1. Introduction
Transition metal dichalcogenides (TMDs) are a family of semiconductor materials highly used in the electronic field due to their electrical properties. There is large interest in TMDs, which have been studied as much as graphene nowadays [
1,
2]. The configuration of the TMDs relies on an MX
2 structure, where M represents the transition metals (Mo, W, V, Nb) and X represents the chalcogen atoms (S, Se, and Te) [
3,
4].
Molybdenum disulfide (MoS
2) is one of the most popular transition metal dichalcogenides. It can be used in the electrical field and as a solid lubricant. Initially, it was used as an oil additive and anti-friction lubricant. Additional applications, for example, those associated with the deposition of MoS
2 as coatings, have recently emerged [
5,
6,
7,
8].
The molybdenum disulfide’s low friction properties are related to its layered crystal structure, where the transition metal stays between the chalcogen atoms, forming a sandwich with strong in-plane bonding. Between these layers, weak van der Waals interactions are maintained, allowing easy shearing on the basal planes (002) parallel to the sliding direction, resulting in low coefficients of friction in the tribological system where the coating is applied [
9,
10,
11]. In this context, high attention has been paid to the tribological behavior of these coatings [
12,
13,
14,
15,
16].
One of the major limitations of MoS
2 is its susceptibility to environmental contaminants such as water and oxygen. On prolonged exposure to moisture, the coating can gradually oxidize in contact with lubricants, converting the MoS
2 to molybdenum oxide (MoO
3), which has very low lubricating properties, increasing the coefficient of friction [
17,
18,
19]. To overcome this problem, doping of these MoS
2 with different oxides, nanomaterials, and metals improves not only the coating’s resistance in humid environments but also the load-bearing capacity, tribological performance, and mechanical properties of the thin film, especially with titanium-doping [
20,
21,
22,
23,
24].
The main contribution of these dopants to MoS
2 coating is their densification, which reduces the oxidation reaction sites inside the coating [
25]. The incorporation of titanium also causes a structural transformation, preventing a crystalline formation and substituting the molybdenum in the structure [
26,
27]. Niobium-doped molybdenum disulfide coatings have also presented interesting tribological behavior under different environments, presenting a low coefficient of friction and a capability to resist humid ambient due to an easier oxygen capture when compared to pure molybdenum disulfide [
28,
29]. However, the number of researchers relating Nb-doping to MoS
2 to the coating tribological behavior is significantly lower than those related to Ti-doping.
Although the doping of molybdenum disulfide is an excellent alternative to improve the properties of the coating, it does not significantly impact the adhesion of these coatings over different substrates, such as steels. Usually, an interlayer of the doped material is deposited between the substrate and the doped MoS
2 film [
30,
31,
32]. Cr is commonly used as an adhesion interlayer when coating tool steels. However, a Nb interlayer may be specified to improve the adhesion of the Nb-doped MoS
2 coating, due to the greater affinity of Nb with the Nb-doped MoS
2 coating. Moreover, multilayers of different characteristics are used, such as one thin metallic adhesion interlayer followed by a nitride reinforcing layer such as TiN or CrN [
33,
34,
35]. These intermediate layers give better mechanical support to the outermost layer and improve the adhesion of the coatings. A NbN intermediate layer could grant lower hardness gradients, and a higher thickness of interlayers keeping detrimental stresses further away from the coating/substrate interface.
This research intends to investigate the role that Nb doping can play in the adhesion of MoS2 coatings to an H13 steel substrate. The experiment design comprised coating steel with a Nb adhesion interlayer, followed by a NbN reinforcing layer, and finally depositing Nb-doped MoS2 coatings with different Nb contents and assessing the effects of niobium-doped molybdenum disulfide coatings on its structure, mechanical properties, and adhesion to the substrate.
2. Materials and Methods
Niobium-doped molybdenum disulfide coatings were deposited by Physical Vapor Deposition (PVD), using a balanced Pulsed Direct Current Magnetron Sputtering (pDCMS) system, on Silicon Wafers (100) and quenched and tempered H13 steel substrates. Mechanical and microstructural characterization of the coatings was carried out on the coatings deposited on the Si substrate, while mechanical behavior was assessed for the coating deposited on the H13 steel.
