Mechanical Properties, Tribological Performance and Oxidation Resistance of HfC_x/a-C:H Coatings Prepared by Pulsed DC Magnetron Sputtering

Abstract

The development of protective coatings with simultaneously enhanced mechanical, tribological, and antioxidant properties remains a major challenge for micro-electro-mechanical systems operating under harsh environments. In this work, HfC_x/a-C:H coatings with varying carbon contents were deposited by magnetron sputtering. Increasing the C₂H₂ flow rate from 12 to 20 sccm drove the coating structure to undergo two-stage evolution, from a composite structure dominated by HfC nanograins with a-C:H distributed at triple junctions of HfC_x grain boundaries to a typical nanocomposite structure with ~8 nm HfC_x nanograins embedded in a continuous a-C:H matrix. The coating deposited at 18 sccm exhibited the highest hardness (31.3 GPa) and effective Young’s modulus (392.3 GPa), owing to enhanced interface-mediated strengthening effect induced by the optimized nanocomposite structure. The coating prepared at 20 sccm showed the lowest friction coefficient (0.28), the lowest wear rate (6.82 × 10⁻⁶ mm³/N·m), and the best oxidation resistance. These improvements were supported by the enhanced mechanical properties and a-C:H fraction, the increased interface density and tortuosity, and the regulation of oxidation kinetics by the a-C:H matrix. This work provides an effective strategy for designing multi-functional protective coatings with balanced mechanical, tribological, and oxidation performance.

1. Introduction

With the continuous development of Micro-Electro-Mechanical Systems (MEMS) [1,2], micro-brakes and micro-mechanical transmission systems have attracted increasing attention in applications of aerospace, precision instruments, and harsh-environment sensing [3]. These devices rely on microscale contact interfaces to achieve motion control, energy transmission and power regulation. Therefore, the performance and reliability of the interface coatings have a crucial impact. Compared with traditional macroscopic mechanical systems, the contact area of the braking and transmission components based on micro-electromechanical systems is much smaller. This results in a high concentration of local contact stress and extremely limited lubrication conditions [4]. As a consequence, the interface coating is more prone to plastic deformation, wear-induced damage, and interfacial failure. Moreover, as the demand for operation in high-temperature environments continues to increase, the oxidation and structural degradation of the coating will be further exacerbated. Therefore, the combination of high contact stress, severe wear, and thermal oxidation is one of the most challenging working conditions faced by interface coatings in micro-braking and micro-mechanical transmission systems. Under such conditions, the mechanical robustness, tribological stability, and oxidation resistance of the coating become decisive factors governing device lifetime and long-term reliability.

Under this failure mode that combines high contact stress, severe wear and thermal oxidation, the existing coatings exhibit notable limitations. Diamond-like carbon (DLC) coating possesses high hardness and elastic modulus, which enable excellent low-friction and wear-resistant performance under low-load conditions [5]. However, in applications such as the tooth surfaces of micro-gears and the raceways of micro-bearings, the highly concentrated local contact stress could induce brittle delamination of the DLC coating. This, in turn, reduces the long-term reliability of the coating. More dangerously, DLC is easily oxidized under moderate temperatures, leading to device instability. Molybdenum disulfide (MoS₂) coating, owing to its layered structure, exhibits extremely low shear strength and outstanding friction-reducing capability. However, its limited load-bearing capacity makes it susceptible to compressive failure under high contact stresses. This further leads to the rapid loss of the structural integrity of the coating. In addition, MoS₂ coating is prone to oxidative decomposition under moderate temperature and humid environments [6]. These drawbacks restrict its applicability in high-frequency and medium-to-high temperature micro-transmission systems. To sum up, the above issues indicate that merely relying on the existing single-phase coatings is insufficient to meet the collaborative requirements of micro-brake and micro-mechanical transmission systems in terms of high hardness, low friction, wear resistance, and oxidation resistance.

Transition metal carbide HfC possesses high stiffness and elastic modulus, which could provide the required load-bearing capacity for microscale contact interfaces and could effectively suppress plastic deformation under high contact stresses [7,8]. Adding amorphous carbon phase (a-C:H) to HfC to form a composite structure enables the tailoring of toughness and damage resistance while maintaining high load-bearing capacity [9]. Moreover, the presence of a-C:H phase can reduce the shear resistance and friction-induced noise during sliding, thereby improving the tribological performance [10,11]. The formed composite structure has the potential to regulate the atomic diffusion pathways. Consequently, it can reduce the risk of oxidation under medium-high temperatures. Therefore, the HfC_x/a-C:H coating has the potential to simultaneously possess high hardness, high wear resistance and oxidation resistance.

Zou et al. [12] investigated the infrared-visible compatible stealth property and tribological performance of the HfC_x films with varying carbon content. They revealed that the amorphous carbon phase located at the nanocolumnar crystals and grain boundary triple junctions plays a key role in achieving both excellent infrared-visible compatible stealth property and good wear resistance. Zhang et al. [13] prepared HfC_x films by reactive magnetron sputtering and examined the influence of microstructure on their mechanical and tribological properties. The results showed that the HfC_x film containing 52 at.% a-C:H exhibits a low friction coefficient of 0.1 and a wear rate of 1.10 × 10⁻⁶ mm³/N·m. Li et al. [14] conducted a study on the mechanical properties (including microhardness and elastic modulus) of HfC coatings by combining first-principles calculations and experimental methods. Ferro et al. [15] investigated the intrinsic hardness of HfC films by separating it from composite hardness. However, to the best of our knowledge, a systematic investigation on the combined mechanical, tribological properties, and oxidation resistance of HfC_x/a-C:H coatings has not yet been studied and reported.

