The Effect of Carbon Structure of DLC Coatings on Friction Characteristics of MoDTC-Derived Tribofilm by Using an In Situ Reflectance Spectroscopy

ta-C and Tantalum-doped ta-C (ta-C:Ta) coatings were deposited using the FCVA (filtered cathode vacuum arc) method, as illustrated in Fig. 1. A SUJ2 (AISI52100) steel disk was used as the substrate for ta-C coating, while a Si wafer was used for ta-C:Ta coatings. Prior to deposition, the substrate underwent a 20-min pre-sputtering process with Ar ions to remove surface contaminants. In the deposition process, carbon ions were generated through arc discharge from a graphite target (with a purity of 99.99%). The discharge current for ta-C deposition was 50 A, while for ta-C:Ta deposition, it was 80 A. The generated carbon ions were accelerated by a bias voltage of \(- 100\) V applied to the substrate. The substrate was placed on a rotation table, rotating at 10 rpm for ta-C deposition and 4 rpm for ta-C:Ta deposition. The FCVA system also featured a magnetron sputtering source used to introduce tantalum into the ta-C coating by sputtering a tantalum target (with a purity of 99.99%) using argon ions. The amount of doped tantalum was adjusted by varying the discharge current. Table 1 presents the properties of the ta-C and ta-C:Ta coatings. In this paper, the ta-C and ta-C:Ta coatings are referred to as ta-C, ta-C:Ta_0.024, and ta-C:Ta_0.103, based on their tantalum-to-carbon ratios measured by XPS (X-ray photoelectron spectroscopy) using the PHI Quantera III instrument from ULVAC-PHI Inc., Japan.

Pure and tantalum-doped graphene nanocrystalline carbon (GNC) films were deposited on p-type < 100 > silicon substrates using a multifunctional ECR (electron cyclotron resonance) plasma sputtering system, as schematically shown in Fig. 2 [27]. This system combines ECR plasma sputtering and magnetron sputtering to achieve its unique hybrid sputtering function. The ECR plasma sputtering method was employed to deposit the GNC film with low-energy electron irradiation, while the magnetron sputtering allowed for independent control of the doping parameters during film deposition. The vacuum chamber maintained a background pressure of 8 × 10^–5 Pa, with argon gas used to maintain a working pressure of 1 × 10^–1 Pa. A mirror-confinement magnetic field was created using magnetic coils, and the ECR plasma was generated by introducing microwave (2.45 GHz, 500 W) through a quartz window. Before deposition, the silicon substrate underwent cleaning by argon ion sputtering for 3 min. Then, the ECR carbon target was sputtered by argon ions with a bias voltage of \(- 500\) V to provide carbon atoms for film growth. During film deposition, low-energy electron irradiation onto the film was achieved by applying a positive bias voltage of + 80 V to the substrate. The electron irradiation energy was approximately 80 eV, and the electron irradiation density was 66.2 mA/cm². Magnetron target currents were set at 100 mA and 300 mA to achieve different tantalum doping concentrations. The deposition time was 30 min, and the film thickness was approximately 150 nm. The properties of the GNC and GNC:Ta coatings are indicated in Table 2, and they are referred as to GNC, GNC:Ta_0.030 and GNC:Ta_0.113 coatings based on their tantalum-to-carbon ratios measured by XPS.

In this study, a pin-on-disk type friction tester, as depicted in Fig. 3, was utilized to evaluate the friction properties of each DLC coating in a lubricant containing MoDTC. The friction tester was equipped with reflectance spectroscopy (OPTM-H2, Otsuka Electronics Co., Ltd.) to enable in situ analysis of the contact point, which was positioned above the contact area. A sapphire hemisphere (diameter of φ 8 mm) was used as the mating material for the DLC coatings due to its high transmittance of over 85% in the visible light range. The specific friction test conditions are detailed in Table 3. The measurement spot of the reflectance spectroscopy had a diameter of approximately φ 10 μm, which was smaller than the Hertzian contact diameter of approximately φ 40 μm under the given friction test conditions.

The calculation of thickness and optical properties of the tribofilm was performed using the OPTM post-analysis software (Otsuka Electronics Co., Ltd., Japan), relying on the in situ observed reflectance, denoted as R. The reflectance, R, was defined as Eq. (1):

The optical model shown in Fig. 4 was utilized for the calculation. The substrate and atmosphere were represented by Si (Si wafer) and sapphire (sapphire hemisphere), respectively. The DLC coating was incorporated in the model, and its optical properties were fixed to as-deposited state. The tribofilm derived from MoDTC was designated as the analysis layer, with both thickness and optical properties set as variable values. Moreover, the roughness layer was considered in the optical model because each DLC coating had a different surface roughness. The roughness in the optical model was treated as a mixture state of two materials (with a ratio of 0.5 and 0.5) according to Eq. (2) [44].

