E ﬀ ects of the Nitrogen Flow Ratio and Substrate Bias on the Mechanical Properties of W–N and W–Si–N Films

: The reactive gas ﬂow ratio and substrate bias voltage are crucial sputtering parameters for fabricating transition metal nitride ﬁlms. In this study, W–N ﬁlms were prepared using sputtering with nitrogen ﬂow ratios ( f ) of 0.1–0.5. W–N and W–Si–N ﬁlms were then prepared using an f level of 0.4 and substrate bias varying from 0 to − 150 V by using sputtering and co-sputtering, respectively. The variations in phase structures, bonding characteristics, mechanical properties, and wear resistance of the W–N and W–Si–N ﬁlms were investigated. The W–N ﬁlms prepared with nitrogen ﬂow ratios of 0.1–0.2, 0.3, and 0.4–0.5 displayed crystalline W, amorphous W–N, and crystalline W 2 N, respectively. The W–N ﬁlms prepared using a nitrogen ﬂow ratio of 0.4 and substrate bias voltages of − 50 and − 100 V exhibited favorable mechanical properties and high wear resistance. The mechanical properties of the amorphous W–Si–N ﬁlms were not related to the magnitude of the substrate bias.


Introduction
W-N films were developed as hard coatings and display remarkable mechanical properties [1][2][3][4][5][6][7]. The phase of W-N films varies from α-W to β-W, β-W 2 N, and δ-WN as the N concentration increases [8,9]. The β-W 2 N phase is a B1 structure, in which N atoms occupy half of the octahedral interstitial sites of cubic close-packed W atoms [9][10][11]. The cubic W 2 N phase formed in W-N films has been extensively discussed in the literature, whereas the hexagonal WN phase, fabricated using sputtering processes, has been only recently observed by applying high reactive gas flow ratios, accompanied with a high N concentration in the films [2,9]. Polcar et al. [3] reported that δ-WN forms in sputtered W-N films containing >55 at.%. Baker and Shah [10] reported that a saturated N concentration of 35 at.% in W-N films for a W 2 N phase could be obtained by sputtering. Similar phase variations from W 2 N to WN were reported for W-N films prepared using the cathodic arc method, with N 2 flow ratios exceeding 0.35 [12]. However, Lou et al. [7] did not report hexagonal or cubic WN phases, even in W-N films fabricated by the superimposed high-power impulse magnetron sputtering and mid-frequency pulses system by using a high N 2 flow ratio of up to 74%, which formed an N/W ratio of 0.60. The hardness (H) levels of α-W, β-W 2 N, and δ-WN were 13, 24, and 28 GPa, where α is the lattice constant for the individual reflection, K is constant, and θ is the diffraction angle. The X-ray diffraction (XRD) patterns for determining the texture coefficients and grain sizes of films followed a Bragg-Brentano (θ-2θ) scan. The texture coefficient, Tc, was defined as [32]: where I m (hkl) is the measured relative reflection intensity of the (hkl) plane, I 0 (hkl) is the relative intensity from the standard reference, and n, the number of reflections, is two in this study. The grain sizes were calculated using the Scherrer formula [31]. The H and E of the films were measured using a nanoindentation tester (TI-900 Triboindenter, Hysitron, Minneapolis, MN, USA) equipped with a Berkovich diamond probe tip. The indentation depth was 50 nm following the 1/10 rule used for determining the mechanical properties of thin films [33]. The H and reduced elastic modulus values were calculated based on the Oliver and Pharr method [34]. The average surface roughness (R a ) values of the films were evaluated by using an atomic force microscope (Dimension 3100 SPM, Nanoscope IIIa, Veeco, Santa Barbara, CA, USA) with a scanned area of 5 × 5 µm 2 . The residual stress of the films on Si substrates was determined using Stoney's equation [35], which was previously illustrated [36]. The thickness of the Si substrate is 525 µm. The bonding characteristics of the elements were examined by using an X-ray photoelectron spectroscope (XPS, PHI 1600, PHI, Kanagawa, Japan) with an Mg Kα X-ray. The wear resistance of films was examined through the pin-on-disk test. A cemented tungsten carbide (WC-6 wt.% Co) ball, 6 mm in diameter, was used as the stationary pin. The normal load was 2 N, the sliding speed was 104.5 mm/s, the wear track diameter was 16 mm, and the wear length was 200 m.

