1

B. D. Ozsdolay, C. P. Mulligan, M. Guerette, L. Huang, and D. Gall, “Epitaxial growth and properties of cubic WN on MgO(001), MgO(111), and Al2O3(0001),”

Thin Solid Films, 590, 276 (2015).

Epitaxial growth and properties of cubic WN on MgO(001), MgO(111), and Al2O3(0001)

B. D. Ozsdolay,1 C. P. Mulligan,2 Michael Guerette,1 Liping Huang1, and D. Gall1,* 1Department of Materials Science and Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180, USA

2U.S. Army Armament Research Development & Engineering Center, Benet Laboratories, Watervliet, NY 12189, USA

Tungsten nitride layers, 1.45-μm-thick, were deposited by reactive magnetron sputtering on

MgO(001), MgO(111), and Al2O3(0001) in 20 mTorr N2 at 700 °C. X-ray diffraction ω-2θ scans, ω-rocking curves, φ scans, and reciprocal space maps show that all layers exhibit a cubic rocksalt structure, independent of their N-to-W ratio which ranges from x = 0.83–0.93, as determined by energy dispersive and photoelectron spectroscopies. Growth on MgO(001) leads to an epitaxial WN(001) layer which contains a small fraction of misoriented grains, WN(111)/MgO(111) is an orientation- and phase-pure single-crystal, and WN/Al2O3(0001) exhibits a 111-preferred orientation containing misoriented cubic WN grains as well as N-deficient BCC W. Layers on MgO(001) and MgO(111) with x = 0.92 and 0.83 have relaxed lattice constants of 4.214±0.005 and 4.201±0.031 Å, respectively, indicating a decreasing lattice constant with an increasing N-vacancy concentration. Nanoindentation provides hardness values of 9.8±2.2, 12.5±1.0, and 10.3±0.4 GPa, and elastic moduli of 240±40, 257±13, and 242±10 GPa for layers grown on MgO(001), MgO(111), and Al2O3(0001), respectively. Brillouin spectroscopy measurements yield shear moduli of 120±2 GPa, 114±2 GPa and 108±2 GPa for WN on MgO(001), MgO(111) and Al2O3(0001), respectively, suggesting a WN elastic anisotropy factor of 1.6±0.3, consistent with the indentation results. The combined analysis of the epitaxial WN(001) and WN(111) layers indicate Hill’s elastic and shear moduli for cubic WN of 251±17 and 99±8 GPa, respectively. The resistivity of WN(111)/MgO(111) is 1.9×10-5 and 2.2×10-5 Ω-m at room temperature and 77 K, respectively, indicating weak carrier localization. The room temperature resistivity is 16% and 42% lower for WN/MgO(001) and WN/Al2O3(0001), suggesting a resistivity decrease with decreasing crystalline quality and phase purity. Keywords: WN; tungsten nitride; epitaxy; hardness; elastic constant; anisotropy; shear modulus; Brillouin scattering

*Corresponding Author: Daniel Gall, galld@rpi.edu, *1-518-276-8471

© 2015. This manuscript version is made available under the CC-BY-NC-ND 4.0 licensehttp://creativecommons.org/licenses/by-nc-nd/4.0/

2

1. Introduction Transition metal nitrides are known for their high hardness, wear and corrosion resistance,

and high temperature stability [1–3], and are therefore widely used as hard wear-protective coatings, diffusion barriers in microelectronics, and optical or decorative coatings [4,5]. There is a great interest in accurate values for the intrinsic mechanical properties like hardness and elastic modulus for all transition metal nitrides. These properties are, in general, a function of composition, phase, and orientation. However, the measured coating properties are typically also strongly affected by the microstructure including grain size, intrinsic strain, porosity and roughness [6]. It is challenging to determine the intrinsic properties using measurements on polycrystalline films, since they depend on the microstructure which, in turn, strongly varies with deposition conditions [7]. An effective approach to deconvolute the intrinsic properties from microstructural effects is to perform property measurements on well characterized single-phase epitaxial single crystal layers, as has previously been done for HfN [8], TiN [9], TaN [10], ScN [11], CrN [12], ZrN [13], and VN [14] and is done in this study for tungsten nitride.

Tungsten nitride has been reported to crystallize in a variety of phases, with the most common being a stoichiometric hexagonal δ-WN or a cubic W2N with a rocksalt structure where 50% of the N-sites are vacant [15,16]. In addition hexagonal and rhombohedral W2N3 and cubic W3N4 phases have also been reported, using high pressure, high temperature synthesis methods [17]. Polycrystalline tungsten nitride layers have been deposited reactive DC magnetron sputtering [18–24], but also by reactive pulsed laser deposition [23,25], RF sputtering [26], cathodic arc deposition [27], atomic layer deposition [28], and chemical vapor deposition [29–32], and have been used to determine the mechanical properties by nanoindentation [21–27]. Reported values for the WNx elastic modulus E range from E = 240-430 GPa [21,23,26], where an increase in the N-to-W ratio from x = 0.61-0.71 leads to a decrease in E from 305-275 GPa [26], or for x = 0.43, 0.89, and 1.38 to E = 380, 380, and 325 GPa [21], with the latter value being further reduced through strain relaxation upon annealing to E = 285 GPa [21]. Relatively small N concentrations yield a reduced modulus Er = 279-224 GPa for samples with x = 0.17-0.5 deposited by PLD [23] and Er = 313-284 GPa for sputtered WNx with x = 0.11-0.25 [26]. In addition, phase and intrinsic stress also affect E, for example, hexagonal WN1.17 with a compressive stress of -12.6 GPa exhibits a high E = 430 GPa [26]. The hardness H of WNx has been reported to range from 5-39 GPa [22–27]. This large range is attributed to variations in microstructure, nitrogen content, layer density, and phase content. For example, H decreases from 32 to 26 GPa as x increases in cubic WNx from 0.35 to 0.7, but H increases again with increasing x > 0.75 due to the formation of hexagonal phase inclusions [26]. In addition, H is reported to increase with deposition power and nitrogen partial pressure but decrease with total gas pressure, providing a large range from 6-30 GPa in a single study [22].

In this paper, we report on the growth and characterization of tungsten nitride layers to determine the mechanical properties of the relatively unexplored cubic phase. For this purpose, epitaxial and polycrystalline WN layers with a cubic phase and a measured N-to-W ratio of 0.83-0.93 were deposited on single crystal MgO(001), MgO(111), and Al2O3(0001) substrates by DC reactive magnetron sputtering, yielding 001 and 111-oriented epitaxial layers on the MgO substrates, and 111-oriented polycrystalline WN layers on Al2O3(0001), with the latter containing some misoriented WN as well as bcc W grains. Nanoindentation results show hardness values and elastic moduli that are lower than reported in most previous literature [21–23,25–27], with the exception of the particularly low hardness reported in Refs. 22 and 25 that can be attributed to underdense microstructures. Our approach of growing epitaxial thin films

3

reduces or eliminates the effects of grain boundaries and different crystal orientations on property measurements. Complementary characterization by nanoindentation and Brillouin spectroscopy provide both Young’s and shear moduli. Electrical resistivity measurements at room temperature and 77 K indicate a negative temperature coefficient, suggesting weak charge carrier localization in WN, while the W grains in WN/Al2O3(0001) reduce the overall thin film resistivity. 2. Experimental procedure

