Mechanical Properties of Zr–Si–N Films Fabricated through HiPIMS/RFMS Co-Sputtering

Abstract

Zr–Si–N films were fabricated through the co-deposition of high-power impulse magnetron sputtering (HiPIMS) and radio-frequency magnetron sputtering (RFMS). The mechanical properties of the films fabricated using various nitrogen flow rates and radio-frequency powers were investigated. The HiPIMS/RFMS co-sputtered Zr–Si–N films were under-stoichiometric. These films with Si content of less than 9 at.%, and N content of less than 43 at.% displayed a face-centered cubic structure. The films’ hardness and Young’s modulus exhibited an evident relationship to their compressive residual stresses. The films with 2–6 at.% Si exhibited high hardness of 33–34 GPa and high Young’s moduli of 346–373 GPa, which was accompanied with compressive residual stresses from −4.4 to −5.0 GPa.

1. Introduction

Nanocomposite nitride films are comprised of nacocrystalline metal–nitride grains scattered in an amorphous Si₃N₄ matrix. These are recognized as nc-Me_nN/a-Si₃N₄ films, and have been widely investigated [1,2,3,4]. A hardness of 50 GPa was attained for nc-TiN/a-Si₃N₄ films, which was attributed to the inhibition of crack propagation by the amorphous tissue surrounding the nanocrystals [1]. The Ti–Si–N films with 8 at.% Si exhibited maximum hardness [1,2]. Previous studies [5,6,7] have applied the nc-Me_nN/a-Si₃N₄ model to interpret the hardness improvement of Nb–Si–N films, which exhibited a hardness of 34 GPa accompanied with the addition of 5–13 at.% Si. However, the hardness enhancement for Nb–Si–N films is restricted in contrast to that for Ti–Si–N films. Moreover, the previous works on Zr–Si–N films [8,9,10,11] also exhibited restricted improvement in hardness as the addition of 2–6 at.% Si achieved a hardness of 30–36 GPa. Additionally, the reported strengthening mechanisms were not consistent. The nc-Me_nN/a-Si₃N₄ model could not be applied to most Si-containing metal-nitride films. In our previous studies on Ta–Si–N [12], Nb–Si–N [13] and Zr–Si–N [14] coatings, fabricated using direct-current magnetron sputtering (DCMS) (introducing high Si contents into nitrides), formed an amorphous structure on X-ray and improved the oxidation resistance; although their mechanical properties deteriorated. Therefore, high-Si-content Ta–Si–N [15] and Zr–Si–N [16] coatings are applied for glass molding dies, which are operated under a low oxygen-containing atmosphere at approximately 600 °C. By contrast, the low-Si-content Zr–Si–N coatings exhibited a crystalline structure accompanied with a hardness level of 23–24 GPa and a compressive stress of less than 2 GPa [14]. High-power impulse magnetron sputtering (HiPIMS) [17,18] was recommended to grow films with a dense structure, high hardness and high residual compressive stress [19,20]. Therefore, exploring the mechanical performance of low-Si-content Zr–Si–N films fabricated by HiPIMS is essential for extending potential practical applications. Because the HiPIMS process resulted in low deposition rates [18], hybrid HiPIMS/DCMS [21] and HiPIMS/RFMS (radio-frequency magnetron sputtering) [22] were utilized to raise the deposition rate. In a previous study [23], the ZrN_x films prepared using HiPIMS system exhibited a nanoindentation hardness level of 26–27 GPa. In this study, Zr–Si–N films were fabricated through HiPIMS/RFMS hybrid process, before the mechanical properties of these films were investigated.

2. Materials and Methods

Figure 1 shows the schematic of HiPIMS/RFMS co-sputtering system. A target of 99.999% pure Si was connected to a RF (radio-frequency) power generator (13.56 MHz), while the targets of 99.995% Ti and 99.9% Zr were connected to a pulse power supply (SPIK 2000A; Shen Chang Electric Co., Ltd., Taipei, Taiwan) operated in the unipolar negative mode. The Si, Ti, and Zr targets were 50.8, 76.2 and 76.2 mm in diameter, respectively. The chamber was pumped to a base pressure of 6.7 × 10⁻⁴ Pa using a turbopump (TURBOVAC 600C, Leybold, Hsinchu, Taiwan) and monitored by an ionization gauge (274005, Granville-Phillips Company, Longmont, CO, USA). The substrate-to-target distance was 12 cm. The substrate holder without heating was rotated at 10 rpm and was grounded. The substrate holder was maintained at less than 70 °C during the deposition, which resulted in an insignificant thermal stress of less than 0.1 GPa. A working pressure of 0.4 Pa was formed after introducing 30-sccm-Ar flow into the chamber, before Ti was deposited at 400 W for 40 min. This creates an interlayer on silicon substrates, which had a thickness of 80–100 nm. Subsequently, Zr–Si–N films were co-sputtered on Ti interlayers after flowing a gas mixture of Ar and N₂ into the chamber. The sum of the Ar and N₂ flow rates were maintained at 30 sccm and the working pressure was maintained at 0.4 Pa. The HiPIMS process was conducted using the pulse on-time (t_on) and pulse off-time (t_off) of 100 and 3900 µs, respectively, which resulted in a duty cycle of 2.5% and a repetition frequency of 250 Hz.

