Influence of Al content on the phase formation, growth stress and mechanical properties of TiZrAlN coatings

Thin Solid Films 538 (2013) 32–41

Contents lists available at SciVerse ScienceDirect

Thin Solid Films journal homepage: www.elsevier.com/locate/tsf

Influence of Al content on the phase formation, growth stress and mechanical properties of TiZrAlN coatings I.A. Saladukhin a, b, G. Abadias a,⁎, A. Michel a, S.V. Zlotski b, V.V. Uglov b, G.N. Tolmachova c, S.N. Dub d a

Institut P', Université de Poitiers-CNRS-ENSMA, Département Physique et Mécanique des Matériaux, SP2MI, Téléport 2, F86962 Chasseneuil-Futuroscope, France Belarusian State University, 220030 Minsk, Belarus Kharkov Institute of Physics and Technology, 61108, Kharkov, Ukraine d Institute for Superhard Materials, NAS of Ukraine, 04074 Kiev, Ukraine b c

a r t i c l e

i n f o

Available online 7 January 2013 Keywords: Nitrides Quaternaries Phase formation Nanocomposite films TiZrAlN Hardness

a b s t r a c t Quaternary (Ti,Zr)1−xAlxN transition metal nitride films, with Al content x ranging from 0 to 0.37, were reactively sputter-deposited from individual metallic targets under Ar+N2 plasma discharges on Si substrates at Ts =270 °C. The influence of Al addition on the crystal structure, phase formation, growth morphology and intrinsic stress development, electrical and mechanical properties was systematically investigated. Three distinct compositional regions were evidenced: i) for 0≤x≤0.07, films develop a columnar structure consisting of cubic TiZr(Al)N grains with (111) and (200) preferred orientation, large compressive stresses up to ~−4 GPa and hardness increase from ~20 to ~24 GPa, ii) for 0.09≤x≤0.16, Al incorporation favors the growth of nanocomposite films consisting of (200)-oriented cubic TiZr(Al)N nanocrystals surrounded by a highly-disordered matrix, accompanied by a decrease of compressive stress, whereas a maximum hardness H~27 GPa and H/E ratio of 0.105 is reached at x~0.12 and x=0.14, respectively, and iii) x>0.16, XRD amorphous films are formed, with reduced mechanical properties. The structure–stress-properties relationship is discussed based on evolutionary growth regimes induced by incorporating a high-mobility metal in a refractory compound lattice. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Multicomponent alloying of transition metal (TM) nitride systems is attracting considerable interest to improve further the performance of hard and wear resistant coatings [1–6]. In particular, alloying TiN, ZrN or CrN with AlN is known to increase oxidation resistance, while retaining high hardness. But these favorable properties depend on the crystal structure of as-grown TM–Al–N thin films [7–15] and subsequent phase transformations occurring under thermal work load conditions, e.g. during cutting or milling operations [9,10,15–20]. It is well known that metastable cubic (c) solid solution Ti1−xAlxN films with NaCl structure can be synthesized by physical vapor deposition, such as cathodic arc evaporation [7,8] or magnetron sputtering [9–12], for AlN mole fractions up to 0.6–0.7, while higher Al contents (x>0.7) favor the hexagonal ZnS-wurtzite (w) structure, resulting in lower mechanical properties. Ab initio calculations [21–23] confirmed that the maximum (metastable) solubility limit of AlN in c-Ti–Al–N is ~ 0.7, while the predicted existence field of the cubic phase in the isovalent Zr1−xAlxN and Hf1 −xAlxN systems is comparatively reduced

⁎ Corresponding author at: Institut P', Dpt Physique et Mécanique des Matériaux, Université de Poitiers, SP2MI, Téléport 2, Bd Pierre et Marie Curie, 86962 ChasseneuilFuturoscope, France. Tel.: +33 5 49 49 67 48; fax: +33 5 49 49 66 92. E-mail address: [email protected] (G. Abadias). 0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2012.12.090

to x ≤ 0.5 [23,24]. Strategies aiming at stabilizing the cubic structure of Ti1 − xAlxN films at higher Al content have been proposed [25,26], among which substitutional alloying with TM like Cr [1,2,21], Ta [5,27], Zr [21,28,29] or Hf [3,4] has received recent interest. In addition to improved thermal stability (age hardening), quaternary Ti–Al–TM–N films with TM= W or Mo are also predicted to exhibit toughness enhancement [6]. Ternary Ti–Zr–N thin films grown by vacuum arc [30,31] or magnetron sputtering [32] have shown to exhibit enhanced hardness compared to TiN, due to solid solution strengthening. But the oxidation behavior was not improved compared to TiN, since the ternary alloys were found to fully oxidize at temperatures between 500 and 600 °C [33]. Similarly to TiAlN, it is therefore expected that addition of Al into TiZrN would yield a higher oxidation resistance by promoting the growth of an outer, passive Al2O3 layer. In a recent study, Chen et al. [29] have shown that incorporation of Zr (up to z = 0.29) in Ti1 − x−zAlxZrzN (with 0.37 ≤ x ≤ 0.55) delays the formation of detrimental w-AlN phase upon annealing in vacuum and assists the formation of a dense oxide scale for z = 0.05. However, a systematic investigation of phase formation upon alloying Al to Ti–Zr–N has not yet been reported. In the present work, we use in situ stress measurements, X-ray diffraction (XRD) and X-ray reflectommetry (XRR), transmission electron microscopy (TEM), electrical resistivity and nanoindentation measurements to study the influence of Al incorporation on the structure, phase

