- Email: [email protected]
Mechanical properties optimization of tungsten nitride thin films grown by reactive sputtering and laser ablation
Vacuum 85 (2010) 69e77
Contents lists available at ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Mechanical properties optimization of tungsten nitride thin films grown by reactive sputtering and laser ablation E.C. Samano*, A. Clemente, J.A. Díaz, G. Soto CNyN-Universidad Nacional Autónoma de México, A. Postal 356, Ensenada, BC, Mexico
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 July 2009 Received in revised form 22 March 2010 Accepted 3 April 2010
Transition metal nitrides coatings are used as protective coatings against wear and corrosion. Their mechanical properties can be tailored by tuning the nitrogen content during film synthesis. The relationship between thin film preparation conditions and mechanical properties for tungsten nitride films is not as well understood as other transition metal nitrides, like titanium nitride. We report the synthesis of tungsten nitride films grown by reactive sputtering and laser ablation in the ambient of N2 or N2/Ar mixture at various pressures on stainless steel substrates at 400 C. The composition of the films was determined by XPS. The optimal mechanical properties were found by nanoindentation based on the determination of the proper deposition conditions. As nitrogen pressure was increased during processing, the stoichiometry and hardness changed from W9N to W4N and 30.8e38.7 GPa, respectively, for films deposited by reactive sputtering, and from W6N to W2N and 19.5e27.7 GPa, respectively, for those deposited by laser ablation. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: Hard coatings Nanoindentation Tungsten nitride Reactive sputtering RPLD
1. Introduction The term of hard coatings is applied to structures which improve wear resistance and extend the lifetime of the structure. Hard coatings are industrially used onto cutting tools, automotive engine parts, turbine blades, structural components, etc [1,2]. Traditionally, materials utilized as hard coatings had a single composition, crystalline phase, and microstructure. Some of the most common hard coatings used as thin films are diamond-like carbon (DLC), BN, B4C, SiC, Al2O3, Si3N4 and WC [1]. A growing interest in the synthesis of transition metal nitrides has risen due to their chemical inertness and high hardness to be used as wearresistant and hard protective coatings. In particular, TiN thin films and nanocomposites using TiN satisfy these requirements and have been extensively investigated by a variety of deposition techniques in the recent literature [1e5]. For instance, there is a close relationship between film composition and its mechanical properties for TiN films [3]. However, there are not as many reports as expected on other transition metal nitrides. Tungsten nitride films have generated considerable interest in the manufacturing of semiconductor devices because they are an excellent barrier for Cu diffusion into Si at high temperatures [6]. The deposition of diffusion barriers motivated the synthesis of WNx
* Corresponding author. Tel.: þ52 646 1744602; fax: þ52 646 1744603. E-mail address: [email protected] (E.C. Samano). 0042-207X/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.vacuum.2010.04.004
thin films by several techniques like CVD [7e9], dc reactive sputtering [10e13], ALD [14], rf reactive sputtering [15,16], ion beam sputtering [17], reactive laser ablation [18,19] and cathodic arc [20]. There are reports concerning to the mechanical properties of alloys containing tungsten nitride sputtered coatings, like CreWeN [21], TieWeN [22] and SieWeN [23]. WNx thin films are also a good candidate to be used as a hard coating [9,13,16,19,20]. The composition, mass density, electrical and optical properties of WNx thin films on silicon wafers at room temperature grown by reactive laser ablation have been already published by our group but the study of the mechanical properties was missing [18]. This study investigates the best possible mechanical properties as a function of film composition and microstructure for WNx coatings on stainless steel substrates synthesized by two different deposition techniques: reactive sputtering and laser ablation. 2. Experiment As previously mentioned, the WNx thin films were grown by two different deposition methods: RPLD and dc-sputtering. The experiment was carried out in a laser ablation system described elsewhere [24]. The system consists of three UHV stainless steel chambers: sample introduction, growth and analysis, as shown in Fig. 1. The base pressure of the chambers is approximately 107 Pa. A KrF excimer laser is focused onto a high purity (99.9% at.) tungsten disc of 5 cm in diameter at an angle of 50 off the surface normal in the growth or deposition chamber. The values of laser
70
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 1. Cross sectional view of the reactive pulsed laser deposition system. The gas control system is also displayed.