Table 1. Sputtering parameters and thicknesses of Zr–Si–N films.

Sample	Ar (sccm)	N2 (sccm)	HiPIMS PowerWZr (W)	RF PowerWSi (W)	Ave. Target Voltage (V)	Peak Current (A)	Time (min)	Thickness (nm)	Rate (nm/min)
Batch I
Zr52Si8N40	28.0	2.0	300	50	660	67	210	831	4.0
Zr48Si9N43	27.5	2.5	300	50	643	80	200	720	3.6
Zr45Si10N45	27.0	3.0	300	50	600	90	240	732	3.0
Zr44Si10N46	26.5	3.5	300	50	588	105	310	836	2.7
Zr43Si10N47	26.0	4.0	300	50	588	110	360	898	2.5
Batch II
Zr71N29	28.0	2.0	300	0	640	58	400	1183	3.0
Zr62Si2N36	28.0	2.0	300	25	618	67	285	937	3.3
Zr62Si3N35	28.0	2.0	300	30	617	50	270	932	3.4
Zr58Si5N37	28.0	2.0	300	35	615	70	255	926	3.6
Zr56Si6N38	28.0	2.0	300	40	615	67	230	847	3.8
Zr52Si8N40	28.0	2.0	300	50	660	67	210	831	4.0
Zr50Si10N40	28.0	2.0	300	60	660	70	210	876	4.2

lists the main variables for depositing Zr–Si–N films. Batch I samples were prepared using various nitrogen flow rates, while Batch II samples were prepared using various sputter powers. The chemical compositions of the films were analyzed using a field-emission electron probe microanalyzer (FE-EPMA, JXA-8500F, JEOL, Akishima, Japan). The PET (pentaerythritol), TAP (Thallium acid phthalate), and LDE1 (layered diffracting elements of W and Si) crystals were used to analyze Zr, Si, as well as N, and O, respectively. The samples were labeled as Zr_xSi_yN_100−x−y. The film thicknesses determined on the fracture surface were evaluated by field emission scanning electron microscopy (FE-SEM, S4800, Hitachi, Tokyo, Japan). Grazing-incident X-ray diffraction (GIXRD) was performed by using an X-ray diffractometer (XRD, X’Pert PRO MPD, PANalytical, Almelo, The Netherlands) to identify the phases. The incidence angle was 1°. Moreover, the XRD patterns were obtained by the Bragg–Brentano (θ–2θ) scan to calculate lattice parameters and texture coefficients. The nanostructure of the films was observed using transmission electron microscopy (TEM, JEM-2010F, JEOL, Akishima, Japan). The TEM samples with a protective Pt layer were produced using a focused ion beam system (FEI Nova 200, Hillsboro, OR, USA). The mechanical properties, hardness and Young’s modulus of the films were evaluated using a nanoindentation tester (TI-900 Triboindenter, Hysitron, Minneapolis, MN, USA) and calculated using the Oliver and Pharr method [24]. An atomic force microscope (AFM, DI 3100, Bruker, Santa Barbara, CA, USA) recorded the surface roughness. The residual stress of the films was measured by the curvature method [25].

3. Results and Discussion

3.1. Zr–Si–N Films Fabricated Using Various Nitrogen Flow Rates

Table 2. Chemical compositions, surface roughness, mechanical properties and residual stress values of Zr–Si–N films.

