1. Introduction
Solid lubricants have been used to make sliding systems effective. In the past few decades, solid lubricants made up of layered crystalline structures and displaying low-friction properties have been reported. These properties have been ascribed to the low resistance of the atomic shear plane to shear force resulting from weak van der Waals interactions between the neighboring layers [
1]. Recently, even single-layer graphene, hexagonal boron nitride and molybdenum disulfide (MoS
2) have been reported to exhibit low friction when strongly anchored onto an underlying substrate [
2,
3]. These observations imply that the low shear strength of the atomic shear plane is not the only determinant of the outstanding lubricities of these materials.
Of these layered materials, graphite is the most widely used solid lubricant because of its superb lubricity, low cost, and abundance in nature [
4]. As large-area synthesis and transfer of single-layer graphene have become feasible, graphene has been emerging as a promising candidate for serving as an atomically thin solid lubricant [
5,
6,
7]. Atomically thin solid lubricants can be effectively utilized for demanding applications such as microelectromechanical and nanoelectromechanical systems (MEMS and NEMS), and bio-implants, where sliding occurs over small distances, specifically in the sub-micrometer range. Graphene, in particular, has been reported to show excellent tribological performance in micro-scale dry contact [
8].
When large-area graphene is synthesized by carrying out chemical vapor deposition (CVD), mainly copper substrates are used as a catalyst in the reaction [
9] to utilize graphene as a protective film for a surface, therefore, a transfer of the graphene from the copper substrate to the target surface is necessarily required. A temporary sturdy substrate such as polymethyl methacrylate (PMMA) and polydimethylsiloxane needs to be used to support the highly compliant graphene and hence prevent it from becoming damaged during its transfer. However, a problem in this regard is that several wet chemical steps, which can contaminate the graphene surface, are incorporated in the transfer process [
10]. While, as mentioned above, single-layer graphene displays outstanding lubricity when strongly bound to its underlying substrate [
3], contaminants that become adsorbed onto the graphene surface during the transfer process can interfere with the strong bonding between the graphene film and the substrate. Therefore, strong bonding between the transferred graphene and its underlying substrate might not be sufficient to produce low friction. Also, when sliding surfaces make a micro-scale contact area, contaminants would have an undesirable effect on the tribological performance of the graphene.
Another issue limiting the application of graphene as a solid lubricant is the difficulty of controlling its thickness over a large area. Among tribology specialists, atomically thin graphene film is considered to be an excellent solid lubricant for MEMS or NEMS applications, but not so much so for typical macro-scale sliding systems. However, in such small-scale systems, controlling film thickness is crucial because the thickness determines the clearance between contacting surfaces. Therefore, an effective way to control the thickness of graphene needs to be developed in order to utilize it as a solid lubricant in MEMS and NEMS applications [
7].
MoS
2 is known as an effective lubricant like graphite and can be directly synthesized on an arbitrary target substrate by reacting a Mo metal source on the substrate with H
2S gas [
11]. The direct synthesis without a transfer process can offer advantages from the tribological perspective. First, it can overcome the contamination problems expected in the case of graphene. Accordingly, CVD-grown MoS
2 film would be expected to strongly bind its substrate and display excellent intrinsic tribological characteristics in sliding systems. Second, when carrying out such a direct synthesis, it is possible to control the produce a film with uniform thickness over a large area. Therefore, lubrication engineers can control the clearance between the contacting surfaces in MEMS and NEMS applications. In this context, CVD-grown MoS2 would be considered to be an excellent atomically thin solid lubricant for tribological applications if it can be made to have frictional properties surpassing those of CVD-graphene.
Many types of wear mechanisms have been proposed over the past few decades. For example, adhesion, plowing, corrosion, erosion, surface fatigue and seizure have all been proposed for explaining wear [
12,
13]. A sliding system is rarely dominated by any one wear mechanism, and instead generally more than two wear processes occur simultaneously. Hence, the friction and wear properties of any particular sliding system are very hard to predict. For designers who wish to select a suitable solid lubricant, the wear rate of the sliding surfaces and the life of the sliding system are of particular interest. Generally, they broadly classify the wear of various materials as either ‘mild’ or ‘severe’ [
14,
15,
16]. Mild wear results in a smooth surface and severe wear gives a rough surface with a high wear rate.
In the present study, we compared the dry friction and wear properties of CVD-grown MoS2 with various thicknesses to those of CVD-grown graphene at low (72 MPa) and high (378 MPa) contact pressures. These two contact pressures were selected to observe contrasting wear behaviors of MoS2 films. SiO2/Si was chosen as a substrate for the graphene and MoS2 films because it is a widely used material for MEMS/NEMS applications and electronic devices. In our investigation, CVD-grown MoS2 effectively reduced the friction and showed better resistance to wear than did CVD-grown graphene. Furthermore, a transition from mild to severe wear of MoS2 was observed.