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

formation, stress development, electrical and mechanical properties of magnetron sputtered (Ti,Zr)1−xAlxN films with x ranging from 0 to 0.37. 2. Experimental details 2.1. Thin film growth and real-time stress evolution (Ti,Zr)1 − xAlxN thin films were deposited on (001) Si wafer covered with native (~ 2 nm) oxide using reactive unbalanced magnetron co-sputtering under Ar + N2 plasma discharges. Deposition was carried out at the substrate temperature Ts = 270 °C in a high vacuum chamber (base pressure ~ 10 −5 Pa) equipped with 7.5 cm diameter water-cooled planar magnetron sources. Metallic Ti (99.995% purity), Zr (99.2% purity) and Al (99.999% purity) targets, placed in a confocal configuration at a distance of 18 cm from the substrate holder, were used for co-sputtering. Prior to deposition, all targets were sputtercleaned for 3 min in pure Ar plasma discharge, while the substrate was shielded by a shutter. The atomic fraction of Al relative to metallic elements, x, in the films was varied from 0 to 0.37 by tuning the rf power supply of the Al target from 0 to 200 W, while maintaining the dc power supply of Ti and Zr targets constant at 300 and 220 W, respectively. The Ti and Zr target powers were chosen so as to obtain a Ti:Zr concentration ratio in the synthesized coating of about 1:1. A constant bias voltage of −56 V was applied to the substrate using a rf power supply. The working pressure was fixed at 0.20 Pa (10 sccm of Ar flow), while the N2 partial pressure was systematically adjusted using MKS MicroVisionPlus mass spectrometer to be in a range of 1.8–2.5× 10−3 Pa, corresponding to optimized conditions to synthesize stoichiometric TiZrN films in metallic target mode [32]. The substrate stage was rotated at 15 rpm to ensure thickness and composition uniformity during all deposition. Details on the deposition process parameters are reported in Table 1. The film thickness hf was varied between ~50 to ~350 nm by adjusting the deposition time. The growth rate (R) was determined from accurate determination of film thickness using X-ray Reflectivity (XRR) measurements (see Section 2.2) on a ~50 nm thick film series (total deposition time of 150 s). A multi-beam optical stress sensor (MOSS) developed by kSA was implemented into the deposition chamber, so that in situ stress measurements could be carried out, enabling one to obtain real time information on the stress evolution during growth [34]. The MOSS relies on the monitoring of the spacings between an array (3 × 3 in our geometry) of laser beams reflected off the sample surface and recorded on a CCD camera. The change in curvature, Δκ, is deduced from the mean value of the differential spot spacing, δD/D0, using the following expression, valid for small angle approximation, δD cosα , where α is the incident angle of the laser beam meaΔκ = D0 2L sured with respect to the sample normal and L the optical path length from the substrate to the CCD camera. The geometrical factor

33

cos α/2L was calibrated using a set of flat and reference mirrors with known radius of curvature. The change in curvature is related to the average biaxial film stress, σavg, according to the Stoney equation 1 2 σ avg hf ¼ Δκ M s hs , where hf is the film thickness, hs the substrate 6 thickness and Ms the biaxial elastic modulus of the substrate (Ms =180.5 GPa for Si (001)). By convention, a negative value of σavg corresponds to a compressive stress. The acquisition time, in static mode (no rotation of the substrate holder), was 0.8 s, corresponding to an average of 4 data points. With our own experimental set-up, a typical curvature resolution of 0.2 km−1 is attained, giving a stress sensitivity of 0.29 N/m for a Si substrate with hs = 200 μm. 2.2. Microstructural characterization and electrical resistivity X-ray diffraction (XRD) was employed for structural identification using a D8 Bruker AXS X-ray diffractometer operating in Bragg–Brentano configuration and equipped with CuKα1 wavelength (0.15418 nm) and an energy dispersive Si(Li) detector (Sol-X detector) defined with a 0.2 mm opening angle slit. Appearance of new phases was analyzed by deconvolution procedure of diffraction lines using Lorentz's function. The grain size was evaluated according to the broadening of the (200) diffraction line using the Scherrer's equation, i.e. ignoring in a first approximation the contribution of microstrain. Integrated intensities, Ihkl, of the (111) and (200) peaks of (Ti,Zr)1−xAlxN films were normalized with respect to the corresponding integrated intensities of the Al-free film (x= 0), after background subtraction. The (200) texture coefficient, I ∗ T200, was calculated according to T 200 ¼  200  where Ihkl refer to I200 þ I111 normalized intensity. The atomic fraction of metallic elements (Ti, Zr and Al) in films deposited on glassy carbon substrates was determined by energy dispersive X-ray spectroscopy (EDX) with a global accuracy of ±2%, using an Oxford Instruments AZTek EDX unit attached to a JEOL 7001F-TTLS scanning electron microscope (SEM) operated at 20 kV. The nitrogen concentration was quantified by the method of Rutherford backscattering (RBS) using He + ions with the energy of 1.6 MeV at the High Voltage Engineering tandetron system accelerator. The obtained spectra were fitted using SIMNRA software. The mass density, ρ, film thickness and surface and interface roughness were quantitatively determined from XRR measurements performed on a Seifert XRD 3000 apparatus using a channel-cut Ge (220) monochromator to select the CuKα1 line (0.15406 nm) exclusively, with a low divergence. An optical model based on Parratt's formalism [35] and assuming a three-layer stacking (topmost surface layer, film layer and film/substrate interface layer) was used for the refinement procedure. The relative error in the determination of film thickness and mass density from XRR is estimated to be ±1% and ±3%, respectively. Studies of the film's surface topography were performed in a Multimode Digital Instrument atomic force microscope (AFM) working in tapping mode, while fracture

Table 1 Process parameters and elemental composition determined by EDX and RBS of as-deposited magnetron sputtered TiZrAlN coatings at fixed working pressure (0.20 Pa), substrate temperature Ts = 270 °C and bias voltage Vs = −56 V. The dc power supply of Ti and Zr targets was fixed at 300 and 220 W, respectively. Al target RF power supply (W)

Al target discharge voltage (V)

N2 flow (sccm)

pN2 (mPa)

Growth rate, R (Å/s)

[Ti]/([Ti] + [Zr]) (EDX)

[Al]/([Al] + [Ti] + [Zr]) (EDX)

N concentration, at.% (RBS)

0 20 30 40 50 60 70 80 100 120 200

0 130 155 185 205 230 250 274 310 350 480

1.3 1.3 1.3 1.4 1.4 1.5 1.5 1.5 1.5 1.6 1.6

1.8 1.9 1.9 1.8 2.2 2.0 2.3 1.9 2.0 2.2 2.4

3.1 3.1 3.1 / 3.2 3.2 / / / 3.6 3.9

0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.49

0 0.03 0.05 0.07 0.09 0.12 0.14 0.16 0.21 0.25 0.37

50 54 / / / / 42 / / 33 28

34

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

cross-sectional SEM observations were carried out at 20 kV to reveal the growth morphology. Plan-view TEM was performed on selected samples, of two types. A thin (~ 60 nm) TiZrAlN film was deposited either directly on a TEM copper grid covered with an amorphous SiOx-on-carbon membrane, or on crystalline rock-salt. The first type of samples could be directly observed, while it was necessary for the second type to dissolve the NaCl in distilled water in order to collect the thin film on a Cu grid. A JEOL 3010 ARP microscope, operating at 300 kV was used to observe the microstructure of the thin films, in either conventional bright field, selected area electron diffraction (SAED) or high resolution (HR) mode (point resolution 0.185 nm). In this latter case, as the (200) and (111) interplanar distances of the nitride system are above the point resolution limit, the images obtained were directly interpreted without further simulation. The film resistivity was measured using a standard four-point probe technique, with typical currents of 20 to 60 μA.

Log Intensity (arb. units)

fit x=0.05 (data) x=0.25 (data) 0.5

1

1.5

2.5

3

3.5

Al rf power (W) 0

3. Results

2

2 θ (Deg.)

