WC nanocrystals are firmly held together by LM and do not separate under the high contact pressure. Besides, the maximum value of electrostatic potential at the interface is below the Fermi level (Fig. 3b, inset, Supplementary Data 1), suggesting that there is no tunneling barrier at the ideal WC/LM interface. Meanwhile, the projected local density of states (PLDOS) between the Fermi levels of the two WC electrodes are ≈10⁻²eV⁻¹Å⁻³, which is of the same order of magnitude as that inside the LM (Supplementary Fig. 15). These electronic states contribute as carriers during electrical transport, which facilitate the high conductivity of LMNC. It should be noted that the real phase interfaces in LMNC film are nanocrystalline and oxidized, which results in lower conductivity than that of an ideal system. However, the conductivity of LMNC can still reach 4 × 10⁶S/m, which is between that of WC and LM.

On the other hand, we calculated the capillary suction pressure⁵⁹ of this liquid-bridged structure:

$$P_{\mathrm{c}}=k\gamma V\frac{F^{*}}{d}$$ 1

$k$: average coordination number ($k=2e^{2.4V}$ if the particles are spheres with an equal diameter); $\gamma$: liquid surface tension; $V$: volume fraction of particles; $F^{*}$: dimensionless interparticle force; $d$: particle diameter. $F^{*}=F/\left(\pi \gamma d\right)$, where $F$ is the capillary force between particles. Due to the construction from loosely packed micron-sized particles, the existing LM composites and other liquid-bridged materials (e.g., wet soil) possess the capillary pressure of merely 0.1–1 MPa⁶⁰, which makes them prone to forming a discontinuous structure. In contrast, the LMNC film formed under the mechanochemical stimulation features a high fraction of nano-sized WC crystals (Fig. 3d and Supplementary Fig. 14), and the capillary pressure generated by the high-surface-tension LM is estimated to be around 200 MPa (Supplementary Discussion 1 and Supplementary Fig. 16). This enables it to maintain a dense structure (Fig. 3h, j) while exhibiting high hardness and modulus. As a result, the LMNC proposed in this work combines the high deformability and conductivity of LM, and the high mechanical strength of conductive ceramics.

Formation mechanism of LMNC

Finally, we investigated the mechanism and key influencing factors of the LMNC film formation. The formation and mechanism of various anti-wear tribofilms have been reported in literature⁶¹. For example, when the steel-steel contact pair slides in oil with added ionic liquid, the reaction products of iron and ionic liquid will deposit on the surface, forming a tribofilm through pressing and tribochemical reactions⁶². These studies provide inspiration for analyzing the formation mechanism of the LMNC films. Unlike the solid-state tribofilm reported previously, the synthesis of LMNC involves the participation of LM and the formation of a conductive nanogranular structure, which contains both solid and liquid phases. This indicates that the FIIS process proposed in this work has a mechanism different from that in other literature.

We found that the ambient pressure during the friction process is crucial for the formation of LMNC. In the test of changing the ambient pressure during sliding, the ECR of the contact pair began to significantly increase after the pressure increased to ~1 Pa, and the ECR returned to a low level when the pressure decreased by 10⁻⁵Pa again (Fig. 4a). To further analyze the influence of environmental pressure on ECR and LMNC film formation, we conducted tests at different ambient pressure. The interface can only exhibit protruded LMNC films under 10⁻⁴ ~ 10⁻²Pa (Fig. 4c and Supplementary Fig. 17). Furthermore, the contents of Ga and O elements at the contact zone vary nonmonotonically with the increase of ambient pressure, indicating that an appropriate amount of oxygen can assist the formation process. Besides, the ECR can be maintained at ~3 mΩ within the range of 10⁻⁵Pa to 10⁻²Pa, but its average and fluctuation increase significantly after the ambient pressure exceeds 1 Pa (Supplementary Figs. 18, 19).

Fig. 4. Oxygen assisted liquid metal (LM) release and liquid-metal-bridged nanogranular composite (LMNC) film formation.

a Electrical contact resistance (R_c) variation under different ambient pressure at a sliding velocity of 5 mm/s and an electric current of 2 A. b Ab initio molecular dynamics (AIMD) simulation of oxygen promoting LM release. The blue dotted line shows the original position of the LM surface, which protrudes after oxygen is absorbed by LM. c Element content of the LMNC film formed under different ambient pressure. The dotted lines mark the element content of the substrate surface. Insets are the surficial morphology, height profile and Ga element distribution at the contact zone obtained under the corresponding ambient pressure. Source data are provided as a Source Data file.