In the present work, driven by the need to enhance the reliability of MEMS devices under high-frequency and elevated-temperature conditions, a series of HfC_x/a-C:H coatings with tunable carbon content was fabricated by pulsed DC magnetron sputtering. The composition, chemical bonding states, phase composition, and microstructure of the HfC_x/a-C:H coatings were systematically analyzed using XPS, TEM, XRD, and SEM/EDS. Based on these characterizations, the mechanical properties, tribological behavior, and oxidation resistance were comprehensively investigated. Furthermore, the underlying mechanisms governing the relationships among composition, microstructure, and the resulting properties were analyzed and discussed in detail.

2. Experimental Procedure

2.1. Coating Preparation

HfC_x/a-C:H coatings were deposited in a Balzers 640 R deposition system operating in a reactive atmosphere of Ar and acetylene (C₂H₂). The schematic diagram of coating deposition is shown in Figure 1. A pulsed DC power supply (Pinnacle Plus, Advanced Energy) was employed to sputter an Hf target with a purity of 99.9% and dimensions of 150 mm in diameter and 8 mm in thickness. The target-to-substrate distance (D_t-f) was maintained at 12 cm. During the entire deposition process, the substrate holder was kept under floating potential conditions. A variety of substrates were utilized depending on the intended characterization. M2 cylindrical substrate (63 HRC, ∅ = 30 mm × 8 mm), stainless steel plates (40 mm × 20 mm × 4 mm), silicon wafers (10 mm × 10 mm), and iron sheets (400 mm × 4 mm × 0.25 mm) were selected for evaluating mechanical properties, tribological behavior, oxidation resistance, microstructure, chemical bonding states, and residual stress.

Prior to deposition, the chamber was evacuated to a base pressure of 1 × 10⁻⁴ Pa. The substrates were then subjected to Ar ion etching for 15 min under a bias voltage of −700 V in order to remove surface contaminants of the substrates. To enhance coating adhesion, an Hf interlayer with a thickness of ~200 nm was deposited in advance. Following the deposition of the interlayer, C₂H₂ was introduced as the reactive gas, with its flow rate systematically varied from 12 to 20 sccm in increments of 2 sccm. The Ar flow rate was maintained at a constant of 200 sccm. Under these conditions, the total working pressure was approximately 0.39 Pa. The substrate holder was rotated at a constant speed of 30 revolutions per minute (rpm) during deposition. The cathode discharge parameters were kept unchanged, including an average current of 2.2 A, a discharge voltage of 212 V, and a pulse frequency of 50 kHz.

2.2. Experimental Tests

The elemental composition, stoichiometric ratio of crystalline phase, and chemical bonding states of the HfC_x/a-C:H coatings were characterized by X-ray photoelectron spectroscopy (XPS). The measurements were carried out under ultra-high vacuum conditions (~10⁻⁸ Pa) using an Al Kα radiation source with an energy of 1486.7 eV. Before analysis, the sample surfaces were sputter-cleaned by 1 keV Ar ion bombardment for 10 min to remove surface contaminants introduced during air exposure, while minimizing any alteration to the intrinsic composition of the coatings. The phase structures were investigated using an X-ray diffractometer (BRUKER D8, Karlsruhe, Germany, with Co source) operated at a scanning rate of 0.1° s⁻¹. The crystallite size was estimated according to the Scherrer equation. The surface morphology and fracture cross-sectional features of the coatings were examined by field emission scanning electron microscopy (FE-SEM, JEOL JSM-7800F, Akishima, Tokyo, Japan), which was also used to measure the coating thickness. Detailed information on the coating microstructure was obtained through TEM (FEI Talos F200X-G2, Hillsboro, OR, USA, equipped with Super-X energy spectrometer). The cross-section of the sample used for HR-TEM analysis was prepared using a FIB-SEM double beam system (FEI, Helios Nano Lab 600i, Hillsboro, OR, USA) operating at 2–30 kV. The gallium ion beam was operated at a voltage of 30 kV.

Residual stress was evaluated by measuring the curvature radius of coatings deposited on the stress-relieved iron bar with dimensions of 400 mm × 4 mm × 0.25 mm. The stress values were calculated based on Stoney’s equation. Before the coating deposition, the iron bar substrate was annealed for 2 h to eliminate the initial residual stress, thereby ensuring a zero-curvature condition and avoiding potential errors in the subsequent curvature measurements. Upon completion of the coating deposition, the curvature radius (R) of the coated iron bar was measured using an Altysurf profilometer (AltiSurf 500, ALTIMET SAS, Marin, France) equipped with an optical probe. The residual stress in the as-deposited HfC_x/a-C:H coating was then calculated from the measured curvature radius using Stoney’s equation, as expressed in Equation (1) [16].

$$\sigma ={\displaystyle \frac{E_s}{6(1-{\upsilon }_s)}}{\displaystyle \frac{d_s^2}{d_c}}{\displaystyle \frac{1}{R}}$$

(1)

In this equation, E_s, d_s, and ν_s represent the Young’s modulus, thickness, and Poisson’s ratio of the substrate, respectively, whereas d_c denotes the coating thickness and R corresponds to the measured curvature radius. The nanohardness (H) and effective Young’s modulus (E*) of the coatings were measured using a nanoindentation tester (CSM Instruments) with an applied load of 5 mN. The tests were performed at room temperature using the continuous stiffness measurement mode with the Berkovich diamond indenter. The tribological properties were assessed using a pin-on-disk tribometer (CSM, Peseux, Switzerland). All tests were conducted under dry sliding conditions against 6 mm WC/Co balls in ambient air with a relative humidity of 38%. The testing parameters included a normal load of 5 N, a sliding speed of 5 cm/s, a wear track radius of 3 mm, a total sliding cycle of 10,000, and a total sliding distance of 188.5 m. After the tribological tests, the cross-sectional profiles of the wear tracks were measured using an Altysurf profilometer equipped with an inductive probe, while the worn surfaces were examined by SEM and EDS. The oxidation resistance of the coatings was evaluated in air at temperatures ranging from 200 °C to 600 °C using a furnace (Nabertherm GmbH, Lilienthal, Germany), with a holding time of 2 h at each temperature. After annealing, the samples were allowed to cool naturally to room temperature within the furnace. The oxidation degree of the coatings was subsequently characterized by XRD and SEM/EDS analyses.

