Tribological Characteristics of Single-Layer h-BN Measured by Colloidal Probe Atomic Force Microscopy

Abstract

The tribological characteristics of single-layer (1L) hexagonal-boron nitride (h-BN) were systematically investigated using colloidal probe atomic force microscopy, with an aim to elucidate the feasibility as a protective coating layer and solid lubricant for micro- and nanodevices. The experiments were performed to detect the occurrence of failure of 1L h-BN for up to 10,000 cycles under various normal forces. The failure of 1L h-BN did not occur for 10,000 cycles under a 10 μN normal force, corresponding to a contact pressure of about 0.34 GPa. However, the complete failure of 1L h-BN occurred faster with an increasing normal force from 20 to 42 μN. It was observed that the SiO₂/Si substrate was locally exposed due to defect formation on the 1L h-BN. The Raman spectroscopy measurement results further suggest that the failure was associated with the compressive strain on 1L h-BN. The friction of 1L h-BN before failure was orders of magnitude smaller than that of a SiO₂/Si substrate. The overall results indicate the feasibility of atomically thin h-BN as a protective coating layer and solid lubricant. In particular, the results of this work provide fundamental tribological characteristics of pristine h-BN as a guide, which may be helpful in other practical deposition methods for atomically thin h-BN with enhanced tribological characteristics.

1. Introduction

Atomically thin hexagonal-boron nitride (h-BN) has attracted remarkable attention due to its outstanding material properties [1,2]. For example, the elastic modulus of single-layer (1L) h-BN was found to be about 865 GPa [3], which is comparable to that of graphene [4]. Additionally, given that high-quality 1L h-BN can resist oxidation at temperatures up to 800 °C in air, high-temperature applications have been proposed [5,6]. Thermal and chemical inertness [6,7] of atomically thin h-BN further suggest applications for metal surface protection from corrosion [8,9]. In addition, it can be applied as a dielectric layer in electronic and optical devices [10]. To expand the applicability of atomically thin h-BN based on these properties, large-area growth using a chemical vapor deposition (CVD) method has been demonstrated [11,12].

Conventionally, bulk h-BN has been used as solid lubricants [13,14] and lubricant additives [15,16] based on its low friction characteristics originating from the layered structure. Atomically thin h-BN also exhibits low friction characteristics, similar to other two-dimensional (2D) materials such as graphene and MoS₂ [17,18]. Several studies have been performed to understand the friction characteristics of atomically thin h-BN. The fundamental frictional characteristics were explored with respect to the normal force, sliding speed, and environmental conditions [19]. The interfacial shear strength and surface damage characteristics have been exploited via progress and constant force scratch tests using atomic force microscopy (AFM) [20]. It was also demonstrated that CVD-grown 1L h-BN with a few μm-sized domains can provide low friction characteristics, along with oxidation and electric resistance [21]. Recently, the friction characteristics of graphene/h-BN heterojunctions have been investigated for superlubricity via molecular dynamics simulations [22,23]. In addition, applications such as a reinforcement for components of the composite [24,25] and a lubricant additive for water [26,27] as well as oil [28] have been proposed. Furthermore, the feasibility as protective coating layers and solid lubricants has been proposed for micro- and nanodevices. However, the friction characteristics of atomically thin h-BN have not been extensively investigated compared to graphene and MoS₂ [29,30,31,32,33]. In particular, recent studies have been more focused on the friction characteristics and the wear characteristics have not been pursued intensively. Determination of the degree of tribological performance of each atomically thin 2D material can provide a useful way to optimize them as a protective coating layer and solid lubricant with enhanced properties.

In general, AFM has been widely used to experimentally investigate the friction characteristics of 2D materials including h-BN, given that it can map friction characteristics associated with the surface structure and topography at the nanoscale [20,29,31,32,33]. The results from previous works were often obtained under relatively high contact pressures up to several GPa due to the use of conventional AFM probes with radii of a few tens of nm, which could in turn cause premature failure of 2D materials. However, data should be obtained across a wide spectrum of experimental conditions for the applicability of atomically thin h-BN. In particular, assessment of the tribological performance of atomically thin h-BN under various contact pressures may be useful in the design of the protective coating layer and solid lubricant depending on the operational conditions.