Sample	Chemical Composition (at.%)				Ra (nm)	Ha (GPa)	Eb (GPa)	H3/E2 (GPa)	Stress c (GPa)
	Zr	Si	N	O
Batch I
Zr52Si8N40	52.0 ± 1.1	7.6 ± 0.2	40.1 ± 1.3	0.3 ± 0.1	3.5 ± 0.1	27.4 ± 1.6	256 ± 7	0.31	−2.8 ± 0.1
Zr48Si9N43	47.5 ± 0.7	9.0 ± 0.2	42.3 ± 0.9	1.2 ± 0.1	2.2 ± 0.2	20.4 ± 1.4	246 ± 9	0.14	−1.6 ± 0.2
Zr45Si10N45	44.6 ± 1.1	10.2 ± 0.2	44.4 ± 1.0	0.8 ± 0.2	0.3 ± 0.0	14.5 ± 0.2	184 ± 3	0.09	−0.4 ± 0.1
Zr44Si10N46	43.1 ± 1.3	10.1 ± 0.4	45.8 ± 1.3	1.0 ± 0.2	1.1 ± 0.1	13.8 ± 0.5	187 ± 7	0.08	−0.2 ± 0.1
Zr43Si10N47	42.4 ± 1.3	10.0 ± 0.2	46.6 ± 0.9	1.0 ± 0.3	1.1 ± 0.4	14.0 ± 0.6	190 ± 8	0.08	−0.6 ± 0.1
Batch II
Zr71N29	70.6 ± 0.5	0.1 ± 0.0	28.7 ± 0.7	0.6 ± 0.2	4.4 ± 0.2	21.5 ± 1.1	283 ± 13	0.12	−0.7 ± 0.1
Zr62Si2N36	61.1 ± 1.9	2.2 ± 0.1	35.8 ± 2.1	0.9 ± 0.3	1.3 ± 0.2	33.1 ± 1.0	346 ± 11	0.30	−4.9 ± 0.3
Zr62Si3N35	61.0 ± 0.4	3.2 ± 0.1	34.9 ± 0.4	0.9 ± 0.2	1.4 ± 0.1	34.4 ± 1.0	369 ± 11	0.30	−5.0 ± 0.4
Zr58Si5N37	57.5 ± 0.2	4.5 ± 0.2	37.1 ± 0.1	0.9 ± 0.2	1.1 ± 0.0	34.0 ± 1.4	367 ± 13	0.29	−4.7 ± 0.2
Zr56Si6N38	55.8 ± 0.2	5.6 ± 0.1	37.9 ± 0.3	0.7 ± 0.2	0.8 ± 0.1	33.2 ± 1.0	373 ± 12	0.26	−4.5 ± 0.6
Zr52Si8N40	52.0 ± 1.1	7.6 ± 0.2	40.1 ± 1.3	0.3 ± 0.1	3.5 ± 0.1	27.4 ± 1.6	256 ± 7	0.31	−2.8 ± 0.1
Zr50Si10N40	49.9 ± 0.6	9.7 ± 0.1	40.2 ± 0.8	0.2 ± 0.3	3.3 ± 0.0	21.5 ± 1.8	230 ± 7	0.19	−1.9 ± 0.1

^a: Hardness; ^b: Young’s modulus; ^c: Residual stress.

shows the chemical compositions of the Batch I films fabricated using various nitrogen flow rates. The O contents of the Zr–Si–N films were in the range of 0.3–1.2 at.%. The N content of the films increased from 40 to 43, 45, 46, and 47 at.% when the nitrogen flow rate was increased from 2.0 to 2.5, 3.0, 3.5, and 4.0 sccm, respectively. The Si content was changed from 8 to 9, 10, 10, and 10 at.%, respectively. The film thickness was maintained at levels of 720–898 nm by adjusting the deposition times. Figure 2 exhibits an evident decrease in the deposition rate from 4.0 to 2.5 nm/min as the nitrogen flow rate is increased from 2.0 to 4.0 sccm. The decrease was attributed to the highly ionized reactive gas used in the HiPIMS process, which had a severe poisoning effect of the Zr target [26,27,28].

Figure 3 depicts the GIXRD patterns of the Zr–Si–N films. The Zr₅₂Si₈N₄₀ and Zr₄₈Si₉N₄₃ films displayed a face-centered cubic (fcc) ZrN (ICDD 35-0753) phase; the Zr₄₅Si₁₀N₄₅ films exhibited mixed fcc ZrN and X-ray amorphous phases; and the Zr₄₄Si₁₀N₄₆ and Zr₄₃Si₁₀N₄₇ films were found to have an X-ray amorphous phase. Although the differences in Si and N contents between these films were not large, the structure varied evidently. The films with Si content of less than 9 at.% and N content less than 43 at.% exhibited a fcc structure, while the films with Si and N contents above the limitations exhibited an amorphous phase-dominated structure.