energy, number of pulses, and repetition rate were optimized in a previous experiment and were kept fixed at 260 mJ, 18,000 and 5 Hz, respectively [18]. The film composition was studied by introducing N2 gas in the 0e10.0 Pa range. The ablated species were deposited on stainless steel ANSI304 substrates kept at 400 C. Every film was in situ characterized by XPS at the end of the film processing. The films were also grown in a reactive dc magnetron sputtering system, as shown in Fig. 2. A current-regulated dc power supply was used to provide a discharge with an input power of 100 W and a target-substrate separation of 3 cm. The target diameter and thickness are 5 cm and 0.6 cm, respectively. The power density is approximately 5 W/cm2 with an input current of 170 mA. The WNx films were grown on stainless steel ANSI304 substrates at 400 C by using the same tungsten target described above. Prior to deposition, the system was pumped down to a base pressure of 103 Pa. Afterwards, a flow of an Ar/N2 gas mixture, independently regulated by mass-flow controllers, fed the deposition chamber. The partial Ar pressure was kept constant at 0.53 Pa for the whole experiment, while the N2 partial pressure was varied from 0 to 1.6 Pa. Before film processing, the target was sputter cleaned for 5 min, with a shutter shielding the substrate. A deposition time of 10 min was used for every experimental run, the same for all depositions. The total pressure, Pt, as read from a Pirani-gauge was led to an analog input in a computer. This signal is compared to the preset pressure; the error signal is supplied to a proportional, integral, derivative (PID) algorithm. The PID output is used to compensate the gas flow controlled by the mass-flow controllers. The partial pressure of the reactive gas in the deposition chamber was estimated by knowing the total pressure, chamber volume, pumping speed, elapsed time, and gas flow of both gases [25]. This scheme is a closed-loop partial pressure regulator and it can be implemented to produce gas mixtures in a controlled way. For RPLD the same scheme was applied, even if nitrogen was the only gas. A Dektak profilemeter is utilized to measure the thickness of all samples. The thickness was between 0.5 and 0.7 mm for films grown by RPLD, and between 1.5 and 1.8 mm by reactive sputtering. A Philips difractometer, model X’pert, was employed to determine the crystallinity of the samples by X-ray diffraction (XRD) using the Cu Ka line with a 0.154 nm wavelength. The film morphology and roughness were determined by atomic force
microscopy (AFM) using a NanoScope III, Veeco Instruments, in contact mode. Several images of the samples surface were taken at scales of 10, 5, 2, 1 mm and 500 nm. The mechanical properties of the samples were measured using a TriboScope nanoindenter, Hysitron. The nanoindenter is used in conjunction with the AFM so that the area of indentation can be visualized with the same probe. The probe is a commercial Berkovich diamond tip attached to a transducer. This unit replaces the conventional optical head with a cantilever of the NanoScope III AFM. The tip penetrates into the sample from zero up to a maximum predetermined force (loading region); then, the tip returns to its original position (unloading region). The “golden rule” in nanoindentation to be used in the study of the mechanical properties of thin films is that the maximum depth of penetration in a test must be no more than 10% of film thickness [26,27]. The graph showing the relationship between force or load and penetration or displacement is called the indentation curve. The hardness, H, and reduced elastic modulus, Er, were found from the unloading region of the indentation curve by means of the theory developed by Oliver and Pharr [28,29], and corrections to this theory for very shallow indentations [30]. After choosing a region on the sample, 4 indentations (for 4 different loads) were performed. These loads were 200, 400, 800 and 1500 mN; and 800, 1500, 2000 and 3000 mN for films processed by RPLD and reactive sputtering, respectively. A number of 4 different regions, corresponding to a total of 16 indentations, were analyzed. 3. Results The WNx thin films were deposited on stainless steel disks kept at 400 C on both deposition methods. The nitrogen partial pressure was varied from 0 to 10.0 Pa in the RPLD system and from 0 to 1.6 Pa in the sputtering system. The analysis was in situ performed immediately after each deposition for the films grown by RPLD, while it was done ex situ for those grown by reactive sputtering. The only elements present in a typical XPS spectrum of a WNx film grown by RPLD were W, N, and O. Meanwhile, W and N were just observed in the XPS spectrum for the film synthesized by reactive sputtering. Fig. 3 shows the relative atomic concentrations of tungsten, nitrogen and oxygen obtained by XPS as a function of nitrogen pressure for WNx films grown by RPLD. The relative atomic concentration of nitrogen steadily grows in steps from 0 to
E.C. Samano et al. / Vacuum 85 (2010) 69e77
71
Fig. 2. Cross sectional view of the reactive dc magnetron sputtering system. The gas control system is also shown.