3. Results and Discussion
Figure 2 shows coefficients of friction of the various samples determined using the low contact pressure. As shown in
Figure 2a, friction coefficient for graphene was observed to gradually increase from 0.23 to 0.45 as the number of sliding cycles was increased to ten. The initial value of the coefficient of friction (0.23) was similar to that of graphene on Cu against a fused silica (0.22) reported in Reference [
8]. The increased friction coefficient value of 0.45 was comparable to the friction coefficient of the bare SiO
2/Si substrate. These results taken together indicated that the graphene film was removed and the SiO
2/Si substrate was exposed by the time ten cycles of sliding were performed.
To verify that the graphene film was indeed removed, an optical image of the sample was taken after 10 cycles of sliding as shown in
Figure 3a. From the optical contrast between worn and unworn areas, it is possible to define the boundary between worn and unworn areas. For clarity, an AFM image, shown in
Figure 3b, was obtained at the boundary between the worn and unworn areas. The difference between the height of the worn area and that of the unworn area was measured to be about 0.5 nm (
Figure 3c), corresponding to the thickness of monolayer graphene, and hence indicating that graphene film was removed from its substrate. A Raman spectrum image (2D peak) of the sample was acquired (
Figure 3d), and inspection of this image also revealed that the graphene film was removed from the substrate after 10 cycles of sliding, but that small graphene particles remained on the worn track. These results taken together demonstrated the instability of the graphene film exposed to the applied low contact pressure.
Compared to the graphene sample, the tested MoS
2 samples exhibited lower coefficient of friction values. After 10 cycles of sliding, these values were only 0.1 to 0.2 for the MoS
2 samples, as shown in
Figure 2a. Moreover, the coefficient of friction of the 12L MoS
2 film remained at a low value of 0.2 even up to 200 cycles of sliding, as shown in
Figure 3b. These results verified the superior frictional properties and wear resistance of the MoS
2 films compared to those of the graphene film.
After 10 cycles of sliding, scratch lines on the centers of wear track of each on the 2L and 4L MoS
2 film can be seen as shown in
Figure 4a,b. The color of the scratches is same to the bare SiO
2/Si substrate. In contrast to theses, blue color remained on the wear track of 12L MoS
2 sample after the sliding test. This result indicated that the SiO
2/Si substrate was not exposed completely even after 10 cycles of sliding. Some of the wear debris piled up at the end of the wear track of the coated flat samples and the rest adhered to the counterpart (fused silica ball) as shown in
Figure 4d. Although the scratch lines formed and the underlying substrate was exposed for the 2L and 4L MoS
2 samples, there was no significant change in the coefficient of friction after 5 cycles of sliding. Perhaps the debris transferred to the counter-surface of the fused silica functioned as the tribofilm. Previously, debris transferred from CVD-grown graphene was observed to form the tribofilm and to show low friction, comparable to that of unworn graphene [
8].
Slips can occur at the interlayer of a layered material when a friction force is released parallel to the interlayer plane [
1]. If wear occurs in a layer-by-layer fashion, the underlying layer would be exposed to the sliding surface after detachment of the top layer. The newly exposed layer should have a surface morphology essentially identical to that of the original top layer if there is no plastic deformation during sliding, and the wear depth should correspond to integer multiples of the monolayer thickness. To test this hypothesized layer-by-layer wear mechanism, we obtained an AFM image of the boundary between the worn and unworn areas of the 12 L MoS
2 sample subjected to 5 cycles of sliding, as shown in
Figure 5a. The difference between the height of the worn area and that of the unworn area was measured to be about 0.8 nm (
Figure 5b), corresponding to the thickness of a monolayer of MoS
2. Also, the worn area here showed an RMS roughness of 0.31 nm (
Figure 5c), close to the RMS roughness values measured for the unworn MoS
2 films shown in
Figure 1b–d. Based on these height difference and RMS roughness results, we concluded that the wear of the tested MoS
2 film occurred in a layer-by-layer fashion.
Figure 6 shows coefficients of friction of our various samples obtained under high contact pressure. The measured coefficient of friction of the bare SiO
2/Si substrate was 0.48 ± 0.07 immediately after the onset of siding and the friction coefficient value of 1L graphene rapidly increased to a very similar value of 0.52 ± 0.08. The similarity of the results for the 1L graphene film and the bare SiO
2/Si were thought to be caused by the removal of the graphene film and the resulting exposure of the underlying SiO
2/Si substrate very soon after the sliding was commenced. To test this explanation, we stopped the sliding test only after 20 sliding cycles and inspected the wear track of the 1L graphene sample as shown in
Supplementary Figure S2. Severe wear damage on the underlying SiO
2/Si substrate was observed. This observation demonstrated that the graphene film immediately detached from the SiO
2/Si substrate under the conditions of high contact pressure.
