Atomic-scale reconstruction of sapphire induced by selective tribochemical removal of surface atoms

Abstract

Although biomimetic nanostructured surfaces can impart exceptional properties, their inherent fragility makes them prone to failure under mechanical wear. This study presents a method for friction-induced atomic-scale surface reconstruction of sapphire, resulting in durable, scale-like nanostructures that retain the hardness of bulk the sapphire (27.2 GPa). Friction experiments and reactive force field molecular dynamics simulation reveal that surface reconstruction is driven by the selective removal of atoms from less chemically inert crystal planes of sapphire via interfacial bridge bonds during friction with SiO₂. Subsurface characterizations demonstrate that the nanostructures maintain a consistent single-crystal structure with the original sapphire, thereby ensuring their exceptional mechanical properties. To evaluate its triboelectric performance and wear resistance, nanostructured sapphire is integrated into triboelectric nanogenerators (TENGs) with a diamond-like carbon (DLC) friction layer. Triboelectric tests indicate that after 100k cycles of friction, the discharge performance of TENGs exhibits no significant decline, and the nanostructures show negligible wear, demonstrating their extraordinary durability. This study introduces an atomic-scale strategy for nanostructure fabrication, providing an approach to surface reconstruction of sapphire and insights into the development of wear-resistant nanostructures.

Nanostructure fragility limits applications. The authors develop a tribochemical surface reconstruction strategy to build single-crystal nanostructures on sapphire, which retain bulk-level wear resistance and improve triboelectric device performance.

Introduction

Biological surfaces in nature often exhibit unique functions due to their distinct micro/nano structures, such as the anti-reflective properties of insect eyes¹, the superhydrophobicity of lotus leaves², and the wear resistance of snake scales³. Inspired by nature, various bio-inspired nanostructures with specific functionalities have been designed and widely applied in fields such as self-cleaning surfaces⁴, antireflective optical windows⁵, and triboelectric nanogenerators⁶. Despite their functional advantages, nanostructured surfaces are inherently fragile and susceptible to mechanical wear under external loads, particularly shear forces, which significantly compromise their durability^7–9. It is reported that surface wear is a major hindrance in the solid/solid interface of triboelectric nanogenerators¹⁰. The simultaneous achievement of robust charge generation and surface durability in triboelectric interfaces presents ongoing technological hurdles. Although certain bio-inspired nanostructures (e.g., scale-like ordered patterns) can enhance wear resistance, their long-term applicability in wear-prone environments is constrained by inherent material mechanical properties.

Enhancing the mechanical strength of materials is a fundamental approach to improving the durability of bio-inspired nanostructures. Thus, the selection of an anti-wear material is a prerequisite for fabricating nanostructure surfaces. Sapphire, a highly durable transparent ceramic with an exceptional hardness (Mohs: 9), is commonly used in optical components such as protective windows for optical devices and transparent windows for spacecraft, which is an ideal material for fabricating wear-resistant nanostructures¹¹. However, its extreme hardness and chemical inertness also make it difficult to process¹². The main methods currently used for sapphire nanostructure fabrication are chemical etching and laser processing. Chemical etching can obtain highly ordered periodic nanostructures through the combination of advanced nano-lithography and imprinting techniques¹³. Chien et al. fabricated a nanostructure with a period of 330 nm, however its hardness is only 3.7 GPa, much lower than that of unstructured bulk sapphire¹⁴. Femtosecond laser micromachining, a widely used tool for nanostructure fabrication, can create bio-inspired nanostructures with high aspect ratios on sapphire surfaces¹⁵. But the high-throughput laser ablation induces thermal damage to the sapphire, resulting in the formation of amorphous and polycrystal phases, which also reduces the mechanical strength of the sapphire^16,17. It is evident that, although these methods can produce sapphire nanostructures of various forms, they are unable to achieve the mechanical strength of the original sapphire.

Aside from these traditional methods mentioned above, surface reconstruction is another approach for forming nanostructures on flat sapphire surface. Surface reconstruction can be defined as the process in which the breaking of chemical bonds at the material’s surface leads to the rearrangement of surface atoms, resulting in the formation of new surface structures¹⁸. In recent years, energy-field-based surface reconstruction of sapphire has received extensive research. High-temperature annealing can transform a flat sapphire surface into nanoscale stepped structures, and by adjusting the annealing time and temperature, steps of specific heights can be tailored^19,20. These nanoscale stepped structures are widely used as substrates for epitaxial growth^21–23. Additionally, similar stepped structures can be obtained on sapphire surfaces through plasma-enabled atomic-scale reconstruction, with processing times reduced from several hours to just a few minutes²⁴. Previous studies have shown that sapphire nanostructures obtained via surface reconstruction exhibit near-perfect crystallinity, which greatly preserves the original mechanical strength of sapphire, offering significant potential for wear-resistant nanostructure fabrication^20,24. However, current research on sapphire surface reconstruction is still limited to energy-field-based approaches, and can only produce stepped nanostructures. Research into other forms of surface reconstruction remains unexplored.