To elucidate the effect of oxygen on film growth, we conducted ab initio molecular dynamics (AIMD) simulations. The results show that when an oxygen molecule is present, it will be adsorbed by the LM and subsequently diffuse into its interior, inducing a protrusion of the LM (Fig. 4b, Supplementary Data 2 and 3) that promotes its release from the substrate. However, when excessive oxygen is present around the contact interface, a dense oxide layer rapidly forms on the LM surface⁶³, thereby hindering its subsequent release. This explains the variation of Ga and O contents with environmental pressure in Fig. 4c.

The applied load also has a significant impact on the stability of the LMNC film (Supplementary Fig. 20). When the load increases from 0.5 N to 2.6 N, the cross-sectional area of the LMNC film formed on the surface of the WC-LM composite increases, indicating that the external stress promotes formation of the LMNC film. However, as the load continues to increase, the cross-sectional area of the film decreases and signs of wear appear. In addition, the impact of counterpart hardness, sliding velocity, and interfacial electric current were also studied (Supplementary Figs. 21–23).

Based on above, the FIIS process for film fabrication involves three simultaneously occurring processes:① the gradual release of LM from the WC-LM substrate under the influence of oxygen molecules (Fig. 4); ② friction leading to slight wear of the WC matrix and the formation of WC nanocrystals (Fig. 3j, k); ③ the thorough mixing of LM and WC nanocrystals under high contact pressure (Fig. 3g–i). This mechanochemical process can be regulated by the oxygen content, resulting in the uniform, dense, and controllable film growth.

In summary, we propose a FIIS strategy to fabricate LMNC film that simultaneously possess excellent electrical conductivity and mechanical strength, realizing a low contact resistivity and a high hardness of 2.6 GPa. Among the reported electrical contact systems, LMNC film achieves the highest interfacial current density of 1.5 GA/m². The FIIS approach enables the concurrent realization of LM release, WC nanocrystallization, and thorough mixing under GPa-level contact pressure. Such mechanochemical processes promote the formation of metallic bonds across the WC/LM interfaces and effectively eliminate the interfacial tunneling barrier, forming the unique material composed of high-fraction (~66 vol%) ceramic nanocrystals bridged by LM. Other materials that are chemically stable with LM and can form strong interfacial bonding with LM, such as tungsten, iron, aluminum nitride, etc., are also expected to be used as the solid fillers to form composites with an LMNC structure through the FIIS method, achieving various functional properties.

Methods

Materials

Tungsten carbide powders with an average particle size of 50 nm (WC, 99.9%) were purchased from Zhongkeleiming Daojin Technology Co., Ltd (China, Beijing). The nanoscale powders help decrease the required sintering temperature and reduce defects such as pore formation within the composites. Eutectic gallium-indium alloy containing 75.5% gallium and 24.5% indium by weight (EGaIn, 99.99%) was selected as the LM in this work and purchased from Sino Santech Materials Technology Co., Ltd (China, Changsha).

Fabrication of the WC-LM substrate and pure WC

To prepare the WC-LM substrate, WC powders and LM were evenly mixed with a mass ratio of 85:15 via mechanical stirring. Subsequently, the mixture was placed in a graphite die and sintered under the protection of argon atmosphere with the heating rate of 100 °C/min, axial pressure of 70 MPa and temperature of 1350 °C, followed by holding for 15 min and then cooling to room temperature in the furnace (FHP-828, HAATN, China). The argon atmosphere suppressed vaporization and oxidation of LM during the high-temperature sintering. Finally, the sintered composite underwent grinding and polishing to reduce surface roughness to below Ra 0.05 μm. The LM content of the produced WC-LM substrate was reduced and lower than that in the mixed powder (Supplementary Fig. 2). This reduction is attributed to partial LM being squeezed out of the die during the hot-pressed sintering. The process of preparing pure WC was similar, except that the raw materials for sintering were replaced by pure WC powders, and the sintering temperature and pressure were 1450 °C and 90 MPa, respectively.

Hardness and elastic modulus

The Vickers hardness of the materials was measure by a microhardness tester (Qness Q60A+, ATM Qness GmbH, Austria). The holding time of load was 10 s. The test was repeated at least five times, and the average value was taken as the hardness measurement result. The nano-indentation test was conducted by a nanoindenter (Bruker, USA), obtaining the nano-hardness and elastic moduli. The indentation depth was controlled to be less than 1/10 of the film thickness to reduce the influence of substrate effects on measurement.

Conductivity

The electrical conductivity of WC-LM composite was measured by the four-probe method, using the digital micro-ohm meter (UT620C, UNI-Trend Technology, China) with a four-point probe. And the conductivity of the LMNC film was calculated based on the Holm formula using the measured ECR data.

Component characterization

The elemental composition of materials was analyzed by energy dispersive spectrometer (EDS, XPLORE30, Oxford Instruments, UK) with acceleration voltage of 15 kV and scanned area of ~4.5 mm × 3.4 mm. X-ray diffraction (XRD, Ultima IV, Rigaku, Japan) was used to obtain the phase components of samples with the scanning speed of 2°/min. X-ray photoelectron spectroscopy (XPS, K-Alpha, Thermo Fisher Scientific, USA) was used to determine the chemical states of the elements. The subsurface composition of the samples was obtained by argon ion etching.