W-N Films Prepared with Various Nitrogen Flow Ratios
The atomic compositions of the Batch W films fabricated with f levels of 0.1-0.5 are displayed in Table 1. These samples were labeled W01, W02, W03, W04, and W05 for the f level sets of 0.1, 0.2, 0.3, 0.4, and 0.5, respectively. These samples had Si concentrations of 0.5-1.9 at.%, which originated from Si substrates because these films had smaller thicknesses (511-675 nm). The N concentration increased from 4.7 to 8. 3, 12.4, 22.5, and 25.3 at.% as the f level increased from 0.1 to 0.2, 0.3, 0.4, and 0.5, respectively, whereas the W concentration decreased from 93.6 to 85.9, 80.5, 75.4, and 72.8 at.%, respectively. Studies [1,7,37] have reported that an increase in the f level causes an elevation in the N/W ratio of W-N films. The N/W ratios were 0.05, 0.10, 0.15, 0.30, and 0.35 for the W01, W02, W03, W04, and W05 films, respectively. This variation was accompanied by the film phase changing from α-W to amorphous and β-W 2 N, as illustrated in Figure 1a. Similar phase variations with increasing f were reported without applying substrate bias [17]. Moreover, the increase in compressive residual stress from 0 to 2.4 GPa was related to an increase in the N/W ratio. The GIXRD pattern of W01 films displayed a body-centered cubic phase with (110), (200), and (211) reflections, whereas the GIXRD pattern of the W02 films only exhibited a W(110) reflection, with an intensity decrease relative to that of the W01 films. The GIXRD patterns of the W04 and W05 films exhibited a face-centered cubic (FCC) phase with (111), (200), (220), and (311) reflections. Figure 1b presents the Bragg-Brentano scan XRD patterns of the Batch W samples. The grain sizes of the W01 and W02 films were determined to be 10.2 and 4.4 nm, respectively, by using the full width at half maximum (FWHM) of (110) reflections. The reduction in the deposition rate from 8.4 to 7.7 and 7.1 nm/min with an increase in the f level from 0.1 to 0.2 and 0.3 could be attributed to the reduction in sputtering yield. The standard intensity ratio of (111):(200) for the W 2 N phase was 100:47, whereas the W04 and W05 films exhibited the (200) texture, as determined using Equation (2) and reflection intensities from Figure 1b. The texture coefficients (T c s) were 0.46 and 1.54 for T c (111) and T c (200) of the W04 films, respectively, whereas the texture values were 0.76 and 1.24 for T c (111) and T c (200) of W05 films, respectively. The deposition rates were 7.3 and 6.4 nm/min for the W04 and W05 films, respectively. These values were affected by the target poisoning effect.     [38] and H/E* (E*: effective Young's modulus) [39] represent the elastic strain to failure. Films with H/E* > 0.1 and We ≥ 60% displayed high toughness [39]. H 3 /E 2 [40] and H 3 /E* 2 [41] reflect the resistance to plastic deformation, and are used to evaluate wear resistance [42]. The H/E and H 3 /E 2 ratios and elastic recovery (We) of the Batch W samples increased with increasing f levels. The wear tests of the W01 and W04 films were examined in Section 3.4. Studies [43][44][45] have reported that the films with high surface roughness values exhibit larger deviations and lower averages of mechanical properties evaluated using a nanoindentation technique. Except for the W02 sample with a high Ra of 13.23 nm, all the films in this study exhibited a low Ra level (<5 nm). The effects of Ra on the mechanical properties were insignificant. A nitrogen flow ratio f of 0.4 was sufficient to fabricate films with a W2N phase which exhibited satisfactory mechanical properties. Therefore, an f of 0.4 was utilized in the next section to evaluate the substrate bias effects on the characteristics of W-N films.   