In this study, we have presented the friction-induced atomic-scale surface reconstruction of sapphire, leading to the formation of a single-crystal, periodically ordered, scale-like sapphire nanostructure with hardness matching that of flat sapphire. By combining friction experiments and reactive force field molecular dynamics simulation, we comprehensively reveal the mechanism of selective tribochemical removal on the miscut sapphire surface, as well as the reconstruction process of the nanostructured surface, driven by friction with SiO₂. Both theoretical and experimental results indicate that during friction, the *OH groups at the interface undergo condensation reaction, forming Si-O-Al bridge bonds, which cause sapphire atoms to be removed under the shear force of these interface bridge bonds. The varying tribochemical removal rates of different atomic planes lead to the reconstruction of the flat sapphire surface into a unique scale-like nanostructure. Furthermore, by pairing the nanostructured sapphire with diamond-like carbon (DLC) to form a triboelectric nanogenerator (TENG), the wear resistance of the nanostructures was tested. After 100k cycles of friction, the discharge energy of the TENG remained near its peak value, with no significant wear observed on the nanostructures. This study offers a method for surface reconstruction of sapphire and produces a durable sapphire nanostructure, which is suitable for applications in fields requiring long-term wear resistance, such as triboelectric energy harvesting.

Results

Friction-induced fabrication of scale-like nanostructures in sapphire

During the cutting sapphire from ingots, a slight misorientation may exist between the actual surface and the theoretical crystal plane²⁵. The miscut angle is the angle between the workpiece surface and the intended crystal plane. The sapphire workpiece surfaces in this study have a certain miscut angle. To evaluate the specific value of miscut angles, we used high-resolution x-ray diffraction (XRD) to measure the miscut angle along the friction direction of the sapphire, as shown in Fig. 1a. There is a quantitative relationship between the miscut angle and the diffraction angles in the two detected directions²⁶. By rotating the sapphire horizontally by 180°, the diffraction angles on both sides of the friction direction can be measured. Due to the presence of the miscut angle, the diffraction angles deviate from the Bragg angle.

$$\mathit{a}=\frac{{\mathit{\theta }}_{\mathbf{2}}-{\mathit{\theta }}_{\mathbf{1}}}{\mathbf{2}}$$ 1

Fig. 1. Sapphire miscut angle measurement method and surface morphology of sapphire after friction with different materials.

a Spatial relationship between the XRD measurement direction and the friction direction. The colored square beneath the Al₂O₃ ball indicates the Al₂O₃ friction region, while the square beneath the SiO₂ ball indicates the SiO₂ friction region. b Relationship between the miscut angle and the direction of atomic step, the blue spheres represent atoms in the sapphire lattice. c Original surface morphology and corresponding cross-sectional profile. d Surface morphology and corresponding cross-sectional profile after friction with Al₂O₃ ball. e Surface morphology and corresponding cross-sectional profile after friction with SiO₂ ball. All AFM data shown in this figure were collected from different regions of the same sapphire substrate.

Here, α represents the miscut angle, and θ₁ and θ₂ are the diffraction angles in the two directions, respectively. Additionally, the presence of the miscut angle leads to the formation of atomic steps on the sapphire surface. For convenience, in this study, we define the positive direction as the one in which the atomic step height and the nanostructure height increase. The relationship between the step direction and the miscut angle is shown in Fig. 1b, where the side with the larger diffraction angle corresponds to the positive direction of the step.

Besides, friction experiments were conducted using a self-constructed device (Supplementary Fig. 1), and atomic force microscopy (AFM) was used to examine the surface morphology of the sapphire before and after friction, with results shown in Fig. 1c–e. The initial sapphire surface is atomically smooth after chemical-mechanical polishing (Fig. 1c). To elucidate the role of friction-induced chemical reactions in the surface reconstruction of sapphire, comparative experiments were performed using Al₂O₃ ceramic balls and SiO₂ balls on a same sapphire sample under identical processing parameters. Details of the friction experiments are provided in the Methods and Supplementary Section 1. The experimental conditions were as follows: sapphire miscut angle 1°, sliding speed 50 mm/min, sliding duration 30 min, and normal loads 2 N. Al₂O₃ balls are used as the control group because Al₂O₃ ceramics have essentially the same chemical composition as sapphire. Thus, the friction between Al₂O₃ balls and sapphire is dominant in mechanical interactions. On the other hand, the friction-induced chemical reactions between SiO₂ and sapphire are well-documented in polishing studies^27,28. Although macroscopic SiO₂ balls differ from nanoscale SiO₂ abrasives in their physicochemical properties, their consistent chemical composition provides a theoretical basis for the tribochemical reaction between the SiO₂ ball and sapphire. Our experimental results clearly demonstrate that the sapphire surface after friction with Al₂O₃ ball (Fig. 1d) exhibits typical plowing wear marks, while the surface after friction with SiO₂ ball (Fig. 1e) undergoes reconstruction and forms a periodically ordered nanostructure. This comparison confirms that friction-induced chemical reactions are a necessary condition for the surface reconstruction of sapphire, and pure mechanical interactions alone cannot produce nanostructures. Furthermore, after extensive validation through repeated experiments, the surface reconstruction exhibits high uniformity and reproducibility (Supplementary Section 2) and we observe an interesting phenomenon: the positive direction of the nanostructures consistently aligns with that of atomic steps caused by the miscut angle (Fig. 1b and e). This indicates an intrinsic relationship between the nanostructures and the miscut angle.