Figure 1A shows schematically the deposition system used, and
Figure 1B shows the architecture of the deposited coatings.
Figure 1A schematically depicts the configuration of the reactor used for depositing the Nb:MoS
2 coatings. High purity (99.999 wt.%) Nb and MoS
2 targets were used to avoid contamination of the deposited films by other elements. The power applied to the Nb target was changed, while the power applied to the MoS
2 target was maintained constant. Changing the applied power to the Nb target made it possible to change the stoichiometry of the coating, as it will be later demonstrated. The specimens were placed on a 100 mm round table rotating at a speed of 18.5 rpm.
Table 1 shows the geometric and deposition parameters of the process.
Figure 1B shows the different architectures that have been deposited. The first one presents only a thin metallic Nb interlayer and the functional Nb:MoS
2 layer was used to evaluate the influence of different Nb-doping concentrations on the coating’s mechanical properties, microstructure, and adhesion. The second one was deposited to evaluate the mechanical and microstructural properties of the NbN layer deposited over a metallic Nb interlayer. The third one, a combination of the first two architectures, underwent the same characterization as the first. However, in this case the focus was observing the impact of the NbN layer on the adhesion of the coating with a fixed Nb concentration.
Before the deposition, the H13 samples were ground with sandpaper from #120 to #2000 and polished with 9 and 3 µm abrasive particles. The Si (100) samples were obtained polished on one side.
Samples were cleaned with an ultrasonic cleaner and embedded in ethanol for 10 min before placing on the sample holder. A plasma cleaning step was performed with the samples inside the reactor chamber with a power of 150 W, pressure of 2 mTorr, and 10 sccm of Ar for 10 min to remove oxides and impurities that remained on the surface of samples.
After the deposition of the Nb:MoS
2 coatings, a pure layer of MoS
2 was deposited for 5 min, with the same conditions applied to the molybdenum disulfide target, presented in
Table 1.
Table 2 shows the samples’ nomenclature according to the target power applied to the Nb target and the presence of the NbN interlayer.
For the evaluation of the coatings’ microstructure, a Jeol JSM-6010LA Scanning Electron Microscope (SEM) (JEOL Ltd., Tokyo, Japan) with Electron Dispersive Spectroscopy (EDS) with 10 kV was used to measure the concentration of Nb, S, and Mo elements in a semi-quantitative way.
Further, a single sample with a Nb interlayer, NbN layer, Nb:MoS2 layer, and pure MoS2 outer layer was taken to a dual beam high-resolution Thermo Fischer Scientific SCIOS 2 SEM, with a combination of a Field Emission Gun (FEG) and a focused ion beam (FIB) of gallium. The FIB was used to prepare the sample, which was further taken to a JEOL JEM-2100F Transmission Electron Microscope (TEM) (JEOL Ltd., Tokyo, Japan), to observe the cross-section of the sample in high resolution and to observe the nanostructure of the different layers and corresponding selected area diffraction patterns (SAED). During preparation, a platinum layer was deposited on top of the specimen to protect the film from being damaged by the bombardment with gallium ions. Higher-resolution Energy Dispersive Spectrometry was also performed in the 10 keV energy range.
A Bruker Hysitron Ti-950 nanoindenter, with a Berkovich diamond indenter tip, was used to evaluate the hardness and elastic modulus of the coatings according to the Oliver and Pharr method [
36]. Each coating with different Nb concentrations and the NbN layer were measured with 98 indentations per sample, divided into two 7 × 7 matrices, in two distinct regions of the sample with a 100 to 1300 µN load range to reduce significantly substrate and tip influence in the measurement of the coating hardness.
Raman spectroscopy was carried out to characterize the structure of the NbN layer, in a Horiba Xplora One, with a 538 nm laser wavelength and 1% laser power, which is approximately equal to 8 mW.