2.3. Nanoindentation 30

60

90

120

200

8.0

b)

ZrN

mass density,ρ (g/cm3)

The mechanical properties of the thicker (Ti,Zr)1−xAlxN films (hf ~320–350 nm) deposited on Si substrates (hs ~650 μm) were studied by nanoindentation using a Nano Indenter-G200 system (Agilent Technologies, USA) equipped with a continuous stiffness measurement attachment option. This attachment offers a continuous measurement of the contact stiffness via a superimposed alternating current signal during loading, which in turn provides a continuous measurement of the elastic modulus E and average contact pressure (ACP)—Meyer's hardness H as function of the penetration depth during a single loading segment [36]. Eight indentations were made on each sample. A diamond Berkovich tip with some tip blunting was used. AFM measurements of the tip shape have been made, and the results showed that the Berkovich tip can be described as a sphere with an effective radius R of 230 nm when the indentation depth is less than 30 nm. Load (P) and displacement (h) were continuously recorded up to a maximum displacement of 200 nm at a constant indentation strain rate of 0.05 s−1. The values of E and H were determined from the ACP–h and E–h data, and corrected according to tip blunting using real area function calibrated on reference sample (fused silica). A Poisson ratio value of ν=0.25, typical for such TM nitrides [6], was assumed for calculation of E.

a)

XRR TiZrAlN films

7.0

6.0

5.0

TiN

4.0

c-AlN w-AlN

3.0

0

0.1

0.2

0.3

0.4

0.5

Al content, x Fig. 1. a) Typical XRR scans (symbols) of ~50 nm thick (Ti,Zr)1−xAlxN films deposited on Si with x = 0.05 (30 W) and x = 0.25 (120 W) and corresponding best-fit curves (red line) using optical formalism of Parratt. b) Evolution of the mass density, ρ, as a function of x content. Solid symbols refer to quaternary (Ti,Zr)1 − xAlxN films, while empty symbols correspond to mass density of TiN and ZrN films deposited under similar growth condition (see Ref. [32]). The dashed line is a linear interpolation between c-TiZrN and c-AlN mass densities.

3.1. Structural and morphological evolution with Al content Fig. 1a shows representative XRR scans of (Ti,Zr)1−xAlxN films with x = 0.05 and 0.25 together with the best-fit scans. Well defined Kiessig fringes are observed, attesting of the rather low surface roughness of these layers. The quantitative adjustment of the theoretical model to experimental XRR scans yields surface roughness less than ~1 nm for a total film thickness of ~50 nm. The growth rate is found to increase from R = 3.1 to 3.9 Å/s when the rf power supply delivered to the Al target increases in the range 0–200 W (see Table 1). Note that with increasing Al power supply from 0 to 200 W, the [Ti]/([Ti] + [Zr]) ratio was found to remain almost unchanged and equal to 0.50. The N content was found to be around 50 at.% for 0 ≤ x ≤ 0.09, while for higher Al contents, a N depletion was observed in the films. In the present sputtering conditions, the N2 partial pressure was fixed at ~2×10−3 Pa, corresponding to optimized conditions to obtain stoichiometric TiZrN films [32]. As Al has less affinity with N atoms than Ti or Zr, a higher N2 partial pressure would be required to form stoichiometric AlN-rich TiZrAlN films. The evolution of the film mass density, ρ, is reported in Fig. 1b as a function of Al target power and corresponding Al atomic fraction, x. For x = 0, the mass density of the TiZrN film is found to be 6.45 g/cm3, close to the value of 6.34 g/cm3 extrapolated for a Ti:Zr ratio of 1:1

from TiN and ZrN reference powders. A gradual decrease in ρ is observed with increasing Al content from x = 0 to x =0.37. However, the obtained values are lower than the linear interpolation between c-TiZrN and c-AlN (see dashed line in Fig. 1b), suggesting that the coordination number of metals (and related electronic structure) is altered upon incorporation of Al in the TiZrN lattice. It may be anticipated from this evolution that cubic TiZrAlN solid solutions are not stabilized over the whole Al-range investigated, and a tendency towards a more covalent bonding character as Al content is increased (ρw-AlN b ρc-AlN). The evolution of XRD patterns of ~ 150 nm thick (Ti,Zr)1 − xAlxN (x = 0.03 to 0.25) films is reported in Fig. 2. For low Al contents, x = 0.03 to 0.09 (Fig. 2a), diffraction lines are found at 2θ ~ 35.0° and 40.6°, corresponding to (111) and (200) Bragg reflections from c-TiZrAlN. Note that no significant diffracted intensity could be detected in the angular regions corresponding to (220) and (311) reflections. A close inspection on the shape and angular position of the (111) and (200) peaks reveals fine structural evolutions with Al content. A stronger decrease in the intensity of the (111) XRD line is initially observed upon Al incorporation. At x = 0.07 the intensity of both (111) and (200) lines drops and, afterwards, the (200) line rises again in intensity for 0.09 ≤ x ≤0.16. This result suggests a change in preferential

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

1.0

a)

a) Texture coefficient, T200

Intensity (arb. units)

35

0.8

0.6

0.4

0.2

region 1

region 2

0 0

0.05

0.10

0.15

0.20

Al content, x

34

36

38

40

42

44

0.470

20

b)

2 θ (Deg.)

0.460

b) Intensity (arb. units)

Grain size (nm)

15 0.450

0.440

10

0.430

5

lattice parameter (nm)

32

0.420

region 1

region 2 0.410

0 0

0.05

0.10

0.15

0.20

Al content, x

32

34

36

38

40

42

44

2 θ (Deg.) Fig. 2. XRD patterns of ~150 nm thick (Ti,Zr)1−xAlxN films with various Al contents x. The vertical dashed lines refer to the (111) and (002) reflections of metastable c-TiZrN films (x = 0). Also indicated are the Bragg reflections of wurtzite w-AlN (JCPDS card no. 25-1133) and cubic c-AlN (JCPDS card no. 25-1495) phases.

grain orientation with Al incorporation. As shown in Fig. 3a, it is clear that increased values of x promote the formation of a (200) preferred orientation. From this figure, the transition from region 1 to region 2 is revealed at x ~ 0.07. This change in preferential orientation is accompanied by a significant broadening of the (200) peak, especially for x = 0.12 and 0.14 (Fig. 2b). With the further increase of Al content, the scattered intensity becomes very broad in the 34–42° angular region. For x = 0.25, one can also observe a shift in angular position towards lower angles. Based on additional samples not shown here, it can be suggested that (Ti,Zr)1−xAlxN films are ‘XRD amorphous’ at x >0.16, defining region 3. Examples of XRD patterns deconvolution for various x contents are given in Fig. 4. The XRD pattern of the film with x=0.03 (Fig. 4a) is representative of a single TiZr(Al)N phase with rocksalt structure, while an additional broad contribution with weak intensity, centered around 2θ~ 37°, is required to reproduce the intensity profile of the films with x=0.07 and 0.12 in the region between the main cubic (111) and (200) lines (Fig. 4b and c). Reflections from w-AlN (JCPDS card no. 25-1133) and metastable c-AlN (JCPDS card no. 25-1495) phases exist in the 36–38° angular region; however, due to the lack of c-AlN (200) reflection at ~44°, it is unlikely to consider the isostructural decomposition into two cubic phases, i.e. TiZrN-rich and AlN-rich c-TiZrAlN structures. A

