Synthesis and characterisation of Mo-B-C thin films deposited by non-reactive DC magnetron sputtering

Surface & Coatings Technology 309 (2017) 506–515

Contents lists available at ScienceDirect

Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat

Synthesis and characterisation of Mo-B-C thin films deposited by non-reactive DC magnetron sputtering P. Malinovskis a,⁎, J. Palisaitis b, P.O.Å. Persson b, U. Jansson a, E. Lewin a a b

Department of Chemistry-Ångström Laboratory, Uppsala University, SE-751 21 Uppsala, Sweden Thin Film Physics Division, Department of Physics, Chemistry, and Biology (IFM), Linköping University, SE-581 83 Linköping, Sweden

a r t i c l e

i n f o

Article history: Received 30 June 2016 Revised 30 November 2016 Accepted in revised form 1 December 2016 Available online 05 December 2016 Keywords: Mo-B-C Magnetron sputtering Films Hard coatings Molybdenum diboride Nanocomposite

a b s t r a c t Thin films in the Mo-B-C system with varying carbon content (up to 37 at.%) were deposited using non-reactive DC magnetron sputtering. The phase composition and microstructure were determined and the potential use of the films in sliding electrical contact applications was evaluated. Films with lower than 23 at.% carbon content consisted of nanocrystalline MoB2 − x grains surrounded by an amorphous tissue phase (a-B for binary, and aBCx for ternary films). With increasing carbon content grain sizes was found to decrease (from 16 to 5 nm), and above 23 at.% carbon the films deposited at room temperature were X-ray amorphous. Scanning transmission electron microscopy and energy dispersive X-ray spectroscopy reveal that these films contain Mo-rich and Mo-poor regions, and thus are two-phase amorphous nanocomposites. Low-carbon content samples exhibited a friction coefficient against the steel counter surface of 1.1; this was reduced to 0.8 for high carbon-content films. Analysis of the tribofilm revealed formation of molybdenum oxide and amorphous carbon, however without significant lubricating effect at room temperature. Hardness and elastic modulus decrease with carbon content from ~29 to ~22 GPa and ~526 to ~326 GPa. These values give an H/E ratio of 0.06 to 0.07, indicating brittle material. Resistivity was found to increase with carbon content from ~175 μΩ cm for binary Mo-B to ~395 μΩ cm for Mo-B-C thin film with 37 at.% of C. Therefore all the above results suggest that the Mo-B-C films are not suitable for sliding electrical contacts. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Transition metal diborides combine attractive properties which includes high hardness and wear resistance, high electrical conductivity and oxidation resistance [1,2]. The properties' enhancement can be further achieved if material is synthesized in the form of a thin film. Sputtered TiB2 films have been observed to be were superhard with values more than twice that of the bulk crystal value, which was explained by Mayrhofer et al. [3], who attributed the hardening to (001)-textured TiB2 grains surrounded by an amorphous tissue phase of boron. This tissue phase hinders dislocation movements, thus yielding the hardening. Recent studies of magnetron sputtered Nb-B and Cr-B films by Nedfors et al. have demonstrated the presence of such a boron tissue phase also in these materials systems, as well as a significant hardness increase compared to bulk materials [4,5]. We observed a similar behaviour in magnetron-sputtered Mo-B films [6]. Consequently, nano-sized diboride grains surrounded by a boron tissue phase seem to constitute the typical thin film microstructure for magnetron sputtered Me-B films. ⁎ Corresponding author. E-mail address: [email protected] (P. Malinovskis).

http://dx.doi.org/10.1016/j.surfcoat.2016.12.003 0257-8972/© 2016 Elsevier B.V. All rights reserved.

A potential application for metal boride films is as sliding electric contacts, which require a low wear rate, a high electrical conductivity, a low contact resistance combined with a low friction coefficient. So far the limiting factor for magnetron sputtered metal diboride film applications as electric contacts – is the high friction coefficients (N0.5) which has been observed in several systems [5,7,8]. An exception is the Nb-B system where Nedfors et al. observed a friction coefficient of only 0.12 versus a steel counter surface, however these coatings also exhibited a high contact resistance due to their high hardness [4]. In our recent Mo-B study [6], we predicted, based on stability trends in the diborides, an extensive formation of molybdenum oxides, which potentially could reduce the friction in a tribocontact. An extensive molybdenum oxide formation was indeed observed but did not lead to reduced friction at room temperature in ambient environment. An alternative approach is to tailor the properties by introducing a third element like carbon [5,8–11] or nitrogen [11–13]. In general the introduction of carbon or nitrogen to a material based on a metal diboride has led to a decrease in grain-size and eventually an amorphisation of the material. This has been coupled with a decrease in hardness [5,10–13]. The approach of studying increasing carbon content of coatings around the MeB2-C tie line has been explored in several systems in connection with possible applications in electrical contacts:

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

Nb-B-C [10] and Cr-B-C [5]. It was found that the ternary films exhibited a slightly lower hardness than the binary Me-B films but lower contact resistances and friction coefficients making them more suitable for a contact application. Also the Ti-B-C system has been studied with regards till electrical contact applications, but then the starting point was TiC and the tie-line TiC-TiB2 was investigated [9]. The reduced friction coefficients for the Me-B-C films could be explained by the formation of a thicker a-BCx tissue around the crystalline grains thereby softening the material (hardness ~20 GPa [14]), and increasing ductility, thus preventing brittle fracture and the formation of debris in the wear track [8,10]. Furthermore, carbon is well-known to be able to form lubricating tribolayers, which further can reduce the friction coefficients [8,9, 15]. There are also some studies on thin films from the Mo-B-C system, however these have been centred on the ternary phase Mo2BC, and thus not on studying the effect of carbon alloying on a MoB2 material. Furthermore, the Mo2BC phase is not on the tie-line MoB2-C, and generally requires high substrate temperature (or an energetic process) to be formed [16–18]. At lower temperatures, amorphous or poorly crystalline coatings have been formed, in agreement with the other Me-B-C systems mentioned above [17,18]. In the present study the microstructure and properties of DC magnetron sputtered Mo-B-C thin films were investigated, and are presented together with data from previously deposited binary Mo-B coatings [6] in order to clearly demonstrate the effect of carbon alloying. The aim is to investigate if the added carbon can improve the tribological and mechanical properties (COF b 0.6 and H/E N 0.1) of the Mo-B films for a potential use as sliding electric contacts. Mo-B-C films were grown with varying carbon contents and at two different temperatures (300 °C and 600 °C), using a compound MoB1.5 and a graphite targets. Structures, as well as mechanical and tribological properties of these materials, were evaluated using X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), transmission and scanning electron microscopy (TEM and SEM), Raman spectroscopy, nanoindentation and tribological ball-on-disk tests. A single high temperature experiment was carried out in order to investigate the temperature effect to film structure and properties. 2. Experimental details The Mo-B and Mo-B-C films were deposited by non-reactive DCmagnetron sputtering from circular targets (diameter 51 mm, thickness 3.2 mm): MoB1.5 (stated composition of 2:1 could not be confirmed by our analysis of the target, nor the stated purity of 99.9%, see below) and C (stated purity 99.9%). XRD showed that the MoB1.5 target consisted of the rhombohedral hR18-MoB2 phase. However, elastic recoil detection analysis (ERDA) of the target showed B/Mo ratio ~ 1.5 (equivalent to 61 at.% of B). The ERDA analysis also showed that target contains carbon, oxygen and hydrogen, therefore it is assumed to originate from adsorption of water and other oxygen-containing species in pores as observed in SEM. The significant outgassing in the sputter chamber from a freshly mounted target supports this conclusion. All experiments were carried out after baking and outgassing of the targets and chamber. A DC-magnetron sputtering system (ultra-high vacuum chamber with base pressure b1 · 10−7 Pa) with a confocal sputter-down configuration with three targets and a target to substrate distance of 150 mm was used. The targets were mounted with 24° angle between substrate and target normal. The substrates were mounted on rotational table with integrated heater, preheated before deposition and kept at constant temperature of 300 °C, as measured by thermocouple (calibrated with optical pyrometer). To investigate if a full temperature study was called for, a single deposition at high temperature the substrates was conducted, in this case the substrates were kept at 600 °C. An Ar+ plasma was ignited at 0.4 Pa, using a 42 sccm Ar gas flow. The MoB1.5 target current was kept constant at 150 mA (yielding ~2.2 W/cm2), while the C target current was varied between 0 and 250 mA (yielding 0 to ~8.5 W/cm2) to attain the ternary alloys with different carbon content. Prior to