3. Results and Discussion

3.1. Chemical Bonding State

XPS analysis is performed for HfC_x/a-C:H coatings deposited under different C₂H₂ flow rates to determine their chemical bonding states. The charge calibration is performed by setting the binding energy of the C 1 s photoelectron peak to 285 eV. The high-resolution C 1 s spectra and the related fitting components are shown in Figure 2.

For all HfC_x/a-C:H coatings, the C 1 s spectra are deconvoluted into two low binding energy peaks associated with the Hf-C bonds and three high binding energy peaks corresponding to the amorphous carbon bonds and the C-O bond. They are Hf-C bond (281.2 eV), Hf-C* bond (282.5 eV), sp²-C (284.5 eV), sp³-C (285.4 eV), and C-O bond (286.8 eV), respectively. The presence of Hf-C* bond here originates from the core electrons at the interface between the carbide crystalline phase and the amorphous carbon phase, which has been reported multiple times in published works [17,18,19]. The appearance of the C-O bond mainly stems from the surface adsorption caused by long-term exposure to air. Its content is very low, and its impact on the main structure of the coating can be negligible. Therefore, from the chemical environment in which Hf-C, HfC*, and C-C (sp² and sp³) coexist, it can be determined that the coating exhibits a two-phase composite structure consisting of a crystalline HfC phase and an a-C:H phase. In addition, the relative fractions of the different chemical bonds in the HfC_x/a-C:H coatings are quantitatively evaluated from the high-resolution C 1 s XPS spectra. The contribution of each chemical bond is calculated by integrating the corresponding fitted peak area and normalizing it to the total C 1 s envelope, as summarized in

Table 1. Relative contents of Hf-C, Hf*-C, sp²-C, and sp³-C bonds, a-C:H content, and stoichiometric ratio x of HfC_x nanocrystals in the HfC_x/a-C:H coatings with different C₂H₂ flow rates.

C2H2 Flow Rate (sccm) No.	Composition [at.%]		Types of Bonding for C 1 s XPS [at.%]					a-C:H [at.%]	HfCx
	Hf	C	Hf-C	Hf*-C	sp2-C	sp3-C	C-O
12	53.8	46.2	60.4	12.5	14.9	9.7	2.5	11.7	0.64
14	51.4	48.6	56.1	12.4	17.9	10.6	3	14.3	0.67
16	47.7	52.3	52.1	9.8	21.2	13.4	3.5	18.7	0.70
18	42.6	57.4	45.5	7.1	24.7	19.6	3.1	26.2	0.73
20	40.9	59.1	40.2	10.1	25.6	20.5	3.6	28.2	0.75

. This quantitative analysis enables a direct comparison of the evolution of carbide-related bonds and a-C:H-related bonds as a function of C₂H₂ flow rate.

Table 1 shows that the chemical bonding state of the HfC_x/a-C:H coating undergoes a systematic evolution with increasing C₂H₂ flow rate. As the C₂H₂ flow rate increases from 12 sccm to 20 sccm, the total atomic concentration of Hf decreases from 53.8 to 40.9 at.%, while the C content continuously increases from 46.2 to 59.1 at.%. This results in a decrease in the relative fractions of Hf-C and Hf-C* bonds from 60.4 at.% and 12.5 at.% to 40.2 at.% and 10.1 at.%, respectively. Meanwhile, the proportion of the a-C:H phase increases progressively, with sp²-C and sp³-C components increasing by 71% and 110%, respectively. This evolution is closely associated with the reduced fraction of Hf species in the plasma and the enhanced carbon chemical potential under high C₂H₂ supply, the latter intrinsically favoring the formation of carbon-rich amorphous networks [20].

In addition, the growth rate of sp³-C in a-C:H matrix is higher than that of sp²-C, which is caused by stronger energetic bombardment of coating-forming particles provided by the higher flow rate of C₂H₂ [21]. It is noteworthy that, despite the pronounced increase in the a-C:H fraction and the overall reduction in Hf-C bonding content, a portion of carbon atoms is still incorporated into the crystalline HfC phase. This is evidenced by the increased substoichiometric value x of HfC_x from 0.64 at 12 sccm C₂H₂ to 0.75 at 20 sccm C₂H₂. It indicates that within the investigated C₂H₂ flow rate window of 12–20 sccm, the flux of Hf atoms arriving at the substrate surface remains consistently sufficient. Moreover, the kinetic conditions allow carbon atoms to be effectively captured by Hf sites and accommodated within the carbide lattice. Consequently, the incorporation of C in the HfC_x is enhanced. Simultaneously, the carbon supply is also abundant and monotonically increasing. Thus, in addition to the lattice-incorporated carbon, other carbon atoms enter the amorphous matrix, leading to an increase in sp²- and sp³-C.