Here, ε represents the complex dielectric constant, which is calculated using the refractive index n and the extinction coefficient k with Eq. (3):

The reflectance of the optical model R₀₁₂₃₄ was calculated using Eqs. (4–10):here, r_ij represents the amplitude reflectivity between the interfaces of i and j, β_m is the interference phase angle, N_m, n_m, and k_m are the complex refractive index, the refractive index, and the extinction coefficient of each layer, respectively, and t_m is the thickness of each layer. The suffixes 0, 1, 2, 3, and 4 indicate sapphire, the tribofilm derived from MoDTC, the roughness layer, the DLC coatings, and the Si substrate, respectively. The calculated reflectance R₀₁₂₃₄ (Eqs. (4)-(10)) was fitted to the in situ measured reflectance R (Eq. (1)) using the nonlinear least-squares method, resulting in the determination of the thickness t, the spectrum of refractive index n, and the extinction coefficient k of the tribofilm.

Raman spectroscopy was employed for the assessment of coating properties, utilizing a Raman spectroscopy equipment (RENISHAW, in Via Reflex) with a laser wavelength of 532 nm. The acquired Raman spectrum was subjected to deconvolution to isolate two characteristic DLC peaks: the G band, located between 1500 and 1600 cm⁻¹, and the D band, located around 1350 cm⁻¹, utilizing a Gaussian function. The acquired Raman spectra and its deconvolution were indicated in Fig. 5. The ta-C and ta-C:Ta coatings exhibited broad Raman spectra, while the GNC and GNC:Ta coatings displayed relatively sharp G and D peaks. The intensity ratio of the G and D peaks, represented as the I_D/I_G ratio, is depicted in Fig. 6. The I_D/I_G ratio of ta-C:Ta coatings increased from 0.19 to 1.00 with an increasing Ta/C ratio. In contrast, the I_D/I_G ratio of GNC:Ta coatings remained approximately 1.2 regardless of the Ta/C ratio. The G peak originates from the vibration of graphite, whereas the D peak originates from defects in the sp² structure [45‐47]. Consequently, it is inferred that specimens with higher I_D/I_G ratios, such as GNC:Ta coatings and ta-C:Ta_0.103 coating, encompass a larger number of graphite defects.

Moreover, the ratio of C–C sp² bonding to C–C sp³ bonding in each coating was investigated using X-ray Photoelectron Spectroscopy (XPS) equipment. The C1s narrow peak (280 eV—295 eV) was acquired, and its background was determined using the Shirley method, as illustrated in Fig. 7 [48]. In the case of ta-C pure and GNC pure coatings, the C 1 s narrow peak underwent deconvolution into C–C sp² (284.4 eV), C–C sp³ (285.4 eV), and C-O (287.6 eV) [49]. For Ta-doped DLC coatings, the consideration of C-Ta (283.6 eV) was added to the deconvolution, alongside C–C sp², C–C sp³, and C–O [50]. The results of peak deconvolution are depicted in Fig. 8, and the calculated \({{\text{sp}}}^{2}/\left({{\text{sp}}}^{2}+{{\text{sp}}}^{3}\right)\) ratio from Fig. 8 is presented in Fig. 9. The ta-C pure coting exhibited the smallest \({{\text{sp}}}^{2}/\left({{\text{sp}}}^{2}+{{\text{sp}}}^{3}\right)\) ratio of 0.33, indicating a predominantly sp³-rich structure. In both ta-C:Ta and GNC:Ta coatings, the \({{\text{sp}}}^{2}/\left({{\text{sp}}}^{2}+{{\text{sp}}}^{3}\right)\) ratio increased with a higher Ta/C ratio. Consequently, the doped Ta contributed to the augmentation of sp² bonding in the DLC coatings.