Fig. 3. Influence of Al content, x, on (Ti,Zr)1−xAlxN film texture (a), lattice parameter and grain size of (200)-oriented c-TiZr(Al)N grains (b). The bulk lattice parameter of binary compounds with cubic structure (data from JCPDS files) is also reported on the right-hand axis.

more plausible scenario would be the formation of a nanocomposite film, consisting of cubic and hexagonal phases (the corresponding contributions are shown as green solid lines in Fig. 4b and c). This additional phase is referred hereafter to ‘hex-nanophase’, as its corresponding XRD line is very broad (5–6°). For x=0.16 (Fig. 4d), the asymmetric shape of the broad peak centered at 2θ~ 39° is satisfactorily reproduced when considering the contribution of this ‘hex-nanophase’ having reflections around the (0002) and (1011) lines of w-AlN, in addition to c-TiZr(Al)N. A shift of the (200) XRD line position of c-TiZr(Al)N towards lower angles is observed when x rises from 0 to 0.09 (see Figs. 2а and 4). The position of the (111) XRD line seems comparatively less sensitive to the Al content. This could be due to two overlapping contributions, the one from (111) c-TiZr(Al)N and the second from (0002) w-AlN diffraction lines (Fig. 4b). The out-of-plane lattice parameter of the cubic TiZr(Al)N grains was therefore calculated from the position of the (200) peak only. A gradual increase in the lattice parameter of the c-TiZr(Al)N solid solution is clearly observed in regions 1 and 2 (see Fig. 3b). Since a decrease of the lattice parameter would be expected upon incorporation of Al in c-TiZrN (a = 4.10 Å for metastable c-AlN [37] and a = 4.45 Å for metastable c-TiZrN [38]), the observed lattice expansion would suggest either an increase of compressive stress or a Zr enrichment with increased x values. As seen from Fig. 2, the addition of Al during growth of TiZrAlN films leads to the broadening of the (002) diffraction peak of the TiZr(Al)N

36

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

a)

32

34

36

38

40

42

44

b)

46

32

34

36

2θ (deg.)

38

40

42

c)

32

34

36

38

40

44

46

2θ (deg.)

42

44

2θ (deg.)

d)

46

32

34

36

38

40

42

44

46

2θ (deg.)

Fig. 4. Deconvolution procedure of XRD patterns of ~150 nm thick (Ti,Zr)1−xAlxN films for Al content x = 0.03 (a), x = 0.07 (b), x = 0.12 (c) and x = 0.16 (d). Vertical dashed lines refer to w-AlN (JCPDS card no. 25-1133), c-ZrN (JCPDS card no. 35-753) as well as metastable c-TiZrN film (x = 0), while green solid lines show the contribution of c-TiZr(Al)N and ‘hex-nanophase’ to total diffracted intensity.

solid solution. Scherrer's equation was used to obtain a rough estimate of the grain size of the cubic phase. A pronounced decrease from 16 to 3 nm is observed in the 0–0.16 Al content range (Fig. 3b). Let us mention that these data are lower bound values of the grain size, since other contributions of XRD line broadening, like e.g., microstrain, are here neglected. The as-deposited films exhibit distinct microstructural features with increasing Al content. Cross-sectional SEM views (not reported here) indicate a change from columnar to more globular morphology, from region 1 to region 3. Films with low Al content (x≤ 0.09) exhibit the characteristics of a competitive columnar growth, reported for magnetron sputtered TiN films, with rough, facetted top surfaces. Some triangular-shaped mounds are visible in the AFM surface topography image, corresponding to the emergence of (111) columns with threefold symmetry. A drastic change is observed for the film with x = 0.16 (region 3): it exhibits a smooth, featureless surface (Fig. 5b), consistent with the growth of a nanocrystalline/nanocomposite films. The change in microstructure from region 1 to region 3 is also clear from plan-view TEM observations, of which Fig. 6 shows two selected (Ti,Zr)1−xAlxN samples with x = 0 (region 1) and x = 0.14 (region 2),

a)

b)

Fig. 5. Surface topography AFM images of ~330 nm thick (Ti,Zr)1−xAlxN films with a) x = 0.09 and b) x = 0.16.

respectively, The microstructure of the Al-free sample (Fig. 6a) is remarkable, since the grain boundaries are systematically associated with a bright contrast, indicating low-density grain boundaries with the possible formation of voids. This feature disappears as soon as a small Al content (region 1) is introduced, giving a hint of a preferential Al incorporation at such boundaries. Increase in Al content is also associated with a drastic decrease in grain size, from average values of ~10 nm (x= 0) to ~5 nm (x≈ 0.14), and subsequently with a progressive amorphization of the whole thin film (region 3). This effect can also be deduced from the SAED patterns (aperture diameter ~200 nm), as the dotted diffraction rings at low Al content gradually evolve towards uniform and broad rings. The film with x = 0.14 (Fig. 6b) consists of crystalline grains embedded in a surrounding highly disordered or ‘amorphous’ matrix. A close inspection of the image contrast suggests that the matrix is such that no crystalline planes can be defined despite the short range order, which leads to the description with the term “amorphous”. However, SAED and HR studies also highlight another microstructural difference, namely the change of texture. Although both 111 and 200-type growth directions are present at low Al content (see inset of Fig. 6a), it can be seen that the (200) texture is predominant at higher Al content. Indeed, in diffraction mode, the increased 200-ring intensity (see inset of Fig. 6b) is the signature for such an (200) texture, as the reduction in intensity of the 220-ring is attributed to reduced 111 growth orientation, and high resolution imaging gives an example of a 200 oriented grain (Fig. 6d). HRTEM further reveals that the grain boundary is not sharp, as coexistence with the disordered matrix can be seen locally (Fig. 6d). Further TEM observations show that samples are “fully amorphous” when Al content is increased above x = 0.16, with a noticeable evolution of the short-range order. Therefore, based on XRD and TEM observations, it is proposed that (Ti,Zr)1−xAlxN films with 0.12≤ x ≤ 0.16 exhibit a dual-phase nanocomposite structure, consisting of (200)-oriented cubic TiZr(Al)N nanograins surrounded by a highly disordered matrix. However, the exact structure and composition of this ‘matrix’ are yet unknown.