All WN layers were deposited in a load-locked ultra-high vacuum (UHV) dc magnetron sputtering system with a base pressure of <10-9 torr [33,34]. Polished 10×10×0.5 mm3 MgO(001), MgO(111) and Al2O3(0001) wafers were cleaned in successive rinses of trichloroethylene, acetone, isopropanol, and deionized water, blown dry with dry N2, attached to a Mo block with silver paint, loaded into the system through the load-lock chamber, and thermally degassed at 900 °C for 1 hour. This procedure has previously been successfully applied to grow epitaxial CrN(001)/MgO(001) [35–37], ScN(001)/MgO(001) [33], Cu(001)/MgO(001) [38–40], Ag(001)/MgO(001) [34], and Al1-xScxN(0001)/Al2O3(0001) layers [41,42]. The target was a 51-mm-diameter, 6-mm-thick water-cooled 99.95% pure W disk positioned 9.3 cm from the substrate at a 45° angle. Before each deposition, the target was sputter etched for 5 minutes with a shutter shielding the substrate. Depositions were performed at 20 mTorr 99.999% pure N2, using a constant dc power of 300 W applied to the magnetron source, yielding a deposition rate of 375 nm/hr. All layers were deposited at 700 °C, as measured by a thermocouple below the sample stage that was cross-calibrated with a pyrometer focused on the sample surface. This temperature was chosen to provide the best crystalline quality for all layers.

X-ray diffraction was done with a Panalytical X’Pert PRO MPD system with a Cu Kα source and a two-bounce two-crystal Ge(220) monochromator, yielding a parallel incident beam with a wavelength of λ = 1.5406 Å, a divergence of 0.0068°, and a width of 0.3 mm. Sample alignment included height adjustment as well as correction of the ω and χ tilt angles were done by maximizing the substrate peak intensity. Symmetric ω-2θ scans were obtained using a 0.27° slit and a 0.04 radian Soller Slit before the diffracted x-ray beam entered a PW1964/96 scintillation point detector. ω-rocking curves and φ scans were obtained using constant 2θ angles corresponding to WN 111 or 113 reflections and using the same parallel beam geometry as used for ω-2θ scans. Asymmetric reciprocal space maps around 113 reflections were obtained using a MediPix2 PIXcel solid-state line detector and a geometry with a small diffracted beam exit angle of 8-12° with respect to the sample surface to cause a beam narrowing which increases the 2θ resolution. In addition, ω-2θ scans with a divergent beam Bragg-Brentano geometry and the PIXcel line detector were acquired over a large 2θ range from 5-90° in order to detect small inclusions of possible secondary phases or misoriented grains.

The N-to-W composition ratio of each sample was determined using a combination of x-ray photoelectron spectroscopy (XPS) and energy-dispersive x-ray spectroscopy (EDS). XPS spectra were acquired using Al Kα radiation (1486.7 eV) in a PHI Versaprobe system with a hemispherical analyzer and an 8-channel detector. All high resolution spectra were collected using a pass energy of 23.5 eV and a step size of 0.2 eV. The surfaces were analyzed by XPS without sputter cleaning. EDS spectra were collected in a FEI Helios Nanolab SEM with an accelerating voltage of 5.0 kV, a working distance of 5.0 mm, and an Oxford Instruments X-MaxN 80 silicon drift detector which is particularly well suited for light element analysis. The EDS spectra were analyzed with the Oxford Instruments AZtec EDS software, and the system

4

was calibrated using the single beam current Oxford Instruments QCAL approach and tested with Micro-Analysis Consultants Ltd. produced BN and CaSiO3 standards, indicating an accuracy of the measured N-to-metal ratios of ±2.3%. The hardness and elastic modulus were measured using a Hysitron TI 900 Tribodenter nanoindenter with a diamond Berkovich tip with a nominal tip radius of 150 nm. Ten indents were performed for each maximum load of 2, 4, 6, 8, and 10 mN, after which loading-unloading curves were inspected for anomalies related to surface contamination or roughness. As a result, a maximum of 1-2 indents were excluded from the data analysis for each set of ten indents [43]. The instrument was calibrated using a fused quartz standard of known hardness and elastic modulus. H and E of the layers were determined with the Oliver and Pharr method [44], using an elastic modulus of 1140 GPa and a Poisson’s ratio of 0.07 for the diamond indenter tip [45], and assuming a Poisson’s ratio ν = 0.25 for WN. The sheet resistance of each layer was measured at both room temperature (293 K) and liquid nitrogen temperature (77 K) using a linear four point probe with 1-mm inter-probe spacings, a Keithley 2182A Nanovoltmeter, and a Keithley 6220 Precision Current Source providing 3 mA. The resistivity was calculated using the layer thickness of 1.45±0.05 μm, as measured by cross sectional SEM on cleaved samples, and correcting for the substrate geometry according to Ref. [46]. Brillouin light scattering (BLS) spectra were obtained using a six-pass high contrast Fabry–Perot interferometer from JSR Scientific Instruments and a monochromatic λ = 532.18 nm Verdi V2 DPSS green laser providing a 65 mW beam that was focused on the sample surface with an average incident angle θ = 60±1° relative to the surface normal. A 1-inch achromatic doublet lens (f = 76.2 mm, NA = 0.15, 90% clear central aperture) was used for laser delivery with TM polarization. The laser light was scattered from surface (Rayleigh) acoustic phonons [47] and collected using the same lens, using a shutter to block most of the intense unshifted light. The surface Rayleigh wave velocity VR was determined from the measured frequency shift Δf using VR = λΔf /(2sinθ) [47]. For analysis, the layers are considered semi-infinite media [47,48], since their thickness of 1.45±0.05 μm is much larger than the wavelength of the surface acoustic wave of 0.3 μm. The local strain associated with a surface wave is primarily a shear strain perpendicular to the direction of wave propagation. Therefore, the measured surface wave velocity is related to the transverse bulk wave velocity Vt through VR = βVt [49], where the factor β is a value close to but smaller than unity, and is a function of the surface orientation and the wave propagation direction and is affected by the elastic anisotropy and the ratio of the effective shear modulus vs bulk modulus. For the analysis in this report we assume a constant β = 0.94, which is within the reported typical range of 0.90-0.97 [48] and corresponds to the value that has been reported for the analysis of sputter deposited Zr1-xTaxN layers [48]. We note that an error in the assumed β of, for example, 0.03 would introduce an error of 6% in our reported values for G. From the transverse bulk wave velocity and the mass density ρ we determine an effective shear modulus G = ρVt

2 which refers to the shear perpendicular to the surface and perpendicular to the wave propagation [50]. The sample orientation for the BLS in these experiments has been set such that the in-plane wave propagation direction is parallel to the substrate edge. This corresponds to the WN[100] direction for the epitaxial WN(001)/MgO(001) layer, such that the G001 obtained from the measurement on WN/MgO(001) corresponds to shear of the (100) plane along [001], and therefore G001 = C44. In contrast, the in-plane wave propagation direction for the 111-oriented epitaxial and polycrystalline layers deposited on MgO(111) and Al2O3(0001) have

5

not been determined, such that the measured G111 refers to an effective shear modulus perpendicular to the (111) surface for these samples. 3. Results and discussion