Moreover, the work function of the DLC surface was assessed using the Kelvin method, employing the Kelvin Force Microscopy (KFM) mode of an atomic force microscope (SPA-400, Hitachi High-Tech). In KFM measurements, the contact potential difference (\(\Delta {\text{V}}\)) between the specimen (DLC coatings) and the probe (Rh-coated cantilever, Si-DF3-R) was measured. The work function of the DLC coatings was then computed from the measured \(\Delta {\text{V}}\) using Eq. (11) as follows:here, e represents the elementary charge, and \({\phi }_{{\text{specimen}}}\) and \({\phi }_{{\text{probe}}}\) denote the work function of the specimen and the probe, respectively. In this experiment, \({\phi }_{{\text{probe}}}\) was set as the work function of Rh, 4.98 eV [51]. The measured work functions of the six types of DLC coatings are illustrated in Fig. 10. The work function of the ta-C pure coating was 4.93 eV, a value consistent with the typical ta-C coating [52]. The work function of GNC coatings exhibited a decrease from 5.14 to 5.00 eV with an increasing Ta/C ratio. In the case of the ta-C:Ta coating, the work function of ta-C:Ta0.024 and ta-C:Ta0.103 coatings was 4.83 eV and 5.01 eV, respectively.

The friction characteristics of DLC coatings in a MoDTC-added lubricant are depicted in Fig. 11. As illustrated in Fig. 11a, the ta-C:Ta_0.113 coating exhibited the lowest friction coefficient among all DLC specimens, reaching around 0.07 by the conclusion of the friction test. Conversely, the ta-C:Ta_0.024 coating displayed the highest friction among all DLC coatings. The friction characteristics of GNC coatings in a MoDTC-added lubricant are compared in Fig. 11b. All GNC coatings demonstrated similar friction coefficients, fluctuating around 0.10, irrespective of the Ta/C ratio (Ta concentration). The results of GNC:Ta coatings indicated that Ta concentration did not appear to have a significant impact on the friction reduction effect of MoDTC. As a crucial coating property, the relationship between the average friction coefficient (averaged over three friction tests for each DLC coating) and the I_D/I_G ratio is presented in Fig. 12. In Fig. 12, coatings with a higher I_D/I_G ratio (such as ta-C:Ta_0.103 and all GNC:Ta coatings) tended to exhibit a lower friction, approximately 0.11, with the MoDTC-added lubricant, which was lower than the average friction coefficient with base oil. These results suggested that DLC coatings with a higher I_D/I_G ratio were effective in reducing friction in MoDTC-added lubricants. Considering the case of the ta-C pure coating, the average friction coefficient with the MoDTC-added lubricant was 0.124, which was comparable to GNC:Ta_0.113 coatings. However, the average friction coefficient of ta-C coating with base oil was 0.114. Therefore, the friction reduction effect of MoDTC did not manifest in the friction test of the ta-C coating.

Raman spectroscopy is a useful technique to evaluate MoDTC-derived tribofilm due to the distinct peaks associated with MoS₂. Notably, the major product of MoS₂ exhibits two characteristic peaks, the A_1g peak at 409 cm⁻¹ and the E_2g peak at 383 cm⁻¹ [32, 53]. In this study, Raman analysis of the tribofilm formed on four specimens, namely ta-C pure, ta-C:Ta_0.024, ta-C:Ta_0.103, and GNC:Ta_0.030, was conducted. The Raman spectra of these specimens are presented in Fig. 13. The A_1g and E_2g peaks were observed in the tribofilm on ta-C:Ta_0.024, GNC:Ta_0.030, and ta-C:Ta_0.103 coatings, with particularly higher intensities in GNC:Ta_0.030 and ta-C:Ta_0.103 coatings. Conversely, the A_1g and E_2g peaks were scarcely observed in the tribofilm on ta-C pure coating, indicating a limited generation of MoS₂. The intensities of the A_1g and E_2g peaks are correlated with the amount of MoS₂ in the tribofilm. Therefore, the relationship between the intensity of the A_1g peak and friction coefficient is illustrated in Fig. 14. Specimens with higher A_1g peak intensity, such as GNC:Ta_0.030 and ta-C:Ta_0.103 coatings, tended to exhibit lower friction coefficients of around 0.10. Consequently, the friction reduction in GNC:Ta_0.030 and ta-C:Ta_0.103 coatings was attributed to the abundant generation of MoS₂. Moreover, the tribofilm with a higher intensity of the MoS₂-derived A_1g peak was formed on specimens with a higher I_D/I_G ratio, such as GNC:Ta_0.030 (I_D/I_G = 1.14) and ta-C:Ta_0.103 (I_D/I_G = 1.02). This suggests that the characteristics of DLC coatings have an influence on the generation of low-shear MoS₂ material.