Table 2 lists the surface roughness, hardness, Young’s modulus and residual stress values of Zr–Si–N films. The crystalline Zr₅₂Si₈N₄₀ and Zr₄₈Si₉N₄₃ films exhibited relatively high surface roughness levels (R_a) of 3.5 and 2.2 nm, respectively. The Zr₄₅Si₁₀N₄₅ films revealed the lowest roughness level of 0.3 nm, while the amorphous Zr₄₄Si₁₀N₄₆ and Zr₄₃Si₁₀N₄₇ films exhibited a roughness level of 1.1 nm. Figure 4 shows the variations of hardness and residual stress related to different N₂ flow rates. The hardness values of crystalline Zr₅₂Si₈N₄₀ and Zr₄₈Si₉N₄₃ films were 27.4 and 20.4 GPa, respectively, which were accompanied with residual stress of −2.8 and −1.6 GPa, respectively. In contrast, the Zr₄₅Si₁₀N₄₅, Zr₄₄Si₁₀N₄₆ and Zr₄₃Si₁₀N₄₇ films exhibited a hardness level of 13.8–14.5 GPa, which was accompanied with the presence of an amorphous. This resulted in a low residual stress level of −0.2 to −0.6 GPa. The elastic moduli and hardness of Zr–Si–N films displayed similar variations related to their chemical compositions. Additionally, the H³/E² values, which represent resistance to plastic deformation [29], were 0.31 and 0.14 GPa for the crystalline Zr₅₂Si₈N₄₀ and Zr₄₈Si₉N₄₃ films, respectively. By contrast, the Zr₄₅Si₁₀N₄₅, Zr₄₄Si₁₀N₄₆ and Zr₄₃Si₁₀N₄₇ films containing an amorphous phase exhibited a low H³/E² level of 0.08–0.09 GPa.

3.2. Zr–Si–N Films Fabricated Using Various Power Levels

The Batch II films were fabricated through HiPIMS/RFMS co-sputtering using an average HiPIMS power of 300 W on the Zr target and various RF powers on the Si target (Table 1). The O contents of the films were 0.2–0.9 at.% (Table 2). The N content of the Batch II samples was changed from 29 to 36, 35, 37, 38, 40, and 40 at.% as the RF power increased from 0 to 25, 30, 35, 40, 50 and 60 W, respectively, and the Si content varied from 0, to 2, 3, 5, 6, 8, and 10 at.%, respectively. Both Batches I and II samples exhibited an atomic ratio N/(Zr + Si) of 0.41–0.89, which is under the stoichiometric ratio of 1.0 for ZrN. In a previous study [22], Ta–Zr–N films fabricated by the same HiPIMS/RFMS co-sputtering also exhibited an under-stoichiometric N/(Ta + Zr) ratio of 0.40–0.68. In another previous work [23], ZrN_x films prepared using HiPIMS system exhibited an under-stoichiometric N/Zr ratio of 0.65–0.78. The deposition times for Batch II processes were adjusted to fabricate films with similar thickness values of 831–1183 nm. The deposition rates increased from 3.0 to 4.2 nm/min when increasing the RF power from 0 to 60 W.

Figure 5 displays the GIXRD patterns of the Zr–Si–N films with various Si content. The Zr₇₁N₂₉ films exhibited a fcc ZrN structure with a strong (111) reflection. The introduction of Si into the films caused the structure to exhibit a strong (200) reflection, while the (111) reflection diminished. The fcc reflections became broadened and shifted toward the right side with further increases in the Si content in the films. Figure 6 and Figure 7 show the XRD patterns under a Bragg–Brentano scan and their texture coefficients [30,31] of the Batch II films, respectively. The reflections of Ti [ICDD 44-1294], TiSi [ICDD 17-0424] and ZrSi₂ [ICDD 32-1499] phases were observed besides a ZrN phase. The Ti and TiSi reflections were presented in another Ti/Si sample (not shown in this paper). The preferred orientations were (111) and (200) for the films with Si contents of <6 at.% and >8 at.%, respectively. Figure 8 exhibits the lattice parameters and grain sizes of the Zr–Si–N films calculated using (111) reflections. The lattice parameters decreased when increasing the Si content from 0 to 8 at.% due to the small atomic size of Si compared to Zr. This implied that 8 at.% Si or part of the 8 at.% Si was solved in the ZrN matrix and the residual Si atoms may reside at grain boundaries. Sandu et al. [7] suggested a solubility limit of 4 at.% Si for Nb–Si–N and Zr–Si–N films. A three-region model interpreted the structural evolution of composite M–Si–N films, which included the solubility of Si, coverage layer of SiN_x at grain boundaries and reduction of crystallite size. In contrast, the lattice parameters of the Zr₅₀Si₁₀N₄₀ films with a higher Si content of 10 at.% were larger than those of the films with 4–6 at.% Si, which could be attributed to the coexistence of Zr₃N₄ [16]. The grain sizes of the Batch II films determined by the Scherrer formula [32] exhibited a decreasing trend from 19.8 to 17.3, 17.7, 10.6 and 8.1 nm when increasing the Si contents from 0 to 3, 6, 8 and 10 at.%.