The measured N-to-W ratio x in the deposited layers, as determined by EDS, is 0.92±0.02 for WN/MgO(001), 0.83±0.02 for WN/MgO(111), and 0.93±0.02 for WN/Al2O3(0001). These values are confirmed by XPS measurements which agree to within 3% with the EDS results. That is, our layers have a relatively high nitrogen concentration in comparison to the previously reported concentration range for cubic WNx. In particular, DC reactive magnetron sputtering in a Ar+N2 mixture has been reported to yield x = 0.33 with a N2 fraction in the gas of fN2 = 10% [19], or x = 0-0.5 with fN2 increasing from 0-75% [51], but also x = 0-1.1 for fN2 = 0-75% [21], x = 0-1.2 for fN2 = 0-63% [24], and x = 0-1.2 for fN2 = 5% with an increasing 17-26 Pa total pressure [52]. However, the layers with high N content x > 0.9 have a reported tendency to exhibit a hexagonal WN phase [24,26]. RF sputter deposition and PLD have been reported to yield even higher nitrogen concentrations (again exhibiting a hexagonal phase for high x), likely due to energetic implantation, with x = 0.6-1.6 for fN2 = 10-75% [53], x = 1.1-1.4 for fN2 = 10-60% [54], and x = 0.35-1.17 for fN2 = 38-83% [26] for RF sputtering, and x = 1.76 for PLD in 75 mTorr N2 [55].

Figure 1 shows sections of x-ray diffraction ω-2θ spectra from three WN layers deposited on MgO(001), MgO(111), and Al2O3(0001). The spectrum in Fig. 1(a) from a layer deposited on MgO(001) exhibits WN 002 and MgO 002 peaks at 2θ = 42.75±0.10° and 42.909±0.001°, respectively. These two peaks are the only detectable peaks within the measured 2θ = 30-70° range, if using a diffraction configuration with a monochromator. However, x-ray diffraction using a divergent beam Bragg-Brentano geometry reveals weak secondary peaks corresponding to 111, 220, and 311 oriented cubic WN grains that are 600, 750, and 4000 times less intense than the 002 peak, indicating that this layer exhibits predominately a 001 out-of-plane orientation with small inclusions of misoriented grains. Additional ω-2θ scans (not shown) acquired with the parallel monochromatic source and using a series of ω-offsets of 0.1-1.0° to suppress the high intensity MgO 002 substrate peak provide a more accurate determination of the WN 002 peak position of 2θ = 42.80±0.05°. This yields an out-of-plane lattice constant of a = 4.222±0.005 Å. The spectrum from the WN/MgO(111) sample in Fig. 1(b) shows a MgO 111 peak at 2θ = 36.947° and a WN 111 peak at 2θ = 37.34°, corresponding to a = 4.168 Å. No other reflections can be detected for this sample neither using the monochromatic parallel nor the divergent x-ray source configuration, indicating a single 111 out-of-plane orientation for WN deposited on MgO(111). The layer deposited on Al2O3(0001) shows a WN 111 XRD peak, as shown in Fig. 1(c). However, its intensity is approximately 30 times weaker than for the layer grown on MgO(111), indicating lower crystalline quality and weaker grain alignment. The spectrum obtained from a divergent beam measurement is also plotted in Fig. 1(c), showing a WN 111 peak at 2θ = 37.3° and a broad approximately 3 times weaker shoulder between 36 and 37°, which suggests a rocksalt structure WNx phase with a varying lattice constant. More specifically, using λ = 1.5418 Å since both Kα1 and Kα2 x-rays contribute to the divergent beam measurement, the peak maximum provides an a = 4.17 Å, but the shoulder peak yields a range for a = 4.21 - 4.32 Å. The divergent beam scan also reveals weak additional peaks (not shown) at 2θ = 40.3±0.1°, 43.3±0.1°, 63.9±0.2° and 75.8±0.3°, which are 5, 35, 12, and 40 times weaker than the WN 111 peak, respectively. The first of those peaks is close to 40.265°, the expected position for BCC W 110, while the three latter peaks are from cubic WN 002, 220, and 311 reflections.

6

This indicates that this WN layer grown on Al2O3(0001) contains phase separated BCC grains which likely have a low N concentration, as well as a small fraction of randomly aligned WN grains, while the majority of the layer exhibits a preferred 111-orientation.

Figure 2 shows results from additional XRD analyses to investigate the crystalline quality, the epitaxial relationship with the substrate, and the strain state of the WN layers. The plots in Figs. 2(a) and (b) are ω rocking curves from WN/MgO(111) and WN/Al2O3(0001) samples, respectively. They are obtained using a constant 2θ = 37.286° for WN/MgO(111) and 2θ = 37.350° for WN/Al2O3(0001), corresponding to the WN 111 reflection, which is the strongest peak for both samples, as shown in Figs. 1(b) and (c). The full-width at half-maximum (FWHM) of the WN 111 rocking-curve peak is 4.3° for WN/MgO(111) and 3.7° for WN/Al2O3(0001). These values indicate a strong preferred 111-orientation for the WN layer on both MgO(111) and Al2O3(0001) substrates. The rocking curve from the layer grown on the MgO(001) substrate is not shown, because its primary peak, attributed to the WN 002 reflection, is too close to the MgO 002 substrate peak, such that its rocking-curve width cannot be deconvoluted from the effect of the tail of the substrate peak. However, reciprocal mapping provides some comparable information for this sample, as discussed below.

Figures 2(c) and (d) show XRD φ-scan plots from WN/MgO(111) and WN/Al2O3(0001), respectively. They are obtained by rotation of the sample about its surface normal by the angle φ, while keeping 2θ and ω constant at 75.679° and 66.830° for WN/MgO(111), and at 75.744° and 65.443° for WN/Al2O3(0001), which corresponds to conditions to detect asymmetric WN 113 reflections for this case of 111-oriented layers. The scan from WN/MgO(111) in Fig. 2(c) exhibits 3 peaks at φ = 56°, near φ = 180°, and at φ = 295°, close to the expected 3-fold symmetric positions at φ = 60°, 180°, and 300° for an epitaxial cubic WN(111) layer. The position of the peak near φ = 180° cannot be known precisely as it appears as a shoulder on the substrate MgO 113 reflection. The background intensity between the peaks is 6 cps, while the peak maxima range from 120-260 cps, which excludes the sharp feature at φ = 180° due to the tail of the MgO 113 reflection. The variation in the peak intensity as well as the presence of the substrate feature only for the φ = 180° peak are attributed to the substrate miscut, more specifically, the substrate rotation in these experiments is about the surface normal, while equal intensity for the three peaks would be expected for rotation about the 111 crystal direction, which is tilted (by the substrate miscut) from the surface normal. Fig. 2(d) shows the corresponding φ-scan obtained from the WN/Al2O3(0001) layer. The curve has a nearly constant intensity of 400 cps, indicating a random in-plane grain orientation for this WN layer.