0.15, from 0.15 to 0.2, from 0.2 to 0.25, and from 0.25 to 0.38 when the nitrogen pressure changes, respectively, from 0 to 2, from 2 to 3.33, from 3.33 to 4, and from 4 to 6.66 Pa. Meanwhile, the relative atomic concentration of tungsten gradually drops from 0.98 to 0.83, from 0.83 to 0.79, from 0.79 to 0.74, and from 0.74 to 0.61 when the nitrogen pressure changes from 0 to 2, from 2 to 3.33,
Fig. 3. Relative atomic concentration of each element in the WNx films processed by RPLD as a function of N2 background pressure.
from 3.33 to 4, and from 4 to 6.66 Pa, respectively. The relative atomic concentration of any of the elements did not change for pressures above 6.66 Pa. The atomic oxygen amount in average never exceeds 0.02 in the whole pressure range from 0 to 6.66 Pa. Because the amount of oxygen is quite low, it is ignored in the estimate of the stoichiometry of the films. Following this reasoning, the stoichiometry of the films could be given approximately by W6N, W4N, W3N and W2N, at the pressure range between 0.66 and 2, 2 and 3.33, 3.33 and 4, 5.33 and 6.66 Pa, respectively, ignoring spurious oxygen. The average deposition rates for the films grown by RPLD vary from 0.4 A/pulse to 0.28 A/ pulse when the nitrogen partial pressure changes from 0 to 6.66 Pa. These results are shown in Table 1. Fig. 4 shows the relative atomic concentrations obtained by XPS of tungsten and nitrogen as a function of nitrogen pressure for WNx films grown by reactive sputtering. Similarly to Fig. 3, the relative atomic concentration of nitrogen slowly grows from 0 to 0.07 when the nitrogen pressure changes from 0 to 0.53 Pa, then it suddenly jumps to 0.2 when the nitrogen pressure changes from 0.53 to 0.66 Pa. The amount of nitrogen remains almost constant for nitrogen pressures above 0.66 Pa. Meanwhile, the relative atomic concentration of tungsten diminishes from 1 to 0.93 when the nitrogen pressure changes from 0 to 0.53 Pa; afterwards it abruptly drops to 0.8 when the nitrogen pressure changes from 0.53 to 0.66 Pa. The amount of tungsten keeps basically unchanged for nitrogen pressures above 0.66 Pa. According to these values, the stoichiometry of the films could be represented approximately by W9N and W4N in the 0 < pN2 0.53 Pa and pN2 0.66 Pa nitrogen pressure ranges, respectively, keeping in mind that the nitrogen
72
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Table 1 Average numerical values of deposition rate (D. R.), roughness (R), hardness (H), reduced elastic modulus (Er) and plastic resistance parameter (H3/E2r ) as a function of N2 nitrogen pressure (film composition) for the most representative samples processed by reactive pulsed laser deposition and reactive sputtering. pN2 (Pa)