To further investigate the relationship between the miscut angle and the nanostructures, C-plane sapphire samples with different miscut angles of 0°, 1°, 3° and 5° were subjected to SiO₂ ball friction experiments. The resulting surface morphologies after balls sliding are shown in Fig. 2. Fig. 2a–d display the surface morphology of sapphires with four different miscut angles after friction. Notably, the sapphire with a 0° miscut angle exhibits no nanostructures after friction, while sapphires with 1°, 3°, and 5° miscut angles form varying-sizes nanostructures. It demonstrates that a certain miscut angle is another essential condition for nanostructure generation. Fig. 2e–h show the magnified images of the regions corresponding to Fig. 2a–d. As the miscut angle increases, the nanostructures become shorter and more densely packed. Moreover, for clarity of description, we define the angle between the longer axis of the nanostructures and the sapphire surface as the nanostructure angle, and Fig. 2j–l show that the nanostructure angle increases with the miscut angle of sapphire surface. These changes align with the trend of surface atomic steps: as the miscut angle increases, the atomic steps become shorter, the angle between steps and the surface increases, and their density rises accordingly^29,30. Similarly, an increase in the miscut vertical to the sliding direction also affects the nanostructure morphology. Specifically, as this vertical miscut increases, the nanostructures gradually evolve from continuous steps to discrete, scale-like nanostructures (Supplementary Fig. 8). In addition, the experiments with varying friction durations on the same sapphire (Supplementary Fig. 9) show that the length and height of the nanostructures increase with friction time, while the angle remains consistent with the miscut angle (with a deviation of no more than 0.5°). These findings strongly suggest that the formation of nanostructures is guided by underlying atomic steps. Accordingly, variations in the miscut angle lead to synchronized changes in the geometrical dimensions of both the nanostructures and the atomic steps. Given that the presence of atomic steps introduces two different crystal planes on the surface, we infer that the difference in chemical reactivity between the crystal planes result in differential atomic removal rates during friction, which leads to the formation of scale-like nanostructures.

Fig. 2. Surface morphology of sapphire with different miscut angles after friction.

a–d AFM images of sapphire with miscut angles of 0°, 1°, 3°, and 5° after 30 min of friction. e–h Magnified images of the regions corresponding to (a–d). i–l Profiles of the corresponding positions in (e–h). The red dotted lines in (i–l) indicate the parallel nanostructures.

Material transfer mechanisms during friction

In order to verify the above hypothesis, the material removal behavior is analyzed during friction process. The surface morphologies of the SiO₂ ball and sapphire were examined by scanning electron microscopy (SEM), and their elemental compositions were characterized by energy dispersive spectroscopy (EDS). It can be seen that the friction region of the SiO₂ ball shows typical round wear scar (Fig. 3a). At the edge of the wear region of the SiO₂ ball, we observe attachments (Fig. 3b). EDS mapping of attachment reveals the presence of Al element (Fig. 3c), which is likely a product of the tribochemical reaction. In addition, we characterized the three-dimensional features of the SiO₂ ball and supplemented the analysis of changes in its surface topography and roughness before and after friction. After sliding, the apex of the SiO₂ ball was worn down by ~3.2 μm, and the surface roughness Sa decreased from 1.40 nm to 0.58 nm (Supplementary Section 5). Notably, owing to the sufficiently large wear depth, even SiO₂ balls with initially rough surfaces are capable of producing regular nanostructures on the sapphire surface. However, wear mark is difficult to be identified in the friction region of the sapphire (Fig. 3d). Upon magnification of the friction region, the distinct scale-like nanostructures (Fig. 3e) are observed. Fig. 3f presents the EDS spectra acquired from the friction regions shown in Fig. 3a and Fig. 3d, corresponding to the SiO₂ ball and the sapphire, respectively. Al was detected in the wear scar of the SiO₂ ball, but no Si was found in wear track of the sapphire. These results suggest that some material of the sapphire surface was removed and transferred to the SiO₂ ball, but no SiO₂ material was transferred to the sapphire surface. We further analyze the sapphire friction region using x-ray photoelectron spectroscopy (XPS) (Supplementary Section 6). The high-resolution spectrum data of Si 2p is discrete, which provides stronger evidence that friction products do not adhere to the sapphire surface. Therefore, the material transfer during the friction process is unidirectional, meaning the bonding force of the friction products is stronger at the SiO₂ end.