The tribofilm components were analyzed using energy dispersive X-ray spectroscopy (EDS). The EDS spectra of the tribofilm are compared in Fig. 15. For the tribofilms formed on ta-C:Ta_0.024, GNC:Ta_0.030, and ta-C:Ta_0.103 coatings, the presence of C and O was confirmed. Distinct peaks of C(Kα) at 0.277 keV and O(Kα) at 0.525 keV were observed, along with convolutions of Ta (M) at 1.71 keV with Si (Kα) at 1.74 keV and Mo (Lα) at 2.29 keV with S (Kα) at 2.31 keV. Notably, the tribofilm on the GNC:Ta_0.030 coating exhibited a stronger Mo (Lα) and S (Kα) peak compared to that on the ta-C:Ta_0.103 coating. This observation suggested that the tribofilm on the GNC:Ta_0.030 coating was thicker than that on the ta-C:Ta_0.103 coating, leading to a higher total intensity of Mo and S in the former. Additionally, a distinct peak corresponding to Al (Kα) was identified in the tribofilm on the ta-C:Ta_0.024 coating. Aluminum was presumed to originate from wear particles of the sapphire hemisphere (Al₂O₃). The ta-C:Ta_0.024 coating, being a relatively hard material with a hardness of 12.3 GPa compared to 5.2 GPa for GNC:Ta_0.030 and 9.5 GPa for ta-C:Ta_0.103 coatings, resulted in the generation of wear particles from the mating sapphire, which were subsequently incorporated into the tribofilm.

The thickness t and optical properties, including refractive index n and extinction coefficient k, of the tribofilm were calculated using the optical interference formula indicated in Eqs. (4–10). The analysis and consideration were applied to the tribofilm formed on ta-C:Ta_0.024, ta-C:Ta_0.103, GNC pure, GNC:Ta_0.030, and GNC:Ta_0.113 coatings at intervals of 60 cycles. The calculated tribofilm thickness is presented in Fig. 16. For the tribofilm on the GNC pure coating, the reflectance at 60 cycles fell below the minimum resolution of 0.01 for the equipment. Consequently, the analysis was conducted after 120 cycles for this case. The tribofilm on the ta-C:Ta_0.103 coating consistently exhibited a thickness of around 6 nm, less than half of the thickness observed in the other tribofilms. The GNC:Ta_0.113 coating, which had a Ta amount close to ta-C:Ta_0.103, showed the largest thickness of around 25 nm. Thus, the Ta amount did not appear to be significant in determining the thickness of the tribofilm. The thickness of the tribofilm is determined by both its generation and removal. Based on in situ observed thickness, the tribofilm was maintained on the sliding surface from the initial to the end of the testing cycles. In Fig. 17, the relationship between tribofilm thickness and friction coefficient is depicted. The three types of GNC:Ta coatings exhibited similar friction characteristics, maintaining around 0.10. However, the tribofilm thickness on GNC:Ta_0.113 was consistently over 25 nm, a value larger than the tribofilm on GNC pure and GNC:Ta_0.030 coatings by 10 nm. Consequently, it is suggested that other characteristics of the tribofilm have a more significant effect on friction properties. Therefore, the effect of the tribofilm composition was considered in the next chapter.

In addition to thickness, optical properties were calculated through the analysis of optical interference. As an illustrative example, Fig. 18 depicts the transition of refractive index n and extinction coefficient k of the tribofilm formed on the ta-C:Ta_0.103 coating. The specimen exhibited a gradual reduction in friction as the test progressed in the MoDTC-added lubricant. In terms of optical properties, both refractive index and extinction coefficient increased with the progression of friction. This outcome suggests that in situ reflectance spectroscopy is effective for observing the evolution of the tribofilm and its tribological properties simultaneously. The optical properties of the tribofilm are presumed to be determined by its constituents, such as MoDTC-derived products and wear particles from the DLC coating. Therefore, the estimation of the tribofilm composition based on its optical properties using effective medium approximation (EMA) was conducted [44].

In the Effective Medium Approximation (EMA) method, the complex dielectric constant of the mixture (tribofilm, in this case) is determined by the complex dielectric constant and volume fraction of each component, as described in Eqs. (12) and (13)here, \({\varepsilon }_{tribofilm}\) represents the complex dielectric constant of the tribofilm, calculated using the refractive index and extinction coefficient from in situ reflectance spectroscopy. \({\varepsilon }_{i}\) represents the complex dielectric constant of each component, calculated using pre-measured refractive index and extinction coefficient values. The complex dielectric constant is determined using the refractive index n and extinction coefficient k as shown in Eq. (13).