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

37

0

Fig. 6. Plan-view TEM images of ~60 nm thick (Ti,Zr)1 −xAlxN films. (a) At x = 0, the bright-field images show elongated grains having bright contrast, like a white border, along the grain boundaries; the corresponding SAED and its linear intensity profile is given in the inset. The white frame corresponds to the area shown in detail in Fig. 6c. (b) At x=0.14, small grains which are in a zone axis condition appear as dark contrast, inside a mostly amorphous matrix (see SAED pattern with intensity profile in inset). The area inside the white square is enlarged in Fig. 6.d. (c) HR image of small disoriented grains at x=0; d) HR image of preferentially 200-oriented grains at x=0.14, inset: fast Fourier transform (FFT) of the grain covering the right-hand side of the image: the square-geometry dots in the FFT correspond to the {200}-type planes perceptible in the image.

3.2. Stress evolution The real-time stress evolution during film growth was studied using MOSS. Fig. 7a shows the average film stress, σavg, as a function of the deposited thickness, hf, for various x contents of (Ti,Zr)1−xAlxN films. A compressive stress development is observed for all films in the early growth stages (hf b 10 nm). With increasing film thickness, three distinct stress behaviors, characteristics of the regions 1, 2 and 3, respectively, are identified. Stress gradients are found for films of regions 1 and 2, while XRD-amorphous films (region 3) have a uniform compressive stress over hf. In region 1, after the initial compressive stress build-up, |σavg| reaches a maximum at hf ~ 50 nm whose magnitude depends on the Al atomic fraction, and then decreases with further increase of hf.

Fig. 7. a) Evolution of average stress, σavg, determined from in situ MOSS technique, during sputter-deposition of (Ti,Zr)1−xAlxN films for representative Al content, x. b) Evolution of σavg as a function of Al content, calculated for hf = 150 nm. The solid line is a guide to the eye.

Films of region 2 exhibit an opposite tendency: the compressive stress steadily increases with hf. To better visualize the influence of Al content on the stress magnitude, the average film stress is plotted vs. x, for a fixed hf =150 nm value (Fig. 7b). A non-monotonous Al content dependency is observed. In region 1, the compressive stress increases with Al content, and a sharp maximum of |σavg| ~4 GPa is reached at x~0.07, i.e. at transition from region 1 to region 2. With further increase of Al content in region 2, the compressive stress decreases by a factor of two, reaching ~2 GPa at the transition from region 2 to region 3. In region 3, a subsequent, though less abrupt compressive stress decrease is found with increasing Al content. 3.3. Electrical resistivity Fig. 8 shows the evolution of electrical resistivity, ρel, of (Ti,Zr)1−xAlxN films as a function of Al content. The value of ρel =72 μΩ.cm obtained for x=0 is consistent with previous values reported for magnetron sputtered

38

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

polycrystalline c-TiZrN films [32] and confirms the metallic character of these ternary alloys. A slight decrease of ρel is first observed at the lowest Al fractions (x=0.03 and 0.05); then ρel is found to increase, in particular a sharp transition from 115 to 230 μΩ.cm occurs between x=0.12 and x=0.14 (region 2). In region 3, the electrical resistivity appears little unaltered upon further Al incorporation. It remains far below the resistivity of bulk wurtzite AlN (ρel ~1015 μΩ.cm), which is a wide bandgap insulator [39].

3.4. Hardness evolution Typical load (P)–displacement (h), ACP−h and E–h curves for hard (Ti,Zr)1 − xAlxN coatings are shown in Fig. 9a and b, respectively. It can be seen that ACP increases linearly with h at the beginning of loading. The dependence of ACP vs. h changed at a displacement higher than 25 nm (point 1 in Fig. 9b), since a leveling off is noticed above point 1. ACP reaches a maximum value of about 27 GPa at h = 60 nm (point 2 in Fig. 9b) and then begins to decrease at displacements higher than 100 nm. Such dependence of ACP on displacement is due to changes of contact deformation regimes (elastic – elastic–plastic – full plasticity) with increasing penetration depth [40]. At h >60 nm the regime of full plasticity is reached and the hardness determination becomes possible. On the other hand, hardness is minimal distorted by substrate effect at h b 80 nm (Fig. 9b). So the hardness of the coatings was determined at the displacement range from 60 to 80 nm. The elastic modulus is more sensitive to the influence of low modulus silicon substrate (Fig. 9b). Therefore the elastic modulus E of the coating was measured at the displacement of about 25 nm. In Oliver and Pharr analysis of load–displacement curve it is supposed that the sample is elastically homogeneous [41]. In the case of 350 nm thick TiZrAlN film (E is about 300 GPa) on Si substrate (ESi ~170 GPa) the sample is not elastically homogeneous and we observed the coupled response of the film and more compliant substrate [42]. So, the hardness reported in Fig. 10 is the composite hardness of TiZrAlN thin film and Si substrate at the displacement of about 60 nm. The intrinsic coatings hardness is likely higher. The nanoindentation hardness and elastic modulus evolutions of (Ti,Zr)1 − xAlxN coatings with an increase in the Al content, x, are presented in Fig. 10a. Chemical composition and microstructure (volume fraction of nanocrystalline and amorphous phase) have significant effect on the mechanical properties of TiZrAlN coatings. In region 1, the hardness of the TiZrAlN coatings increased from 20 to 23 GPa, as x was increased from 0 to 0.05. The maximum hardness (24–27 GPa) is obtained in region 2. In region 3, the hardness of the TiZrAlN coatings decrease rapidly down to 14 GPa, as x was increased

Fig. 9. a) Load–displacement curve for 330 nm-thick (Ti,Zr)1−xAlxN film with x = 0.09 deposited on Si, b) corresponding average contact pressure (ACP) and elastic modulus (E) vs. displacement h.

further from 0.16 to 0.37. The elastic modulus of the TiZrAlN coatings increased from 284 to 304 GPa in region 1 and then decreases rapidly to 180 GPa as the Al content was increased from 0.09 to 0.37. It has been shown that the coating wear resistance is improved with by both high hardness and low Young's modulus [43,44]. In this respect, a high H/E ratio is desirable (this ratio characterizes the value of elastic recovery during unloading for nanocontact interaction [45]). As shown in Fig. 10b, the H/E ratio of TiZrAlN coatings increased from 0.072 to 0.106 as the Al atomic fraction was increased up to x=0.14. In the x=0.14 to 0.16 range, the hardness of TiZrAlN remains rather high (about 24 GPa) but the elastic modulus drops drastically to about 235 GPa. Therefore, a maximum value of H/E ~0.1 is reached at the transition between regions 2 and 3. Then, the H/E ratio decreased with a further increase in the Al content (region 3). 4. Discussion

Fig. 8. Evolution of electrical resistivity (measured at room temperature) with Al content, x. The solid line is a guide to the eye.