3.2. Microstructure and Morphology

Figure 3a shows the XRD patterns of the HfC_x/a-C:H coatings deposited under different C₂H₂ flow rates. Figure 3b presents the evolution trends of grain size, a-C:H content (obtained based on XRD and XPS results) and coating compressive stress. Three diffraction peaks can be observed at 39.2$°$, 45.5$°$ and 66.4$°$, corresponding to the (111), (200), and (220) crystal planes of the face-centered cubic HfC (JCPDS: 03-065-8751), respectively. In the C₂H₂ flow rate range of 12 to 16 sccm, the HfC phase exhibits a distinct (111) preferred orientation, accompanied by a slight increase in peak intensity. The quantitative analysis of the FWHM of the (111) peak (Figure 3b) reveals that the HfC nanocrystals exhibit a slight coarsening, with the average grain size increasing from 15.3 nm to 17.0 nm. The (111) preferred orientation is attributed to the surface-energy-dominated growth mechanism of the coating within this C₂H₂ flow rate range. In this case, the system tends to minimize its total free energy by favoring the development of (111)-oriented grains, thereby enhancing the (111) texture [22]. In addition, in the range of 12–16 sccm, the carbon supply flux in plasma increases nearly linearly, whereas the flux of sputtered Hf species remains essentially constant. The increasing carbon supply flux promotes the increase in a-C:H content. However, its content may not yet reach the threshold required to inhibit the aggregation and growth of HfC grains. Therefore, the grain size increases slightly [23].

As C₂H₂ flow rate further increases to 18 sccm, the crystalline phase transitions to a random orientation composed of (111), (200) and (220) planes. The overall diffraction peak intensity decreases markedly, demonstrating a sudden refinement of the HfC grains (to 8.8 nm). Moreover, the content of the a-C:H phase significantly increases at this C₂H₂ flow rate (Figure 3b). The refined HfC grain and increased a-C:H content result in an increase in the proportion of the two-phase interface (between HfC and a-C:H). Under such conditions, the inherent surface free energy difference between different HfC crystal planes no longer dominates the total energy of the coating system. Instead, the interfacial energy between HfC and the isotropic a-C:H matrix takes over gradually. Meanwhile, the residual stress measurement (the bar chart in Figure 3b) demonstrates a monotonic increase trend within the C₂H₂ flow rate range of 12–18 sccm. The above two points indicate that the surface energy is no longer the primary driving force determining crystal orientation, while the roles of interfacial energy and strain energy are becoming significant [24]. Therefore, the crystal nucleation and growth behavior are governed by the combined effects of strain energy, interfacial energy, and kinetic constraints, giving rise to the observed random orientation of the HfC phase. In addition, the higher a-C:H content at the 18 sccm C₂H₂ flow rate more effectively encapsulates the HfC grains, thereby impeding the necessary adatom diffusion required for grain coalescence and growth. This consequently led to a sharp reduction in the HfC grain size.

After further increasing the flow rate of C₂H₂ to 20 sccm, the HfC phase still maintains its randomly oriented character. However, the intensity of the (111) peak slightly recovers, corresponding to a moderate increase in grain size to 12.4 nm. This behavior is attributed to the partial relaxation of compressive stress and the slowdown of the increase rate of a-C:H content [25,26]. The partial relaxation of compressive stress allows the intrinsically lower surface free energy of the (111) crystal plane to re-engage in the competition for crystal orientation, thus resuming the growth of the (111) peak. The slowdown of the increase rate of a-C:H content weakens its inhibitory effect on HfC grain growth, leading to a renewed increase in grain size. In addition, it is noted that from 12 to 20 sccm, the diffraction peaks shift towards lower angles. This is attributed to the continuous lattice expansion caused by the stoichiometric changes in HfC_x grains and the accumulative compressive stress inside the HfC_x/a-C:H coating.

To further elucidate the detailed composite structure of the HfC_x/a-C:H coatings, high-resolution TEM together with the selected area electron diffraction (SAED) analyses were conducted for the coatings deposited at the C₂H₂ flow rates of 14 sccm and 18 sccm, as shown in Figure 4. For the HfC_x/a-C:H coating deposited at 14 sccm, the cross-sectional HRTEM (Figure 4a1) image reveals the coexistence of well-defined lattice fringe regions and atomically disordered regions. These regions are attributed to well-crystallised HfC grains and an a-C:H matrix lacking long-range order, respectively. It can be observed that a-C:H does not form a continuous network structure. Instead, it is located at the triple junctions of HfC grain boundaries. Correspondingly, the SAED pattern consists of several distinct diffraction rings with non-uniform intensity distribution along the ring circumference, which is attributed to the crystallized HfC phase [27]. The weak diffuse scattering is also observed. This indicates a limited content of a-C:H matrix. As a result, it is demonstrated that the coating deposited at 14 sccm C₂H₂ exhibits a nanocomposite structure. It is mainly composed of the crystalline phase HfC and a small amount of a-C:H distributed at triple junctions of HfC_x grain boundaries. For the coating deposited at 18 sccm C₂H₂, shown in Figure 4b1, the large-scale lattice fringes disappear, and the proportion of atomically disordered regions increases. Meanwhile, the disordered region evolves into a continuous matrix that separates the crystalline regions. The corresponding SAED pattern is characterized by continuous, broadened diffraction rings with slightly diffuse boundaries. Moreover, the discrete diffraction spots are scarcely observable on the diffraction rings. These features indicate a refinement of HfC grains and an increased a-C:H volume fraction. Therefore, through HETEM and SEAD, it is determined that the HfC_x/a-C:H coating deposited at 18 sccm possesses a typical nanocomposite structure, which consists of a continuous a-C:H matrix and HfC grains dispersed within it.

Overall, combining the results of XPS, XRD, and TEM, it can be inferred that within the carbon supply window investigated in the present work, all HfC_x/a-C:H coatings possess a two-phase composite structure consisting of HfC_x nanocrystals and a-C:H matrix. The microstructure evolution of the HfC_x/a-C:H coatings is governed by the competitive growth between HfC_x grains and a-C:H matrix. In the carbon supply range of 12–16 sccm C₂H₂, both the size of HfC_x nanocrystals and the content of a-C:H matrix increase in the coatings. The a-C:H phase is located at HfC_x grain-boundary triple junctions. When the C₂H₂ flow rate reaches 18 sccm, the formation rate of the a-C:H phase exceeds the growth rate of HfC_x grains, causing the amorphous network to encapsulate the HfC_x grain boundaries. Consequently, a reduction in HfC_x grain size is observed. It thus results in a typical nanocomposite structure in which HfC_x grains are embedded in a continuous a-C:H matrix. As the C₂H₂ flow rate further increases to 20 sccm, the growth rate of the a-C:H matrix slows down, while the HfC_x grains exhibit a slight coarsening. It suggests that the nanocomposite structure still exists, accompanied by a variation in the average grain separation.