507

deposition the substrates were plasma etched using a DC-bias of − 300 V and plasma ignited on Mo target (stated purity 99.95%) which were run at 25 mA, thereafter a thin adhesion layer of Mo was deposited, by lowering the bias to −50 V and increasing the current on Mo target to 100 mA. All depositions were performed with a DC bias voltage of − 50 V on the substrate table with low ion current (b 0.2 mA/cm2). Coatings were deposited simultaneously on Si(001), α-Al2O3(001) and 316L stainless steel substrates for structural properties (Si(001) 0.5 mm thick), stress measurements (Si(001) 0.3 mm thick); mechanical properties (α-Al2O3) and tribological (316L steel) analysis, respectively. To attain a coating thickness of around 1 μm the deposition time of the Mo-B-C coating was varied between 250 and 350 min. The binary Mo-B sample was studied in-depth and the results have been published in [6], but are included in this paper acting as a reference in order to reveal the effect of carbon alloying into the Mo-B system. Sample compositions were determined by Time-of-Flight Energy ERDA (ToF-E-ERDA) experiments, performed at the Tandem Accelerator Laboratory at Uppsala University. Experiments were performed in a vacuum chamber at a base pressure of b1 · 10−4 Pa. 36 MeV 127I8+ ions were used as primary projectiles. Scattered recoils were detected in an angle of 45° with respect to the primary beam, with both incidence and exit angle of beam and detected particles at 22.5° with respect to the sample surface. Further details on the detection systems can be found in Ref. [19]. Calculation of depth profiles from time-energy coincidence spectra of recoils was done using the CONTES program package [20]. The chemical bonding in the films was analysed with XPS using a Physical Electronics Quantum 2000 with monochromatic Al Kα radiation and a 45° photoelectron take-off angle. High resolution spectra were acquired on the surface (to analyse tribological behaviour) as well as after sputter-etching (1 keV Ar+ ions for 10 min, yielding an approximate depth of 42 nm) to analyse the bulk of the coating material, i.e. below the surface oxidised region. The analysis region was 200 μm in diameter and the ion beam for sputter-etching was rastered over a 1 × 1 mm area. Compositional and chemical bonding analysis was carried out on coatings deposited on Si(001)-substrates. For analysis of crystalline phases XRD was performed with a Philips X'Pert MRD diffractometer with Cu Kα radiation and parallel beam geometry using a Göbel mirror on the primary side and a parallel plate collimator with an 0.27° acceptance angle on the secondary side. Both θ/2θ and grazing incidence scans performed, the latter with a 2° incident angle. Peak positions and widths were determined by curve fitting of the peaks and measured full width at half maximum. Grain sizes were estimated using Scherrer's equation, assuming spherical grains [21]. Lattice parameters were determined using the observed (001) and (100) peaks, assuming a hexagonal crystal structure. Coating morphology and wear were studied using a Zeiss Merlin SEM instrument. As-deposited samples on Si(001) were used to attain fractured cross-sections. Surface morphology as well as wear was studied on coatings deposited on steel substrates. In both cases in-lens detector and a 5 keV beam were used. EDX (energy dispersive X-ray spectroscopy) analysis was performed at 20 keV beam energy and collected with EDX X-Max 80 mm2 Silicon Drift Detector with high sensitivity and analysed with AZtec (INCA energy) software for X-ray mapping and element analysis. Raman spectroscopy was carried out with Renishaw Invia Raman spectrometer equipped with 532 nm laser to identify Raman-active phases at the surface of the films. Raman spectra were normalised according to the highest intensity in each spectrum. Electron transparent cross-section TEM samples were prepared from as-grown films using the traditional “sandwich” method. The films were cut into small pieces, mounted into a Ti grid, glued with a high temperature glue (Gatan G-1 epoxy) and polished to a thickness of approximately 50 μm. Ar+ ion milling followed with 5 keV and 2 keV (to minimize the damage) ion energy at a 5° milling angle from both sides in a Gatan precision ion polishing system (PIPS). TEM images

508

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

and diffraction patterns were acquired in Tecnai G2 TF20 UT, operated at 200 kV. STEM-EDX (energy dispersive X-ray spectroscopy) spectrum images combined with high angle annular dark field imaging (HAADF) analyses were performed in double-corrected Linköping FEI Titan3 60– 300, operated at 300 kV. STEM-EDX mapping was undertaken with the embedded high sensitivity Super-X EDX detector. Mechanical properties (hardness and Young's modulus) were measured on the CSM Instruments Ultra Nano Hardness Tester (UNHT) equipped with Berkovich diamond tip. Load-displacement curves were measured in 20 different spots on the sample with 70 nm in displacement control and unloading and loading rates of 1.5 mN/min (with maximum load of b 4 mN). Hardness and elastic modulus were determined by Oliver-Pharr method and averaged from at least 15 indentation spots over sample [22]. Electrical resistivity was measured with four point probe AIT CMT-SR2000N Sheet Resistance/Resistivity Measurement System, averaging measurements from 5 different spots. Tribological measurements were performed using a ball-on-disk setup. Ball bearing steel (100Cr6) and alumina balls were used as the counter surfaces. The track radius was 3 mm, sliding speed 0.1 m/s and the distance up to 80 m (4000 laps) with a load of 1 N and 0.5 N for steel and alumina balls, respectively in order to maintain the same contact pressure of 0.8 GPa. No lubrication was used; the measurements were carried out in ambient atmosphere (22 °C) with 55% relative humidity. The wear rates were estimated after 4000 laps by measuring the wear track depth with WYKO NT1100 optical profilometer and assuming sphere cut shape [23]. The same equipment was used to measure the surface roughness, curvature and estimate the stress in the film, by using Stoney's equation [24,25] on 0.3 mm Si (001) substrates within 200 × 200 μm area. 3. Results ERDA measurements of the samples revealed that the diboride film (see Fig. 1 and Table I) is substoichiometric with a B/Mo ratio of 1.65 (equivalent to 62 at.% B). This is only slightly higher than the target with a B/Mo ratio of about 1.5 (equivalent to 61 at.% B) and is well within the maximum homogeneity range (61–66 at.% B) of the hP3-MoB2 phase [26] observed in the samples, see XRD results below. The slight increase in B content compared to the target is typical for non-reactively sputtered metal borides, and is due to the low collision cross section of boron compared to Mo which causes more boron atoms arrive at the substrate [2]. As the carbon content is increased from no carbon