Scratch tests were conducted on a UMT2—Bruker equipment, with a Rockwell C diamond tip, a stroke of 10 mm, and a ramp load up to 40 N. The scratches were analyzed in an Olympus BX51M Optical Microscope, to observe the full length of the scratch and to compare the surface with the data of the coefficient of friction, normal, and tangential load given by the scratch test equipment. Specific regions of the scratch were also analyzed by SEM and EDS, to understand the behavior of the coating during these tests.
3. Results and Discussion
Figure 2 presents SEM images related to the film morphologies obtained after the Nb/Nb:MoS
2 coating deposition. All the Nb:MoS
2 films presented a columnar structure, typical of MoS
2 sputtered coatings [
32,
37], regardless of the amount of Nb incorporated into the coating.
The thickness of the obtained coatings is presented in
Table 3. One can see that the increasing applied power to the Nb target did not promote significant variations in thickness. The average Nb:MoS
2 coating thickness was 0.53 ± 0.03 µm and the metallic Nb interlayer thickness was approximately 0.12 ± 0.01 µm.
Table 4 shows the chemical composition of the coatings obtained from the EDS analysis. One can see that there is almost a linear correlation between the power of the target and the Nb content in the MoS
2 coating, with the S/Mo ratio staying practically constant at around 2.3. The literature usually reports a 1.6 to 1.8 S/Mo ratio, but as niobium substitutes Mo on the structure of MoS
2, the S/Mo ratio increases. On the other hand, the Nb/Mo ratio, which gives information about the doping of Nb on the coating, increases steadily due to the increasing Nb content of the coating. This information corroborates with the increase in the S/Mo ratio for the Nb:MoS
2 film because the concentration of niobium is inversely proportional to the concentration of molybdenum, which means that when Nb is added, the concentration of Mo decreases, indicating its substitution. The literature reports that titanium also substitutes molybdenum in the microstructure, allowing the same interpretation of the effects of niobium doping [
26,
27].
Figure 3 shows the microstructure of the Nb25NbN300 sample. Three different layers can be seen, namely, the metallic niobium interlayer, the NbN layer, and the Nb:MoS
2 functional layer.
The thickness of the Nb interlayer, NbN layer, and Nb:MoS2 functional layer were 0.12 ± 0.01 µm, 0.26 ± 0.01 µm, and 0.48 ± 0.02 µm, respectively.
Columnar structures in the NbN layer are typical of nitride layers deposited by PVD [
38,
39,
40]. FEG-STEM images were taken from the sample Nb30NbN300, prepared by ion beam milling in FIB equipment, for further characterization of the coatings, as indicated in
Figure 4. This figure confirms the metallic Nb, the NbN layer, the Nb:MoS
2 film, and a pure MoS
2 outermost layer. The light gray layer corresponds to Pt deposited on the surface of the specimen to protect the film from being damaged by the focused ion beam.
The thickness of deposited layers is presented in
Table 5.
Figure 5 presents EDS analyses of the films. Both the image’s bottom and top portions are related to Si as the substrate and the Pt-protecting layer deposited to perform the FIB process, respectively. Oxygen also appears, mainly above the Si substrate, due to the oxidation of the substrate, which was not completely removed during the coating deposition.
Oxygen was detected along the layers, especially in the layers containing Nb, which may be explained by the oxidation of the target before the deposition, since no oxygen could be detected in the pure MoS2 outermost layer.
Figure 5 shows the Mo, S, and Nb concentration maps across the multilayer film. It is worth noting the NbN reinforcing interlayer beneath the Nb:MoS
2 layer.
It is important to mention that
Figure 5 indicates the presence of Nb at the region corresponding to the FIB Pt protecting layer. This result is explained by the similar radiation these elements present, misleading the EDS technique. The indication of Mo and S in the NbN layer is due to the superimposition of Mo Lα and Lβ with the Nb Lα and Lβ and S Kα and Kβ radiation energies. During the window integration for creating the EDX maps, when the energy difference is less than 140 eV (resolution of the detector), the peaks overlap. This explains the presence of a small fraction of Mo and S or Pt signal in NbN EDX maps.