In this work, the tribological characteristics of 1L h-BN were systematically investigated using colloidal probe AFM. A colloidal probe was prepared for the experiments under relatively low contact pressures, given that the colloidal probe can provide a controlled geometry. The experiments were performed under various normal forces for up to 10,000 cycles while monitoring the variation of the friction force to determine the occurrence of failure of 1L h-BN. In addition, friction force microscopy (FFM) images were obtained after the experiments to clearly observe the failure of 1L h-BN. Furthermore, Raman spectroscopy was employed to gain a better understanding of the failure mechanism of the 1L h-BN. The finding may be useful to elucidate the feasibility of 2D h-BN as a protective coating layer and solid lubricant for micro- and nanodevices.

2. Materials and Methods

2.1. Materials

1L h-BN flakes from bulk h-BN (HQ graphene, Groningen, The Netherlands) were deposited onto a SiO₂/Si substrate with about a 300 nm thick SiO₂ layer by mechanical exfoliation. Confocal microscopy (VK-X200, Keyence, Osaka, Japan) was used to identify the locations of 1L h-BN based on optical contrast. In contrast to 1L graphene and MoS₂, it was relatively difficult to find the exact location of 1L h-BN due to the small optical contrast [34]. However, the 1L h-BN was often located near thick flakes and could be identified after careful focus. Then, AFM images were obtained to characterize the topography and thickness of the 1L h-BN using commercialized AFM (MFP-3D, Asylum Research, Santa Barbara, CA, USA) prior to the experiments. The AFM images were obtained in the intermittent contact mode using a Si probe with a nominal normal spring constant of 2 N/m (AC240TS, Olympus, Tokyo, Japan). FFM images were obtained under a normal force below 1 nN using a Si cantilever with a nominal normal spring constant of 0.2 N/m (PPP-LFMR, Nanosensors™, Neuchâtel, Switzerland) to minimize surface damage during the contact mode imaging.

After the preparation and characterization of the specimens, the experiments were conducted using the colloidal probe. For the colloidal probe, a silica microsphere (Polysciences Inc., Warrington, PA, USA) was selected due to its relatively high hardness. It was also carefully glued onto the center of the free end of a tipless probe with a nominal normal spring constant of 42 N/m (TL-NCH, Nanosensnors, Neuchâtel, Switzerland). The radius of the microsphere measured using confocal microscopy was about 6.6 µm. After the preparation of the colloidal probe, a preliminary wear test for the running-in of the contact area of the colloidal probe was conducted with the goal of minimizing the effect of probe wear on the measurement results. In the preliminary test wear test, the colloidal probe was slid against a bare Si under 20 µN normal force for about 1 mm sliding distance.

2.2. Methods

The experiments with 1L h-BN were performed under various normal forces ranging from 10 to 42 µN. The experiments were repeated three times at each normal force for reproducibility. The maximum speed at the mid-section of the reciprocating motion was 2 µm/s. Additionally, the length of the stroke was set to 1.2 µm. The variation of the friction force was monitored during the experiment to evaluate the friction force as a function of the number of cycles and to address the failure of 1L h-BN. The number of cycles until a significant increase of the friction force was observed ranged from 880 to 10,000 cycles, depending on the normal force. An additional experiment was conducted to observe the 1L h-BN surface before the occurrence of the failure, with an aim to understand the failure mechanism of 1L h-BN. To obtain such a 1L h-BN surface, this experiment was repeated until no significant increase in friction was observed under a 42 µN normal force for the number of cycles comparable to that required for the failure. Furthermore, the experiment with the SiO₂/Si substrate was carried out under a 42 µN normal force for comparison. To minimize the thermal drift during the experiments, the AFM system was equilibrated for several hours after loading the probe and the specimen. After the experiments, AFM images were obtained with intermittent contact to observe the wear and surface damage of the 1L h-BN and SiO₂/Si substrate. FFM images under a normal force below 1 nN were also obtained to clearly observe the exposure of the Si/SiO₂ substrate due to the failure of the 1L h-BN, based on the difference of the friction characteristics. For clarity, the scan direction for FFM imaging was set perpendicular to the wear track. All experiments were conducted at ambient conditions with a temperature of 23 ± 3 °C and relative humidity of 40% ± 5%.