Fig. 1. Ternary phase diagram of the Mo-B-C system [60,61] with thin film sample compositions shown as solid violet dots (line is guide to the eyes). The hollow triangle represents the composition of the high temperature (HT) sample. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table I Compositional data from ERDA and result of XPS peak fittings. C target current (mA)

Elemental concentrations (at.%)

B/Mo ratio

XPS peak fitting results (%) B1s

0 50 100 150 200 200 HT 250

Mo

B

C

O

37 34 31 29 26 26 24

62 57 52 47 42 42 38

≤1 8 16 23 31 31 37

≤2 ≤1 ≤1 ≤1 ≤1 ≤1 ≤1

1.65 1.68 1.68 1.62 1.62 1.62 1.58

B-Mo

B-B

B-C

94 68 66 62 64 59 60

6 14 15 16 11 10 11

– 18 19 22 25 31 29

(b 1 at.%) to 37 at.% of C, the B/Mo ratio stays approximately constant in all samples. A small amount of oxygen (b 2 at.%) was detected in the deposited samples as well. Unintentional carbon in the binary MoB sample can be attributed to the impurities present in the compound target [6]. X-ray diffractograms of the samples are shown in Fig. 2 together with reference peak positions for the hexagonal hP3-MoB2 phase [27]. The binary sample shows good agreement with the reference. As the carbon content increases in the deposited films, the characteristic peak becomes broader and decrease in intensity. Finally, at 23 at.% C the films become X-ray amorphous. When carbon is added to the films there is also a slight shift of the peaks towards lower 2θ values, indicating cell parameter expansion, which can be attributed to stresses or a solid solution of carbon into the diboride phase. The only other observed peaks in the diffractograms (marked “x”) originate from the Mo adhesion layer. Thus, it can be concluded that the only crystalline phase in the films is hexagonal hP3-MoB2, and that the films at higher carbon contents are X-ray amorphous. Since the films are substoichiometric with respect to boron (B/Mo ratio b 2), the hexagonal phase will henceforth be denoted as MoB2 − x. Estimated grain sizes and calculated lattice parameters are given in Table II. The grain sizes are found to vary between ~ 5 to ~ 16 nm; while the lattice parameters show a slight increase with increasing carbon content: a = 3.05– 3.10 Å and c = 3.07–3.10 Å, respectively. The sample grown at elevated temperature (600 °C) but the same composition (31 at.% C), exhibited broad diffraction peaks originating from the hexagonal MoB2 − x

Fig. 2. GI-XRD of the Mo-B and Mo-B-C samples with C content up to 37 at.%, including HT sample at 31 at.% C. Red dashed lines and x are attributed to the hexagonal MoB2 phase and Mo adhesion layer, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

509

Table II XRD data and residual stresses. Grain sizes are averages from 001 and 101 reflections. Sample (at. %)

0 8 16 23 31 31 HT 37

Lattice parameters a (Å)

c (Å)

3.05 3.10 3.12 – – 3.10 –

3.07 3.09 3.10 – – 3.09 –

Grain size (nm)

Cell volume (Å3)

Residual stress (GPa)

16 10 5 – – 4 –

24.7 25.7 26.2 – – 25.7 –

−0.1 ± 0.1 0.4 ± 0.2 0.8 ± 0.2 1 ± 0.4 0.9 ± 0.2 1 ± 0.1 0.8 ± 0.2

phase compared to the X-ray amorphous behaviour of the low temperature deposited sample. Fractured cross-section SEM pictures of Mo-B-C coatings with different C-content are shown in Fig. 3. The microstructure of the films is dense and the films appear homogeneous throughout the thickness of the coating with an evolution from a columnar structure for the binary sample to a more fine grained and random morphology At the highest carbon contents the cross section almost appear glass like. The hightemperature sample exhibits a well pronounced columnar structure which is coarser than the binary sample's morphology. There is also a small variation in sample thickness from 0.8 μm (0 at.% C) to 1.2 μm (37 at.% C). Separate measurements of the surface roughness showed that the roughness was b 5 nm for all samples. Fig. 4 shows the cross section HRTEM (high resolution TEM) images of a binary MoB2 − x film (top) and the Mo-B-C films with 31 at.% C, grown at 300 °C (middle) and 600 °C (bottom) with some grains outlined. SAED (Selective area electron diffraction), DF (dark field) and HRTEM revealed that the binary film exhibits large amounts of nanocrystalline particles, with a grain size of ~ 10 nm. Alloying by C reduced the grain size to ~3 nm and grains were observed to be embedded in an amorphous tissue phase in the 31 at.% films, which consequently resulted in broad rings in the electron diffraction patterns. The high temperature sample exhibits slightly larger (~5 nm) and more pronounced grains (also embedded in amorphous tissue phase), which are supported by sharper electron diffraction rings. DF–TEM corroborates the decrease in crystallite size with a strong contrast grains in the binary sample and weak contrast in the samples with carbon, for both Mo-B-C films grown at 300 °C and 600 °C. These results correlate well with the XRD results above. Fig. 5 shows STEM images with EDX maps of the same three samples. There are clearly two phases present in the samples: bright areas that can be associated with the crystalline phase grains and darker areas associated with the tissue phase, which increases in volume with increased carbon content. Matrix phase is an additional description, however the term tissue phase is used throughout this paper to avoid confusion. This contrast suggests a higher average atomic number in the crystalline phase than in the tissue phase. EDX maps were performed on these areas (see Fig. 5), but due to tailing of B peak into the C peak, the separation of C and B is not possible, therefore the C map contains contribution from B. Nevertheless, particles enriched with Mo and a-BCx tissue phase from EDX C map (with contribution from B) are clearly visible. It is however unclear how B and C are distributed throughout the tissue phase. EELS on the binary sample (not shown here, see Ref. [6]) confirms that the B is present in both regions, and that the B bonding differs between particles and tissue phase. It can be assumed that the situation is similar for the ternary samples, but cannot be confirmed due to the small grain size making it impossible to isolate respective phase in these samples. XPS was used to study the chemical bonding in the films. In all samples Mo3d region (spectra not shown) exhibits only one distinct contribution with the Mo3d5/2 peak at average position of 228.0 eV. This agrees reasonably well with literature values for both Mo-B (227.9 eV

Fig. 3. SEM cross sections of the Mo-B and Mo-B-C samples with increasing C content, including HT sample. Scale bar is valid for all images.