Fig. 3. Surface morphology and elemental analysis of sapphire and SiO₂ friction regions.

a SEM image (SE mode) of the friction region of SiO₂. b Magnified image of the edge of the SiO₂ ball friction region. c EDS mapping of the attachment on the SiO₂ ball. d SEM image (SE mode) of the friction region of sapphire. e Magnified image of the sapphire friction region. f EDS spectrum of the SiO₂ ball and sapphire friction regions. g–l XPS spectra of the SiO₂ ball before and after friction. Peak fitting was performed using Gaussian–Lorentzian functions after Shirley background subtraction: g Survey, h Al 2p and i Si 2p spectra before friction. j Survey, k Al 2p and l Si 2p spectra after friction.

Subsequently, XPS analysis is conducted on the SiO₂ balls before and after friction to determine the specific composition of the attachments and further reveal the material removal mechanism on the sapphire surface. The procedure for selecting the measurement regions is detailed in Supplementary Section 7. The full spectrum of the original SiO₂ ball is shown in Fig. 3g, where peaks for Si and O are detected, along with a C contamination peak. Fine scans of the Al 2p and Si 2p regions reveal discrete Al 2p data (Fig. 3h), and Si 2p only shows the Si-O peak at 103.0 eV. After friction, the full spectrum of the SiO₂ ball with attachments is shown in Fig. 3j. In addition to the previously detected elements, Al is also detected. Peak fitting of Al 2p and Si 2p reveals peaks for Al-O-Si at 74.5 eV and Al-OH at 75.8 eV, and for Si-O-Al, Si-O, and Si-OH at 101.9, 103.1, and 104.3 eV, respectively^31–34. The appearance of the Al-OH and Si-OH peaks indicates that H₂O molecules dissociate during the friction process, and *OH groups adsorb at the surface of sapphire and the SiO₂ ball. The Si-O-Al bridge bonds formed are products of *OH condensation reaction, capable of transmitting shear stress³⁵. In summary, due to the stronger bonding of attachments (Si-O-Al bridge bonds) at the SiO₂ end, some sapphire atoms are removed from the surface in the form of Si-O-Al under shear stress during the friction process.

Surface reconstruction based on atomic selective tribochemical removal in crystal surface

Furthermore, we analyze the subsurface characteristics of sapphire before and after friction using transmission electron microscopy (TEM) to explore the underlying mechanisms of sapphire surface reconstruction. Given that different miscut angles lead to varying nanostructure lengths, we selected the sapphire sample with a 3° miscut angle for observation to ensure an appropriate number of steps within the field of view. Fig. 4a and b show the sub-surface characteristic of sapphire before and after friction. After friction, periodic nanostructures are observed, with an angle of ~3°. Fig. 4c–f show the HRTEM images of the original sample. Fig. 4c is a high-resolution image corresponding to the blue boxed region in Fig. 4a, and HRTEM images of the yellow and red boxed regions in Fig. 4c are shown in Fig. 4d, e, respectively. As the original sample had undergone chemical mechanical polishing, although it exhibits an atomically smooth surface, an ~2 nm subsurface damage layer remains (Fig. 4d). In addition, the sapphire surface clearly exhibits a miscut angle of ~3° relative to the (0006) crystal plane. Fig. 4f presents the fast Fourier transform (FFT) and inverse FFT (IFFT) results of the (0006) crystal plane shown in Fig. 4e. The FFT images corresponding to Fig. 4e and Fig. 4i, along with the information on the masked spots, are shown in Supplementary Section 8. The measured interplanar spacing is 0.22 nm, and the lattice fringes are regular and well defined, with no discernible defects. Similarly, Fig. 4g–j show high-resolution observations of the nanostructure formed after friction. Fig. 4g corresponds to the blue boxed region in Fig. 4b, and HRTEM images of the yellow and red boxed regions in Fig. 4g are shown in Fig. 4h and 4i, respectively. As seen in Fig. 4h, no damage features similar to those in Fig. 4d are observed in the subsurface. Fig. 4j shows the FFT and IFFT results of the (0006) crystal plane derived from Fig. 4i. Consistent with the original sample, an interplanar spacing of 0.22 nm is obtained, and no obvious lattice defects are observed. These results demonstrate that the sapphire surface reconstruction strategy proposed in this study represents a low-damage, atomic-scale fabrication approach. It removes the pre-existing subsurface damage layer without introducing additional lattice defects, thereby preserving the intrinsic crystalline integrity of sapphire. Additionally, we further analyzed the elemental composition of the nanostructure sub-surface through EDS (Supplementary Section 9), and the results confirm that it consists only of Al and O elements, consistent with the conclusions in XPS.

Fig. 4. Subsurface analyses of sapphire ((0006) plane with a miscut angle of 3° along the $\left[11\overline{2}0\right]$ axis) before and after friction.

Subsurface morphology of sapphire (a) before and (b) after friction. c High-resolution TEM (HRTEM) image of the blue box region in (a). d HRTEM image of the yellow boxed region in (c). e HRTEM image of the red boxed region in (c). f FFT and inverse FFT of (e). g HRTEM image of the blue box region in (b). h HRTEM image of the yellow boxed region in (g). i HRTEM image of the red boxed region in (g). j FFT and inverse FFT of (i). k The surface reconstruction process of the sapphire scale-like nanostructure.