0 0.53 0.8 1.06 1.33 1.6 2.66 4 5.33 6.66 8
Reactive pulsed laser deposition
Reactive sputtering
Stoichiometry
D. R. ( A/p.)
R (nm)
W e e e W6N e W4N W3N W2N W2N W2N
0.40 e e e 0.36 e 0.34 0.33 0.30 0.28 0.28
0.1 e e e 0.1 e 0.2 0.2 0.3 0.4 0.5
0.02
0.02
0.03 0.02 0.02 0.03 0.03
H (GPa)
H3/E2r (GPa)
Stoichiometry
D. R. ( A/s)
R (nm)
19.2 e e e 19.5 e 20.3 21.5 27.0 27.7 22.2
0.104 e e e 0.095 e 0.114 0.156 0.331 0.338 0.218
W W9N W4N W4N W4N W4N e e e e e
30.0 29.0 27.0 27.0 26.0 25.0 e e e e e
2.1 2.3 2.8 3.2 3.3 3.5 e e e e e
1.2
1.3 1.6 1.8 2.1 2.2 2.4
content is underestimated due to argon ion sputtering during processing. The average deposition rates for the films grown by reactive sputtering vary from 30 A/s to 25 A/s when the nitrogen partial pressure changes from 0 to 1.6 Pa. The results are shown in Table 1. The films processed by sputtering required a cleaning procedure to remove physisorbed species with an argon gun prior to surface characterization by XPS. The argon ions preferentially sputter nitrogen atoms which might cause an underestimation of the nitrogen content in the WNx films. Oxygen contamination might be expected to occur for the films grown by magnetron sputtering but oxygen atoms were also removed due to the sputter cleaning effect. XRD was used to examine the crystallinity of WNx films deposited at various nitrogen pressures by both film processing techniques. XRD spectra using the BraggeBrentano geometry of representative WNx films by RPLD with a substrate at 400 C are shown in Fig. 5. The corresponding diffractogram of the stainless steel substrate has been added as a comparison, Fig. 5(a). A peak corresponding to a-W(110) planes [12,17,19,20] can be observed for the film grown in high vacuum, 0 Pa. Its intensity and width is a clear sign of its crystalline structure, as shown in Fig. 5(b). However, this peak slightly shifts to lower diffraction angles by increasing its width and decreasing its intensity as nitrogen is incorporated into the film, as shown in Fig. 5(c)e(g), until it completely vanishes at pN2 ¼ 8 Pa, as shown in Fig. 5(h). The shift to lower diffraction angles is interpreted as an increase in the
Fig. 4. Relative atomic concentration of each element in the WNx films processed by reactive sputtering as a function of N2 background pressure.
0.05 0.05 0.07 0.06 0.04 0.08
H (GPa) 30.3 30.8 31.6 32.6 38.2 38.7 e e e e e
3.2 3.5 4.1 3.3 3.9 3.1
H3/E2r (GPa) 0.214 0.298 0.373 0.451 0.726 0.719 e e e e e
tungsten specific volume, or the average WeW distance, as expected for nitrogen incorporation in a tungsten matrix [31]. The width and intensity loss of the peaks in Fig. 5(c)e(g) suggest that the nitrogen atoms were incorporated into the films in a disordered manner and, therefore, their initial long-range order has been lost. This is confirmed by observing a blunted peak in the curve given in Fig. 5(c). This assumption was based on the WNx film densities already reported in a previous work [18]. A small but clear feature corresponding b-W2N (200) plane is found for the film grown at pN2 ¼ 8 Pa, Fig. 5(g) and (h) [13,19,20]. Tiny crystallites oriented along the (200)-planes start to grow in films with high nitrogen content, W2N stoichiometry. This might affect the mechanical properties of the WNx films grown by RPLD. Fig. 6 shows representative qe2q diffractogram for films grown by reactive sputtering at a 400 C substrate temperature. Again, a peak corresponding to a-W (110) planes can be observed for the film grown in the absence of N2 gas, only Ar, as a clear sign of the crystalline structure, as shown in Fig. 6(a). Nevertheless, this peak disappears as a small amount of nitrogen is incorporated into the film and two new peaks corresponding to b-W2N (111) and b-W2N (200) appear, as shown in Fig. 6(b), at pN2 ¼ 0.66 Pa [13,19,20]. As the nitrogen pressure
Fig. 5. XRD qe2q scans spectra showing the evolution of the main peaks at various pressures: (a) stainless steel substrate, (b) pN2 ¼ 0 Pa, (c) pN2 ¼ 1.33 Pa, (d) pN2 ¼ 2.66 Pa, (e) pN2 ¼ 4.0 Pa, (f) pN2 ¼ 5.33 Pa, (g) pN2 ¼ 6.66 Pa and (h) pN2 ¼ 8.0 Pa for WNx films grown by RPLD.