[28]) and Mo-C (227.8 eV [28]), two chemical environments which thus have too small difference in binding energy to be separated in our measurements. Moreover, no changes in peak position or shape are observed with varying carbon content. Fig. 6 shows XPS spectra of the B1s and C1s regions from the Mo-B-C films in the present study. It is clear that a curve fitting is required to clarify the contribution from different bonding states. Starting with the B1s spectra, we know that the binary film consists of MoB2 − x and small amounts of a boronrich tissue phase. We can then fit these spectra to two peaks at 188.5 eV and 187.7 eV, attributed to B-Mo and B-B, respectively [29– 31]. When carbon is added to the films, we can see an increase in intensity on the high-energy side of the B1s spectra. This fits well with a peak

510

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

Fig. 4. Cross section TEM results of binary Mo-B and Mo-B-C samples with 31 at.% of C grown at 300 °C and 600 °C. The binary sample is crystalline. Some of the few crystalline grains in the Mo-B-C films are outlined. Here SAED – selective area electron diffraction, DF – dark field and big picture is HR-TEM. Scale bar valid for all images.

at 188.9 eV, which can be attributed to B-C contributions [15]. A peak at this energy is in agreement with other XPS studies on magnetron sputtered B-C films [9,32]. The different contributions of the B-Mo, B-B and B-C peaks to the total B1s spectra are summarised in Table I. As expected the B-C peak increases and the B-Mo peak is reduced with increasing C content in the films. There is also a slight shift of the B-Mo contribution to lower binding energies as carbon content increases, stabilising at 188.2 eV for carbon contents of 16 at.% and higher. The curve fitting of the C1s peak is more difficult to an extensive overlap of different possible contributions. The C1s region is shown in Fig. 6b and as expected the signal grows as carbon content increase. The observed C1s region exhibits one broad peak which cannot be fitted with one contribution. Actually, three contributions are expected in this

region: carbon bonded to carbon (C\\C), boron (C\\B) and molybdenum (C\\Mo). The latter contribution corresponds to e.g. a solid solubility of C into the boride and is often neglected in XPS studies of Me-B-C films. The C-C contribution is situated at 284.6 eV [28,33] and an increase with carbon content is observed. Two further peaks are necessary to fit the C1s spectra, these are located at 283.4 eV (I) and 282.7 eV (II). The literature value for C1s in Mo2C is 282.7 eV [28], which matches peak II in our spectra. However, from studies on Ti-B-C coatings [9] it is known that the C-Me peak can be shifted to higher binding energies when carbon dissolves into the boride structure [34,35]. Therefore peak I may also have contribution from C-Mo [36,37]. In the XPS B1s spectra (Fig. 6a) a clear B-C contribution is observed, thus a C-B contribution must be present in the C1s region. The literature indicated that C in B surrounding should appear at ~283 [35,38,39] or ~284 eV [38–40], hence both peaks I and II can have contribution from carbon bonded to boron. This is also supported by the trends: as the carbon increases the relative contribution of peak I increases and peak II decreases. If there was only C-B contribution in peak I, the much higher B-C contribution would be expected in the B1s spectra, which is not observed (see Fig. 6a). Hence both peak I and II must both have contributions from C\\Mo and C\\B bonds. Due to the extensive overlapping, we prefer to not assign the peak I and II to a specific binding environment. Finally, the XPS C1s spectra also have a small tailing at high binding energies and could therefore contain a small C-O contribution [41], which in intensity matches the total oxygen content of b2 at.%. In addition to XPS, the as-deposited sample surfaces were also analysed with Raman spectroscopy, which spectra are shown in Fig. 7a. No really strong Raman features are observed from sample surface only exposed to air. There are only weak peaks of molybdenum oxide and weak indications of boron oxide, as well as the D and G peaks of carbon, visible for samples with a carbon content of 16 at.% or greater. The carbon D and G band peaks increase with carbon content and are observed at 1350 and 1582 cm− 1, respectively. The presence of the D and G peaks, which are typical of amorphous carbon phase [42] has also been reported for a-B4C [43]. The results of nanoindentation measurements, in the form of hardness and elastic modulus are shown in Fig. 8a. The hardness value for the binary sample is 29 ± 2 GPa. As the carbon content increases the hardness monotonically decreases to 22 ± 1 GPa for the highest amount of C (37 at.%). The elastic modulus also decreases from 526 ± 26 GPa to 326 ± 7 GPa. The sample grown at higher substrate temperature exhibits a decrease in hardness and elastic modulus, from 24 ± 1 GPa to 21 ± 1 GPa and from 355 ± 6 GPa to 336 ± 16 GPa, respectively. A material is considered ductile if H/E ratio is above 0.1 [44–47]; the here studied Mo-B-C materials exhibit a low H/E value, which increases from 0.06 to 0.07 (see Fig. 7, star markers) with the addition of C. Hence, the ductility is not strongly influenced by the carbon addition. Furthermore, the film hardness is directly related to the amount of B-C contributions to the B1s XPS spectra (see Fig. 8b), which is an indirect measure of a-BCx tissue phase. Also the H3/E2 ratio (a measure of the resistance to plastic deformation) was investigated, but no interesting trends could be observed for the present series of samples. One of the possible parameters that influence the measured hardness compared to the “true” material hardness is stress in the film. The calculated stresses are given in Table II. As can be seen, the binary sample exhibits a tensile stress of 0.1 GPa. As carbon is introduced, the stress changes from tensile to compressive and increases until 23 at.% of C, where a tensile stress of about 1 GPa is observed. As more carbon is introduced, no significant changes occur. Considering these, relatively low stresses, hardening due to stress or strain must be considered to be negligible. Results from the tribological tests are summarised in Table III, where it can be seen that all samples exhibit rather high friction. Initially the COF is low, but rises quickly (after 100 to 200 revolutions) and becomes rather high (N 0.8). In Table III the average COF values for the last 4000 laps are given. Two groups of films that can be distinguished with

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

511

Fig. 5. Cross section STEM and EDX maps of binary Mo-B and Mo-B-C samples with 31 at.% of C grown at 300 °C and 600 °C.

respect to friction behaviour: one group with a carbon-content below ~20 at.%, and another group with carbon-contents above that. The low carbon-content group shows a rapid increase in the friction coefficient,

which eventually stabilises (~500 revolutions) at about 1.1. The group with higher carbon-content shows a somewhat longer initial stage (up to ~ 1000 revolutions), and the friction coefficient eventually reaches

Fig. 6. High resolution XPS B1s and C1s spectra from Mo-B and Mo-B-C samples up to 37 at.% C, including HT sample. Dashed lines are the guide for an eye to indicate the fitted peak positions.

512

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

Fig. 7. Normalised Raman spectra of surfaces (a) and wear tracks (b) for binary Mo-B and Mo-B-C samples with reference peaks for MoO3 [62], Fe2O3 [63] and B2O3 [64].

0.9. The 31 at.% of C sample grown at higher temperature (600 °C) exhibits the same friction coefficient as the comparable sample grown at lower temperature, but there is much more oscillations around the value indicating uneven wear. All the samples exhibited similar depth and width of the wear tracks (~500 nm and 150 μm, respectively, as measured after ~ 4000 revolutions. The observed wear rate (see Table III) is similar for all the samples with a combined average value of 1.5 ± 0.2 × 10−5 mm3/Nm (assuming there is no significant ball wear). It should be noted that the given wear rate is approximate, as the actual wear tracks did not have the ideal shape, partially due to transfer from the steel ball. In order to analyse the cause of tribological properties the wear tracks were analysed with Raman, SEM/EDX and XPS. Fig. 7 shows normalised Raman spectra of the as-deposited film (a) and the wear tracks (b) of all samples. The wear tracks of the binary sample, as well as the ternary samples with up to about 20 at.% C, exhibit strong and distinct Raman features. These are identified as molybdenum oxide, iron oxide and minor indications of boron oxide. On the other hand, the ternary samples with a carbon-content higher than about 20 at.% exhibit only increased carbon D- and G-bands in the wear track compared to the original surface, and the spectra are dominated by the wide Dand G-bands. For these samples, only minor peaks from MoO3 are visible and no peaks from B2O3. It is thus likely that the two regions of