Figure 5 presents the SEM micrographs of the HfC_x/a-C:H coatings deposited at different C₂H₂ flow rates. The upper and lower rows correspond to the brittle fracture cross-sectional morphologies and the surface morphologies, respectively. All HfC_x/a-C:H coatings exhibit a typical columnar morphology growing along the deposition direction. With increasing C₂H₂ flow rate, the columnar structure undergoes progressive densification, characterized by narrowed intercolumnar gaps and diminished contrast of column boundaries. Meanwhile, the surface morphology evolves from relatively coarse granular features to finer and more uniform particles, indicating a gradual reduction in surface roughness. This morphological evolution is closely associated with the progressive enrichment of the a-C:H fraction. At low C₂H₂ flow rates (12 and 14 sccm), the a-C:H content is limited and the growth of the crystalline phase dominates, resulting in well-developed columnar morphology with clearly defined boundaries. As the C₂H₂ flow rate increases from 16 to 20 sccm, the a-C:H phase not only segregates to grain boundaries, but also tends to concentrate in the intercolumnar regions [28], thereby filling structural voids and weakening the contrast of columnar structures. Consequently, the columnar structure becomes more compact. In addition, the coating thickness is obtained by measuring the brittle cross-section using SEM. The evolution trend of coating thickness with C₂H₂ flow rate is shown in Figure 6. It can be seen that at the C₂H₂ flow rate of 12 sccm, the coating thickness is approximately 4.6 μm. As the C₂H₂ flow rate increases from 14 sccm to 20 sccm, the coating thickness remains around 4.0 μm, without a significant decreasing trend.

3.3. Mechanical Properties

Figure 7a shows the hardness (H) and effective Young’s modulus (E*) of the HfC_x/a-C:H coatings with different C₂H₂ flow rates. It can be observed that H and E* exhibit the same evolutionary trend. From 12 to 20 sccm C₂H₂, the H and E* increase first and then decrease; the maximum values of both are obtained at 18 sccm C₂H₂. The H from 16.3 to 31.3 GPa and then decreases to 28.2 GPa, while E* increases from 242.9 to 392.3 GPa and then decreases to 362.7 GPa.

From 12 to 16 sccm, the grain size of the coating exhibits slight coarsening, accompanied by a nearly linear increase in a-C:H content. The increasing amount of a-C:H leads to a reduction in the direct contact between adjacent HfC_x grains. Consequently, the plastic deformation must traverse a greater number of interfaces, including both grain boundaries and phase boundaries, promoting a transition from dislocation-dominated to interface-controlled deformation. This transition effectively enhances the resistance to plastic flow. At the 18 sccm C₂H₂ flow rate, the coating develops a typical nanocomposite structure in which refined HfC_x (~8 nm) nanocrystallites are dispersed within the a-C:H matrix. Under this structural configuration, the number of dislocation sources and dislocation movement within individual HfC_x nanocrystals are significantly restricted. Furthermore, the plastic deformation needs to be coordinated among multiple isolated HfC_x nanocrystals, which maximizes the interface-mediated strengthening effect. These two factors jointly contribute to the HfC_x/a-C:H coating achieving its highest hardness. Meanwhile, from 12 to 18 sccm, the hardness enhancement is also contributed to by the increased ratio of sp³/sp²-C in a-C:H matrix [29], the increased stoichiometric ratio of HfC_x nanograins [30], and the rise in residual compressive stress (Table 1). The substoichiometric hardening mechanism reported in [31] indicates that non-metal deficiencies are the primary origin of sub-stoichiometry, and the induced vacancies serve as pinning centers, thereby improving coating hardness. When the C₂H₂ flow rate further increases to 20 sccm, the grain size coarsening of HfC_x promotes the recovery of dislocation motion space. Moreover, the continued growth of the a-C:H phase alters the structural balance between nanocrystalline and amorphous regions. The enhanced continuity of the a-C:H phase may facilitate localized shear deformation [32], leading to a reduction in hardness. The decrease in compressive stress at 20 sccm also contributes to the hardness decline. In addition, although the sp³/sp²-C ratio continues to increase, together with enhanced densification of the columnar morphology at 20 sccm C₂H₂ flow rate, these changes do not lead to a corresponding increase in hardness.

Based on the evolution trends of H and E*, the semi-empirical wear-resistance indicators H/E* and H³/E*² of the HfC_x/a-C:H coatings as a function of C₂H₂ flow rate are presented in Figure 7b. With increasing C₂H₂ flow rate, both H/E* and H³/E*² exhibit a non-monotonic evolution characterized by an initial increase followed by a subsequent decrease, reaching their maximum values at 18 sccm. Such behavior reflects the changes in the elastoplastic response induced by the evolution of the nanocomposite structure of the HfC_x/a-C:H coatings [33].

3.4. Tribological Behavior

Figure 8a shows the dynamic friction curves of the HfC_x/a-C:H coatings deposited at different C₂H₂ flow rates during the sliding tests. The specific values of steady-state friction coefficient and wear rate of the HfC_x/a-C:H coatings are summarized in Figure 8b. To highlight the low friction and wear resistance of the HfC_x/a-C:H coating, the results of friction tests on the uncoated M2 substrate are also presented in Figure 8 for comparison. It can be observed that the application of the HfC_x/a-C:H coating significantly reduces the coefficient of friction and wear rate.