Table 3 lists the mechanical properties of the W-N films. The W01 films with zero residual stress exhibited an H of 22.4 GPa and an E of 335 GPa. The H and E of the W02 films increased to 26.2 and 388 GPa, respectively, which can be attributed to the lower grain size and residual stress of −0.7 GPa. The W03 films exhibited a lower H and E of 22.1 and 308 GPa, respectively, caused by the formation of an amorphous phase. The W04 and W05 films with a W 2 N phase exhibited H levels of 25.9 and 29.6 GPa and residual stresses of −1.6 and −2.4 GPa, respectively, as well as E values of approximately 350 GPa. H/E [38] and H/E* (E*: effective Young's modulus) [39] represent the elastic strain to failure. Films with H/ E* > 0.1 and We ≥ 60% displayed high toughness [39]. H 3 /E 2 [40] and H 3 /E* 2 [41] reflect the resistance to plastic deformation, and are used to evaluate wear resistance [42]. The H/E and H 3 /E 2 ratios and elastic recovery (We) of the Batch W samples increased with increasing f levels. The wear tests of the W01 and W04 films were examined in Section 3.4. Studies [43][44][45] have reported that the films with high surface roughness values exhibit larger deviations and lower averages of mechanical properties evaluated using a nanoindentation technique. Except for the W02 sample with a high R a of 13.23 nm, all the films in this study exhibited a low R a level (<5 nm). The effects of R a on the mechanical properties were insignificant. A nitrogen flow ratio f of 0.4 was sufficient to fabricate films with a W 2 N phase which exhibited satisfactory mechanical properties. Therefore, an f of 0.4 was utilized in the next section to evaluate the substrate bias effects on the characteristics of W-N films.    Table 1 displays the atomic compositions and thicknesses of the W-N films (Batch A) prepared using a power of 150 W, an f level of 0.4, and substrate bias voltages varying from 0 to −150 V. The samples A0 and W04 were the same. The Batch A samples were prepared using only a W target and displayed a low Si concentration of 0.7-3.9 at.%, which was attributed to low thickness values (521-610 nm). Furthermore, the Si concentrations originated from the Si substrates. Except for the A25 samples, the thicknesses and deposition rates of the films prepared through bias sputtering were lower compared with those of the films prepared at grounded state (A0), indicating the occurrence of re-sputtering. Figure 3a depicts the GIXRD patterns of the Batch A films, which exhibit an FCC    Table 1 displays the atomic compositions and thicknesses of the W-N films (Batch A) prepared using a power of 150 W, an f level of 0.4, and substrate bias voltages varying from 0 to −150 V. The samples A0 and W04 were the same. The Batch A samples were prepared using only a W target and displayed a low Si concentration of 0.7-3.9 at.%, which was attributed to low thickness values (521-610 nm). Furthermore, the Si concentrations originated from the Si substrates. Except for the A25 samples, the thicknesses and deposition rates of the films prepared through bias sputtering were lower compared with those of the films prepared at grounded state (A0), indicating the occurrence of re-sputtering. Figure 3a depicts the GIXRD patterns of the Batch A films, which exhibit an FCC W 2 N phase. Most W-N films fabricated using sputtering processes were reported to crystallize into a β-W 2 N phase. Polcar et al. [3,8] have reported mixed δ-WN and β-W 2 N phases in W 42 N 58 coatings, with an N/W ratio of 1.38. In the present study, the N/W ratios of Batch A films were 0.30-0.44 (Table 1), with a β-W 2 N phase. Moreover, an expanded lattice formed because all W 2 N reflections shifted toward lower 2θ angle levels. Figure 4 displays the lattice parameters of these crystalline W-N coatings, determined using Equation (1). The lattice constants of the A0, A25, A50, A100, and A150 samples were 0.4199, 0.4198, 0.4202, 0.4201, and 0.4208 nm, respectively. The standard lattice constant of the β-W 2 N phase was 0.4126 nm [ICDD 00-025-1257] (ICDD: International Centre for Diffraction Data). The lattice constants of the Batch A films exhibited expanded lattices accompanied with compressive residual stresses. Figure 3b illustrates the Bragg-Brentano scan XRD patterns of the Batch A films. Figure 5 displays the T c s of the Batch A samples, determined using Equation (2), and the T c s of the W05 samples are also illustrated. All the crystalline W-N films displayed a (200) orientation accompanied by high compressive stresses ranging from −1.6 to −3.2 GPa.