In conclusion, based on the above results and the discussion on material removal mechanisms, the surface reconstruction process is illustrated in Fig. 4k. For sapphire with a certain miscut angle on the (0006) plane, the surface contains two crystal planes due to the presence of atomic steps: the close-packed (0006) plane and the non-close-packed plane. As the close-packed plane is the most chemically stable surface of sapphire, the atoms on this plane are less susceptible to be removed in the form of Si-O-Al bonds during friction compared to those on non-close-packed plane^36,37. Consequently, there is a higher removal rate of atoms on the non-close-packed plane, resulting in an unequal material removal between the two crystal planes and ultimately leading to the formation of scale-like nanostructures. This explains why the nanostructure angle exactly equals the miscut angle. Since sapphire atoms are removed in the form of Si-O-Al bonds under shear stress, significant amorphization is avoided, allowing the nanostructure to retain its single-crystal form and preserving its mechanical strength consistent with the original sapphire. Therefore, friction-induced surface reconstruction of sapphire presents a nanostructure forming method with atomic-scale fabrication capabilities.

Atomic interaction and material removal mechanism during the friction process

To further clarify the atomic interaction mechanism between different atoms at the friction interface, we conduct a reactive force field molecular dynamics (ReaxFF-MD) simulation of amorphous SiO₂ friction with miscut sapphire in the presence of H₂O molecules. The simulation details can be found in the “Methods” section and Supplementary Section 10. The peak positions of the radial distribution function (RDF) reflect the most probable bond length, and when the displacement of an atom relative to its initial position exceeds the cutoff distance where the RDF is largest, the atom is considered to have been removed due to bond breaking^38,39. From the RDF of sapphire (Supplementary Fig. 21), it is clear that atoms displaced by more than 2.1 Å can be regarded as having been removed from the sapphire substrate. Fig. 5b shows the difference in atomic removal on the sapphire surface during forward and reverse friction. Due to the miscut angle, during forward sliding, a significant number of atoms are removed at the surface step, while during reverse sliding, only a few atoms from the close-packed plane are removed. Supplementary Movie 1 and 2 show the atomic motions during forward and reverse friction, respectively, where these phenomena can be clearly observed. The statistical analysis of atomic removal quantities in different friction directions (Fig. 5f) also reflects this phenomenon. Supplementary Section 11 shows the surface morphology of sapphire after unidirectional cyclic friction, where nanostructures appear during forward friction, but not in reverse friction. This demonstrates that the removal of non-close-packed plane atoms is a necessary condition for the formation of nanostructures, further confirming the surface reconstruction mechanism based on atomic selective tribochemical removal from crystal planes.

Fig. 5. ReaxFF-MD simulation of friction between amorphous SiO₂ and sapphire.

a Friction interface before friction. b Differences in sapphire surface atoms removal during forward and reverse friction. c Material removal mechanism of sapphire. Non-transparent spheres represent a set of atoms involved in the tribochemical reaction, showing the same atoms tracked across the time series. d Changes in the number of H₂O molecules during friction. e Changes in the number of *OH groups and Si-O-Al bridge bonds during friction. f Variation in the number of sapphire surface atoms removed during forward and reverse friction.

Figure 5c illustrates the sapphire material removal mechanism induced by friction. Initially, *OH groups dissociated from H₂O molecules adsorb on the SiO₂ and sapphire surfaces. As friction progresses, the *OH groups release a hydrogen atom, which subsequently condense to form Si-O-Al bridge bonds. The released hydrogen atom then combines with free hydroxyl groups in the environment to form H₂O molecules. Subsequently, under the action of shear stress, the Si–O–Al bridge bond ruptures at the interface with the sapphire and detach from its surface. The free *OH groups in the environment terminate the vacancies exposed by the removed Al atoms, leading to the formation of new Al-OH groups. The presence of Al-O-Si and Al-OH bonds in Fig. 3k supports this mechanism. Meanwhile, the oxygen atoms on the sapphire surface are also concurrently removed during friction, with two distinct removal modes observed: one in which O atoms are removed together with the Si–O–Al bridge bonds, and another in which they desorb from the surface in the form of H₂O molecules; details are provided in Supplementary Section 12. In addition, to verify the critical role of water molecules in the material removal process, we investigated the material wear behavior of SiO₂ balls and sapphire under a water-free environment through vacuum friction experiments (Supplementary Section 13). These results confirm that the participation of water molecules is a necessary condition for the occurrence of the aforementioned tribochemical reactions. Further analysis of the changes in the quantities of different groups during the friction process can shed light on the atomic interaction mechanism. The number of H₂O molecules continuously decreases during friction (Fig. 5d), as H₂O molecules decompose into *OH groups and adsorb on the friction interface (Fig. 5a). This results in an increase in the number of *OH within the first 250 ps (Fig. 5e). After 250 ps of friction, the number of Si-O-Al bridge bonds rapidly rises, consuming the *OH adsorbed at the interface, leading to fluctuations in the number of *OH groups. These variations further support the material removal mechanism based on friction-induced *OH condensation reaction and shear-driven breaking of interfacial bridge bonds.