As shown in
Figure 6, MoS
2-coated SiO
2/Si samples showed relatively low and stable coefficients of friction, in the range 0.18 to 0.24, for the first few dozens of sliding cycles, with the thickest tested MoS
2 film showing this relatively low friction coefficient for the most cycles. But after several additional cycles, the coefficients of friction did increase, to 0.48 ± 0.05, comparable to the friction coefficient of bare SiO
2/Si. The eventual similarity of the friction levels of the MoS
2-coated SiO
2/Si to that of bare SiO
2/Si was thought to be caused by the removal of the MoS
2 film from its substrate. To test this explanation, optical images of the MoS
2-coated SiO
2/Si samples were taken after 50 sliding cycles, at which point the coefficient of friction was still low and stable. The optical image of the sample coated with the 2L MoS
2 film (
Figure 7a) showed that the MoS
2 film was totally removed from its substrate after 50 cycles of sliding, despite the sample having still exhibited the relatively low level of friction, i.e., lower than that of the bare SiO
2/Si. After the sliding test, the film appeared to have transferred onto the counter-surface of the SiC ball, according to the image of the ball shown in
Figure 7b. These results taken together suggested that the transferred MoS
2 functioned as a solid lubricant on the ball to retain the lower friction level for a few additional cycles of sliding [
4,
8].
Inspection of optical and AFM images of the 12L MoS
2 sample under high contact pressure after 50 cycles of sliding (
Figure 8a,b) suggested that residues of the MoS
2 film remained on its substrate after this sliding. The E
2g (left peak in
Figure 8c) and A
1g (right peak in
Figure 8c) Raman signals obtained from these residues verified that they were composed of MoS
2. The heights of the MoS
2 residues were measured using tapping mode AFM as shown in
Figure 8d, and found to be about 8 nm, closely corresponding to the thickness of 12L MoS
2. Both the MoS
2 residues and the exposed SiO
2/Si surfaces showed nanoscale variations in height. Also, the exposed SiO
2/Si surfaces did not show significant damage. Based on these results, we speculated that the wear under high contact pressure occurred as a result of the repeated frictional stresses that exceeded the limit of the bonding strength between the 12L MoS
2 films and the underlying substrates. Cracks at the boundary between the films and the underlying substrates may have formed and propagated, resulting in a chipping off of the wear fragments. Such a wear process is totally different than the layer-by-layer wear mode observed under conditions of low contact pressure.
4. Conclusions
In this study, we compared the friction and wear properties of 2L, 4L, and 12L CVD-grown MoS2 to those of the CVD-grown graphene at low (72 MPa) and high (378 MPa) contact pressures. We found that, regardless of thickness, the MoS2 films showed lower friction than did the graphene film both at low and high contact pressures. We derived two explanations for the relatively low friction levels of the MoS2 films we produced compared to that of the graphene film: One being that the MoS2 films exhibited a clean surface because they were synthesized directly on SiO2/Si substrates without requiring a transfer process, in contrast to contaminants on the graphene film formed during its transfer process having weakened its lubricity; and the other being the conformal contact between MoS2 and the SiO2/Si substrates due to the direct growth of the MoS2 films on SiO2/Si.
The MoS2 films also showed resistance to wear superior to that of the graphene film both at low and high contact pressures. At low contact pressure, graphene was rapidly removed from the SiO2/Si substrate but MoS2 was not. It was observed that the MoS2 film occurred in a layer-by-layer fashion and the friction did not immediately increase after removal of the MoS2 film. Perhaps the MoS2 film transferred to the counterpart during wear and then functioned as a tribofilm.
At high contact pressure, the friction of the sample coated with the MoS2 film did not increase immediately after the MoS2 film fully detached from its substrate. Similarly, in the condition of low contact pressure, MoS2 film transferred to the counterpart served as a lubricant. However, at high contact pressure, the friction of the graphene-coated sample showed an immediate increase after the onset of sliding. A layer-by-layer wear process did not occur on the MoS2 films, in contrast to observations in the low contact pressure tests. Instead, wear fragments were generated due to the failure at the boundary between the MoS2 films and the underlying substrates.
We concluded that MoS2 films have superior potential for reducing friction and wear compared to graphene films. Further tribological studies need to be carried out under various sliding conditions, e.g., loads, speeds, temperatures, and counter materials, because different modes of wear were indicated in the current work for the different loading conditions. Our results can be used for constructing a wear-mechanism map of MoS2 films, which would be a helpful design guide for lubrication engineers in the NEMS/MEMS fields.