tribological behaviour are connected to the formation of oxides in the wear tracks. SEM imaging (Fig. 9) confirm that the coatings are not worn through after 4000 laps, and reveal that the wear track of the binary sample exhibits a rough surface and cracks, while both samples with 31 at.% C have a much smoother wear track, indicating a smoother wear behaviour. EDX maps for binary MoBx is shown in Ref. [6]. These reveal that the samples are not worn through. Also it shows that the small amount of steel counter surface which can be seen in the wear-track (Fig. 9) consisting mainly of chromium and iron oxides. This is also true for Mo-B-C samples with different carbon content (not shown here). An interesting observation is that the small particles mainly consist of chromium oxide, while big ones are iron oxides with chromium oxides. In the Raman measurement, on the other hand, only Fe2O3 is detected, which most likely depend on the negligible amount of Cr-O particles which is assumed to be too small to give an observable signal. In both cases it is a clear indication of high friction and destruction of the steel counter surface, as neither Fe, nor Cr are present in the coating material and are not observed after the friction test with alumina counter surface. XPS analysis of the wear tracks was also performed; however, the results were similar for all coatings, so no spectra are shown. The XPS B1s peak was not detectable, indicating no B in the wear tracks. The Mo3d region exhibited clear Mo-O contributions in all samples. Finally, the

Fig. 8. a) Hardness (circles), elastic modulus (triangles) and H/E ratio (stars) of Mo-B-C samples up to 37 at.% of C b) Hardness (circles) and resistivity (triangles) plotted against the relative B-C contribution to the B1s XPS spectra (see Fig. 6 and Table I). Solid markers used for samples deposited at 300 °C, open for the sample deposited at 600 °C, lines are intended as guides to the eye.

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515 Table III Tribological properties of MoBx and Mo-B-C thin films: COF - coefficient of friction, as derived from, pin-on-disk measurements. Given values are averages and 95% confidence intervals from the last 4000 revolutions in respective test. Wear rate was calculated from optical profilometry data, assuming even wear and no ball wear. Sample (at. %)

(COF) against steel (100Cr6)

COF against alumina (Al2O3)

Wear rate (mm3/Nm)

0 8 16 23 31 31 HT 37

1.10 1.08 1.13 0.98 0.93 0.91 0.90

0.60 ± 0.02

(1.3 (1.5 (1.6 (1.7 (1.3 (1.1 (1.9

± ± ± ± ± ± ±

0.03 0.03 0.02 0.06 0.03 0.0.73 0.05

0.75 ± 0.02

± ± ± ± ± ± ±

0.2) 0.3) 0.3) 0.3) 0.3) 0.2) 0.2)

· · · · · · ·

10−14 10−14 10−14 10−14 10−14 10−14 10−14

C1s region of the unworn surface and in the wear tracks of all samples contained C\\O and C\\C bonds contributions, however more pronounced in the wear track. To further investigate the tribological behaviour, and to exclude that the observed carbon based tribofilm originates from the steel counter surface, additional tests using an alumina ball was performed on the binary and the 31 at.% C samples. The measured friction values were 0.6 and 0.75, respectively (Table III), thus clearly showing a lower (but still high) friction coefficient against alumina. No significant differences in the wear track were observed by SEM, XPS or Raman, compared to when sliding against steel, thus showing that the observed tribofilm primarily is formed by the Mo-B-C coatings. To investigate the film electrical properties, resistivity measurements were performed (Fig. 8b) and it was found that the resistivity is increasing with carbon content. The binary sample exhibits a resistivity of 175 ± 40 μΩ cm. For the samples deposited at 300 °C (filled triangles) the resistivity is linearly increasing with B-C contribution to the B1s XPS spectra, reaching 395 ± 33 μΩ cm for the 37 at.% C sample. The HT sample (Fig. 8b open triangle) exhibit higher resistivity than the rest of the films (499 ± 16 μΩ cm) and does not follow the same trend as the other samples.

513

4. Discussion The Mo-B thin films consist of hexagonal hP3-MoB2 − x grains surrounded by a boron-rich tissue phase as reported previously [6]. The precise chemical composition of this tissue phase is unclear and small amounts of Mo may be dissolved into the boron-rich phase as well. The sputter-deposition of sub-stoichiometric metal diboride coatings with a boron rich tissue phase has been reported for several systems previously, including NbB1.8 [4] and CrB1.87 [13] and CrB1.5 [5]. Upon alloying with carbon, the crystallinity of the Mo-B-C films is reduced compared to MoB2 − x The grain size of the MoB2 − x grains decreases rapidly with C content and approaches ~ 2 nm at 23 at.% C (Table II). Above 23 at.% C, the films are X-ray amorphous. The general microstructural evolution observed for the Mo-B-C materials has been observed upon carbon or nitrogen alloying in other sputter-deposited materials based on metal borides as well [5,10–13]. In particular the microstructural evolution of the present material show large similarities with what has been observed previous studies on Nb-B-C [10] and CrB-N [12,13]. In the following, we will discuss the carbon-rich (N23 at.%C) and the carbon-poor (b23 at.% C) films separately. 4.1. Carbon-poor (b 23 at.% C) Mo-B-C The diffractograms in Fig. 2 show that the films deposited at 300 °C with b23 at.% C contain nanocrystallites of MoB2 − x. A significant amount of the carbon is bonded in a BCx-phase surrounding the MeB2 − x nanocrystallites. The observed reduction in grain size with increased carbon content could then be explained by a so-called grain boundary decoration process [48–50], where the limited solubility of carbon in the boride grains leads to an enrichment of carbon at the boride grain surface thereby reducing grain growth leading to re-nucleation of new boride grains. Similar microstructures have previously been observed in similar magnetron sputtered Me-B-C systems, where Me_Ti [9], Nb [10] and Cr [5]. It is possible that the amorphous BCxphase also contains Mo but this cannot be established from available data. A recent APT (Atom Probe Tomography) study of Cr-B-C films have showed that the amorphous BCx-phase can contain up to 15 at.% of Cr [51]. There are strong evidences that some C can be dissolved in the MoB2 − x phase as well. As the total carbon content increases, there is an increase in unit cell volume from 24.7 to 26.12 Å3, see Table II. This expansion is larger than would be expected from the observed stresses in the films. Furthermore, the XPS results suggest a C-Mo contribution in the C1s spectra, which is expected for a solid solution of C in MoB2 − x. At equilibrium, the solid solubility of C in metal diborides is very low but since magnetron sputtering is carried out at non-equilibrium conditions, a significant amount of carbon may be dissolved e.g. on vacant boron sites. Such a solid solution could also be the explanation for the shift observed in the B-Mo contribution to the B1s spectra, which indicate that the bonding in the boride phase is altered at lower carbon contents. A solid solubility of C in transition metal borides has been suggested also in other magnetron sputtered Me-B-C systems but no detailed study of maximum solubility carbon in metal diborides and its effect on properties have yet been performed. 4.2. Carbon-rich (N 23 at.% C) Mo-B-C

Fig. 9. SEM images of wear tracks after completed pin-on-disk measurements: a) MoBx sample, b) Mo-B-C sample with 31 at.% C and c) HT Mo-B-C sample with 31 at.% C.