W-N Films Prepared with Various Bias Voltages
Coatings 2020, 10, x FOR PEER REVIEW 6 of 13 W2N phase. Most W-N films fabricated using sputtering processes were reported to crystallize into a β-W2N phase. Polcar et al. [3,8] have reported mixed δ-WN and β-W2N phases in W42N58 coatings, with an N/W ratio of 1.38. In the present study, the N/W ratios of Batch A films were 0.30-0.44 (Table  1), with a β-W2N phase. Moreover, an expanded lattice formed because all W2N reflections shifted toward lower 2θ angle levels. Figure 4 displays the lattice parameters of these crystalline W-N coatings, determined using Equation (1). The lattice constants of the A0, A25, A50, A100, and A150 samples were 0.4199, 0.4198, 0.4202, 0.4201, and 0.4208 nm, respectively. The standard lattice constant of the β-W2N phase was 0.4126 nm [ICDD 00-025-1257] (ICDD: International Centre for Diffraction Data). The lattice constants of the Batch A films exhibited expanded lattices accompanied with compressive residual stresses. Figure 3b illustrates the Bragg-Brentano scan XRD patterns of the Batch A films. Figure 5 displays the Tcs of the Batch A samples, determined using Equation (2), and the Tcs of the W05 samples are also illustrated. All the crystalline W-N films displayed a (200) orientation accompanied by high compressive stresses ranging from −1.6 to −3.2 GPa.    Coatings 2020, 10, x FOR PEER REVIEW 6 of 13 W2N phase. Most W-N films fabricated using sputtering processes were reported to crystallize into a β-W2N phase. Polcar et al. [3,8] have reported mixed δ-WN and β-W2N phases in W42N58 coatings, with an N/W ratio of 1.38. In the present study, the N/W ratios of Batch A films were 0.30-0.44 (Table  1), with a β-W2N phase. Moreover, an expanded lattice formed because all W2N reflections shifted toward lower 2θ angle levels. Figure 4 displays the lattice parameters of these crystalline W-N coatings, determined using Equation (1). The lattice constants of the A0, A25, A50, A100, and A150 samples were 0.4199, 0.4198, 0.4202, 0.4201, and 0.4208 nm, respectively. The standard lattice constant of the β-W2N phase was 0.4126 nm [ICDD 00-025-1257] (ICDD: International Centre for Diffraction Data). The lattice constants of the Batch A films exhibited expanded lattices accompanied with compressive residual stresses. Figure 3b illustrates the Bragg-Brentano scan XRD patterns of the Batch A films. Figure 5 displays the Tcs of the Batch A samples, determined using Equation (2), and the Tcs of the W05 samples are also illustrated. All the crystalline W-N films displayed a (200) orientation accompanied by high compressive stresses ranging from −1.6 to −3.2 GPa.    Coatings 2020, 10, x FOR PEER REVIEW 6 of 13 W2N phase. Most W-N films fabricated using sputtering processes were reported to crystallize into a β-W2N phase. Polcar et al. [3,8] have reported mixed δ-WN and β-W2N phases in W42N58 coatings, with an N/W ratio of 1.38. In the present study, the N/W ratios of Batch A films were 0.30-0.44 (Table  1), with a β-W2N phase. Moreover, an expanded lattice formed because all W2N reflections shifted toward lower 2θ angle levels. Figure 4 displays the lattice parameters of these crystalline W-N coatings, determined using Equation (1). The lattice constants of the A0, A25, A50, A100, and A150 samples were 0.4199, 0.4198, 0.4202, 0.4201, and 0.4208 nm, respectively. The standard lattice constant of the β-W2N phase was 0.4126 nm [ICDD 00-025-1257] (ICDD: International Centre for Diffraction Data). The lattice constants of the Batch A films exhibited expanded lattices accompanied with compressive residual stresses. Figure 3b illustrates the Bragg-Brentano scan XRD patterns of the Batch A films. Figure 5 displays the Tcs of the Batch A samples, determined using Equation (2), and the Tcs of the W05 samples are also illustrated. All the crystalline W-N films displayed a (200) orientation accompanied by high compressive stresses ranging from −1.6 to −3.2 GPa.    