Triboelectric and wear resistance properties of scale-like nanostructures

Triboelectric nanogenerators (TENGs) are sustainable energy harvesting technologies that convert mechanical energy into electrical energy through contact electrification and electrostatic induction⁴⁰. Previous studies have shown that the nanostructures on the friction surface can enhance the contact area, thereby effectively improving the discharge efficiency of TENGs⁴¹. In this study, we fabricate sapphire with nanostructures that hold promise for constructing TENGs with both high discharge efficiency and wear-resistant. Herein, we assembled two sliding-mode TENGs: one with flat sapphire and the other with nanostructured sapphire, both paired with DLC coated glass, and performed long-term friction discharge tests. A stable normal load and reciprocating motion were provided by a tribometer, with the normal load set to 2 N, a reciprocating stroke length of 20 mm, and a sliding frequency of 2 Hz. The detailed composition and electrical performance measurement method are provided in the Supplementary Section 14. Fig. 6a–c show the open-circuit voltage, current density, and charge density of flat sapphire-DLC (FS-DLC) TENG during 100, 50k, and 100k cycles of friction. It can be observed that the discharge performance of the TENG remains largely unchanged during the friction process. This indicates that although the excellent mechanical properties of sapphire prevent significant wear, the flat sapphire surface fails to effectively enhance the microscopic contact area with the DLC layer, resulting in persistently low charge transfer efficiency. In contrast to FS-DLC TENG, the nanostructured sapphire-DLC (NS-DLC) TENG exhibits a 2-3 times increase in open-circuit voltage, current density, and charge density during the friction (Fig. 6d–f), indicating that the nanostructures significantly enhance the microscopic contact area between sapphire and DLC. Moreover, after 100k cycles of friction, the discharge performance shows no noticeable decrease, demonstrating that nanostructured sapphire exhibits comparable wear resistance to that of flat sapphire.

Fig. 6. Discharge performance testing and surface morphology changes of sapphire-DLC triboelectric nanogenerators (TENG) before and after friction.

a–c Open-circuit voltage, current density, and charge density variations of flat sapphire-DLC (FS-DLC) TENG in the early, middle, and late stages of friction. d–f Open-circuit voltage, current density, and charge density variations of nanostructured sapphire-DLC (NS-DLC) TENG in the early, middle, and late stages of friction. In (a–f), different colored curves represent measurements collected at 100, 50k, and 100k friction cycles, respectively. Each colored curve corresponds to 5 s of data acquired at the specified cycle count. Surface morphology of DLC in the (g) FS-DLC TENG and (h) NS-DLC TENG after friction, along with the AFM image of the friction region. The inset of AFM image shows the profile along the white dashed line. AFM images of flat sapphire (i) before and (j) after friction. AFM images of nanostructured sapphire k before and (l) after friction. TENG operating conditions during friction with a normal load of 2 N, a relative sliding distance of 20 mm, and a sliding frequency of 2 Hz.

Furthermore, the surface morphology of DLC and two different sapphire surfaces before and after friction is characterized (Fig. 6g–l). The original DLC surface morphology is shown in Supplementary Section 15. After friction with flat sapphire, uneven friction region appears on the DLC surface (Fig. 6g), which is due to the difficulty in achieving complete surface-to-surface contact between flat sapphire and DLC. The AFM image reveals that the DLC surface exhibits protruding plastic deformations, caused by compression and shear force during the friction process. These deformations do not significantly increase the contact area with sapphire, and thus the FS-DLC TENG maintains a consistently low discharge efficiency. In contrast, after friction with nanostructured sapphire, uniform friction region appears on the DLC surface (Fig. 6h), indicating that the nanostructures effectively enhance the uniformity of contact between sapphire and DLC, thereby increasing the overall contact area. The AFM image further reveals numerous nanoholes in the friction region of DLC, whose formation is attributed to the difference in mechanical properties between DLC and sapphire. Supplementary Section 16 presents the mechanical properties test results of sapphire and DLC. It is evident that the hardness of nanostructured sapphire (27.2 GPa) is comparable to that of flat sapphire (26.9 GPa), and significantly higher than that of DLC (11.6 GPa), which ensures that the sapphire surface exhibits no observable wear even after 100k cycles of friction (Fig. 6i–l). It indicates that the sapphire nanostructures obtained by friction-induced atomic-scale surface reconstruction retain mechanical strength nearly identical to the original sapphire, which aligns with the findings in TEM that the nanostructures are single-crystal. Moreover, when the indentation depth is less than 6 nm, the DLC undergoes elastic deformation, whereas plastic deformation occurs at a depth of 12 nm (Supplementary Fig. 31b). This indicates that the elastic–plastic transition point of the DLC lies between 6 and 12 nm, which is comparable to the height of the nanostructures. Therefore, when the nanostructures come into contact with the DLC, they embed into its surface, induce elastic–plastic deformation⁴², and leave behind nanoholes. This process significantly increases the microscopic contact area between the DLC and sapphire, provides more charge transfer sites, and thereby enhances the discharge performance of the TENG. In addition, the surface chemical compositions of sapphire and DLC after friction were analyzed by XPS (Supplementary Section 17). The results indicate that any carbon transfer from the DLC to the sapphire surface during the friction process is negligible and has no discernible influence under the present experimental conditions, and the friction interface maintained high chemical stability, which likewise contributes to reduced wear in the TENG. In summary, the nanostructured sapphire increases the microscopic contact area by inducing elastic–plastic deformations in the DLC, while its mechanical properties—comparable to those of flat sapphire—ensure excellent wear resistance. As a result, it not only enhances the discharge performance of the TENG but also maintains long-term discharge reliability. By comparison with the reported output performances of sliding-mode TENGs based on other material systems (Supplementary Section 18), although the TENG fabricated in this work does not achieve the highest reported values for every performance parameter, its overall output reaches a level comparable to that reported in the literature. Moreover, given that the present device adopts a relatively simple structural configuration, further improvements in output performance are expected through subsequent structural optimization.

