*3.2. Internally Oxidized Ru–Zr Coatings*

Figure 4 shows the cross-sectional SEM image of the annealed Ru0.50Zr0.50(R1) coating, which exhibited a laminated structure with an equilibrated laminated layer period of 55 nm. However, the features of the other coatings could not be identified through SEM. Figure 5 presents the XRD patterns of the Ru–Zr coatings after annealing in 1% O2–99% Ar at 600 ◦C for 30 min; all patterns exhibited monoclinic ZrO2 [ICDD 32-1484], tetragonal ZrO2 [ICDD 42-1164], and Ru phases. The Ru:Zr atomic ratios were maintained at levels similar to those of the as-deposited coatings (Table 1), implying that no volatile oxides were formed during annealing. The O content in the annealed coatings increased to within 58–62 at.%, indicating that extra O was trapped because the stoichiometric ratio of ZrO2 was two, enabling partial Ru atoms to be oxidized.

**Figure 4.** Cross-sectional SEM image of the Ru0.50Zr0.50(R1) coating after annealing in 1% O2–99% Ar at 600 ◦C for 30 min.

**Figure 5.** XRD patterns of the Ru–Zr(R*y*) coatings after annealing in 1% O2–99% Ar at 600 ◦C for 30 min.

Figure 6a–c illustrates the XPS spectra of O 1s, Zr 3d, and Ru 3d core levels, respectively, at various thickness levels of the annealed Ru0.50Zr0.50(R1) coating. The detected depth crossed six periods of the laminated layers. The O and Zr species were identified as O2<sup>−</sup> and Zr4+, whereas Ru was identified as Ru<sup>0</sup> except for the spectra near the surface region (depth < 13 nm), where the Ru*x*<sup>+</sup> and Ru4+ signals were split. The binding energy value of Ru0 3d5/2 (279.96 ± 0.08 eV) was consistent with that of other coatings (279.69–280.16 eV) reported in the literature [13,16,17,27], whereas the binding energies of Ru*x*<sup>+</sup> and Ru4+ 3d5/2 were 280.45 ± 0.11 and 282.57 ± 0.15 eV, respectively. Previous studies reported 281.4–282.2 eV [26,28–30] for the binding energy of Ru4+ 3d5/2. Ru of 17%–20% exhibited the Ru4+ state at a depth of 0–13 nm. Ru atoms remained in its metallic state beneath the near surface region. Figure 6d shows the intensity variations of O2<sup>−</sup> 1s, Ru<sup>0</sup> 3d5/2, and Zr4+ 3d5/2 signals along the depth, which indicates that the variation trend of the

O2<sup>−</sup> 1s profile coincides with that of the Zr4+ 3d5/2 profile and is in contrast to that of the Ru0 3d5/2 profile, implying that ZrO2 is the dominant oxide. Therefore, the annealed Ru0.50Zr0.50(R1) coating comprised alternating oxygen-rich and oxygen-deficient layers stacked along the O-diffusion direction. The binding energy value of Zr4+ 3d5/2 deviated within 182.05–183.35 eV (Figure 6e). Moreover, this range decreased to 182.71–183.35 eV after the data in the first laminated period had been excluded. Previous studies have reported 182.75 [31], 182.8 [32], and 182.9 eV [33] for the binding energy value of Zr4+ 3d5/2. The binding energy value of O<sup>2</sup><sup>−</sup> 1s demonstrated a variation pattern similar to that of the binding energy value of Zr4+ 3d5/2 (Figure 6e). The charging effect of analyzing insulators [34] caused substantial deviation in binding energy. The binding energy difference <sup>Δ</sup> = (O2<sup>−</sup> 1s − Zr4+ 3d5/2) was 347.92 ± 0.05 eV at the analyzed depth of 19.5–318.5 nm. This difference was highly consistent with the reported difference of 348.01 and 348.2 eV, calculated using 530.76 and 182.75 eV [31] or 531.1 and 182.9 eV [33] for O2<sup>−</sup> 1s and Zr4+ 3d5/2, respectively. The periodic changes of nonoxidized metallic Ru suggested the influence of oxygen in the Zr-deficient sublayers.

**Figure 6.** XPS spectra of (**a**) O 1s, (**b**) Ru 3d, and (**c**) Zr 3d core levels of the Ru0.50Zr0.50(R1) coating after annealing in 1% O2–99% Ar at 600 ◦C for 30 min; variation patterns of (**d**) intensity and (**e**) binding energy of O2−1s, Zr4+3d5/2, and Ru0 3d5/2.