The XPS analysis results of the Batch A films are displayed in Table 2. The intensity ratios of W-W:W-N varied from 72:28 to 80:20, 89:11, 88:12, and 76:24 as the bias voltage increased from 0 to −25, −50, −100, and −150 V, respectively. The N/W ratio exhibited an increasing trend. The W-W:W-N ratio exhibited an increasing trend for the W-N bonds related to the increased substrate bias, which indicated that N atoms were incorporated into the interstitial sites of the W 2 N structure, causing an increase in the lattice constant. Table 3 presents the mechanical properties of the Batch A films. The H value increased from 25.9 GPa for the A0 samples to 31.5 GPa for the A25 samples and then decreased to 30.9, 29.9, and 27.1 GPa for the A50, A100, and A150 samples, respectively, despite the compressive residual stress successively increasing. This finding can be attributed to the inverse Hall-Petch effect [46,47] because the grain sizes of the Batch A films were approximately 10 nm. The grain sizes of A0, A25, A50, A100, and A150 films were 10.9, 11.5, 10.8, 10.3, and 9.6 nm, respectively, determined using the FWHM of (200) reflections. The E level decreased from 347 GPa in the A0 samples to 327 and 303 GPa in the A25 and A50 samples and then increased to 341 and 340 GPa in the A100 and A150 samples. The variations between E values were narrow. The H/E and H 3 /E 2 ratios exhibited the maximum values of 0.102 and 0.321 GPa, respectively, for the A50 samples. The We value exhibited a high level of 66%-67% in the A25, A50, and A100 samples. The R a values of the Batch A samples were 3.10-4.93 nm, which is comparable with the range of 3.09-4.54 nm for the Batch W samples, except for the W02 films. Table 4 lists the atomic compositions and thicknesses of the W-Si-N films. The Batch B and C samples, with high Si concentrations of 13.5-16.5 and 27.0-30.9 at.%, respectively, exhibited near-amorphous XRD patterns ( Figure 6) and low residual stresses of −0.6 to −0.9 and −0.5 to −0.7 GPa, respectively (Table 4). Studies [19][20][21][22] have reported that Si 3 N 4 is more stable than W 2 N, WN, and W-silicide based on thermodynamic considerations. Therefore, excess N or Si atoms reacted with W after Si 3 N 4 formation [20]. The high Si and low W concentrations indicate that Batch C samples were amorphous Si 3 N 4 -dominated films.    Figure 7 summarizes the mechanical properties of the W-N and W-Si-N films relative to their residual stress. The mechanical properties of the W-Si-N and W-N films prepared with various powers, assessed in our previous work [13], are also presented for comparison. Figure 8 illustrates the relationship between H and E for the samples displayed in Figure 7. The amorphous films (Batches B and C) exhibited H/E levels of approximately 0.080 ± 0.004, whereas the crystalline W2N films exhibited high H/E values. The Ra values of these amorphous W-Si-N films were 2.10-4.00 nm, which was lower than those of the crystalline W-N films.     [48,49]. Crystalline films, W04, W05, and Batch A films with a W 2 N phase exhibited high H values of 25.9-31.5 GPa, high E values of 303-351 GPa, and high residual stresses, ranging from −1.6 to −3.2 GPa. Figure 7 summarizes the mechanical properties of the W-N and W-Si-N films relative to their residual stress. The mechanical properties of the W-Si-N and W-N films prepared with various powers, assessed in our previous work [13], are also presented for comparison. Figure 8 illustrates the relationship between H and E for the samples displayed in Figure 7. The amorphous films (Batches B and C) exhibited H/E levels of approximately 0.080 ± 0.004, whereas the crystalline W 2 N films exhibited high H/E values. The Ra values of these amorphous W-Si-N films were 2.10-4.00 nm, which was lower than those of the crystalline W-N films. Coatings 2020, 10, x FOR PEER REVIEW 8 of 13 Figure 6. Grazing incident X-ray diffraction (GIXRD) patterns of W-Si-N films. Table 5 displays the mechanical properties of W-Si-N films prepared using various substrate bias voltages. The X-ray amorphous films, Batches B and C, exhibited low H values of 17.1-22.3 GPa, low E values of 218-253 GPa, and low residual stresses, ranging from −0.5 to −0.9 GPa. Moreover, the H values of the Batch C samples were similar to the reported value of 19 GPa for SiNx films [48,49]. Crystalline films, W04, W05, and Batch A films with a W2N phase exhibited high H values of 25.9-31.5 GPa, high E values of 303-351 GPa, and high residual stresses, ranging from −1.6 to −3.2 GPa. Figure 7 summarizes the mechanical properties of the W-N and W-Si-N films relative to their residual stress. The mechanical properties of the W-Si-N and W-N films prepared with various powers, assessed in our previous work [13], are also presented for comparison. Figure 8 illustrates the relationship between H and E for the samples displayed in Figure 7. The amorphous films (Batches B and C) exhibited H/E levels of approximately 0.080 ± 0.004, whereas the crystalline W2N films exhibited high H/E values. The Ra values of these amorphous W-Si-N films were 2.10-4.00 nm, which was lower than those of the crystalline W-N films.

Wear Test
The wear resistance of W-N and W-Si-N films of various categorizations was examined. W01 was the crystalline W-N films with a W phase; W04, A50, and A100 were the crystalline W-N films with a W2N phase; B100 and C100 were the amorphous W-Si-N films. Figure 9 presents the coefficients of friction (COFs) of the samples against the cemented tungsten carbide ball after a sliding distance of 200 m. The relationship between COF and Ra was unclear, which can be attributed to the distinct categorizations of the tested films. Therefore, further research is warranted to investigate this relationship. The abrupt drops of COF in the W01 and C100 samples can be attributed to the exhaustion of films and exposure of the substrates. Table 6 summarizes the wear test results. The wear depths of the W01 and C100 samples were 1334 and 1404 nm, respectively, which were larger than their film thicknesses of 675 and 574 nm, respectively. The other samples exhibited wear depths that were smaller than their film thickness. Figure 10 illustrates the wear scars of the tested samples. Chipping debris was observed along the wear scar of the W01 samples (Figure 10a). The H/E and H 3 /E 2 values of the W01 samples were 0.067 and 0.100, respectively, which were the lowest levels of the tested samples and indicated that W01 was brittle. The wear scars on the C100 samples displayed extrusions of fractured films on part of the wear track ( Figure 10f). The wear track widths of the A50, A100, and W04(A0) samples were 89, 96, and 99 μm, respectively, which was narrow compared with the other tested samples and can be attributed to their high H values [50]. The wear rates of the A50 and A100 films were 1.0 × 10 −6 and 7.5 × 10 −7 mm 3 /Nm, with respective low COFs of 0.49 and 0.46 and high H 3 /E 2 values of 0.321 and 0.230. The amorphous B100 sample exhibited a smooth wear scar and a broad track width of 138 μm because of a low H (22.3 GPa) and a high wear rate (4.9 × 10 −6 mm 3 /Nm).