Discussion

In this work, an atomic-scale surface reconstruction method for sapphire is proposed. The surface reconstruction mechanism based on selective removal of crystal surface atoms is revealed, and the periodic, ordered scale-like sapphire nanostructures are fabricated. By combining ReaxFF-MD simulations and friction experiments, the atomic interaction mechanisms and the formation process of nanostructures during the friction between amorphous SiO₂ and sapphire are investigated. During the friction process, Si-O-Al bridge bonds are formed through dehydration condensation of *OH groups at the interface, and the removal of sapphire atoms is achieved by the shear stress transmitted through these bridge bonds. Due to the miscut angle, the (0006) sapphire surface simultaneously contains another non-close-packed plane. Owing to higher chemical inertness of atoms on the (0006) plane, atoms from the non-close-packed plane are removed at a higher rate, leading to the formation of nanostructures. Based on this atomic-scale low-damage material removal mode, the nanostructures remain single-crystal, ensuring that their mechanical properties are consistent with those of the original sapphire. Furthermore, by combining nanostructured sapphire with DLC, a TENG with high discharge efficiency is created. After 100 k cycles of friction, no significant wear is observed on the nanostructures, and their electrical performance remains stable, indicating stable wear resistance of the obtained nanostructures. As a surface reconstruction process driven by tribochemical reactions, the outcome is expected to be strongly influenced by the tribochemical reaction rate. This issue is discussed in Supplementary Section 4, where the results show that the reconstruction behavior is highly sensitive to the contact pressure at the friction interface. In the future, the morphology of the nanostructures is expected to be tuned by adjusting the sapphire miscut angle and the interfacial pressure during friction, thereby further optimizing the discharge efficiency of the TENG. This work presents an approach for sapphire surface reconstruction and introduces a biomimetic, wear-resistant sapphire surface, which is expected to maintain long-term stability in extreme environments such as high temperature, high pressure, and prolonged friction.

Methods

Sample preparation

Single crystal sapphire (0006) with a miscut angle of 0°, 1°, 3°, 5° in $\left[11\overline{2}0\right]$ axis was purchased from Jingjing Technology Co., Ltd (Hubei, China). The size of the sample is 15 × 10 × 3 mm, polished by chemical mechanical polishing (Sa <0.2 nm). The SiO₂ ball (99.99 % silica content) and the Alumina ceramic ball (purity > 99.5 %) were purchased from Quanzhou Hanyi Trading Co., Ltd (Fujian, China) and Chenfu Hardware Bearing Trade Co., Ltd (Hebei, China), and the diameter of balls was 8 mm. All samples were ultrasonically cleaned in anhydrous ethanol and ultra-pure water for 10 min, and dried with nitrogen before the friction experiment. The DLC layer is deposited onto glass slides via magnetron sputtering, with a 5 mm wide aluminum foil covered at the center of the glass to create a gap between the two electrodes, serving as the bottom of the TENG. Large-area scale-like nanostructures were fabricated on 2-inch (0006) sapphire substrate (miscut angle of 1° in $\left[11\overline{2}0\right]$ axis) purchased from Jiangyin Jingmu Optoelectronic New Materials Co., Ltd, and a 25 × 25 mm² section was cut out by a dicing machine (Veeco Optium ADS-800) to serve as the top of the TENG.

Friction experiment

The friction experiments were carried out by the self-built experimental device shown in Supplementary Fig. 1. All friction experiments were carried out at room temperature (25 °C ± 2 °C) and atmospheric environment, the relative humidity (RH) was ~58.5%. The SiO₂ ball (diameter 8 mm) was kept stationary, while the sapphire sample moved in a reciprocating linear motion along the $\left[11\overline{2}0\right]$ direction at a speed of 50 mm/min. The friction duration was 30 min, and the applied normal load was 2 N. (the maximum Hertzian contact stress was estimated to be ~422 MPa). During friction, the interface is continuously rinsed with pure water, which supplies water molecules for interfacial reactions while simultaneously removing wear debris in a timely manner. Comparative experiments were conducted using alumina balls on sapphire with a 1° miscut angle according to the above parameters. Sapphire samples with 0°, 1°, 3°, 5° miscut angle were tested according to the above parameters. The reciprocating sliding movement of TENG was provided by a multi-functional tribometer (MFT-5000, Rtec Corporation, America) with the frequency of 2 Hz and the load of 2 N.