For quantitative measurement of the forces using AFM, force calibrations were performed. The thermal noise [35] and improved wedge calibration methods [36] were employed in the normal and lateral directions, respectively, for the Si probe used for the FFM measurements. The normal deflection sensitivity was obtained from the force–distance curve on a bare Si substrate, and then, the normal spring constant was determined from the thermal noise spectrum. A grating with trapezoidal steps with a 54°7′ slope (TGF11, Mikromasch, Tallinn, Estonia) was used for the wedge calibration. In particular, friction loops were obtained on the grating under various normal forces for accuracy [37]. Additionally, the wedge calibration was conducted after the experiments to eliminate the effect of probe wear on the measurements, considering that the wear of the probe during the calibration may be significant. The normal and lateral force sensitivities of the Si tip were 23.8 and 1.9 mV/nN, respectively. For the colloidal probe, the thermal noise method was used for the normal force calibration [35,38]. However, the wedge calibration using the grating was not available for the colloidal probe used in this work, due to the relatively large probe radius. Hence, the lateral deflection sensitivity was obtained from the frictional loading and the lateral spring constant was calculated using the beam theory [39,40]. The normal and lateral force sensitivities of the colloidal probe were determined to be 0.22 and 0.16 mV/nN, respectively.

Raman spectroscopy (Alpha 300R, Witec, Ulm, Germany) was further employed to understand the failure mechanism of 1L h-BN. The Raman spectra were obtained from the 1L h-BN before and after the experiment using laser excitation with a 532 nm wavelength, and a 100× objective lens. The resolution of the Raman spectrum was set at 1.4 cm⁻¹ (1800 lines/mm grating).

3. Results and Discussion

3.1. Characterization of 1L h-BN

Figure 1a shows a confocal microscope image of 1L h-BN deposited onto the SiO₂/Si substrate. In contrast to 1L graphene and MoS₂, 1L h-BN located near the thick flake was faintly identified due to the small optical contrast [34], as shown in Figure 1a. The AFM topographic images of 1L h-BN obtained in the intermittent contact mode are presented in Figure 1b along with the cross-sectional height profiles. The black squares in Figure 1a indicate the locations of the AFM topographic images. The locations of the cross-sectional profiles are indicated by the white dashed lines in the topographic images in Figure 1b. AFM often produces slightly different thickness values of 2D materials, which is associated with the adsorbents and measurement uncertainties of AFM [41]. In addition, environmental conditions such as the relative humidity may affect the thickness of 2D materials as the thickness may increase due to water intercalation [19]. From the data shown in Figure 1b, the thicknesses of 1L h-BN were determined to be 0.36 and 0.32 nm, which are slightly smaller than those obtained in previous studies [3,6,20]. However, these thickness values are in good agreement with the interlayer spacing of h-BN [42]. The FFM images of the 1L h-BN obtained under a 1 nN normal force are shown in Figure 1c along with the friction loops. The locations of the friction loops are indicated as the white dashed lines in the FFM images. The darker contrast indicates higher friction in the FFM images in Figure 1c. It can be seen from the figure that the 1L h-BN can be clearly discerned from the SiO₂/Si substrate due to the significantly low friction characteristics. As can be seen from the friction loop, the friction force on h-BN was 0.38 nN whereas that on the SiO₂/Si substrate was 4.1 nN. The AFM and FFM measurement results shown in Figure 1 clearly indicate that the 1L h-BN was well prepared by mechanical exfoliation without significant defect formation.