Wear Test
The wear resistance of W-N and W-Si-N films of various categorizations was examined. W01 was the crystalline W-N films with a W phase; W04, A50, and A100 were the crystalline W-N films with a W 2 N phase; B100 and C100 were the amorphous W-Si-N films. Figure 9 presents the coefficients of friction (COFs) of the samples against the cemented tungsten carbide ball after a sliding distance of 200 m. The relationship between COF and Ra was unclear, which can be attributed to the distinct categorizations of the tested films. Therefore, further research is warranted to investigate this relationship. The abrupt drops of COF in the W01 and C100 samples can be attributed to the exhaustion of films and exposure of the substrates. Table 6 summarizes the wear test results. The wear depths of the W01 and C100 samples were 1334 and 1404 nm, respectively, which were larger than their film thicknesses of 675 and 574 nm, respectively. The other samples exhibited wear depths that were smaller than their film thickness. Figure 10 illustrates the wear scars of the tested samples. Chipping debris was observed along the wear scar of the W01 samples (Figure 10a). The H/E and H 3 /E 2 values of the W01 samples were 0.067 and 0.100, respectively, which were the lowest levels of the tested samples and indicated that W01 was brittle. The wear scars on the C100 samples displayed extrusions of fractured films on part of the wear track (Figure 10f). The wear track widths of the A50, A100, and W04(A0) samples were 89, 96, and 99 µm, respectively, which was narrow compared with the other tested samples and can be attributed to their high H values [50]. The wear rates of the A50 and A100 films were 1.0 × 10 −6 and 7.5 × 10 −7 mm 3 /Nm, with respective low COFs of 0.49 and 0.46 and high H 3 /E 2 values of 0.321 and 0.230. The amorphous B100 sample exhibited a smooth wear scar and a broad track width of 138 µm because of a low H (22.3 GPa) and a high wear rate (4.9 × 10 −6 mm 3 /Nm).

Wear Test
The wear resistance of W-N and W-Si-N films of various categorizations was examined. W01 was the crystalline W-N films with a W phase; W04, A50, and A100 were the crystalline W-N films with a W2N phase; B100 and C100 were the amorphous W-Si-N films. Figure 9 presents the coefficients of friction (COFs) of the samples against the cemented tungsten carbide ball after a sliding distance of 200 m. The relationship between COF and Ra was unclear, which can be attributed to the distinct categorizations of the tested films. Therefore, further research is warranted to investigate this relationship. The abrupt drops of COF in the W01 and C100 samples can be attributed to the exhaustion of films and exposure of the substrates. Table 6 summarizes the wear test results. The wear depths of the W01 and C100 samples were 1334 and 1404 nm, respectively, which were larger than their film thicknesses of 675 and 574 nm, respectively. The other samples exhibited wear depths that were smaller than their film thickness. Figure 10 illustrates the wear scars of the tested samples. Chipping debris was observed along the wear scar of the W01 samples (Figure 10a). The H/E and H 3 /E 2 values of the W01 samples were 0.067 and 0.100, respectively, which were the lowest levels of the tested samples and indicated that W01 was brittle. The wear scars on the C100 samples displayed extrusions of fractured films on part of the wear track (Figure 10f). The wear track widths of the A50, A100, and W04(A0) samples were 89, 96, and 99 μm, respectively, which was narrow compared with the other tested samples and can be attributed to their high H values [50]. The wear rates of the A50 and A100 films were 1.0 × 10 −6 and 7.5 × 10 −7 mm 3 /Nm, with respective low COFs of 0.49 and 0.46 and high H 3 /E 2 values of 0.321 and 0.230. The amorphous B100 sample exhibited a smooth wear scar and a broad track width of 138 μm because of a low H (22.3 GPa) and a high wear rate (4.9 × 10 −6 mm 3 /Nm).

Conclusions
The effects of the f and substrate bias on the mechanical properties of W-N and W-Si-N films were investigated. The main findings were as follows: (1) W-N films varied in phase from α-W to amorphous and β-W2N as f increased from 0.1 to 0.5, accompanied by increases in the compressive residual stress, H/E ratio, H 3 /E 2 ratio, and elastic recovery (We).