Figures 2(e) and (f) show reciprocal space maps around asymmetric 113 reflections from the WN layers grown on MgO(111) and MgO(001), respectively. The maps show iso-intensity contour lines in a logarithmic scale where the inter-line-spacing corresponds to an intensity change by a factor of 2, plotted within k-space where k ⊥ = 2sinθcos(ω-θ)/λ and k|| = 2sinθsin(ω-θ)/λ [56] correspond to directions perpendicular and parallel to the substrate surface. The reciprocal space map in Fig. 2(e) shows a sharp 113 peak from the MgO substrate with a FWHM of 0.002 nm-1 in both the parallel and perpendicular directions as well as a WN peak which is 15 times broader in 2θ and 64 times broader in ω, and is located at Δk⊥= 0.126 nm-1 and Δk|| = -0.077 nm-1 relative to the MgO 113 substrate peak. Therefore this WN/MgO(111) layer has a lattice constant that is smaller in the out-of-plane and larger in the in-plane direction than the lattice constant of MgO of 4.212 Å, indicating tensile stress in this WN layer with a = 4.137±0.032 Å and a|| = 4.297±0.106 Å. The former value agrees with a = 4.168±0.001 Å, determined from the ω-2θ scan in Fig. 1(b) within experimental error.

7

The map in Fig. 2(f) from the WN/MgO(001) layer also shows a sharp, intense 113 peak from the MgO substrate with a FWHM of 0.003 nm-1 in the k|| direction and 0.002 nm-1 in the k

direction, and a 180 times less intense peak from the WN layer which appears as shoulder below and slightly to the right of the substrate peak, with Δk = -0.07 nm-1 and Δk|| = 0.02 nm-1.

Therefore, this WN/MgO(001) layer is nearly relaxed with a = 4.217±0.001 Å and a|| = 4.210±0.011 Å, in good agreement with a = 4.222±0.005 Å from the WN 002 peak in the ω-2θ scan in Fig. 1(a).

We determine the relaxed lattice constant ao = (a ‐ a ν + 2a||ν)/(1 + ν) from the measured a and a||, assuming a Poisson’s ratio of ν = 0.25±0.04 as typical for many transition metal nitrides [8]. The uncertainty in the Poisson’s ratio introduces an error of only ±0.008 Å and ±0.0003 Å for the WN/MgO(111) and the (less strained) WN/MgO(001) layers, respectively, resulting in ao = 4.201±0.031 Å for the WN/MgO(111) layer and ao = 4.214±0.005 Å for the WN/MgO(001) layer. That is, the WN/MgO(111) layer has a 0.3% lower lattice constant than the WN/MgO(001) layer. We attribute this difference to the lower N-to-W ratio of x = 0.83±0.02 for WN/MgO(111), in comparison to x = 0.92±0.02 for WN/MgO(001), consistent with a previous report indicating an increasing ao with increasing x [18]. The polycrystalline microstructure of the WN/Al2O3(0001) layer precludes acquisition of a reciprocal space map and therefore determination of a|| and ao. However, based on the spectrum in Fig. 1(c), we determine the out-of-plane lattice constant to range from a = 4.172 Å to a = 4.321 Å, This range can be directly compared to a⊥ = 4.168 Å from the WN/MgO(111) layer. Therefore, comparing the two 111-oriented layers we find that the WN/Al2O3(0001) layer with x = 0.93±0.02 has larger lattice constant that the WN/MgO(111) layer with x = 0.83±0.02, confirming the trend of increasing lattice constant with increasing x.

The biaxial in-plane strain ε|| = (a|| - ao)/ao is 2.3±2.0 % (tensile) for WN/MgO(111) and ε|| = -0.1±0.2 % (relaxed within experimental uncertainty) for WN/MgO(001). This corresponds to a biaxial stress of 5.9±5.7 GPa and -0.2±0.6 GPa for WN/MgO(111) and WN/MgO(001), respectively, as determined using elastic moduli of E = 257±13 GPa for WN/MgO(111) and E = 240±40 GPa for WN/MgO(001), obtained from nanoindentation measurements presented below. The measured stress in the WN/MgO(111) layer is relatively high. However, stresses of similar magnitude have been reported in previous studies of tungsten nitride layers [24].

The room-temperature electrical resistivity for the WN layers grown on MgO(001), MgO(111), and Al2O3(0001) is 1.6×10-5, 1.9×10-5, and 1.1×10-5 Ω-m, respectively. At 77 K, we measure 10-20% higher values of 2.0×10-5, 2.2×10-5, and 1.2×10-5 Ω-m. That is, all WN layers exhibit a slightly negative temperature coefficient of resistivity (TCR), suggesting a weak carrier localization which is attributed to random point defects associated primarily with N vacancies. A similar negative TCR has been reported for other epitaxial transition metal nitrides including TaNx(001) [57], HfNx(001) [58], and NbNx(001) [43], as well as ternary nitrides including Sc1-

xTixN(001) [59], Ti1-xWxN(001) [60], and Sc1-xAlxN(001) [61], and has been attributed to a weak Anderson localization due to N vacancies and anti-site substitutions for the binaries, and random cation solid solutions for the ternaries.

Our WN room temperature resistivity values are in the middle of the large range of previously reported WN resistivities of 0.1-5.0×10-5 Ω-m [62,63]. We attribute the large range of values to variations in composition, phases, and crystalline quality. In fact, even the three samples from this study, for which nominally the only difference is the substrate, show a considerable variation in the resistivity. The lowest value is measured for the WN/Al2O3(0001) layer, which exhibits a 111 textured polycrystalline microstructure with a broad range of lattice

8

constants and inclusions of misoriented WN and BCC W grains. The layer deposited on MgO(001) has a 50% higher resistivity, which is attributed to it being an epitaxial WN(001) layer with only a small fraction of misoriented grains which exhibit the same cubic WN phase as the epitaxial matrix. The resistivity of the layer grown on MgO(111) is another 20% higher. This WN layer contains no detectable secondary phases or misoriented grains, but a higher N-vacancy concentration. In summary, the resistivity data suggests that WN exhibits weak carrier localization and that the resistivity increases with phase purity, crystalline quality, and possibly also N-vacancy concentration.

Figures 3(a) and (b) show the measured elastic modulus and hardness vs maximum nanoindentation contact depth, from the three WN samples presented above. Each data point represents the average value obtained from analyzing loading and unloading curves of 8-10 indents with the same maximum load, with the error bars indicating the standard deviation in both contact depth and E or H. Five different load functions with maximum loads of 2, 4, 6, 8, and 10 mN yield indentation depths ranging from 50 - 170 nm, which corresponds to 3 - 12 % of the WN layer thickness, indicating that substrate effects can be neglected for all data points.

The measured elastic moduli are 240±40 GPa for WN/MgO(001), 257±13 GPa for WN/MgO(111), and 242±10 GPa for WN/Al2O3(0001). These values are determined from the four data points obtained with maximum loads of 4-10 mN. The E values from the 2 mN indents are excluded in this analysis because they are 14-40% higher than the above cited averages, which is attributed to surface effects which cause an increased hardness due to dislocation starvation in the small plastic deformation region and strain gradients near the surface [64–67]. The measured hardness, determined from indents with maximum loads of 4-10 mN, is 9.8±2.0 GPa for WN/MgO(001), 12.5±1.0 GPa for WN/MgO(111), and 10.3±0.4 GPa for WN/Al2O3(0001). These three hardness values are identical within the experimental uncertainty. The uncertainty, as indicated by the vertical error bars in Fig. 3(b), is larger for the WN/MgO(001) layer than the other two samples, which we attribute to the small inclusions of misoriented grains and associated surface roughness or surface strain gradients for this sample. The values of E = 240-257 GPa are within the lower end of the previously reported range of 240-430 GPa [21,23,26] for tungsten nitride deposited by physical vapor deposition, while the measured H = 10-13 GPa is within the large published range of 5-39 GPa [22–27]. We attribute the large range of published values for E and H to variations in microstructure, composition, and crystal orientation between samples from different researchers.