For HfC_x/a-C:H coatings, as the C₂H₂ flow rate increases from 12 to 20 sccm, the dynamic friction curves exhibit three distinct evolution features. It includes (i) a significant shortening of the initial running-in stage, (ii) a monotonic decrease in the steady-state COF value (from 0.70, 0.68, 0.37, 0.30, to 0.28), and (iii) a gradual reduction in the amplitude of friction fluctuations. Such evolution is associated with the increase in sp²-C content (from 14.9 to 25.6 at.%) and the enhanced mechanical properties. These promote the formation of an easily sheared tribolayer (acting as the solid lubricant) and enhance resistance to plastic deformation [34], respectively.

After tribological testing, the coating deposited at 12 sccm C₂H₂ was completely worn. For other coatings, it can be observed that as the C₂H₂ flow rate increases, the wear rate gradually decreases in the range of $2.06\times {10}^{-5}\to 6.82\times {10}^{-6} {\mathrm{m}\mathrm{m}}^3/\mathrm{N}\mathrm{m}$, consistent with the development of the friction coefficient. However, the hardness, H/E*, and H³/E*² ratios all exhibit a decreasing trend when the C₂H₂ flow rate exceeds 18 sccm, indicating that the improvement in wear resistance cannot be solely attributed to the enhanced resistance to indentation and crack propagation of the coatings. Instead, it is governed by the synergistic interplay between mechanical properties and tribochemical processes.

By combining SEM wear track morphologies, the wear mechanism of HfC_x/a-C:H coating is further analyzed. As shown in Figure 9, with the increase in C₂H₂ flow rate from 12 to 20 sccm, the wear tracks transition from complete failure to a smooth and intact state. The surface of the wear track becomes narrower and shallower, accompanied by fewer furrows. At 12 sccm C₂H₂ flow rate, the wear track is severely damaged and characterized by extensive flake-like delamination, through-thickness cracks, and deep ploughing grooves. These features indicate that material removal is dominated by severe abrasive wear [35], primarily attributed to the insufficient a-C:H content and poor mechanical properties. The limited hardness and toughness facilitate the generation of large wear debris during the running-in stage, which subsequently act as third-body abrasives and accelerate material removal. Meanwhile, under cyclic contact loading, the elevated interfacial shear stress, coupled with the low crack resistance, promotes the nucleation and propagation of subsurface cracks. The progressive coalescence of these cracks eventually leads to large-scale delamination and premature coating failure. When the C₂H₂ flow rate increases to 14 sccm, the extent of delamination is significantly reduced and is mainly confined to the central region of the wear track. Meanwhile, the ploughing grooves become noticeably shallower. The magnified image of the region marked by the yellow dashed line (Figure 9b) reveals multiple microcracks perpendicular to the sliding direction along the fracture edges. These observations suggest that the dominant failure mechanism shifts to localized brittle fracture coupled with abrasive wear, while large-scale delamination is effectively suppressed. This improvement in wear resistance is attributed to the increase in a-C:H fraction and sp²-C content, as well as the increased hardness.

With the C₂H₂ flow rate increasing to 16 and 18 sccm, the wear tracks evolve into shallow and continuous ploughing grooves. No cracks or delamination are observed. The dominant wear mechanism transitions to mild abrasive wear [36]. This improvement originates from the combined effects of the excellent mechanical properties (H, E*, H/E* and H³/E*²) and a sufficient amount of sp²-C. On the one hand, the superior mechanical properties enhance the coating’s ability to resist plastic deformation. On the other hand, the higher sp²-C content facilitates the formation of a stable lubricious tribolayer at the sliding interface. This tribolayer effectively reduces the shear stress and minimizes the accumulation of plastic shear during repeated sliding, thereby further inhibiting the wear of the coating. When the C₂H₂ flow rate further increases to 20 sccm, the wear track surface becomes smoother and is accompanied by very shallow grooves. This observation suggests that material removal is mainly controlled by interfacial shear sliding [37], rather than severe plastic deformation. Therefore, the wear behavior at this stage is more consistent with shear-dominated mild abrasive wear. Moreover, a higher C₂H₂ flow rate promotes the formation of a more continuous a-C:H matrix, accompanied by a higher fraction of sp²-C bonding states. Although the hardness, H/E*, and H³/E*² ratios slightly decrease at 20 sccm C₂H₂ flow rate, the sp²-C-rich phase continuously reduces the interfacial shear resistance, while facilitating the formation of a more stable low-shear tribo-layer. This tribochemical effect compensates for the slight reduction in mechanical properties, leading to a further decrease in wear rate. In addition, the thickness of the coating has a beneficial effect on the wear failure behavior generally, especially in terms of the coating’s ability to resist wear-through failure. A thicker coating can provide a longer wear path and delay substrate exposure during the wear process. Under the same friction test conditions, the coating prepared at 12 sccm C₂H₂ flow rate, although exhibiting the largest thickness (4.6 μm), showed complete wear-through in the wear track. In contrast, the coating prepared at 14 sccm, despite having a lower thickness, only showed localized damage in the center of the wear track, demonstrating significantly better wear resistance. Furthermore, as the C₂H₂ flow rate was further increased to 20 sccm, the coating thickness was consistently lower than that of the coating deposited at 12 sccm, but the wear resistance continued to improve. These results indicate that for the HfC_x/a-C:H coating, the thickness is not the determining factor for tribological durability and wear failure behaviour.