The addition of Al during magnetron sputter-deposition of TiZrN films leads to the development of different growth microstructures, which span from columnar, single-phase c-TiZr(Al)N solid solutions (region 1) to nanocomposites consisting of c-TiZr(Al)N grains surrounded by a disordered matrix (region 2), and finally to amorphous films (region 3). In a recent report by Chen et al. [29], the stabilization of a single phase c-TiZrAlN structure was clearly evidenced from XRD patterns in magnetron sputter-deposited films with a composition Ti0.34Al0.37Zr0.29N, which is similar to the sample with the highest Al content (x= 0.37) investigated here. In our case, (Ti,Zr)1−xAlxN films become ‘amorphous’ for x >0.16, as revealed by XRD and TEM. This

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

Fig. 10. a) Evolution of nanoindentation hardness and elastic modulus of (Ti,Zr)1−xAlxN films with varying Al content, x. b) Evolution of H/E and H3/E2 ratio of (Ti,Zr)1−xAlxN films with varying Al content. Solid lines are a guide to the eye.

amorphization is likely due to the low-adatom mobility encountered in the present deposition conditions, as the substrate temperature was relatively low (Ts = 270 °C) compared to conditions used in Ref. [29] (Ts = 500 °C). The N under-stoichiometry observed for these films (see Table 1) could also contribute to destabilize the cubic phase. In region 2, the films consist of crystalline grains surrounded by a disordered/amorphous matrix. In Zr0.52Al0.48N1.11 films deposited by cathodic arc evaporation at 200 °C, a similar nanocomposite structure was reported [15]. The large miscibility gap of the ZrN-AlN system makes deposition of c-ZrAlN possible only for low Al content, and films having AlN contents higher than 50 at.% typically result in the formation of a mixture of cubic, hexagonal and amorphous parts. In the present work, no unequivocal conclusion could be drawn on the exact structure and chemical composition of the additional ‘hex-nanophase’ inferred from XRD results (Fig. 4b–d). However, TEM observations give hints that this ‘matrix’ phase first forms at the grain boundary of c-TiZr(Al)N grains. Upon progressive incorporation of Al, its volume contribution increases as the size of c-TiZr(Al)N grains is drastically reduced (Figs. 3b and 6), and finally a fully amorphous film is formed, as discussed above. Additional experiments using sub-nanoscale chemically sensitive techniques, such as 3D-atom probe tomography [20] or STEM-EDX mapping [17], would give insights on the nature of this interfacial phase. Concomitantly with the grain size reduction observed in regions 1 and 2, the incorporation of Al, up to x = 0.16, favors the development

39

of (200) preferred orientation of c-TiZrAlN grains (Fig. 3a). The promotion of (200) texture with increasing Al content was also revealed in arc-deposited [8] and magnetron sputtered Ti–Al–N films [11]. For magnetron-sputtered quaternary Ti1 − x−zAlxZrzN films studied by Chen et al. [29], such a change in preferred orientation was not clearly observed, but this can be related to the different chemical composition fields investigated (0.37 ≤ x ≤ 0.52 and z ≤ 0.29). The origin of texture development in binary TM nitride films has been largely discussed in the literature. A recent study of Mahieu and Depla [46], based on the quantification of particle and energy flux reaching the substrate during TiN magnetron sputter-deposition, has provided a comprehensive picture on the relation between growth morphology and texture development, as rationalized in terms of extended structure zone models. In particular, for ‘zone-T’ regime, the development of ‘V-shaped’ (111)-oriented columns at low N2 partial pressure conditions is the result of a competitive columnar growth process due to different kinetically-limited diffusion pathways on (111) planes vs. (002) ones, in agreement with the calculations of Gall et al. [47]. At high N2 partial pressure, the presence of atomic N reduces the diffusion length of Ti adatoms on (002) planes, resulting in faster vertical growth rate along the [200] direction. When the Ti diffusion length is further increased, then a ‘zone II’ microstructure with (200)-oriented straight columns is formed, as recrystallization process can be allowed. For ternary TM nitrides, recent results by Koutsokeras et al. [48] on Ti–Ta–N films have shown that texture development could be understood based on the same kinetics mechanisms proposed for binary counterparts. By comparing magnetron sputter-deposited Ti–Zr–N and Ti–Ta–N films, Abadias et al. [33] have provided evidence for the prime role of growth energetics in governing textural change from (111) to (002) upon incorporation of TM (TM= Zr or Ta) in TiN. Specifically, the deposited energy Edep scaled with the MMe/MAr molar mass ratio, where Me is the alloying Me element. In the present study, the incorporation of Al in (Ti,Zr)1 − xAlxN films was obtained by increasing the flux of Al sputtered atoms, relatively to the Ti and Zr ones (Ti and Zr targets were operated at fixed power), at low N2 partial pressure conditions. As calculated from Monte-Carlo simulations using the SIMTRA code [49], the growth energetics is found to decrease with increasing fraction of Al flux [33], all the more that no Ar ions are backscattered as neutrals from the Al target (MAr > MAl). Therefore, neither grain size refinement nor enhancement of (200) preferred orientation can be ascribed to a defectinduced renucleation process, which would be favored under larger deposited energy conditions, such as in the case of adding Ta to TiN [33,50]. A more plausible scenario, as discussed further below to explain the stress evolution of Fig. 7, is the preferential incorporation of Al at the grain boundary, which leads to disrupt the columnar growth. In region 2, grain growth (of the cubic phase) is further inhibited due to the formation of an interfacial ‘hex-nanophase’: this secondary phase contributes to reduce the grain boundary mobility, similarly to a Zener drag effect [12], and acts as preferential nucleation sites for new grain formation. A gradual increase in the lattice parameter of cubic TiZr(Al)N grains was observed in regions 1 and 2 (see Fig. 3b). In region 1, this can be correlated to the observed increase of compressive stresses (Fig. 7b). However, no straightforward conclusions can be drawn for films of region 2, as the stress evolution measured from wafer curvature reflects the global stress state of the film, i.e. including all phases, while the lattice parameter was measured from the cubic phase only. Since the atomic radius of Al atom is smaller than that of Ti and Zr, a decrease of the lattice parameter would be expected, as derived theoretically [22–24] and confirmed experimentally [9,12,13] for Ti–Al–N and Zr–Al–N films. Although the interplanar spacing cannot be simply described by atomic size considerations, other contributions, like a decrease of valence electron density or a reduced charge transfer between Ti and N when substituting Al for Ti [9], would also play in favor of a reduced cell dimension. Therefore, the observed structural evolution may suggest that Al atoms preferably incorporate as interstitials rather than being