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 6. XRD qe2q scans spectra showing the evolution of the main peaks at various pressures: (a) pN2 ¼ 0 Pa, (b) pN2 ¼ 0.66 Pa, (c) pN2 ¼ 1.07 Pa and (d) pN2 ¼ 1.6 Pa, for WNx films grown by reactive sputtering.
increases, this peak remains unchanged in position but the intensity gradually decreases, as shown in Fig. 6(c)e(d). This means that the films tend to lose long-range crystallinity as the nitrogen content increases, as it occurs in the films grown by RPLD, with a possible influence in the mechanical properties. Fig. 7(a) shows a typical 2 2 mm scan of a 0.5 mm thick of a WNx film grown by RPLD at pN2 ¼ 6.66 Pa before indentation. The smoothness and uniformity of the coating can be observed; in fact, the average roughness resulted to be 0.35 nm. This smooth surface topography with a roughness comparable to the initial surface (stainless steel) indicates that the film is growing with negligible roughening. Fig. 7(b) shows the same spot area as in Fig. 7(a) just after indentation, a clear view of the Berkovich tip geometry is viewed in the impression. The corresponding indentation curve with the measured mechanical properties is also shown in Fig. 7(b). Fig. 8(a) shows a typical 1 1 mm scan of a 1.5 mm thick WNx film deposited by reactive sputtering pN2 ¼ 1.6 Pa before indentation. This surface is rougher, average roughness of 3.5 nm, than the WNx thin film grown by RPLD. The corresponding indentation curve with the measured mechanical properties is shown in Fig. 8(b); a very shallow impression on the WNx film can also be viewed. The mechanical properties for the films grown by both deposition methods were determined by means of the corresponding loadingeunloading curves, similar to those observed in Figs. 7 and 8. Figs. 9 and 10 show the hardness, H, and H3/E2r ratio as a function of nitrogen gas pressure for films grown by RPLD and reactive sputtering, respectively. As mentioned on the previous section, these data points were obtained after averaging the indentations on four different regions on each sample. Then, the quantitative analysis shown in Figs. 9 and 10 correspond to the mean statistical values of the measurements for a given deposition condition per sample. If the hardness values for hard coatings are plotted as a function of penetration depth, as obtained from the indentation curves, the hardness increases at low penetration depths. This is due to the fact that the probe-sample contact produces an elastic indentation due to the roundness of the tip [27]. As the penetration depth increases, a plateau region is found and the hardness values
73
remain almost constant. However, as the contact depths or loads increase to larger values, the hardness decreases due to the substrate influence on the film mechanical properties [27]. The data shown in Figs. 9 and 10 correspond to the average values on the plateau region; the corresponding reduced elastic modulus is also obtained from this graph. The plateau corresponds to a fully developed plastic zone for intermediate contact depth values [27]. The values of the plastic resistance parameter, H3/E2r , have been included in both figures for having a better understanding of the mechanical characteristics of the WNx thin films. K.L. Johnson found the required contact load to initiate plastic deformation when two solid surfaces touch each other [32]. Tsui et al. [33] and Musil [34] used the analysis developed by Johnson to show that this contact load is proportional to the H3/E2r ratio; i.e., plastic deformation might be impeded in materials with high H and low Er. Therefore, the plastic resistance parameter determines the most favorable combination of hardness and reduced elastic modulus for obtaining an optimum protective coating, which is the ultimate goal of this investigation. As observed from Fig. 9, the hardness and plastic deformation ratio for films grown by RPLD are almost constant at pN2 2.66 Pa. Suddenly, these values increase in the 2.66 Pa pN2 5.33 Pa pressure range, and reach maximum values at pN2 z 6.5 Pa. The maximum hardness and plastic resistance ratio values resulted to be 27.7 GPa and 0.34, respectively. By another hand, as observed from Fig. 10, the hardness and plastic deformation ratio for films grown by reactive sputtering are almost