Overall, with increasing C₂H₂ flow rate, the tribological behavior of the HfC_x/a-C:H coatings gradually evolves from severe abrasive wear dominated by high shear stress and brittle spallation to mild abrasive wear regulated by a sp²-induced tribolayer. This transition is mainly driven by the combined effects of the increased sp²-C content and the improved mechanical properties, which systematically optimize the interfacial shear stress and load-bearing capacity, thereby enabling the simultaneous reduction in the friction coefficient and wear rate. Looking back at the discussion of the ratios H/E* and H³/E*², it is inferred that these ratios cannot fully predict the wear performance of HfC_x/a-C:H coatings since they do not account for the regulation of interfacial shear behavior by tribochemical reactions.

3.5. Oxidation Resistance

Figure 10 presents the XRD patterns of the HfC_x/a-C:H coatings after annealing at 500 °C and 600 °C, respectively. When the annealing temperature is below 400 °C, no oxidation-related peaks are detected in the coatings deposited at each C₂H₂ flow rate. After 500 °C annealing, a weak anomalous peak located at 2θ ≈ 34.67° appears in the coatings deposited at low C₂H₂ flow rates (12 and 14 sccm), while no Hf-related oxide peaks are observed. When the annealing temperature increases to 600 °C, all coatings exhibit a mixture of HfO₂ and HfC phases, indicating that the oxidation of the HfC_x/a-C:H coating begins to be activated.

After annealing at 500 °C, the anomalous peak observed at 2θ ≈ 34.67° is assigned to the (220) plane of the Fe₃O₄ phase. This peak is detected in the coatings deposited at 12 and 14 sccm and coexists with the HfC peaks. Moreover, the lower the C₂H₂ flow rate, the higher the intensity of the Fe₃O₄ phase. It is worth noting that no signs of peeling were observed in the coating where the Fe₃O₄ peak appeared, indicating that the Fe₃O₄ does not originate from the coating peeling off and the subsequent oxidation of the exposed M2 layer. Based on this, it is inferred that the coexistence of Fe₃O₄ and HfC phases is caused by the outward diffusion of Fe atoms from the substrate into the coating. According to the morphology analysis in Figure 5, the coatings deposited at low C₂H₂ flow rates (12, 14, and 16 sccm) exhibit a relatively loose columnar structure with pronounced intercolumnar gaps. Furthermore, a higher density of interconnected HfC grain boundaries, as confirmed by TEM, is present in these coatings due to their larger grain size and lower a-C:H content. Both of these may provide a rapid diffusion pathway for the outward diffusion of Fe atoms [38]. Therefore, during the annealing process, Fe atoms diffuse from the M2 substrate into the HfC_x/a-C:H coatings through the loose intercolumnar gaps and interconnected grain boundaries. Then, the oxygen preferentially reacts with Fe, resulting in the formation of Fe₃O₄. Additionally, the Fe₃O₄ peak is notably broad, corresponding to fine grain size. This is because the Fe atoms are dispersed in the grain boundaries and intercolumnar gap regions rather than concentrated in localized areas. As a result, the growing Fe₃O₄ grains are difficult to aggregate and connect with adjacent grains, and thus cannot coarsen. For coatings deposited at 18 and 20 sccm C₂H₂ flow rates, no diffraction peaks are detected except for the HfC phase after annealing at 500 °C. This suggests that the outward diffusion of Fe atoms is significantly suppressed and HfC_x/a-C:H coating itself remains unoxidized, which is attributed to the denser microstructure and reduced grain boundary connectivity.

After annealing at 600 °C, the oxidation resistance of the HfC_x/a-C:H coatings is observed to increase with increasing C₂H₂ flow rate. The coatings deposited at 12 and 14 sccm are completely delaminated from the M2 substrate. Their XRD patterns exhibit a strong Fe peak corresponding to the M2 substrate (located at 2θ ≈ 52°, marked by grey squares) and five weaker HfO₂ peaks. The HfO₂ peaks originate from the oxidation of the pure Hf buffer layer after the detachment of HfC_x/a-C:H coating. The HfC_x/a-C:H coating deposited at 16 sccm C₂H₂ flow rate remains intact, with Fe₃O₄, HfO₂, and HfC phases coexisting in its XRD pattern, indicating partial oxidation of the coating. It is inferred that the Fe₃O₄ phase, HfO₂ phase, and HfC phase represent the oxidized Fe atoms (diffused from the substrate), the partially oxidized HfC_x/a-C:H coating, and the stable HfC_x/a-C:H coating underneath, respectively. To verify this phase composition suggested by XRD, Figure 11 shows the EDS mapping spectrogram of the cross-sections of the HfC_x/a-C:H coatings deposited at 16 sccm after 600 °C annealing. For comparison, the cross-sectional EDS of the coating deposited at 18 sccm under the same annealed condition is also included in Figure 11. By comparing the Fe elemental maps (highlighted by yellow dashed boxes), a pronounced outward diffusion of Fe atoms is observed in the coating deposited at 16 sccm. This confirms that the Fe₃O₄ phase is formed by the outward diffusion and subsequent oxidation of the Fe atoms. Furthermore, it is noted that the intensity and shape of the Fe₃O₄ peak appearing here (16 sccm,600 °C annealing) are very similar to those observed in the coatings deposited at 12 and 14 sccm C₂H₂ after 500 °C annealing. This similarity further supports the previous inference that the Fe₃O₄ phase is formed through the outward diffusion and preferential oxidation of Fe atoms. For the coatings deposited at 18 and 20 sccm, the coexistence of HfC and HfO₂ phases is clearly observed, with the relative intensity of the HfC phase being higher than that of HfO₂ phase. This indicates that the oxidation resistance of the coating is further enhanced at higher C₂H₂ flow rates. In addition, no Fe₃O₄-related peaks are detected, suggesting that the diffusion of Fe atoms is effectively suppressed. This behavior is attributed to the denser morphology and fewer grain boundary interconnections of HfC grains. Moreover, the coating deposited at 20 sccm exhibits a higher relative intensity of the HfC phase than that deposited at 18 sccm, indicating superior oxidation resistance. To further quantify the oxidation degree of the HfC_x/a-C:H coatings deposited at each C₂H₂ flow rate, the cross-sectional morphologies after 600 °C oxidation are systematically analyzed and compared using SEM and EDS.