For carbon contents N23 at.%, the films deposited at 300 °C are X-ray amorphous although some nanocrystallites of diboride were observed by TEM. This is in concord with the Cr-B-C system which shows a microstructure of two amorphous phases; a Cr-rich amorphous phase and aCr poor amorphous phase. Mo also belongs to the group 6 transition metals and obviously exhibits a similar behaviour. The TEM results in Figs. 4 and 5 suggest the presence of a Mo-rich phase which most likely also contain both B and C and a second amorphous Mo-poor phase probably dominated by B\\C bonds although some Mo cannot be

514

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

excluded in this phase in analogy with the Cr-B-C case [5]. The presence of the small amount of crystallites may indicate that the formation of metal diboride grains during film growth is more favourable in the Mo-B-C system than for Cr-B-C. However, it cannot be excluded that the nanocrystallites are formed during the TEM analysis due to beam damage effects [52,53]. The sample deposited at higher temperature (600 °C) compared to low temperature (300 °C) equivalent showed an increased volume of the crystalline diboride phase and a larger amount of a-BCx tissue phase. Furthermore, the crystalline cell volume decreases to a level comparable with the 8 at.% C sample. Our observations can be explained by the increased adatom mobility at higher temperatures, allowing extended diffusion lengths with a more accentuated phase separation where C is not trapped in an amorphous Mo-rich phase and instead contributes to an increased amount of a-BCx tissue phase and a more extensive crystallization of MoB2 − x. 4.3. Functional properties In the Mo-B-C thin films, both the hardness and elastic modulus decrease with increasing carbon content, similar to what is observed in the Cr-B-N system [12,13]. This is not surprising considering how similar they are in terms of structure, as discussed before. As the structure progresses with the same trends so does the properties. The hardness decrease correlates well with increase in relative amount of a-BCx tissue phase (see Fig. 8b), which thus is the likely cause of the softening [54, 55]. Crystalline B4C is the third hardest material after diamond and cubic boron nitride, however amorphous material is known to exhibit rather low hardness (8–9 GPa) [54,55]. Since the sample deposited at higher temperature fall into the same linear trend (see Fig. 8b) it is likely that the amount of tissue phase is what primarily is determining the material hardness for these samples. This decrease in hardness is also in agreement with what has been observed in related metal boride based systems [5,10–13]. As noted above, the H/E ratio increases with addition of carbon, and thus indicates a slightly more ductile behaviour. This is an interesting observation, considering that both Nb-B-C and Cr-B-C in the work of Nedfors et al. (which exhibit similar values) show the opposite trend [5,10]. There is presently no firm explanation for this difference and the possibility of difference in energy of ion bombardment [56] can be excluded as Nedfors et al. [5,10] conducted their experiments in the same equipment, and thus has the same low ion bombardment energy. However, the low H/E index (b0.1) and tensile stress suggest that the films would have low resistance to cracking, which probably is worse for the high temperature sample due to its columnar morphology [56]. One of the aims of this study was to investigate the tribological properties, or rather change of them with the addition of carbon to the binary molybdenum diboride. Carbon-containing films are well known to form lubricating graphitic tribofilms. Furthermore, it has been suggested that MoO3 and B2O3 [57,58] can act as low friction materials. Having all these ingredients in the Mo-B-C thin films, as well as the metastable hexagonal hP3-MoB2 − x phase, which could decompose in a tribological contact, give large potential for the formation of a lubricating tribolayer. However, despite a significant formation of molybdenum and boron oxides, as well as carbon at the tribo-surface, the observed friction coefficient of the Mo-B films was very high (COF N 1), and decreased only to 0.9 for highest carbon content (see Table III). The similar behaviour was observed in Cr-B-N system, where incorporation of N resulted in COF 0.6 or higher [12,13]. Hence, the coatings are hardly low friction materials although there is a small reduction of the friction coefficient as carbon is introduced. Raman studies of the wear tracks revealed that MoO3 and B2O3 formation is greatly reduced in the samples with higher than 20 at.% C. This correlates well with a lower friction coefficient and shows that under the tested circumstances (ambient temperature) the oxides are not low lubricating, but rather the opposite. Raman studies of the wear tracks also reveal stronger carbon D and G bands, which correlates with lower friction coefficient. It thus seems

likely that the carbon at least partially have the intended effect, preventing the formation of high friction oxides, and thus lowering the friction. The previous lubricating experiments of molybdenum oxides were conducted at higher temperatures (N400 °C) [57]. Hence, it may still be possible that the Mo-B-C coatings exhibit low friction coefficient at higher temperatures. The observed electrical resistivity of 175 ± 40 μΩ cm for the binary sample which is clearly higher than for bulk MoB2 (45 μΩ cm) [59], but very similar to what have been observed for other binary transition metal boride films [4,5]. Also the span of the ternary materials, with resistivity up to 499 ± 16 μΩ cm are similar to what has been observed for Nb-B-C thin films [10]. The observed increase in resistivity for the samples deposited at 300 °C correlates well B-C contribution to the B1s XPS spectra (see Fig. 8b), and could therefore be explained by an increasing amount of covalent bonds in the material. However, as the sample deposited at 600 °C does not follow this trend, additional variables are crucial for the properties. One of the motivations for this study was potential use in sliding electrical contacts. The observed resistivity in combination with the lowering of the hardness could give a low contact resistance. However, the observed coefficient of friction (~ 0.8) and wear rate (1.6 ± 0.1 × 10−5 mm3/Nm) are both high, and the requirement of a coefficient of friction below 0.2 is clearly not met. Thus no further analysis into the suitability was performed. 5. Conclusions Sputter deposited Mo-B films can be described as nanocomposites of nanocrystalline hP3-MoB2 − x grains surrounded by a tissue phase of aB. Upon alloying with carbon, the crystallite size decreases and the films are mainly amorphous above 23 at.% C. Addition of carbon is mainly forming an a-BCx tissue phase surrounding the boride phase. The compositions of the a-B and a-BCx tissue phases could not be quantified, but it is not excluded that tissue phases can contain metal. XRD results suggest that some carbon is dissolved into the metal-boride structure. Formation of an amorphous boride phase at high carbon contents is also observed in the Cr-B-C system and may be typical for group 6 transition metals. The observed film hardness and elastic modulus decrease with increased carbon content. This trend was directly correlated to the amount of a-BCx tissue phase in the films and suggests that the mechanical properties are strongly affected by the amount of tissue phase. The H/E ratio increases with addition of carbon, and, thus, indicates a slightly more ductile behaviour. Binary MoB2 − x thin film exhibited high friction coefficient against the steel counter surface of 1.1. Addition of carbon was investigated in order to reduce friction, and a lower friction coefficient (~0.8) was observed for carbon contents above about 23 at.%. Raman measurements on wear tracks revealed that this was correlated to the presence of stronger carbon D and G bands typical of amorphous carbon based phases. Previous observations in literature of a lubricating effect of molybdenum oxides and boron oxides were conducted at higher temperatures (N400 °C) [57], hence the Mo-B-C coatings may exhibit low friction coefficient at higher temperatures. Resistivity measurement revealed moderate resistivity spanning from 175 ± 40 μΩ cm for binary films to 499 ± 16 μΩ cm for HT Mo-B-C films with 31 at.%. Resistivity increase was also directly correlated to the amount of BCx tissue phase. Our study showed that Mo-B-C films have a combination of mechanical and tribological properties making them less useful as sliding electric contacts. Acknowledgements The authors acknowledge The Swedish Research Council for funding under Grant Nos. 621-2012-4359 and 622-2008-405, as well as the Knut and Alice Wallenbergs Foundation for support of the electron microscopy laboratory in Linkoping. Daniel Primetzhofer at the Tandem