TEM analyses were performed on each layer to understand their microstructures and crystallography.
Figure 6 presents the transition from the Pt protection layer deposited for the FIB milling, the pure MoS
2 outer layer, and the Nb:MoS
2 layer.
It is interesting to observe that pure MoS
2 presents a crystalline character and a lamellar structure composed of alternating S-rich and Mo-rich layers [
41,
42]. On the other hand, the Nb:MoS
2 layer beneath the outermost MoS
2 coating is amorphous, indicating that doping the MoS
2 film with Nb amorphized the film.
The literature reports that Ti-doped MoS
2 and MoSe
2 coatings [
26,
42,
43,
44,
45], which are part of the TMD family, react differently for amorphization depending on the concentration of the dopant. For low Ti content (~8 at. %), the amount of titanium is not enough to passivate the (100) plane (Type I structure of molybdenum disulfide), which is highly reactive to environmental elements, so the structure remains crystalline. When introducing higher Ti concentrations (15–19 at. %), amorphous clusters are formed since this titanium content is enough to passivate the (001) planes. However, it is insufficient to inhibit the formation of crystals along planes (002) (Type II of molybdenum disulfide), which are significantly less reactive, and do not react with the dopant as the (001) plane would react, keeping most of the coating formed by crystals and amorphous clusters. Contrary to the two situations above, for even higher Ti concentrations (~25 at. %), the dopant completely disrupts the formation of crystals due to the distortion on the MoS
2 lattice, turning the coating into an amorphous layer [
26,
42,
43,
44,
45].
In
Figure 6, one can observe that, without a dopant, the deposition of MoS
2 is predominantly of Type II, with a (002) plane (parallel to the substrate). When Nb is added, the film becomes predominantly amorphous as indicated in the lower right part of the image). One can assume that, for niobium, the concentration of ~13 at. % is enough to disrupt the formation of crystals, in the same way that this behavior occurs for titanium doping with concentrations around 25 at. %.
Some works in the literature [
26,
28,
29] emphasize the densification that Nb and Ti introduced as dopants promote to the MoS
2 coatings, being one of the greater benefits since it enhances the mechanical properties of the coatings. This fundamental parameter occurs when titanium and niobium are introduced into the film’s structure. However, even though these two elements present very similar atomic radii, with 140 and 145 pm for titanium and niobium, respectively, their atomic masses are significantly different, with niobium having almost twice the atomic mass of titanium (92.90 over 47.86). This value represents the same order of magnitude for sufficient concentration of niobium and titanium dopants to transform the MoS
2 film into amorphous, and less Nb is required to disrupt the formation of MoS
2 crystals.
An additional analysis of film structures is presented in
Figure 7. In
Figure 7A, the Mo and S layers are identified, with the light gray layers composed of Mo atoms, while the S atoms compose the dark gray layers above and below. The S atoms cannot be resolved in the image, as the Mo atoms can, due to their smaller size requiring higher resolution [
11,
41,
42,
46]. For better understanding, a scheme of a monolayer of MoS
2 is presented in
Figure 7C, explaining the observation of
Figure 7A.
In
Figure 7B, small regions of short-range order, immersed in a predominantly amorphous matrix, can be seen. Hudec et al. [
43] reported that for Ti concentrations of 18 at. % on MoSe
2 coatings, amorphous clusters formed inside a crystalline MoSe
2 structure, which is an opposite behavior as that seen in the Nb30NbN300 sample. This difference indicates that the concentration of ~13 at. % Nb was not enough to completely disrupt the formation of MoS
2 crystals. If the Nb concentration was lowered, the same behavior reported by Hudec et al. related to the formation of clusters of amorphous film embedded in a crystalline matrix would possibly be seen. This result suggests that by increasing the niobium content in the MoS
2 coating, no crystalline clusters would be found, with a complete disruption of the crystalline structure. Thus, Nb concentration of ~13 at. % stays between a medium and high concentration, according to the effects reported by the literature [
43].