3.2. Friction and Wear Characteristics of 1L h-BN

The experiments using the colloidal probe were performed after characterization of the topography and thickness of 1L h-BN. Figure 2 shows the variation of the friction forces between 1L h-BN and the colloidal probe with respect to the number of cycles under normal forces of 10, 20, 30, and 42 µN. The contact pressures estimated using the Derjaguin–Muller–Toporov (DMT) contact model [43] were 0.34, 0.41, 0.47, and 0.52 GPa under normal forces of 10, 20, 30, and 42 µN, respectively. From Figure 2a, the friction force at the initial stage of contact sliding (i.e., initial 100 cycles) was found to be about 0.094 ± 0.007 µN under a 10 µN normal force. Then, the friction force decreased gradually to about 0.080 ± 0.005 µN and no significant alteration of the friction force was observed until 10,000 cycles. For the experiments with a 20 µN normal force, the initial friction force was about 0.17 ± 0.03 µN. However, as shown in Figure 2b, it was observed that the friction force increased abruptly after 7520, 5170, and 7330 cycles of reciprocation for the three replicate samples. This outcome suggests that the contact between the colloidal probe and the SiO₂/Si substrate occurred after the failure of 1L h-BN, given that the friction on the substrate was significantly higher than on 1L h-BN. The average number of cycles for the failure of the 1L h-BN was calculated to be 6700 ± 1300 cycles under a 20 µN normal force. It is also worthy to note that gradual increases of the friction force were generally observed before the occurrence of failure of 1L h-BN under a 20 µN normal force. For the experiments with 30 and 42 µN normal forces, the friction forces at the initial 100 cycles were calculated to be 0.24 ± 0.01 and 0.45 ± 0.02 µN, respectively. As clearly shown in Figure 2c,d, these friction forces increased gradually as the number of cycles increased, similar to the results obtained with a 20 µN normal force. Then, abrupt increases of the friction forces were observed after 2500 ± 900 and 1700 ± 1200 cycles for 30 and 42 µN normal forces, respectively. The earlier occurrence of the abrupt increase of the frictional force under higher normal forces indicates the premature failure of 1L h-BN under a higher normal force.

Figure 2e shows the friction force between the 1L h-BN and the colloidal probe before the occurrence of the failure of the 1L h-BN as a function of the normal force. It can be seen from the figure that the friction force increased with increasing normal force, as expected. When the adhesion force was not included in the friction coefficient calculation, the average friction coefficient of the 1L h-BN against the colloidal probe used in this work was calculated to be about 0.008−0.012 from the data shown in Figure 2e. It should be noted that the friction coefficient may significantly vary, depending on the quality of the specimen, counter materials, and measurement and environmental conditions. However, these friction coefficients are generally comparable with those of pristine 1L graphene and MoS₂ (0.01–0.02), and smaller than those of CVD-grown graphene (~0.04–0.1), graphene oxide (~0.25–1.1), and MoS₂ (~0.2) [33,44,45,46]. Additionally, they are significantly smaller than friction coefficients of conventional solid lubricants at the macroscale [47]. This outcome clearly suggests that 1L h-BN can be feasibly used as a solid lubricant, particularly for the case where electrical insulation is needed.