P. Malinovskis et al. / Surface & Coatings Technology 309 (2017) 506–515

Laboratory, Uppsala University, is acknowledged for the assistance with ERDA measurements. The Swedish Foundation for Strategic Research, SSF (FunCase, RMA11-0029) is also acknowledged for financial support. References [1] A.L. Ivanovskii, Mechanical and electronic properties of diborides of transition 3d– 5d metals from first principles: toward search of novel ultra-incompressible and superhard materials, Prog. Mater. Sci. 57 (2012) 184–228. [2] C. Mitterer, Borides in thin film technology, J. Solid State Chem. 133 (1997) 279–291. [3] P.H. Mayrhofer, C. Mitterer, J.G. Wen, J.E. Greene, I. Petrov, Self-organized nanocolumnar structure in superhard TiB2 thin films, Appl. Phys. Lett. 86 (2005) 131909. [4] N. Nedfors, O. Tengstrand, J. Lu, P. Eklund, P.O.Å. Persson, L. Hultman, et al., Superhard NbB2 − x thin films deposited by dc magnetron sputtering, Surf. Coat. Technol. 257 (2014) 295–300. [5] N. Nedfors, D. Primetzhofer, L. Wang, J. Lu, L. Hultman, U. Jansson, Characterization of magnetron sputtered Cr–B and Cr–B–C thin films for electrical contact applications, Surf. Coat. Technol. 266 (2015) 167–176. [6] P. Malinovskis, J. Palisaitis, P.O.Å. Persson, E. Lewin, U. Jansson, Synthesis and characterization of MoB2 − x thin films grown by nonreactive DC magnetron sputtering, J. Vac. Sci. Technol. A 34 (2016) 31511. [7] E. Kelesoglu, C. Mitterer, Structure and properties of TiB2 based coatings prepared by unbalanced DC magnetron sputtering, Surf. Coat. Technol. 98 (1998) 1483–1489. [8] J.-T. Ok, I.-W. Park, J.J. Moore, M.C. Kang, K.H. Kim, Syntheses and mechanical properties of Ti–B–C coatings by a plasma-enhanced chemical vapor deposition, Surf. Coat. Technol. 200 (2005) 1418–1423. [9] J. Lauridsen, N. Nedfors, U. Jansson, J. Jensen, P. Eklund, L. Hultman, Ti-B-C nanocomposite coatings deposited by magnetron sputtering, Appl. Surf. Sci. 258 (2012) 9907–9912. [10] N. Nedfors, O. Tengstrand, P. Eklund, L. Hultman, U. Jansson, Nb-B-C thin films for electrical contact applications deposited by magnetron sputtering, J. Vac. Sci. Technol. A 32 (2014) 41503. [11] M. Stüber, V. Schier, H. Holleck, Properties and performance of new metastable Ti-BC-N hard coatings prepared by magnetron sputtering, Surf. Coat. Technol. 74–75 (1995) 833–837. [12] K.P. Budna, P.H. Mayrhofer, J. Neidhardt, É. Hegedũs, I. Kovács, L. Tóth, et al., Effect of nitrogen-incorporation on structure, properties and performance of magnetron sputtered CrB2, Surf. Coat. Technol. 202 (2008) 3088–3093. [13] B. Mallia, M. Stüber, P.A. Dearnley, Character and chemical-wear response of high alloy austenitic stainless steel (Ortron 90) surface engineered with magnetron sputtered Cr-B-N ternary alloy coatings, Thin Solid Films 549 (2013) 216–223. [14] J.C. Qian, Z.F. Zhou, C. Yan, D.J. Li, K.Y. Li, S. Descartes, et al., Tailoring the mechanical and tribological properties of sputtered boron carbide films via the B1 − xCx composition, Surf. Coat. Technol. 267 (2015) 2–7. [15] T. Hu, L. Steihl, W. Rafaniello, T. Fawcett, D.D. Hawn, J.G. Mashall, et al., Structures and properties of disordered boron carbide coatings generated by magnetron sputtering, Thin Solid Films 332 (1998) 80–86. [16] J. Emmerlich, D. Music, M. Braun, P. Fayek, F. Munnik, J.M. Schneider, A proposal for an unusually stiff and moderately ductile hard coating material: Mo2BC, J. Phys. D. Appl. Phys. 42 (2009) 185406. [17] H. Bolvardi, J. Emmerlich, S. Mráz, M. Arndt, H. Rudigier, J.M. Schneider, Low temperature synthesis of Mo2BC thin films, Thin Solid Films 542 (2013) 5–7. [18] S. Djaziri, S. Gleich, H. Bolvardi, C. Kirchlechner, M. Hans, C. Scheu, et al., Are Mo2BC nanocrystalline coatings damage resistant? Insights from comparative tension experiments, Surf. Coat. Technol. 289 (2016) 213–218. [19] Y. Zhang, H.J. Whitlow, T. Winzell, I.F. Bubb, T. Sajavaara, K. Arstila, et al., Detection efficiency of time-of-flight energy elastic recoil detection analysis systems, Nucl. Instrum. Methods Phys. Res., Sect. B 149 (1999) 477–489. [20] M.S. Janson, CONTES Instruction Manual, Intern. Rep, Uppsala, 2004. [21] P. Scardi, M. Leoni, R. Delhez, Line broadening analysis using integral breadth methods: a critical review, J. Appl. Crystallogr. 37 (2004) 381–390. [22] W.C. Oliver, G.M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, J. Mater. Res. 7 (2011) 1564–1583. [23] J.F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys. 24 (1953) 981–988. [24] G.G. Stoney, The tension of metallic films deposited by electrolysis, Proc. R. Soc. A Math. Phys. Eng. Sci. 82 (1909) 172–175. [25] G.C.A.M. Janssen, M.M. Abdalla, F. van Keulen, B.R. Pujada, B. van Venrooy, Celebrating the 100th anniversary of the Stoney equation for film stress: developments from polycrystalline steel strips to single crystal silicon wafers, Thin Solid Films 517 (2009) 1858–1867. [26] K.E. Spear, P.K. Liao, The B-Mo (boron-molybdenum) system, Bull. Alloy Phase Diagr. 9 (1988) 457–466. [27] S. Okada, T. Atoda, I. Higashi, Y. Takahashi, Preparation of single crystals of MoB2 by the aluminium-flux technique and some of their properties, J. Mater. Sci. 22 (1987) 2993–2999. [28] J.J.F. Moulder, W.W.F. Stickle, P.E. Sobol, K.K.D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, 1995. [29] G. Mavel, J. Escard, P. Costa, J. Castaing, ESCA surface study of metal borides, Surf. Sci. 35 (1973) 109–116. [30] Z. Wang, Y. Shimizu, T. Sasaki, K. Kirihara, K. Kawaguchi, K. Kimura, et al., Fabrication of crystallized boron films by laser ablation, J. Solid State Chem. 177 (2004) 1639–1645.