Figure 7a,b shows the cross-sectional TEM images of the annealed Ru0.48Zr0.52(R5) coating; the laminated structure was evident. Figure 7c shows a high-resolution TEM image of the near-surface region of the annealed coating. The lattice fringes of particular areas indicated that the annealed Ru0.48Zr0.52(R5) coating comprised ZrO2- and Ru-dominant sublayers, which linked together across the original columnar boundaries such that the annealed Ru0.48Zr0.52(R5) coatings were laminated and the columnar boundaries were unresolved. Figure 8a depicts the cross-sectional TEM image of the annealed Ru0.47Zr0.53(R10) coating. The laminated sublayers were curved, because of which the stacks of sublayers among neighboring columnar structures were disconnected. Figure 8b shows the high-resolution TEM image of the middle region of the annealed Ru0.47Zr0.53(R10) coating. The Ru-dominant sublayers were two-nanometers thick only, and disconnected regions of the sublayers among neighboring columnar structures were observed. The fast variation of cyclical gradient concentration for the R10 coatings prepared with a quick substrate holder rotation speed resulted in the formation of grooved sublayers. For the coatings with thicker Ru sublayers, R1, R3, and R5, the sublayers became flat. The misaligned connections were more evident in the near-surface region (Figure 8c).

**Figure 7.** (**a**,**b**) cross-sectional TEM images of the oxidized Ru0.48Zr0.52(R5) coating; (**c**) high-resolution image of the near surface region of the annealed Ru0.48Zr0.52(R5) coating.

**Figure 8.** *Cont*.

**Figure 8.** (**a**) Cross-sectional TEM image of the oxidized Ru0.47Zr0.53(R10) coating; high-resolution images of the (**b**) middle region and (**c**) near-surface region of the annealed coating.

Figure 9a shows a cross-sectional TEM image of the annealed Ru0.46Zr0.54(R30) coating, in which the original columnar boundaries are evident, but no laminated structures were observed. A high-resolution TEM image (Figure 9b) revealed nanocrystalline grains of ZrO2 and Ru, each approximately five nanometers in diameter, implying that a nanocomposite structure had been constructed. Furthermore, Ru grains, the dark regions in the image, tended to concentrate along the columnar boundaries. Figure 10a–c illustrates the XPS spectra of the annealed Ru0.46Zr0.54(R30) coating. The XPS spectra of Ru 3d core levels indicated the presence of minor Ru4+ (3d5/2: 282.11 eV) in addition to Ru<sup>0</sup> (3d5/2: 280.19 ± 0.07 eV) at the near-surface region (Figure 10b), which was attributed to the incorporation of Ru into the ZrO2 grains because RuO2 and ZrO2 possessed a similar tetragonal structure. Figure 10d shows that the intensities of O2<sup>−</sup> 1s, Ru0 3d5/2, and Zr4+ 3d5/2 signals were constant along the depth due to the limit of XPS analyses. Similar binding energy trends were observed for O2<sup>−</sup> and Zr4+ (Figure 10e). The binding energy difference <sup>Δ</sup> = (O2<sup>−</sup> 1s − Zr4+ 3d5/2) was 348.00 ± 0.02 eV at the analyzed depth (5.7–96.9 nm). Therefore, Zr reacted with O during annealing, and the annealed coating exhibited a nanocomposite comprising ZrO2 and Ru grains.

**Figure 9.** (**a**) Cross-sectional TEM image and (**b**) high-resolution image of the oxidized Ru0.46Zr0.54(R30) coating.

**Figure 10.** *Cont*.

**Figure 10.** XPS spectra of (**a**) O 1s, (**b**) Ru 3d, and (**c**) Zr 3d core levels of the Ru0.46Zr0.54(R30) coating after annealing in 1% O2–99% Ar at 600 ◦C for 30 min; variation patterns of (**d**) intensity and (**e**) binding energy of O2−1s, Zr4+3d5/2, and Ru0 3d5/2.