Figure 3 shows examples of the AFM topographies and FFM images of 1L h-BN after the experiments. In Figure 3, the cross-sectional height profiles and friction loops are also provided for clarity. The images in Figure 3a–d were obtained after 10,000 cycles under a 10 µN normal force, 7330 cycles under a 20 µN normal force, 1710 cycles under a 30 µN normal force, and 1200 cycles under a 42 µN normal forces, respectively. As shown in the topographic image along with the cross-sectional height profile in Figure 3a, no significant signs of surface damage and/or wear track formation were observed after 10,000 cycles under a 10 µN normal force. The FFM image and friction loop in Figure 3a further suggest that no significant defects were formed and the SiO₂/Si substrate was not exposed during the experiment [20]. This may be the reason that no significant increase in the friction force was observed during the experiments with a 10 µN normal force, as shown in Figure 2a. In contrast, as is evident from Figure 3b–d, complete failure of the 1L h-BN occurred after the experiments with 20, 30, and 42 µN normal forces. It was plausible that the 1L h-BN was torn under these normal forces and then folded at the edges and at both ends of the wear track. In this regard, the gradual increases of friction forces during the experiments under 20, 30, and 42 µN normal forces, as shown in Figure 2b–d, respectively, can be presumed to be a sign of the failure of 1L h-BN. The exposure of the SiO₂/Si substrate due to the failure of 1L h-BN can be clearly observed in the FFM images and the friction loops due to the difference of the friction characteristics between 1L h-BN and the substrate. The data in Figure 2; Figure 3 clearly indicate an earlier failure of 1L h-BN with increasing normal force. In particular, it was concluded that a contact pressure of less than about 0.3 GPa can be considered as an upper bound to prevent the failure of h-BN for the experimental conditions applied in this work. However, as the contact pressure is larger than 0.34 GPa, failure of 1L h-BN may occur after a few thousand cycles. The normal force- or contact-pressure-dependent onset of wear was often observed for 2D materials [20,48,49,50], similar to other tribological materials. The critical contact pressure from this work was found to be an order of magnitude smaller than that from previous work [20]. This outcome may be due to the onset of wear after the relatively large number of cycles. It is also worthy to note that the critical contact pressure may be affected by major wear mechanisms and defects [49,50]. Nevertheless, these outcomes are expected to aid in the design to improve the performance of atomically thin h-BN-based protective coating layers and solid lubricants.

For comparison, additional experiments were performed to assess the surface damage characteristics of the SiO₂/Si substrate without a 1L h-BN layer against the colloidal probe. Additionally, the surface of 1L h-BN before the occurrence of the failure was carefully observed using AFM, FFM, and Raman spectroscopy. Figure 4a,b shows the topographies and FFM images of the SiO₂/Si substrate after 1000 cycles and 1L h-BN after 3000 cycles under a 42 µN normal force, respectively. The cross-sectional height profiles and the friction loops are presented in Figure 4a,b, respectively. The variations of the friction forces are also plotted with respect to the number of cycles in Figure 4c. As is evident from Figure 4a, the SiO₂/Si substrate was significantly damaged after only 1000 cycles. The surface damage area on the SiO₂/Si substrate exhibited a relatively large width of about 1 µm, which indicates that the thermal drift of the AFM system was relatively large during this particular experiment. From the topography and cross-sectional profile in Figure 4a, a few lines of hillks were found to be formed along the sliding direction due to the contact sliding with the silica colloidal probe in air, as was observed in a previous study [51]. The height of the hillk was as large as about 1 nm, as shown in the cross-sectional profile in Figure 4a. Additionally, the friction of the damaged area was found to be greater than that of SiO₂/Si substrates which did not experience contact sliding with the colloidal probe, as shown in the FFM image along with the friction loop in Figure 4a. Figure 4c further shows that the friction force between the SiO₂/Si substrate and the colloidal probe during the experiment was kept larger than 12 µN, which is about 24 times larger than that between 1L h-BN and the colloidal probe. Furthermore, the fluctuation of the friction force on the SiO₂/Si substrate during the experiment was found to be large, which may be associated with the large friction characteristics of the damaged area.