The primary difference between the samples in this study to those from previous investigations is the absence of grain boundary effects in our epitaxial layers. In addition, the measured high N-to-W ratio of 0.83-0.93 for a cubic structure is unique, as the most stable composition for the cubic structure is expected to be W2N, while stoichiometric WN typically yields a hexagonal structure [16]. Therefore, the relatively low elastic modulus of our samples in comparison to some other studies may be attributed to the high N-content, which yields a high valence electron density that causes filling of antibonding states, as suggested for Mo2N [68]. Consistent with this argument, the E = 257±13 GPa for WN/MgO(111) with x = 0.83 is slightly larger than the E = 242±10 GPa for WN/Al2O3(0001) with x = 0.93 and E = 240±40 GPa for WN/MgO(001) with x = 0.92.

We note here that the presented elastic moduli are obtained by neglecting a possible elastic anisotropy in WN. Such an anisotropy causes a correction to the modulus measured by nanoindentation for any layer which exhibits a preferred orientation including epitaxial layers. This effect has been discussed in detail by Vlassak and Nix [69,70], who provide parameterized

9

expressions for a correction factor β* that relates the measured indentation modulus Mhkl to the isotropic elastic modulus E by Mhkl = β*(E/1-ν2) [69]. The resulting correction is relatively small, for example, even a material with a relatively strong Zener anisotropy ratio A = 2C44/(C11- C12) of 0.5 or 2.0 and a typical Poisson’s ratio along <001> of 0.25 would result in an approximately ±5% correction. Accordingly, the difference in the measured elastic modulus for the 001 and 111 oriented WN layers grown on MgO(001) and MgO(111) can be explained by the elastic anisotropy of WN. For this purpose, we determine the ratio of the measured indentation moduli M111/M001 = 1.07±0.19, which suggests, based on the predictions in Refs. [69,70] and assuming a Poisson’s ratio along <001> of 0.25, an anisotropy factor A = 1.5. However, because of the relatively weak dependence of the indentation moduli on the anisotropy factor, the uncertainty in M111/M001, which is primarily due to the uncertainty in M001 measured by nanoindentation for WN/MgO(001), causes a very large uncertainty in A, corresponding to a range of A = 0.4-3.8. The correlated analysis of the indentation moduli from the WN(001) and WN(111) layers also provides a value for the isotropic WN elastic modulus of 251±17 GPa.

Figure 4 shows Brillouin light scattering spectra which are used as an alternative method to determine the WN elastic properties, similar to what has previously been reported for TiN [50,71] and ZrTaN [48]. The spectra are from three WN layers deposited on MgO(001), MgO(111), and Al2O3(0001), and are shifted by a constant factor for clarity purposes. Peaks associated with surface acoustic phonons are symmetric around the intense central peak at Δf = 0 from elastically scattered photons. The spectrum from the WN/MgO(111) layer exhibits a pair of peaks with intensities of 95 and 65 cts above the background of 55 cts. They are at frequency shifts Δf = -7.65±0.03 and 7.74±0.03 GHz and are attributed to Stokes and Anti-Stokes scattering by surface acoustic phonons. Similarly, the WN/Al2O3(0001) shows Stokes and Anti-Stokes peaks at Δf = -7.55±0.02 and 7.41±0.04 GHz with intensities of 70 and 55 cts above the background level of 90 cts, while the WN/MgO(001) layer yields peaks at Δf = -7.68±0.03 and 8.04±0.04 GHz with intensities of 75 and 80 cts above the background level of 475 cts.

The velocity VR of the surface acoustic phonons is determined from the average of the positive and negative frequency shifts, yielding VR = 2363±16 m/s for WN/MgO(111), VR = 2298±20 m/s for WN/Al2O3(0001), and VR = 2415±20 m/s for WN/MgO(001). This provides, assuming a fully dense microstructure with a density of ρ = 18.1 g/cm3 and a factor β = 0.94 as discussed in the Experimental section, values for the effective shear moduli of G111 = 114±2 GPa for WN/MgO(111), G111 = 108±2 GPa for WN/Al2O3(0001), and G001 = 120±2 GPa for WN/MgO(001). The G111 for the layer on Al2O3(0001) is slightly lower than G111 for the layer on MgO(111). We attribute the smaller G111 for the layer on Al2O3(0001) primarily to microstructural effects, in particular a softening of the layer grown due to strain fields associated with the inclusion of misoriented cubic WN grains as well as N-deficient bcc W grains. In addition, the boundaries between grains with different in-plane orientations in the polycrystalline WN/Al2O3(0001) as well as the higher N-concentration may effectively decrease G111.

Secondly, we compare the result from the WN(001) and WN(111) layers grown on MgO(001) and MgO(111). For these epitaxial layers, the microstructural effects are expected to be of considerably lower importance, such that the comparison provides some direct information on the anisotropy of the WN elastic properties. In particular, G001 = C44 since the in-plane wave propagation is along [100] for the WN/MgO(001) layer, as described in the experimental section, while G111 = 9A(C44)/((2+A)(1+2A)) is a function of the elastic anisotropy A, as derived in Ref. [72]. Using the measured G001 = 120±2 GPa and G111 = 114±2 GPa, we determine a WN elastic

10

anisotropy of A = 1.6±0.3. This value is used, together with C44 = 120±2 GPa to determine the isotropic shear modulus of 99±8 GPa, which is the Hill’s average of the Voigt and the Reuss approximations of 102 and 97 GPa, respectively [73]. Combining these numbers with the isotropic elastic modulus of 251 GPa from the indentation yields C11 = 276, C12 = 126, and C44 = 120 GPa, or a bulk modulus of 176 GPa and an isotropic Poisson’s ratio of 0.26, in good agreement with the original assumption of 0.25.

This A =1.6±0.3 from BLS is in good agreement with A = 1.5 obtained from the analysis of the indentation moduli. However, this agreement should not be over-emphasized, since the uncertainty in the values, particularly from the indentation analysis, are large. The obtained A =1.6±0.3 is large in comparison to reported values for other transition metal nitrides with a rock salt structure, ranging from 0.11-0.71 from experiments [14,43,74–76] and from 0.03-1.04 from first-principles calculations on mechanically stable first row transition metal nitrides [77]. It is particularly noteworthy that our value for the WN anisotropy is larger than unity, suggesting that the stiffness of WN is smaller along the [100] than the [111] direction, which is opposite to the reported results for the other nitrides. We mention here that this unexpected result could potentially also be explained by the lower nitrogen content in the WN/MgO(111) compared to the WN/MgO(001) layer, which may cause a bond softening and, in turn, a reduction in both the elastic and shear moduli [78]. With that argument, the measured G111 < G001 is explained by the lower N content in the WN/MgO(111) layer, while the anisotropy could be smaller than unity. In that case, A < 1 as well as the reduced modulus at lower N content would result in M111 < M001, which is opposite to the measured moduli of 257±13 and 240±40 GPa, respectively. However, because of the large uncertainty of the latter value, the indentation results are nevertheless consistent with a M111 < M001, such that A < 1 if N vacancies cause considerable bond softening in WN. The data from the polycrystalline 111-oriented layer on Al2O3(0001) which has a N-to-W ratio close to WN(001) provides contradicting input to this argument: It exhibits an indentation modulus close to that of the WN/MgO(001) layer, suggesting that composition is more dominant than crystalline orientation, but has a surface acoustic wave velocity that is even lower than that of the WN/MgO(111) layer, suggesting a bond stiffening rather than softening with decreasing N content. We attribute these opposite trends to microstructural effects associated with the polycrystalline structure of WN/Al2O3(0001) such that this layer does not provide any additional insight into the elastic anisotropy discussion.