40

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41

introduced at substitutional sites in the cubic TiZrN lattice. This may also indicate a possible decrease of titanium concentration in the cubic lattice upon Al addition. Such Ti depletion in the c-TiZr(Al)N solid solution may occur at the expense of the additional ‘hex-nanophase’ formation. Additional XRD experiments aiming at determining the stress-free lattice parameter would be required to fully understand this issue, which was beyond the present scope. Computational of the electronic structure would also give a hint on the complex bonding in such TM nitride systems [4,21,27,51]. As shown in Fig. 7, the in situ stress evolution during nanocomposite formation has been reported. Three distinct stress evolution regimes could be evidenced and associated with the three growth regimes: of single-phase cubic solid solutions (region 1), nanocomposites consisting of cubic grains surrounded by highly-disordered matrix (region 2) and fully amorphous structure (region 3). Since no stress variation could be detected upon growth interrupts, the observed stress gradients in region 1 cannot be due to a stress relaxation mechanism. Similar stress evolutions were reported for competitive columnar growth regimes of magnetron-sputtered TiN films, which developed a (111) preferred orientation. Based on previous findings [50,52], it can be concluded that, in region 1, the intrinsic growth stress is the net sum of two stress generating sources: 1) compressive stress due to atomic peening mechanism, causing the incorporation of defects in the growing layer as a result of kinetics transfer of energetic species (sputtered atoms and backscattered Ar) and 2) tensile stress due to attractive forces at the column boundaries and possible voids formation. It can be seen that, compared to the Al-free film, the incorporation of Al first increases the magnitude of compressive stress (region 1 of Fig. 7b). Since at Ts = 270 °C, Al adatoms are much more mobile (Ts/Tf ~ 0.58) than Ti (Ts/Tf ~ 0.28) or Zr (Ts/Tf ~ 0.26) adatoms, where Tf refers to the melting temperature, they can readily be incorporated as excess atoms in the grain boundaries, thereby increasing in-plane film compression. This is supported by TEM observations which show a densification of the grain boundary structure at low Al content. It can be noticed from Fig. 7a that the development of tensile stress is delayed for the film with x = 0.05 in comparison with the case x = 0, suggesting a subtle competition between compressive and tensile stress generation at the grain boundaries in the presence of an Al flux. For Al fluxes corresponding to region 2, a sharp decrease of the |σavg| stress magnitude is found (Fig. 7b), concomitantly with a drastic reduction of the crystallites sizes (Fig. 3b). This decrease of |σavg| can be partly accounted for by the reduced film elastic modulus in this region, from E ~ 300 GPa at x ~ 0.07 to E ~ 230 GPa at x ~ 0.15 (Fig. 10a). It may be also related to the increasing volume fraction of grain boundary and/or amorphous matrix of such films, favoring the preferential annihilation of the growth-induced defects at the interfaces. In region 3, the real-time film force evolution reveals a compressive steady-state stress, associated with the formation of amorphous layers, and suggests a surface-mediated stress build-up. A clear correlation is found between growth regimes, structural and hardness evolutions as a function of the Al content. The coatings with Al contents within region 2 exhibited the highest hardness (up to 27 GPa). Also, a substantial improvement in the coatings H/E ratio was achieved together with a sharp decrease in the coating elastic modulus to about 235 GPa in this compositional range. It is noteworthy that the maximum H/E ratio of 0.105 is achieved at Al content x=0.14, corresponding to the percolation threshold of amorphous matrix around nanometer-sized (~3 nm) cubic grains, as deduced from the abrupt change in resistivity of Fig. 8. With a further increase in the Al content to levels greater than x = 0.16, amorphization takes place, which led to low hardness and H/E ratio of these TiZrAlN coatings. One of the main results of the present study is that the hardness maximum is not correlated with the maximum of compressive stress. The origin of hardness enhancement in nanocomposite hard coatings has been the subject of many studies since the pioneering work of Veprek et al. in 1995 [53], and the contribution of residual stress to

such hardening was often debated. Our careful determination of stress evolution using real-time diagnostic during growth clearly points out that hardening mechanisms in nanocomposite films are not driven by the development of compressive stress, but are necessarily attributable to the inherent self-organization of the structure at the nanoscale [54], supporting the conclusions that strong cohesive interfacial bonding between nanograins is the prerequisite to achieve superhard or ultra-hard coatings with low level of impurities [55]. 5. Summary and conclusions The present study contributes to a better understanding of evolutionary structural and mechanical properties upon multicomponent alloying in transition metal nitride films by addressing the issue of Al incorporation into TiZrN films deposited by reactive magnetron sputtering. Upon addition of Al, three distinct growth regimes are evidenced: i) cubic TiZr(Al)N solid solutions, ii) nanocomposites consisting of cubic TiZr(Al)N grains surrounded by a highly-disordered ‘matrix’, and iii) amorphous phase. A drastic decrease in grain size, change in degree of preferred orientation from (111) to (200) and a non-monotonous stress evolution are observed when increasing Al atomic fraction up to x = 0.16 in (Ti,Zr)1 − xAlxN films. A maximum hardness of 27 GPa is obtained in regime ii) at a critical Al content of ~0.12, corresponding to the percolation threshold of amorphous phase surrounding cubic nanocrystals, as evidenced from electrical resistivity measurements. The present study reports the in situ stress evolution during nanocomposite formation and shows that hardening effect in nanocomposite films is not imputable to the build-up of compressive stress but rather an intrinsic consequence of self-organized nanostructure and strong cohesive interfacial bonding. Acknowledgments The authors wish to thank Dr. Philippe Guerin for technical assistance during magnetron sputter-deposition. V.V. Uglov acknowledges the University of Poitiers for a one month visiting professorship position under which part of this work has been performed. References [1] H. Lind, R. Forsén, B. Alling, N. Ghafoor, F. Tasnadi, M.P. Johansson, I.A. Abrikosov, M. Odén, Appl. Phys. Lett. 99 (2011) 091903. [2] R. Forsén, M. Johansson, M. Odén, N. Ghafoor, J. Vac. Sci. Technol. A 30 (2012) 061506. [3] R. Cremer, D. Neuschütz, Surf. Coat. Technol. 146–147 (2001) 229. [4] R. Rachbauer, A. Blutmager, D. Holec, P.H. Mayrhofer, Surf. Coat. Technol. 206 (2012) 2667. [5] R. Rachbauer, D. Holec, P.H. Mayrhofer, Appl. Phys. Lett. 97 (2010) 151901. [6] D.G. Sangiovanni, V. Chirita, L. Hultman, Thin Solid Films 520 (2012) 4080. [7] Y. Tanaka, T.M. Gür, M. Kelly, S.B. Hagstrom, T. Ikeda, K. Wakihira, H. Satoh, J. Vac. Sci. Technol. A 10 (1992) 1749. [8] A. Hörling, L. Hultman, M. Oden, J. Sjolen, L. Karlsson, Surf. Coat. Technol. 191 (2005) 384. [9] U. Wahlström, L. Hultman, J.-E. Sundgren, F. Adibi, I. Petrov, J.E. Greene, Thin Solid Films 235 (1993) 62. [10] K. Kutschej, P.H. Mayrhofer, M. Kathrein, P. Polcik, R. Tessadri, C. Mitterer, Surf. Coat. Technol. 200 (2005) 2358. [11] L. Chen, M. Moser, Y. Du, P.H. Mayrhofer, Thin Solid Films 517 (2009) 6635. [12] V. Chawla, R. Chandra, R. Jayaganthan, J. Alloys Compd. 507 (2010) L47. [13] H. Hasegawa, M. Kawate, T. Suzuki, Surf. Coat. Technol. 200 (2005) 2409. [14] R. Lamni, R. Sanjinés, M. Parlinska-Wojtan, A. Karimi, F. Levy, J. Vac. Sci. Technol. A 23 (2005) 593. [15] L. Rogström, M. Ahlgren, J. Almer, L. Hultman, M. Odén, J. Mater. Res. 27 (2012) 1716. [16] P.H. Mayrhofer, A. Hörling, L. Karlsson, J. Sjölén, T. Larsson, C. Mitterer, L. Hultman, Appl. Phys. Lett. 83 (2003) 2049. [17] A. Knutsson, M.P. Johansson, L. Karlsson, M. Odén, Surf. Coat. Technol. 205 (2011) 4005. [18] L. Rogström, J. Ullbrand, J. Almer, L. Hultman, B. Jansson, M. Odén, Thin Solid Films 520 (2012) 5542. [19] N. Norrby, M.P. Johansson, R. M'Saoubi, M. Odén, Surf. Coat. Technol. 209 (2012) 203.