Figure 9a, Figure 10a and Figure 11a show the cross-sectional TEM images of the Zr₇₁N₂₉, Zr₆₂Si₃N₃₅ and Zr₅₀Si₁₀N₄₀ films, respectively. The selected area electron diffraction (SAED) patterns of the Zr₇₁N₂₉ and Zr₆₂Si₃N₃₅ films in Figure 9a and Figure 10a indicated a fcc ZrN phase, while the SAED pattern of the Zr₅₀Si₁₀N₄₀ films consisted of continuous rings, which implies a nanostructure. The high-magnification images of the Zr₇₁N₂₉ and Zr₆₂Si₃N₃₅ films (Figure 9b and Figure 10b) exhibited a columnar structure with widths of 19–22 and 14–17 nm, respectively. Figure 11b shows that the Zr₅₀Si₁₀N₄₀ films exhibits a fine structure and the width of the short columnar structure narrows to 9 nm only. The widths of the columnar structure in the Zr–Si–N films displayed a decreasing trend with increasing the Si content, which were consistent with the calculated grain sizes from XRD evaluation (Figure 8). Moreover, the short columnar structure of the Zr₅₀Si₁₀N₄₀ films that represented the crystalline regions were interrupted by amorphous tissue in the growth direction. Therefore, the lattice parameters of the Zr₅₀Si₁₀N₄₀ films did not follow the variation tendency of the Zr–Si–N films with 0–8 at.% Si, and the solid solution strengthening by introducing Si into ZrN matrix were negligible for the Zr₅₀Si₁₀N₄₀ films.

Table 2 and Figure 12 shows the hardness and residual stresses of the Zr–Si–N films with various Si contents. The Zr₇₁N₂₉ films exhibited a hardness of 21.5 GPa and a low compressive residual stress of −0.6 GPa, while the hardness were 33.1–34.3 GPa for the films with 2–6 at.% Si. This was accompanied with an elevated compressive residual stress level from −4.4 to −5.0 GPa. A further increase in Si content to 8 and 10 at.% resulted in decreased hardness of 27.4 and 21.5 GPa, respectively, which was accompanied with a decrease in the residual stress values to −2.8 and −1.8 GPa, respectively. The low-Si-content films exhibited a solid solution strengthening effect, while the formation of amorphous Si₃N₄ for high-Si-content films decreased the mechanical properties and relaxed the residual stress. Most of the previous studies on Zr–Si–N films reported a maximum hardness, which was accompanied with a certain range of Si contents [7,8,9,10]. Dong et al. [10], Martin et al. [9] and Mae et al. [8] reported maximum hardness of 29.8 GPa at 6.2 at.% Si, 36 GPa at 2 at.% Si and 35 GPa at 3 at.% Si, respectively. In the study conducted by Martin et al. [9], the films with a hardness of 36 GPa exhibited a stress of −5.7 GPa and a grain size of 17 nm. These characteristics seemed similar to the reported data in this study. The variation of hardness was evidently related to that of the residual stress [8,14]. Figure 13a shows the relationship of hardness and residual stress for Batches I and II samples, which exhibits a linear trend except for the hardness of Zr₇₁N₂₉ films. A similar linear relationship between Young’s modulus and residual stress was shown in Figure 13b. Table 2 shows that the Zr–Si–N films with 2–6 at.% Si shows high H³/E² levels of 0.26–0.30 GPa, which was accompanied with low roughness levels of 0.8–1.4 nm.

4. Conclusions

Under-stoichiometric Zr–Si–N films with atomic ratios N/(Zr + Si) of 0.41–0.89 were fabricated through HiPIMS/RFMS co-sputtering. The structural characteristics of Zr–Si–N films was influenced by their constitutions. The films with Si content of less than 9 at.% and N content of less than 43 at.% displayed a face-centered cubic structure. The mechanical properties of Zr–Si–N films exhibited linear tendencies that were related to their residual stresses. The crystalline Zr–Si–N films of 2–6 at.% Si showed a high nanoindentation hardness level of 33.1–34.3 GPa and a high Young’s modulus level of 346–373 GPa, which demonstrated H³/E² values of 0.26–0.30 GPa and surface roughness of 0.8–1.4 nm.