4. Conclusions WN layers with a cubic rock salt structure have been deposited on single crystal MgO(001), MgO(111), and Al2O3(0001) substrates by reactive magnetron sputtering at 700 °C. WN0.92/MgO(001) exhibits a cube on cube epitaxy with a relaxed lattice constant of ao = 4.214±0.005 Å, a hardness H = 9.8±2.0 GPa, an elastic modulus E = 240±40 GPa as measured by nanoindentation, and a shear modulus perpendicular to the surface G001 = 120±2 GPa as determined from the surface acoustic wave velocity measured by Brillouin spectroscopy. WN0.83/MgO(111) is a 111 oriented epitaxial layer with ao = 4.201±0.031 Å indicating an increasing lattice constant with increasing N content, H = 12.5±1.0 GPa, and E = 257±13 GPa, and a shear modulus perpendicular to the surface G111 = 114±2 GPa. WN0.93/Al2O3(0001) is a 111 textured polycrystalline layer which also contains misoriented WN and N-deficient bcc W grains. It has an out of plane lattice constant ranging from a = 4.172 - 4.321 Å, H = 10.3±0.4 GPa, E = 242±10 GPa and G111 = 108±2 GPa. The hardness and elastic moduli values for WN grown on the three different substrates are approximately the same, suggesting only minor

11

effects from grain boundaries and possibly a slight increase in E with decreasing N content. In contrast, G111 measured by Brillouin scattering is lower for the polycrystalline WN0.93/Al2O3(0001) vs the single crystal WN0.83/MgO(111) and WN0.92/MgO(001) layers, suggesting softening associated with grain boundaries or alternatively a weak dependence of the shear modulus on composition. The elastic anisotropy determined by comparing the indentation moduli or the surface acoustic phonon velocities from epitaxial layers grown on MgO(001) and MgO(111) is 1.5 and 1.6, respectively, suggesting that WN is stiffer along [111] than along [100], which is opposite to what is typically known for rock salt transition metal nitrides. This may indicate a unique elastic property of WN but may also be explained by the difference in the sample compositions and microstructures. Electrical resistivity measurements at room temperature and 77 K indicate weak carrier localization and a higher resistivity with higher crystalline quality. Acknowledgements The authors acknowledge support from the US National Science Foundation under grant Nos. 1309490 and 1234872. References: [1] H. Holleck, Material selection for hard coatings, Journal of Vacuum Science & Technology A: Vacuum,

Surfaces, and Films. 4 (1986) 2661. [2] J.-E. Sundgren, A review of the present state of art in hard coatings grown from the vapor phase, Journal of

Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 4 (1986) 2259. [3] J. Musil, Hard and superhard nanocomposite coatings, Surface and Coatings Technology. 125 (2000) 322–

330. [4] M. Georgson, The influence of preparation conditions on the optical properties of titanium nitride based

solar control films, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 9 (1991) 2191.

[5] S. Niyomsoan, W. Grant, D.. Olson, B. Mishra, Variation of color in titanium and zirconium nitride decorative thin films, Thin Solid Films. 415 (2002) 187–194.

[6] A.. Ehiasarian, P.E. Hovsepian, L. Hultman, U. Helmersson, Comparison of microstructure and mechanical properties of chromium nitride-based coatings deposited by high power impulse magnetron sputtering and by the combined steered cathodic arc/unbalanced magnetron technique, Thin Solid Films. 457 (2004) 270–277.

[7] J.-E. Sundgren, Structure and properties of TiN coatings, Thin Solid Films. 128 (1985) 21–44. [8] H.-S. Seo, T.-Y. Lee, J.G. Wen, I. Petrov, J.E. Greene, D. Gall, Growth and physical properties of epitaxial

HfN layers on MgO(001), Journal of Applied Physics. 96 (2004) 878. [9] H. Ljungcrantz, M. Odén, L. Hultman, J.E. Greene, J.-E. Sundgren, Nanoindentation studies of single-

crystal (001)-, (011)-, and (111)-oriented TiN layers on MgO, Journal of Applied Physics. 80 (1996) 6725. [10] C.-S. Shin, D. Gall, P. Desjardins, A. Vailionis, H. Kim, I. Petrov, J.E. Greene, M. Odén, Growth and

physical properties of epitaxial metastable cubic TaN(001), Applied Physics Letters. 75 (1999) 3808. [11] D. Gall, I. Petrov, N. Hellgren, L. Hultman, J.E. Sundgren, J.E. Greene, Growth of poly- and single-crystal

ScN on MgO(001): Role of low-energy N2+ irradiation in determining texture, microstructure evolution, and

mechanical properties, Journal of Applied Physics. 84 (1998) 6034–6041. [12] D. Gall, C.-S. Shin, T. Spila, M. Odén, M.J.H. Senna, J.E. Greene, I. Petrov, Growth of single-crystal CrN

on MgO(001): Effects of low-energy ion-irradiation on surface morphological evolution and physical properties, Journal of Applied Physics. 91 (2002) 3589.

[13] A.B. Mei, B.M. Howe, C. Zhang, M. Sardela, J.N. Eckstein, L. Hultman, A. Rockett, I. Petrov, J.E. Greene, Physical properties of epitaxial ZrN/MgO(001) layers grown by reactive magnetron sputtering, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 31 (2013) 061516.

12

[14] A. B. Mei, R.B. Wilson, D. Li, D.G. Cahill, A. Rockett, J. Birch, L. Hultman, J.E. Greene, I. Petrov, Elastic constants, Poisson ratios, and the elastic anisotropy of VN(001), (011), and (111) epitaxial layers grown by reactive magnetron sputter deposition, Journal of Applied Physics. 115 (2014) 214908.

[15] L.E. Toth, Transition Metal Carbides and Nitrides, Academic Press, Inc., New York, New York, 1971. [16] N. Schonberg, Contributions to the Knowledge of the Molybdenum-Nitrogen and the Tungsten-Nitrogen

Systems, Acta Chemica Scandanavica. 8 (1954) 204–207. [17] S. Wang, X. Yu, Z. Lin, R. Zhang, D. He, J. Qin, J. Zhu, J. Han, L. Wang, H. Mao, J. Zhang, Y. Zhao,

Synthesis, Crystal Structure, and Elastic Properties of Novel Tungsten Nitrides, Chemistry of Materials. 24 (2012) 3023–3028.

[18] M.L. Addonizio, A. Castaldo, A. Antonaia, E. Gambale, L. Iemmo, Influence of process parameters on properties of reactively sputtered tungsten nitride thin films, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 30 (2012) 031506.