constant in the pN2 0.8 Pa pressure range. Suddenly, these values increase for pN2 > 0.8 Pa. The films with the maximum hardness and plastic resistance ratio values resulted to be the films grown at pN2 z 1.6 Pa with values of 38.7 and 0.72, respectively. The mechanical properties apparently do not change at pN2 1.6 Pa, as observed from Fig. 10. Both, the hardness and the plastic resistance ratio are found to be higher for films grown by sputtering compared to those grown by RPLD under the same substrate temperature and nitrogen pressure. A summary of statistical averages of surface roughness, hardness, and reduced elastic modulus, as a function of nitrogen pressure is given in Table 1 for the WNx films. 4. Discussion RPLD and reactive sputtering are non-equilibrium thermodynamic processes. The different tungsten ablated species in RPLD collide with the N2 molecules in the plume expansion process and react with them, after dissociation, to form a WNx compound as a thin film on the substrate [35]. The species are released from their initial bonds in ground state in the reactive gas and free radicals, only in the ablation plume, and they rearrange themselves somewhere in the network of the growing layers during thin film deposition. On the other hand, the sputtered species of WNx, from a “poisoned” tungsten target, in the reactive sputtering process are preferentially transferred to the substrate surface to grow a thin film [36]. A saturation of the “poisoning” of the tungsten target is an indication that the nitrogen content in the films does not change. The species in both deposition methods tend to be mobile and, then, enhance nucleation homogeneously when the substrate is at 400 C. XPS data show that tungsten and nitrogen are the only elements present in the films, except for a negligible oxygen contamination only for the films grown by RPLD. Although the atomic concentration curves look similar, as observed from Figs. 3 and 4, these two different synthesis routes might explain the difference in stoichiometry in the WNx films. As a matter of fact, the saturation pressure value in Fig. 4, 0.66 Pa, is lower than the saturation pressure value in Fig. 3, 5.33 Pa. This is why the nitrogen content in the films synthesized by RPLD is higher than those
74
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 7. AFM images of a 0.35 mm thick WNx film deposited by RPLD in a N2 ambient (pN2 ¼ 6.66 Pa) while the substrate is at 400 C (a) before indentation and (b) after indentation showing the corresponding loadingeunloading curve with H ¼ 31.34 GPa and Er ¼ 345.6 GPa for that particular spot.
synthesized by reactive sputtering when the substrate temperature is 400 C. However, the nitrogen content in the films processed by RPLD at a substrate temperature of 400 C in this report is lower than those previously reported at room temperature [18]. Therefore, the substrate temperature affects on the film composition. Besides, there is an influence of the film composition on the mechanical properties, as observed in Table 1, which will be discussed below in this section. The film surface roughness also depends on the deposition method, as shown in Table 1. The films processed by RPLD are smoother than those prepared by reactive sputtering. In part, this
discrepancy is due to difference in kinetic energy of the species; the lower the kinetic energy, the coarser the grain size. The substrate temperature of the WNx films grown was kept at 400 C while film was processed by both methods. The average roughness of the films synthesized by RPLD in this report is approximately an order of magnitude smaller than those reported in the literature using the same deposition technique [19]. As observed from Figs. 5 and 6, in addition to the substrate temperature, the nitrogen pressure during synthesis is mainly responsible for film crystallinity. M. Bereznai et al. have already noticed this effect in the deposition of WNx films by RPLD at room
E.C. Samano et al. / Vacuum 85 (2010) 69e77
75
Fig. 8. AFM images of a 1.5 mm thick WNx film deposited by reactive sputtering in a N2 ambient (pN2 ¼ 1.6 Pa) while the substrate is at 400 C (a) before indentation and (b) after indentation showing the corresponding loadingeunloading curve with H ¼ 42.46 GPa and Er ¼ 318.3 GPa for that particular spot.