In the topography and cross-sectional profile in Figure 4b, the wear of the 1L h-BN surface was not clearly observed after 3000 cycles. However, a number of spikes along the sliding direction can be clearly observed in the wear track from the FFM image and the friction loop in Figure 4b. It is likely that defects were formed on 1L h-BN, which caused the lal increases of the friction force. It is also postulated that the SiO₂/Si substrate was exposed at the defects, considering that the friction force on the defects was as large as 4 nN, which is comparable to that on the SiO₂/Si substrate. In addition, the gradual increase of the friction force on 1L h-BN was observed with an increasing number of cycles in Figure 4c, which is consistent with the data shown in Figure 2. This outcome suggests that more defects were formed as the number of cycles increased. It should be noted that the failure of 1L h-BN did not occur even after 3000 cycles under a 42 µN normal force in this additional experiment although the number of cycles for the failure of 1L h-BN was estimated to be 1700 ± 1200 cycles under the same condition in the data shown in Figure 2d. However, large scattering in the friction and wear data is often observed, which is associated with the difference in the status of the specimen surface as well as experimental uncertainty.

The Raman spectrum obtained from the wear tack formed on the 1L h-BN in Figure 4b was compared to that obtained from the 1L h-BN before the experiment, as shown in Figure 4d. In Figure 4d, the Raman spectra data were fitted using a Lorentzian equation. The spectra show $E_{2g}$ characteristic peaks of about 1370 cm⁻¹ corresponding to the in-plane vibrations of B–N bonds [1,34]. However, the frequency of the $E_{2g}$ peak obtained from 1L h-BN after the experiment was found to be larger than that before the experiment by 0.7 cm⁻¹ (blue shift), which is consistent with previous work [20]. This outcome plausibly suggests in-plane compressive strain at the wear track on 1L h-BN due to contact sliding with the colloidal probe [11]. Before the failure occurs, the compression strain is expected to accumulate with an increasing number of cycles and be released with defect formation. It can be also seen that the peak was slightly broadened after the experiment. However, this could be associated with the measurement uncertainty, considering that the peak broadening due to the defect formation was not clearly observed from the previous work [20].

4. Conclusions

In this work, the tribological characteristics of pristine 1L h-BN deposited by mechanical exfoliation were investigated using colloidal probe AFM. The experiments were performed for up to 10,000 cycles of reciprocating until failure of 1L h-BN occurred under various normal forces ranging from 10 to 42 µN, corresponding to contact pressures of 0.34 to 0.52 GPa. The variation of the friction force was monitored during the experiments. The friction force of 1L h-BN slid against the colloidal probe was found to be orders of magnitude smaller than that of the SiO₂/Si substrate under the given conditions. The results also showed that, for up to 10,000 cycles under a 10 µN normal force, the failure of 1L h-BN did not occur and the orders of magnitude smaller friction force than the substrate were maintained. However, complete failure of 1L h-BN was observed after 6700 ± 1300, 2500 ± 900, and 1700 ± 1200 cycles for 20, 30, and 42 µN normal forces, respectively. In addition, it was found that the friction force increased gradually before the failure of 1L h-BN. The FFM observation of 1L h-BN after the experiment showed that the increase of the friction force may be associated with the defect formation. Raman spectroscopy measurement results further indicate the residual compressive strain on the 1L h-BN surface due to the contact sliding. Hence, it was concluded that the failure of 1L h-BN may occur from the defects formed due to the accumulated compressive strain with an increasing number of cycles.

The overall results of this work indicate the significant friction reduction and surface protection effect of atomically thin h-BN. However, it should be noted that tribological characteristics are substantially affected by the specimen status, operating conditions, and environmental conditions. Therefore, more data is required across a broad spectrum of experimental conditions. In particular, systematic approaches to understanding the major difference between the tribological performance of CVD-grown and pristine h-BN may be needed given that CVD may enhance the applicability of atomically thin h-BN with controllable size, orientation, and thickness. Nonetheless, the outcomes of this work provide useful information for the implementation of atomically thin h-BN as a protective coating layer and solid lubricant for micro- and nanodevices. In particular, the results of this work may be helpful in the practical fabrication of atomically thin h-BN with enhanced tribological characteristics by providing fundamental tribological characteristics of pristine h-BN as a guideline.