Hudec et al. [
43] provide additional evidence that high Ti concentrations entirely disrupt the growth of the MoSe
2 crystals, and the coatings become amorphous. As mentioned before, for the medium titanium concentrations, the perpendicular growth is passivated by Ti atoms residing on the reactive (1 0 0) MoSe
2 edges, although not wholly suppressing the crystalline character of the film. MoSe2 grows predominantly in a horizontal (0 0 2) direction. Ti atoms cannot react with the (0 0 2) planes so that Ti-rich amorphous clusters may nucleate inside the predominantly crystalline matrix. These additional data further support that a similar hypothesis for the observed amorphization of the MoS
2 film by Nb-doping and to the presence of amorphous Nb can be proposed.
Figure 8A,C presents the regions where the SAED patterns were acquired for the MoS
2 and Nb:MoS
2 layers. The SAED patterns for the MoS
2 and Nb:MoS
2 films are shown in
Figure 8B and
Figure 8D, respectively. The SAED pattern for the pure MoS
2 shows diffraction circles, related to the (100), (002), and (110) planes [
6,
47,
48,
49]. The (100) and (110) are related to the hexagonal structure of MoS
2 and the (002) plane relates to the basal plane perpendicular to the substrate surface, regarding the Type II structure [
49,
50]. Despite the pure MoS
2 layer being crystalline, the SAED pattern differs from that of a perfect crystal, with well-defined spots. This difference is due to the spot size used to perform this analysis, which is greater than the thickness of the pure MoS
2 layer. On the other hand, the SAED pattern of the Nb:MoS
2 film corresponds to an amorphous film containing a few nanosized short-range order (SRO) regions, corroborating the images presented before.
A hypothesis to explain this behavior relates to a possible atomic substitution of the Mo atoms by Nb. Nb has an empirical atomic radius (145 pm) very similar to that of Mo (145 pm), allowing the substitution [
41]. Nevertheless, insufficient niobium is available to perform the substitution because the Nb content is much lower than that of molybdenum (~13 at. % Nb against ~26 at. % Mo). This hypothesis is consistent with the fact that some regions remain crystalline, but for the most part, the structure is amorphous. As before, the presence of short-range order (SRO) was found when different elements were previously introduced as dopants, indicating that the dopants were not enough to suppress the full crystallization of the MoS
2 [
51,
52].
It is also important to point out that the Nb:MoS
2 film is a bulk film with the niobium acting as a dopant and not forming Nb and MoS
2 layers, although the deposition system consists of the sample passing under the alternating targets at different times. Previous works of Ti:MoS
2 deposition reported the formation of Ti and MoS
2 multilayers, with a similar rotative deposition system [
15,
53]. However, this multilayer characteristic of the literature is attributed to the low rotation speed of the specimens (~4 rpm), where there is enough time to form different layers of MoS
2 and Ti. Shi et al. [
53] used a higher sample rotation (10 rpm) to deposit a Ti:MoS
2 composite film. The same occurs in this work, where a rotation of 18.5 rpm was used, forming a relatively homogeneous, almost amorphous, Nb:MoS
2 film.
Figure 9 shows the transition between the Nb interlayer and the NbN layer, where it is possible to observe regions of epitaxial growth of NbN crystal columns from the partially amorphous Nb interlayer.
The SAED patterns acquired on these two layers are shown in
Figure 10.
Figure 10B shows the diffraction pattern of the NbN coating with spots revealing the cubic symmetry for the (220) plane and hexagonal close-packed for (100) and (101) planes [
54,
55]. On the other hand, the SAED pattern of the Nb adhesion interlayer in
Figure 10D shows up as continuous circles, indicating a predominantly amorphous structure but displaying some order, as seen for the Nb:MoS
2 layer.
Table 6 shows the hardness and elastic modulus of the various coatings containing increasing amounts of Nb. Hardness and elastic modulus present an approximately linear increase with the Nb concentration, due to the densification of the film [
56], except for the Nb30 sample. The NbN layer presented a higher hardness and elastic modulus, similar to other sputtered niobium nitrides [
57].