Characterizations

The miscut angle of sapphire was accurately measured by a high-precision x-ray diffractometer (XRD, Smartlab 9KW, Rigaku), where the results were obtained with the scanning step of 0.001°. The atomic-scale surface morphological features were obtained by an atomic force microscope (AFM, Bruker Dimension Edge) in tapping mode. The morphology images of the friction area of sapphire and SiO₂ balls were obtained by a scanning electron microscopy (SEM, Apreo2 S Lovac, Thermo Fisher), and the in-situ element detection of the friction area was carried out by an energy dispersive spectrometer (EDS) integrated on SEM. X-ray photoelectron spectroscopy (XPS) results were obtained using a spectroscopy (PHI 5000 Versaprobe Ⅲ) with a monochromatic Al Kα x-ray source. The XPS data were fitted by XPSPEAK41 software, all peaks have been calibrated by C 1 s (284.8 eV) corresponding to carbon contaminations. Aberration corrected transmission electron microscopy (TEM, Titan Themis G2, Thermo Fisher) was used to detect the subsurface of sapphire before and after friction. In addition, the high-angle annular dark-field (HAADF) and EDS were used to study the element distribution of the sapphire sample after friction. The samples of TEM were prepared by focused ion beam (FIB, Helios 600i, Thermo Fisher). A laser scanning confocal microscopy (LSCM, KEYENCE VK-X1000) was used to observe the optical images of the surface morphologies. The surface forms of SiO₂ ball were obtained by a white-light interferometer (WLI, Zygo, NewView™ 9000). The sapphire with nanostructures serving as the top of the TENG was cut using a dicing machine (Optium ADS-800, Veeco). The electrical output of TENG was measured using an electrometer (Keithley 6514, Tektronix). The nanohardness was measured by nanoindentation test using triboindenter (Hysitron TI950, Bruker).

Reactive force field molecular dynamics simulation

Reactive force field molecular dynamics (ReaxFF-MD) simulations were performed with LAMMPS⁴³ to model the tribological process of amorphous SiO₂ and sapphire employing the ReaxFF potential developed by Alireza et al. ⁴⁴ The amorphous SiO₂ substrate was prepared by thermally annealing crystalline SiO₂, following the procedure described in Ref. ⁴⁵ The sapphire surface was constructed with atomic steps by cutting at a 3°misorientation relative to the (0006) plane, mimicking the typical stepped morphology induced by crystal miscut. Both substrates had dimensions of 85 × 50 × 20 ų and were individually relaxed prior to assembling the tribological system by introducing 800 water molecules into the confined gap between their surfaces. To impose external mechanical conditions, the topmost 5 Å of the amorphous SiO₂ was treated as a rigid body, allowing controlled application of normal force and lateral velocity. The bottommost 5 Å of the sapphire was held fixed to support the system, while the adjacent atomic layers near the fixed and rigid regions served as thermostatic layers for temperature regulation. Periodic boundary conditions were applied in the xy directions, and the z-axis was kept free. An initial relaxing process of 50 ps at 300 K was performed to facilitate the free adsorption and interaction of water molecules with both surfaces. After relaxing, the upper rigid body was driven toward the lower substrate at a speed of 10 m/s to simulate the contact process. The sliding process was implemented by laterally translating the upper substrate along both the positive and negative x-directions at a fixed speed of 10 m/s under normal pressures of 3 GPa. Simulations were conducted under the NVE ensemble, and the Verlet algorithm was employed to numerically integrate the equations of motion with a time step of 0.25 fs. The temperature control was achieved via a Langevin thermostat (damping coefficient: 25 fs) applied to the thermostatic regions, maintaining the system at 300 K. Atomic trajectories and structural changes were analyzed using OVITO⁴⁶. Bond formation and breakage were determined using a bond order cutoff of 0.3⁴⁷.

Author contributions

Y.W., H.D., and J.Z. conceived the idea of the work. J.Z. conducted the friction experiments and material characterization. J.Z. contributed to writing the original manuscript. C.L., Y.Z. and J.Z. conducted the ReaxFF-MD calculations. Z.C. counted the number of different atomic groups in the ReaxFF-MD. J.Z. fabricated the TENGs. J.Z. and J.Y. conducted the experiments of electrical performance evaluations. Y.W. and H.D. supervised the study and revised the manuscript. All authors contributed to the writing and approval of the final manuscript.

Peer review

Peer review information

Nature Communications thanks Manel Rodríguez Ripoll, Peter Sherrell, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are available in the main text or the Supplementary Information. 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/s41467-026-71385-0.