[19] C.C. Baker, S.I. Shah, Reactive sputter deposition of tungsten nitride thin films, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 20 (2002) 1699.

[20] Y.G. Shen, Y.W. Mai, Microstructure and structure characteristics of cubic WNx compounds, Materials Science and Engineering A. 288 (2000) 47–53.

[21] T. Polcar, A. Cavaleiro, Structure, mechanical properties and tribology of W–N and W–O coatings, International Journal of Refractory Metals and Hard Materials. 28 (2010) 15–22.

[22] K.K. Shih, D.B. Dove, Properties of W–N and Mo–N films prepared by reactive sputtering, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 8 (1990) 1359.

[23] E.C. Samano, A. Clemente, J.A. Díaz, G. Soto, Mechanical properties optimization of tungsten nitride thin films grown by reactive sputtering and laser ablation, Vaccum. 85 (2010) 69–77.

[24] M. Wen, Q.N. Meng, W.X. Yu, W.T. Zheng, S.X. Mao, M.J. Hua, Growth , stress and hardness of reactively sputtered tungsten nitride thin films, Surface & Coatings Technology. 205 (2010) 1953–1961.

[25] M. Bereznai, Z. Toth, A.P. Caricato, M. Fernandez, A. Luches, G. Majni, P. Mengucci, P.M. Nagy, A. Juhasz, L. Nanai, Reactive pulsed laser deposition of thin molybdenum- and tungsten-nitride films, Thin Solid Films. 473 (2005) 16–23.

[26] P. Hones, N. Martin, M. Regula, L. Francis, Structural and mechanical properties of chromium nitride , molybdenum nitride , and tungsten nitride thin films, Journal of Physics D: Applied Physics. 36 (2003) 1023–1029.

[27] T. Yamamoto, M. Kawate, H. Hasegawa, T. Suzuki, Effects of nitrogen concentration on microstructures of WNx films synthesized by cathodic arc method, Surface and Coatings Technology. 193 (2005) 372–374.

[28] J.S. Becker, R.G. Gordon, Diffusion barrier properties of tungsten nitride films grown by atomic layer deposition from bis(tert-butylimido)bis(dimethylamido)tungsten and ammonia, Applied Physics Letters. 2239 (2003) 1–4.

[29] M. Nagai, K. Kishida, Thin film synthesis of tungsten nitride by the CVD method, Applied Surface Science. 70/71 (1993) 759–762.

[30] O.J. Bchir, S.W. Johnston, A.C. Cuadra, T.J. Anderson, C.G. Ortiz, B.C. Brooks, D.H. Powell, L. McElwee-White, MOCVD of tungsten nitride (WNx) thin films from the imido complex Cl4(CH3CN)W(NiPr), Journal of Crystal Growth. 249 (2003) 262–274.

[31] T. Nakajima, K. Watanabe, N. Watanabe, Preparation of Tungsten Nitride Film by CVD Method Using WF6, Journal of the Electrochemical Society: Solid State Science and Technology. 134 (1987) 3175–3178.

[32] S. Vepřek, M. Haussmann, S. Reiprich, Superhard nanocrystalline W2N/amorphous Si3N4 composite materials, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 14 (1996) 46.

[33] R. Deng, B.D. Ozsdolay, P.Y. Zheng, S. V. Khare, D. Gall, Optical and transport measurement and first-principles determination of the ScN band gap, Physical Review B. 91 (2015) 045104.

[34] J.S. Chawla, D. Gall, Epitaxial Ag(001) grown on MgO(001) and TiN(001): Twinning, surface morphology, and electron surface scattering, Journal of Applied Physics. 111 (2012) 043708.

[35] X.Y. Zhang, J.S. Chawla, R.P. Deng, D. Gall, Epitaxial suppression of the metal-insulator transition in CrN, Physical Review B. 84 (2011) 073101.

[36] X.Y. Zhang, J.S. Chawla, B.M. Howe, D. Gall, Variable-range hopping conduction in epitaxial CrN(001), Physical Review B. 83 (2011) 165205.

[37] X.Y. Zhang, D. Gall, Surface mound formation during epitaxial growth of CrN(001), Thin Solid Films. 518 (2010) 3813–3818.

[38] J.S. Chawla, X.Y. Zhang, D. Gall, Epitaxial TiN(001) wetting layer for growth of thin single-crystal Cu(001), Journal of Applied Physics. 110 (2011) 043714.

13

[39] J.S. Chawla, F. Gstrein, K.P. O’Brien, J.S. Clarke, D. Gall, Electron scattering at surfaces and grain boundaries in Cu thin films and wires, Physical Review B. 84 (2011) 235423.

[40] J.M. Purswani, T. Spila, D. Gall, Growth of epitaxial Cu on MgO(001) by magnetron sputter deposition, Thin Solid Films. 515 (2006) 1166–1170.

[41] R. Deng, S.R. Evans, D. Gall, Bandgap in Al1-xScxN, Applied Physics Letters. 112103 (2013) 1–5. [42] R. Deng, K. Jiang, D. Gall, Optical phonon modes in Al1−xScxN, Journal of Applied Physics. 115 (2014)

013506. [43] K. Zhang, K. Balasubramanian, B.D. Ozsdolay, C.P. Mulligan, S.V. Khare, W.T. Zheng, D. Gall, Growth

and physical properties of epitaxial NbN(001) films on MgO(001), Thin Solid Films, Under Review. (2015). [44] G.M. Pharr, W.C. Oliver, Measurement of Thin Film Mechanical Properties Using Nanoindentation, MRS

Bulletin. 17 (1992) 28–33. [45] Y.-T. Cheng, C.-M. Cheng, Relationships between hardness, elastic modulus, and the work of indentation,

Applied Physics Letters. 73 (1998) 614. [46] F.M. Smits, Measurement of Sheet Resistivities with the Four-Point Probe, Bell System Technical Journal.

37 (1958) 711–718. [47] X. Jiang, J.W. Zou, K. Reichelt, P. Grunberg, The study of mechanical properties of a-c:H films by Brillouin

scattering and ultralow load indentation, Journal of Applied Physics. 66 (1989) 4729–4735. [48] G. Abadias, M.B. Kanoun, S. Goumri-Said, L. Koutsokeras, S.N. Dub, P. Djemia, Electronic structure and

mechanical properties of ternary ZrTaN alloys studied by ab initio calculations and thin-film growth experiments, Physical Review B. 90 (2014) 144107.

[49] A. Kueny, M. Grimsditch, K. Miyano, I. Banerjee, C.M. Falco, I.K. Schuller, Anomalous Behavior of Surface Acoustic Waves in Cu/Nb Superlattices, Physical Review Letters. 48 (1982) 166–170.

[50] X. Jiang, M. Wang, K. Schmidt, E. Dunlop, J. Haupt, W. Gissler, Elastic constants and hardness of ion-beam-sputtered TiNx films measured by Brillouin scattering and depth-sensing indentation, Journal of Applied Physics. 69 (1991) 3053–3057.

[51] Y.G. Shen, Y.W. Mai, D.R. McKenzie, Q.C. Zhang, W.D. McFall, W.E. McBride, Composition, residual stress, and structural properties of thin tungsten nitride films deposited by reactive magnetron sputtering, Journal of Applied Physics. 88 (2000) 1380.