I.A. Saladukhin et al. / Thin Solid Films 538 (2013) 32–41 [20] R. Rachbauer, S. Massl, E. Stergar, D. Holec, D. Kiener, J. Keckes, J. Patscheider, M. Stiefel, H. Leitner, P.H. Mayrhofer, J. Appl. Phys. 110 (2011) 0235515. [21] H.W. Hugosson, H. Högberg, M. Algren, M. Rodmar, T.I. Selinder, J. Appl. Phys. 93 (2003) 4505. [22] P.H. Mayrhofer, D. Music, J. Schneider, J. Appl. Phys. 100 (2006) 094906. [23] D. Holec, R. Rachbauer, L. Chen, L. Wang, D. Luef, P.H. Mayrhofer, Surf. Coat. Technol. 206 (2011) 1698. [24] S.H. Sheng, R.F. Zhang, S. Veprek, Acta Mater. 56 (2008) 968. [25] R. Prange, R. Cremer, D. Neuschütz, Surf. Coat. Technol. 133–134 (2000) 208. [26] G. Greczynski, J. Lu, M. Johansson, J. Jenssen, I. Petrov, J.E. Greene, L. Hultman, Vacuum 86 (2012) 1036. [27] R. Rachbauer, D. Holec, P.H. Mayrhofer, Surf. Coat. Technol. 211 (2012) 98. [28] O. Knotek, M. Böhmer, T. Leyendecker, F. Jungblut, Mater. Sci. Eng. A 105–106 (1988) 481. [29] L. Chen, D. Holec, Y. Du, P.H. Mayrhofer, Thin Solid Films 519 (2011) 5503. [30] V.V. Uglov, V.M. Anishchik, V.V. Khodasevich, Zh.L. Prikhodko, S.V. Zlotski, G. Abadias, S.N. Dub, Surf. Coat. Technol. 180–181 (2004) 519. [31] A. Hoerling, J. Sjölen, H. Willmann, T. Larsson, M. Odén, L. Hultman, Thin Solid Films 516 (2008) 6421. [32] G. Abadias, L.E. Koutsokeras, S.N. Dub, G.N. Tolmachova, A. Debelle, T. Sauvage, P. Villechaise, J. Vac. Sci. Technol. A 28 (2010) 541. [33] G. Abadias, L.E. Koutsokeras, A. Siozios, P. Patasalas, Thin Solid Films 538 (2013) 56. [34] A. Fillon, G. Abadias, A. Michel, C. Jaouen, Thin Solid Films 519 (2010) 1655. [35] L.G. Parratt, Phys. Rev. 95 (1954) 359. [36] J. Hay, P. Agee, E. Herbert, Continuous Stiffness Measurement During Instrumented Indentation Testing, Experimental Techniques, May/June 2010, p. 86.

41

[37] I.A. Abrikosov, A. Knutsson, B. Alling, F. Tasnadi, H. Lind, L. Hultman, M. Odén, Materials 4 (2011) 1599. [38] G. Abadias, V.I. Ivashchenko, L. Belliard, Ph. Djemia, Acta Mater. 60 (2012) 5601. [39] H. Holleck, Surf. Coat. Technol. 36 (1988) 151. [40] D. Tabor, The Hardness of Metals, Clarendon Press, Oxford, 1951, p. 175. [41] W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564. [42] E.T. Lilleodden, W.D. Nix, Acta Mater. 54 (2006) 1583. [43] N.V. Novikov, M.A. Voronkin, S.N. Dub, I.N. Lupich, V.G. Malogolovets, B.A. Maslyuk, G.A. Podzyarey, Diamond Relat. Mater. 6 (1997) 574. [44] A. Leyland, A. Matthews, Wear 246 (2000) 1. [45] E.H. Lee, Y. Lee, W.S. Oliver, L.K. Mansur, J. Mater. Res. 8 (1993) 377. [46] S. Mahieu, D. Depla, J. Phys. D 42 (2009) 053002. [47] D. Gall, S. Kodambaka, M.A. Wall, I. Petrov, J.E. Greene, J. Appl. Phys. 93 (2003) 9086. [48] L.E. Koutsokeras, G. Abadias, P. Patsalas, J. Appl. Phys. 110 (2011) 043535. [49] K. Van Aeken, S. Mahieu, D. Depla, J. Phys. D: Appl. Phys. 41 (2008) 205307, (see also www.draft.ugent.be). [50] G. Abadias, L.E. Koutsokeras, Ph. Guerin, P. Patsalas, Thin Solid Films 518 (2009) 1532. [51] P. Patsalas, G. Abadias, G.M. Matenoglou, L.E. Koutsokeras, Ch.E. Lekka, Surf. Coat. Technol. 205 (2010) 1324. [52] G. Abadias, Ph. Guerin, Appl. Phys. Lett. 93 (2008) 111908. [53] S. Veprek, S. Reiprich, L. Shizhi, Appl. Phys. Lett. 66 (1995) 2640. [54] J. Patscheider, MRS Bull. 28 (2003) 180. [55] S. Hao, B. Delley, S. Veprek, C. Stampfl, Phys. Rev. Lett. 97 (2006) 086102.