## *3.3. Mechanical Properties of Internally Oxidized Ru–Zr Coatings*

Figure 11 depicts the nanoindentation hardness variations of the as-deposited and internally oxidized Ru–Zr coatings prepared at various substrate holder rotation speeds through sputtering. The hardness of the as-deposited coatings increased from 9.1 to 13.1 GPa with the substrate holder rotation speed and decreasing equilibrated laminated layer period. This hardness increase was attributed to the decrease in crystalline size and structural variation. The nanoindentation hardness of all Ru–Zr coatings increased after annealing in 1% O2–99% Ar at 600 ◦C for 30 min. The hardness variation curve of the internally oxidized Ru–Zr coatings exhibited three divisions representing a laminated structure, a disconnected laminated structure, and a nanocomposite region. The hardness increased from 9.1, 10.3, and 10.5 to 15.5, 16.1, and 17.2 GPa for the annealed Ru0.50Zr0.50(R1), Ru0.49Zr0.51(R3), and Ru0.48Zr0.52(R5) coatings, respectively, which exhibited equilibrated laminated layer periods of 55, 18, and 11 nm, respectively. This result indicates that the hardness of the internally oxidized Ru–Zr coatings, which exhibited crystalline phases identical to those identified through XRD analysis and appropriately maintained their multilayer structures, was affected by the layer period. These internally oxidized Ru–Zr multilayer coatings were categorized as nonisostructural oxide/metal multilayers [1]. Dislocation could not be moved across oxide/metal interfaces because oxides are brittle materials that deform through fracture mechanisms, limiting the hardness enhancement [2]; therefore, the hardness of oxide/metal multilayers approached that of the oxide ZrO2. Gan et al. reported a nanoindentation hardness of 18 GPa for monoclinic ZrO2 thin films [35]. By contrast, the hardness of the annealed Ru0.47Zr0.53(R10) coatings with an equilibrated laminated layer period of 5.6 nm exhibited a relatively low level of 12.3 GPa. Although the internally oxidized Ru0.47Zr0.53(R10) coatings were laminated in each columnar structure, the same sublayers among neighboring columnar structures were misaligned and disconnected, which reduced the hardness. The internally oxidized Ru0.46Zr0.54(R15), Ru0.47Zr0.53(R20), and Ru0.46Zr0.54(R30) coatings exhibited high hardness within 16.1–17.9 GPa and were categorized as nanocrystalline composites consisting of hard ZrO2 grains and

soft Ru grains. Figure 12 shows the variation in Young's moduli of the as-deposited and internally oxidized Ru–Zr coatings. The Young's moduli increased from 130 to 140 GPa for R1, R3, and R5 coatings, to 160 GPa for R10 coatings and 170–180 GPa for R15, R20, and R30 coatings. Because the internally oxidized Ru–Zr coatings exhibited similar phases, ZrO2 and Ru, the differences in Young's moduli among the annealed coatings were limited (i.e., 160–180 GPa). The surface roughness values of the Ru–Zr coatings are shown in Table 1. When evaluating the mechanical properties of coatings, previous studies [36–38] have reported that coatings with higher surface roughness exhibit larger standard deviation values or lower mean values. The effect of surface roughness on the mechanical properties of as-deposited Ru–Zr coatings was unclear. By contrast, the mechanical properties of the annealed coatings revealed larger deviations and higher surface roughness values than did those of the as-deposited coatings.

**Figure 11.** Nanoindentation hardness values of the as-deposited and internally oxidized Ru–Zr coatings.

**Figure 12.** Young's modulus values of the as-deposited and internally oxidized Ru–Zr coatings.

## **4. Conclusions**

Rotation speeds of the substrate holder during sputtering affected the crystalline structure and mechanical properties of Ru–Zr coatings both in the as-deposited and internally oxidized states. Because Ru–Zr coatings were fabricated using a cyclical gradient concentration stacked constitution, the coatings prepared at low rotation speeds (1–10 rpm) exhibited a laminated structure in addition to a columnar structure. The as-deposited Ru–Zr coatings exhibited nanoindentation hardness of 9.1–13.1 GPa, and the coatings prepared at higher substrate holder rotation speeds exhibited higher hardness. After annealing in a 1% O2–99% Ar atmosphere at 600 ◦C for 30 min accompanied by the conduction of internal oxidation, the coatings prepared at a substrate holder rotation speed of one to five revolutions

per minute maintained a laminated structure; this structure comprised alternately stacked Ru-dominant and ZrO2-dominant sublayers whose nanoindentation hardness increased to 15.5–17.2 GPa because of the formation of ZrO2 phase and the maintenance of sublayer interfaces. By contrast, the annealed coatings prepared at a rotation speed of 10 rpm maintained a similar laminated structure; however, the stacks of sublayers among neighboring columnar structures were misaligned and disconnected, resulting in relatively low nanoindentation hardness of 12.3 GPa. The annealed coatings prepared at a substrate holder rotation speed of 15–30 rpm exhibited nanocomposite coatings comprising Ru and ZrO2 grains within evident columnar boundaries and a high nanoindentation hardness of 16.1–17.9 GPa.

**Acknowledgments:** The financial support of this work from the Ministry of Science and Technology, Taiwan, under Contract No. 102-2221-E-019-007-MY3 is appreciated.

**Author Contributions:** Yung-I Chen designed the experiments and wrote the paper; Tso-Shen Lu performed the experiments; Zhi-Ting Zheng analyzed the XPS data.

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.