Figure 11 shows a Raman spectrum acquired on the NbN300 layer. One can see a slight shift to the right for the niobium nitride peak with respect to the standard position. Barshilia et al. [
58] found that heating NbN in air promotes this displacement. This behavior may be attributed to oxidation since, as seen in
Figure 5, oxygen contamination from the target was noticed in the niobium nitride layer.
Figure 12 presents the morphologies of the scratches conducted on the coatings analyzed in this work. The coatings do not break. Conversely, they smear along the scratch distance with an associated plastic deformation [
59]. The smearing happens because the coating is not hard and brittle and does not break abruptly.
Figure 13 shows the results obtained with the scratch tests of the Nb25 to Nb40 coatings. This figure presents the evolution of friction coefficients, tangential forces, and normal force as a function of the sliding distance. In
Figure 13, the cohesive and adhesive failures of the coatings are indicated by black arrows. One can see that the adhesive failure of the coatings does not vary significantly, with the Nb content of the coating occurring at ~3.2 mm from the beginning of the scratch. However, a clear improvement in adhesion was achieved in the system containing the 17 GPa hard NbN interlayer, which granted mechanical support to the Nb:MoS
2 coating. In this case, the adhesive failure occurred in the duplex coating at ~4.8 mm from the beginning.
The cohesive and adhesive failures were obtained by combining the information from the scratch images and the increase in scattering observed in the curves, both on tangential force and the coefficient of friction curves. Some failures of the coating itself, not related to the adhesion part, are presented in
Figure 14. As can be seen, the coating does not detach from the surface; it only presents angular cracks from the coating failure itself [
60,
61].
The absence of coating material has characterized the adhesive failures of the coating occurring at specific regions, as related to increased scattering in the coefficient of friction graph, which are observed in
Figure 15 by the SEM and EDS characterization. In the optical microscope, it is possible to observe some regions with different gray shades, as seen in
Figure 15A, which could correspond to regions with increased scattering on the graph. In
Figure 15B, the EDS analyses confirmed that the coating was removed in these regions.
Figure 15C shows that immediately after the beginning of the serrated behavior of the COF graph, the number of regions (white spots) where the coating was removed increased, indicating the film’s adhesive failure [
60,
61].
In
Figure 13, when comparing the adhesive and cohesive failures of samples without the NbN layer, the cohesive failures are different, as presented in
Table 7, since the introduction of niobium changed the mechanical properties of the coatings and, consequently, their fracture strength. However, when one observes the critical loads for adhesive failures, they are similar, ~15 N. This behavior occurs because both coatings present the same interlayer of metallic niobium. Thus, the adhesion forces are not related to the Nb insertion but to the interlayer.
However, when a NbN interlayer is added to the Nb25 coating, obtaining the Nb25NbN300 sample, one observes that the cohesive failure is similar. Still, the critical load for adhesive failure of the duplex-coated (NbN/Nb:MoS2) sample Nb25NbN300 is 28% higher than that of the single-coated Nb25 sample. The amount of niobium being the same, Nb:MoS2 film properties are the same, leading to similar cohesive failures. However, introducing a reinforcing NbN interlayer changed the adhesion properties of the coatings.
The improvement of adhesion occurs due to a higher hardness of the interlayer, diminishing the difference in hardness between the steel substrate and the coating. This interlayer might be responsible for an increase in the load-bearing capacity of the coating [
33,
34].
Comparing the critical loads for adhesive failure presented in
Figure 6 with those reported for Nb:MoS
2 and Ti:MoS
2 by Barslan, Efeoglu, and co-authors [
62,
63], varying from 15 N to 30 N, one can see that the simple Nb:MoS
2 films deposited without a reinforcing layer stays a little beyond. However, it is often difficult to directly compare these values with literature ones, since critical loads are a result of several variables (e.g., substrate hardness) in addition to the adhesion itself.
The Nb:MoS
2 coatings were tested in pinion–gear pairs in a Gear Back-to-Back Test Rig under ISO 68 lubricant (102 viscosity index and 70.3 cSt viscosity at 40 °C) to assess friction coefficients, wear rates, and resistance to contact and bending fatigue, where they could be used as solid lubricants [
64].