Figure 12 presents the cross-sectional SEM and EDS scans of the HfC_x/a-C:H coatings deposited at C₂H₂ flow rates of 14, 16, 18, and 20 sccm. It is observed that the oxidation resistance of HfC_x/a-C:H coating gradually improves with increasing C₂H₂ flow rate, achieving the best oxidation resistance at 20 sccm. This is consistent with the evolution trend indicated by XRD. For the coatings deposited at 12 and 14 sccm C₂H₂, they are completely oxidized and peeled off from the substrate after annealing at 600 °C. Only the top-view morphology and EDS analysis of the coating deposited at 14 sccm are presented, as representative samples. Based on the concentrations of Hf, O, C, and Fe in the EDS spectra, the observed surface is identified as the oxidation product of the pure Hf buffer layer remaining after the complete oxidation and spallation of the HfC_x/a-C:H coating. This observation is consistent with the characteristics given by XRD, which show weak HfO₂ peaks together with a very strong M2 substrate peak. For the coatings deposited at 16, 18, and 20 sccm C₂H₂ flow rates, the thickness of “layer 1” in the EDS images confirms that approximately 76%, 54%, and 29% of the coatings are oxidized, respectively.

For the HfC_x/a-C:H coatings in the present work, the oxidation resistance is primarily governed by the diffusion pathways and oxidation kinetics caused by the microstructure. The diffusion pathways are mainly determined by two key factors: the compactness of the growth morphology and the nature and density of the interfaces provided by the nanocomposite structure [39]. For low-carbon-content HfC_x/a-C:H coatings, the larger intercolumnar gaps and a higher fraction of interconnected and relatively straight grain boundaries (which originate from the nanocomposite structure formed by large HfC grains and a small amount of a-C:H) are observed. Such a structure facilitates rapid diffusion of Fe atoms from the M2 substrate and oxygen from the environment. The subsequent rapid oxidation reactions between these species then lead to severe degradation of coating stability and oxidation resistance. Meanwhile, the reaction between oxygen and HfC_x/a-C:H coating is accompanied by the release of CO/CO₂ gas. This process results in a loose and porous oxide scale that lacks effective barrier properties. Therefore, the loose morphology and straighter grain boundaries are the fundamental reasons for the weaker oxidation resistance of coatings deposited at low C₂H₂ flow rates.

In contrast, high carbon content HfC_x/a-C:H coatings exhibit a dense nanocomposite structure consisting of refined HfC nanograins embedded in a continuous a-C:H matrix. Although the a-C:H phase is thermodynamically unstable at elevated temperatures [40], it plays a critical role in regulating oxidation kinetics. On one hand, during the initial stage of oxidation, the continuous a-C:H network introduces a more tortuous diffusion pathway and a higher density of interfaces, which effectively block the diffusion of both O and Fe atoms. On the other hand, the a-C:H matrix acts as a sacrificial phase that preferentially reacts with oxygen, thereby consuming local oxygen and reducing the effective oxygen flux penetrating the coating. It is important to note that under this diffusion-limited condition, the gradual consumption of the nanoscale-thick a-C:H phase located between grains does not lead to the formation of interconnected microscopic porosity. Instead, it effectively delays the oxidation progression during the critical early stage, thereby reducing the oxidation rate of HfC grains. This kinetic deceleration effect improves the overall oxidation resistance of the coating. To sum up, the superior oxidation resistance of high carbon content HfC_x/a-C:H coatings stems from the continuous a-C:H matrix and the increased density and tortuosity of the interface it provides. These features act as a sacrificial phase and diffusion barrier, thereby reducing the oxidation rate.

4. Conclusions

HfC_x/a-C:H coatings with varying carbon contents were deposited by reactive pulsed DC magnetron sputtering within a C₂H₂ flow rate window of 12–20 sccm. Increasing the C₂H₂ flow rate continuously regulated the coating microstructure, bonding states, and morphology. The microstructure underwent a distinct two-stage evolution. It transformed from a composite structure dominated by HfC_x nanograins with a small amount of a-C:H localized at the triple junctions of HfC_x grain boundaries to a typical nanocomposite structure consisting of HfC_x nanograins embedded in a continuous a-C:H matrix. Meanwhile, the columnar morphology became denser and finer, accompanied by the gradual elimination of intercolumnar gaps. The increase in C₂H₂ flow rate also promoted the formation of a-C:H phase.

The evolution of coating performance was strongly associated with these microstructural and bonding-state changes. The highest nanohardness and effective Young’s modulus (31.3 GPa and 392.3 GPa) were achieved at 18 sccm. It was attributed to enhanced interface-mediated strengthening induced by the optimized nanocomposite structure. The coefficient of friction and wear rate decreased monotonically with increasing C₂H₂ flow rate, reaching minimum values of 0.28 and 6.82 × 10⁻⁶ mm³/N·m at 20 sccm. This improvement originated from the synergistic effect of a-C:H-induced tribochemical reactions and the enhanced mechanical properties. The oxidation resistance was improved with increasing C₂H₂ flow rate. This enhancement arose from the synergistic effect of structural densification, increased interfacial density and tortuosity, and the regulating role of the a-C:H matrix in oxidation kinetics. These effects effectively suppressed oxygen diffusion and delayed coating degradation. Overall, the improved coating performance originated from the synergistic regulation of nanocomposite structure and bonding states induced by the C₂H₂ flow rate. The optimized HfC_x/a-C:H coatings achieved an effective balance among mechanical properties, tribological performance, and oxidation resistance. This work provides an effective strategy for designing multifunctional protective coatings for harsh operating environments.