[52] L. Boukhris, J.-M. Poitevin, Electrical resistivity, structure and composition of d.c. sputtered WNx films, Thin Solid Films. 310 (1997) 222–227.

[53] P.-C. Jiang, J.S. Chen, Y.K. Lin, Structural and electrical characteristics of W–N thin films prepared by reactive rf sputtering, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 21 (2003) 616.

[54] P. Boher, P. Houdy, P. Kaikati, L.J. Van Ijzendoorn, Radio frequency sputtering of tungsten/tungsten nitride multilayers on GaAs, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 8 (1990) 846.

[55] G. Soto, W. de la Cruz, F.F. Castillon, J.A. Diaz, R. Machorro, M.H. Farıas, Tungsten nitride films grown via pulsed laser deposition studied in situ by electron spectroscopies, Applied Surface Science. 214 (2003) 58–67.

[56] P. van der Sluis, Determination of strain in epitaxial semiconductor layers by high-resolution X-ray diffraction, Journal of Physics D: Applied Physics. 26 (1993) A188–A191.

[57] C.-S. Shin, D. Gall, Y.-W. Kim, P. Desjardins, I. Petrov, J.E. Greene, M. Odén, L. Hultman, Epitaxial NaCl structure δ-TaNx(001): Electronic transport properties, elastic modulus, and hardness versus N/Ta ratio, Journal of Applied Physics. 90 (2001) 2879.

[58] H.-S. Seo, T.-Y. Lee, I. Petrov, J.E. Greene, D. Gall, Epitaxial and polycrystalline HfNx (0.8≤x≤1.5) layers on MgO(001): Film growth and physical properties, Journal of Applied Physics. 97 (2005) 083521.

[59] D. Gall, I. Petrov, J.E. Greene, Epitaxial Sc1-xTixN(001): Optical and electronic transport properties, Journal of Applied Physics. 89 (2001) 401.

[60] F. Tian, J. D’Arcy-Gall, T.-Y. Lee, M. Sardela, D. Gall, I. Petrov, J.E. Greene, Epitaxial Ti1-xWxN alloys grown on MgO(001) by ultrahigh vacuum reactive magnetron sputtering: Electronic properties and long-range cation ordering, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 21 (2003) 140.

[61] R.P. Deng, P.Y. Zheng, D. Gall, Optical and electron transport properties of rock-salt Al1-xScxN, Journal of Applied Physics. (2015) submitted.

[62] P. Jiang, Y. Lai, J.S. Chen, Dependence of crystal structure and work function of WNx films on the nitrogen content, Applied Physics Letters. 89 (2006) 122107.

14

[63] J. Lin, A. Tsukune, T. Suzuki, M. Yamada, Different effect of annealing temperature on resistivity for stoichiometric, W rich, and N rich tungsten nitride films, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 17 (1999) 936.

[64] W.D. Nix, J.R. Greer, G. Feng, E.T. Lilleodden, Deformation at the nanometer and micrometer length scales: Effects of strain gradients and dislocation starvation, Thin Solid Films. 515 (2007) 3152–3157.

[65] Z. Xue, Y. Huang, K.C. Hwang, M. Li, The Influence of Indenter Tip Radius on the Micro-Indentation Hardness, Journal of Engineering Materials and Technology. 124 (2002) 371.

[66] J.R. Greer, W.C. Oliver, W.D. Nix, Size dependence of mechanical properties of gold at the micron scale in the absence of strain gradients, Acta Materialia. 53 (2005) 1821–1830.

[67] M. Alderighi, V. Ierardi, F. Fuso, M. Allegrini, R. Solaro, Size effects in nanoindentation of hard and soft surfaces., Nanotechnology. 20 (2009) 235703.

[68] C.L. Bull, T. Kawashima, P.F. McMillan, D. Machon, O. Shebanova, D. Daisenberger, E. Soignard, E. Takayama-Muromachi, L.C. Chapon, Crystal structure and high-pressure properties of γ-Mo2N determined by neutron powder diffraction and X-ray diffraction, Journal of Solid State Chemistry. 179 (2006) 1762–1767.

[69] J.J. Vlassak, W.D. Nix, Measuring the elastic properties of materials by means of indentation experiments, Journal of the Mechanics and Physics of Solids. 42 (1994) 1223–1245.

[70] J.J. Vlassak, W.D. Nix, Indentation modulus of elastically anisotropic half spaces, Philosophical Magazine A. 67 (1993) 1045–1056.

[71] W. Pang, a. G. Every, J.D. Comins, P.R. Stoddart, X. Zhang, Brillouin scattering from acoustic excitations in TiN films on high speed steel—A stiffening system, Journal of Applied Physics. 86 (1999) 311.

[72] K.M. Knowles, P.R. Howie, The Directional Dependence of Elastic Stiffness and Compliance Shear Coefficients and Shear Moduli in Cubic Materials, Journal of Elasticity. (2014) 1–22.

[73] R. Hill, The Elastic Behaviour of a Crystalline Aggregate, Proceedings of the Physical Society of London. 65 (1952) 349–354.

[74] A.J. Perry, A contribution to the study of Poisson’s ratios and elastic constants of TiN, ZrN and HfN, Thin Solid Films. 193-194 (1990) 463–471.

[75] J.O. Kim, J.D. Achenbach, P.B. Mirkarimi, M. Shinn, S.A. Barnett, Elastic constants of single-crystal transition-metal nitride films measured by line-focus acoustic microscopy, Journal of Applied Physics. 72 (1992) 1805–1811.

[76] Y. Liang, X. Yuan, W. Zhang, The effect of magnetostructural coupling on mechanical behaviors in CrN, Solid State Communications. 150 (2010) 2045–2048.

[77] Z.T.Y. Liu, X. Zhou, S. V Khare, D. Gall, Structural, mechanical and electronic properties of 3d transition metal nitrides in cubic zincblende, rocksalt and cesium chloride structures: a first-principles investigation., Journal of Physics. Condensed Matter : an Institute of Physics Journal. 26 (2014) 025404.

[78] S. Jhi, J. Ihm, S.G. Louie, Electronic mechanism of hardness enhancement in transition-metal carbonitrides, Nature. 399 (1999) 1–3.

15

Figures

Figure 1: Sections of XRD ω-2θ scans from 1.45 μm thick WN layers deposited on (a) MgO(001), (b) MgO(111), and (c) Al2O3(0001). The solid lines are obtained using a monochromatic parallel x-ray source while the dashed curve in (c) is obtained with a divergent beam geometry.

16

Figure 2: XRD ω-rocking curves of the WN 111 reflection from (a) WN/MgO(111) and (b) WN/Al2O3(0001) layers, 113 φ-scans from (c) WN/MgO(111) and (d) WN/Al2O3(0001) layers, and 113 reciprocal space maps from (e) WN/MgO(111) and (f) WN/MgO(001) layers.

17

Figure 3: (a) Elastic modulus E and (b) hardness H vs indentation contact depth from nanoindentation measurement on 1.45 μm thick WN layers deposited on MgO(001), MgO(111), and Al2O3(0001).

Figure 4: Brillouin scattering intensity vs frequency shift Δf, from WN layers deposited on MgO(001), MgO(111), and Al2O3(0001). The arrows indicate the positions of Stokes and anti-Stokes peaks.