Tribofilm formation and sliding tests

The fabrication of LM composite tribofilm was achieved using a self-built tribometer. The commercially available cemented carbide containing 94% WC and 6% cobalt by weight (WC-6Co, grade YG6) balls with the diameter of 8 mm were chosen as the counterparts. The friction process was conducted in a chamber with the controllable ambient pressure. The rotational radius is 2 mm and the experimental temperature is 27 °C. Unless otherwise specified, the ambient pressure is 10⁻³Pa, the applied load is 1.5 N, the electric current is 2 A, and the rotational speed is 24 rpm (equivalent to a linear velocity of 5 mm/s). The four-wire method was used to measure the electrical contact resistance (ECR) during sliding. An external power supply applied constant current between the ball and disk with two wires, and a voltmeter measured the voltage drop at the ball/disk interface simultaneously with another two wires. This method can avoid the influence of loop resistance (wires, slip ring, connectors) on the measurement results. The ambient pressure in the experiment was controlled by changing the rotational speed of the molecular pump connected to the chamber. The calculation method of contact pressure, interfacial current density, and contact resistivity can be found in Supplementary Note 1.

Morphology analysis

The three-dimensional (3D) morphology of the contact zone on the materials was measured via a white-light interferometer (NexView, ZYGO, USA) to calculate the wear rate (i.e., ${k_{\mathrm{wear}}=V}_{\mathrm{wear}}/\left(F_{\mathrm{n}}\cdot L\right)$ and apparent contact area (i.e., $A_{\mathrm{c}}=\pi w^2/4$), where V_wear and L represent the wear volume and sliding distance, respectively. A scanning electron microscope (SEM, Gemini 300, ZEISS Group, Germany) was used to obtain the secondary electron images of samples. A transmission electron microscopy (TEM, JEM-F200, JEOL, Japan) was also used to obtain the microstructure of the samples. The cryo-plasma focused ion beam (Cryo-PFIB, Helios 5 Hydra, Thermo Fisher Scientific, USA) facility was used to prepare the TEM sample of the LM composite film. The freezing condition is conducive to avoiding the melting and outflow of LM during ion beam milling. In addition, the ion species we chose is xenon. Compared with the traditional FIB using gallium ion, this helps to avoid the influence of ion on the LM composition.

Contact angle measurement

The receding contact angle between LM and WC was measured using the volume variation method⁶⁴ by an optical contact angle measuring instrument (Dataphysics Instruments GmbH, Germany).

Ab initio calculations

The calculations of the electron localization function (ELF) and electrostatic potential of the interfaces were performed using the Vienna Ab initio Simulation Package (VASP). The projector augmented-wave (PAW) method was employed to model the core electrons, and the kinetic energy cutoff for the plane-wave basis was set to 400 eV. Before calculating the electronic properties, the atomic configurations of the tribosystems were optimized until the forces on all relaxed atoms were smaller than 0.01 eV/Å. Ab initio molecular dynamics (AIMD), tunneling barrier, and projected local density of states (PLDOS) calculations were carried out using the Quantum ATK software. The Fritz–Haber Institute (FHI) type pseudopotentials were adopted in all calculations. During the molecular dynamics simulations, we used the spin-polarized generalized gradient approximation (SGGA) based on the Perdew–Burke–Ernzerhof (PBE) functional to account for the triplet state of the oxygen molecule. In the AIMD simulations, the temperature was set to 300 K, and the time step was set to 1 fs.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: T.M., A.S. and X.L. Methodology: J.Z. and A.S. Investigation: J.Z., A.S., X.T., J.L., W.L., C.H., H.W. and N.Z. Visualization: J.Z. and A.S. Funding acquisition: T.M., A.S. and H.W. Project administration: T.M., A.S. and X.L. Supervision: T.M. and X.L. Writing – original draft: J.Z. and A.S. Writing – review & editing: J.Z, A.S., X.T. X.L. and T.M.

Peer review

Peer review information

Nature Communications thanks Dariusz Jarząbek, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

The authors acknowledge the support of the National Natural Science Foundation of China (grant no. 52225502 to T.M.), Scientific and Technological Project of Yunnan Precious Metals Laboratory (grant no. YPML-20240502088 to X.L., T.M. and H.W.), the National Natural Science Foundation of China (grant no. U25A20298 to X.L., 52363033 to H.W.), Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (grant no. JYB2025XDXM305, JYB2025XDXM209 to T.M.), New Cornerstone Science Foundation through the XPLORER PRIZE to T.M.

Data availability

The data supporting the findings of this study are available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74665-x.