515

[31] C.W. Ong, H. Huang, B. Zheng, R.W.M. Kwok, Y.Y. Hui, W.M. Lau, X-ray photoemission spectroscopy of nonmetallic materials: electronic structures of boron and BxOy, J. Appl. Phys. 95 (2004) 3527–3534. [32] H.A. Castillo, E. Restrepo-Parra, J.M. Vélez, W. de la Cruz, Substrate temperature influence on boron carbide coatings grown by the PLD technique, Surf. Coat. Technol. 205 (2011) 3607–3612. [33] Y. Xie, Ultrahigh purity graphite electrode by core level and valence band XPS, Surf. Sci. Spectra 1 (1992) 367. [34] M.A. Baker, Advanced characterisation of nanocomposite coatings, Surf. Coat. Technol. 201 (2007) 6105–6111. [35] M.D. Abad, J.C. Sánchez-López, M. Brizuela, A. García-Luis, D.V. Shtansky, Influence of carbon chemical bonding on the tribological behavior of sputtered nanocomposite TiBC/a-C coatings, Thin Solid Films 518 (2010) 5546–5552. [36] A. Furlan, U. Jansson, J. Lu, L. Hultman, M. Magnuson, Structure and bonding in amorphous iron carbide thin films, J. Phys. Condens. Matter 27 (2015) 45002. [37] J. Lu, U. Jansson, Chemical vapour deposition of molybdenum carbides: aspects of nanocrystallinity, Thin Solid Films 396 (2001) 53–61. [38] L.G. Jacobsohn, R.K. Schulze, M.E.H. Maia da Costa, M. Nastasi, X-ray photoelectron spectroscopy investigation of boron carbide films deposited by sputtering, Surf. Sci. 572 (2004) 418–424. [39] J.C. Qian, Z.F. Zhou, W.J. Zhang, K.Y. Li, I. Bello, L. Martinu, et al., Microstructure and tribo-mechanical properties of Ti-B-C nanocomposite films prepared by magnetron sputtering, Surf. Coat. Technol. 270 (2015) 290–298. [40] R. Bao, D.B. Chrisey, Chemical states of carbon in amorphous boron carbide thin films deposited by radio frequency magnetron sputtering, Thin Solid Films 519 (2010) 164–168. [41] A.V. Shchukarev, D.V. Korolkov, XPS study of group IA carbonates, Cent. Eur. J. Chem. 2 (2004) 347–362. [42] J. Hodkiewicz, T.F. Scientific, Characterizing Carbon Materials with Raman Spectroscopy, Appl. Note 51901, 2010 1–5. [43] X.Q. Yan, W.J. Li, T. Goto, M.W. Chen, Raman spectroscopy of pressure-induced amorphous boron carbide, Appl. Phys. Lett. 88 (2006) 131905. [44] A. Leyland, A. Matthews, Design criteria for wear-resistant nanostructured and glassy-metal coatings, Surf. Coat. Technol. 177–178 (2004) 317–324. [45] A. Leyland, A. Matthews, On the significance of the H/E ratio in wear control: a nanocomposite coating approach to optimised tribological behaviour, Wear 246 (2000) 1–11. [46] J. Musil, J. Sklenka, R. Čerstvý, Protection of brittle film against cracking, Appl. Surf. Sci. 370 (2016) 306–311. [47] J. Musil, Hard nanocomposite coatings: thermal stability, oxidation resistance and toughness, Surf. Coat. Technol. 207 (2012) 50–65. [48] J. Patscheider, N. Hellgren, R.T. Haasch, I. Petrov, J.E. Greene, Electronic structure of the SiNx/TiN interface: a model system for superhard nanocomposites, Phys. Rev. B: Condens. Matter Mater. Phys. 83 (2011) 1–7. [49] T. Zehnder, J. Patscheider, Nanocomposite TiC/a–C:H hard coatings deposited by reactive PVD, Surf. Coat. Technol. 133–134 (2000) 138–144. [50] T. Zehnder, P. Schwaller, F. Munnik, S. Mikhailov, J. Patscheider, Nanostructural and mechanical properties of nanocomposite nc-TiC/a-C:H films deposited by reactive unbalanced magnetron sputtering, J. Appl. Phys. 95 (2004) 4327–4334. [51] T. Boll, M. Thuvander, S. Koch, J.N. Wagner, N. Nedfors, U. Jansson, et al., An APT investigation of an amorphous Cr-B-C thin film, Ultramicroscopy 159 (2015) 217–222. [52] O. Tengstrand, N. Nedfors, M. Andersson, J. Lu, U. Jansson, A. Flink, et al., Beam-induced crystallization of amorphous Me–Si–C (Me_Nb or Zr) thin films during transmission electron microscopy, MRS Commun. 3 (2013) 151–155. [53] O. Tengstrand, N. Nedfors, M. Andersson, J. Lu, U. Jansson, A. Flink, et al., Model for electron-beam-induced crystallization of amorphous Me–Si–C (Me_Nb or Zr) thin films, J. Mater. Res. 29 (2014) 2854–2862. [54] S. Liza, N. Ohtake, H. Akasaka, J.M. Munoz Guijosa, Tribological and thermal stability study of nanoporous amorphous boron carbide films prepared by pulsed plasma chemical vapor deposition, Sci. Technol. Adv. Mater. 16 (2016) 35007. [55] J. Braddock-wilking, S. Lin, B.J. Feldman, Evidence for and role of boron carbide icosahedra in the hardness of amorphous hydrogenated boron carbide, Tribol. Lett. 5 (1998) 145–148. [56] J. Musil, Flexible hard nanocomposite coatings, RSC Adv. 5 (2015) 60482–60495. [57] A. Erdemir, A crystal-chemical approach to lubrication by solid oxides, Tribol. Lett. 8 (2000) 97–102. [58] A. Erdemir, A crystal chemical approach to the formulation of self-lubricating nanocomposite coatings, Surf. Coat. Technol. 200 (2005) 1792–1796. [59] W. Martienssen, H. Warlimont, W.M. Prof, H.W. Prof., Springer Handbook of Condensed Matter and Materials Data, Springer, Berlin Heidelberg, 2005. [60] E. Rudy, F. Benesovsky, L. Toth, Mo-B-C phase diagram, ASM alloy phase diagrams database, P. Villars, editor-in-chief; H. Okamoto and K. Cenzual, section editors; ASM Int. Mater. Park. OH. (2016). [61] E. Rudy, F. Benesovsky, L. Toth, Untersuchung der Dreistoffsysteme der Va- und VIaMetalle mit Bor und Kohlenstoff, Z. Met. 54 (1963) 345–353. [62] B. Lafuente, R.T. Downs, H. Yang, N. Stone, The power of databases: the RRUFF project, R100217, in: T. Armbruster, R.M. Danisi (Eds.), Highlights in Mineralogic Crystallography, W. Gruyter, Berlin, Germany 2015, pp. 1–30. [63] B. Lafuente, R.T. Downs, H. Yang, N. Stone, The power of databases: The RRUFF project, R060190, in: T. Armbruster, R.M. Danisi (Eds.), Highlights in Mineralogic Crystallography, W. Gruyter, Berlin, Germany 2015, pp. 1–30. [64] R. Machaka, B.W. Mwakikunga, E. Manikandan, T.E. Derry, I. Sigalas, Raman spectra of hot-pressed boron suboxide, Adv. Mater. Lett. 2 (2011) 58–64.