For the tribofilm formed on ta-C:Ta_0.103, GNC pure, GNC:Ta_0.030, and GNC:Ta_0.113 coatings, the tribofilm is assumed to consist of three components: MoS₂, MoO₃, and wear particles from DLC coatings. MoS₂ and MoO₃ are major products from MoDTC, and their optical properties were measured using pure specimens of MoS₂ and MoO₃. The optical properties of wear particles from DLC coatings utilized the as-deposited state values for each coating. Additionally, for the tribofilm on the ta-C:Ta_0.024 coating, the existence of sapphire was considered, in addition to MoS₂, MoO₃, and DLC particles, due to the inclusion of wear particles from the mating sapphire, as described in Chapter 3.4. The transition of the volume fraction of the tribofilm estimated from optical properties is shown in Fig. 19. In Fig. 19a, the tribofilm on the ta-C:Ta_0.024 coating contained approximately 34 vol.% of sapphire particles. Concerning MoDTC-derived products, the tribofilm contained an almost equal volume fraction of MoS₂ and MoO₃. For the tribofilm on the ta-C:Ta_0.103 coating, which experienced a reduction in friction to 0.07, the volume fraction of MoS₂ in the tribofilm increased from 40.6 to 50.2 vol.% with friction, while the volume fraction of MoO₃ decreased from 32.5 to 15.9 vol.%. To assess the impact of MoDTC-derived products on friction characteristics, the \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio was calculated from the volume fraction, as shown in Fig. 19. The relationship between the \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio and friction coefficient is indicated in Fig. 20. The \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio of the tribofilm on the ta-C:Ta_0.103 coating increased to 0.75, and the friction decreased to 0.07. Conversely, the \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio of the tribofilm on the ta-C:Ta_0.024 coating was around 0.50, and the friction coefficient exceeded 0.12. The plots of the tribofilm formed on GNC:Ta coatings concentrated at a \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio of around 0.65 and a friction coefficient of around 0.10. The correlation coefficient R between the \({{\text{MoS}}}_{2}/({{\text{MoS}}}_{2}+{{\text{MoO}}}_{3})\) ratio and friction coefficient was \(R=-0.85\), indicating a strong correlation. Thus, it is considered that the friction coefficient is strongly influenced by the generation of low-shear material of MoS₂ from MoDTC-derived materials, and the DLC coatings with a higher I_D/I_G ratio enhance the generation of MoS₂.

The relationship between the I_D/I_G ratio and work function is illustrated in Fig. 21. Generally, DLC coatings with a higher I_D/I_G ratio exhibited a higher work function, with the exception of the ta-C pure coating. Akada et al. reported that the work function of highly oriented pyrolytic graphite (HOPG), characterized by a defect-free graphite structure, increased with damage caused by Ar plasma [54]. Hence, it is inferred that DLC coatings with a higher work function incorporate a significant amount of graphite structure defects, specifically graphite edges, which contribute to the increased intensity of the D peak in Raman analysis. Additionally, the ta-C pure coating had a smaller \({{\text{sp}}}^{2}/\left({{\text{sp}}}^{2}+{{\text{sp}}}^{3}\right)\) ratio of 0.33, indicating a sp³-rich amorphous structure. Consequently, the work function of the ta-C pure coating is believed to be influenced by factors other than graphite edges, such as dangling bonds.

Nakashima et al. reported the adhesion of silica to occur on graphite edges with higher work functions [22]. Furthermore, Murashima et al. found that ZnDTP-derived tribofilm was abundantly generated on DLC coatings with a higher sp²/sp³ ratio, including graphite edges [55]. These prior studies suggest that graphite defects play an active role and exert an influence on lubricant additives. Additionally, Murabayashi proposed the decomposition process of MoDTC on a DLC coating through molecular dynamics simulation [56]. According to their findings, MoDTC adheres to a DLC surface, donating electrons to the DLC, thereby enhancing MoDTC decomposition. Therefore, graphite edges are deemed significant as sites where MoDTC adheres and reacts with the DLC surface, elucidating why specimens with higher I_D/I_G ratios exhibit lower friction with MoDTC.

Moreover, the ta-C:Ta_0.103 coating displayed the lowest friction coefficient of 0.07 and consistently maintained a tribofilm with a high concentration of MoS₂ among all DLC specimens. In previous studies by Komori, it was observed that a hard and rough DLC coating promotes the formation of a MoS₂-rich tribofilm due to higher contact pressure at asperities and increased frictional heat [57]. In our study, although the three types of GNC:Ta coatings exhibited a higher I_D/I_G ratio of around 1.2, their hardness ranged from 1.1 GPa (GNC pure) to 6.0 GPa (GNC:Ta_0.113), which is softer than the ta-C:Ta_0.103 coating with a hardness of 9.5 GPa. Consequently, the ta-C:Ta_0.103 coating is considered to be in an optimal state, characterized by the inclusion of active graphite edge sites and sufficient hardness to promote the generation of MoS₂, resulting in friction reduction.