temperature and 500 C [19]. In the present study, the films are crystalline when they are grown in high vacuum, either processed by RPLD or reactive sputtering, as shown by the a-W (110) peak in both curves at the bottom of Figs. 5 and 6. However, the crystallinity of the films is different as nitrogen is incorporated during the deposition process. The WNx films grown by RPLD lose long-range order as the nitrogen pressure increases, as shown in Fig. 5. The only exception to this leaning is the appearance of a small feature associated to tiny crystallites of W2N oriented along the (200)planes for the films grown at pN2 z 8 Pa, Fig. 5(g) and (h). On the contrary, new crystalline phases corresponding to b-W2N (111), small peak, and b-W2N (200), large peak appear in the WNx films grown by reactive sputtering as soon as a small amount of nitrogen
is introduced during film deposition, as shown in Fig. 6. These peaks become smaller at pN2 > 0.66 Pa, and the peak related to a-W (110) suddenly disappears at any nitrogen pressure different to vacuum. Consequently, the WNx films have the tendency to lose long-range crystallinity as the nitrogen content increases for both deposition methods. Regarding the relationship between crystallinity and mechanical properties for the WNx films grown by RPLD and reactive sputtering, the films are composed by tiny crystalline grains embedded in an amorphous matrix. This can be observed from Fig. 5(g), with a W2N stoichiometry, and Fig. 6(d), with a W4N stoichiometry, for the films processed at pN2 ¼ 6.66 Pa and pN2 ¼ 1.6 Pa, respectively. This is in good agreement with Tsui [33]
76
E.C. Samano et al. / Vacuum 85 (2010) 69e77
WNx films reported in this work are comparable to those values published in the literature by reactive sputtering [12,16], RPLD [19]. Our results also agree well with recent reports for WNx coatings with thickness in the order of a few microns, hardness in the 28 and 40 GPa range [37]. 5. Conclusions
Fig. 9. Relationship of hardness and plastic deformation ratio as a function of N2 background gas pressure for WNx films grown on stainless steel by RPLD.
and Musil [34], a decrease of film crystallinity, characterized by Xray peaks with a reduced intensity, results in an improvement of hardness and plastic deformation ratio. Remarkable enhancement in mechanical properties of WNx films are induced by incorporation of nitrogen into interstitial sites of tungsten, producing changes in crystallographic orientation of grains in the material. However, an increment of the nitrogen content beyond a threshold value, pN2 > 6.66 Pa, for films grown by RPLD produces a decrement on hardness and elastic recovery probably due to voids from unreacted nitrogen. A similar effect in the mechanical properties might occur for films grown by reactive sputtering at pN2 > 1.6 Pa. In overall, the WNx films grown by reactive sputtering resulted to have hardness and plastic deformation parameter values higher than those grown by RPLD at any stoichiometry or any nitrogen pressure, as shown in Table 1. The WNx films with the most favorable mechanical properties depend on the deposition method. The film grown by RPLD having an approximated stoichiometry of W2N has H and H3/E2r values of 27.7 and 0.34 GPa, respectively, as observed from Fig. 9. Meanwhile, the film grown by reactive sputtering having an approximated stoichiometry of W4N has H and H3/E2r values of 38.7 and 0.72 GPa, respectively, as observed from Fig. 10. These are the coatings with the best mechanical properties for each film processing technique. The mechanical properties obtained by indentation methods of the
Fig. 10. Relationship of hardness and plastic deformation ratio as a function of N2 background gas pressure for WNx films grown on stainless steel by reactive sputtering.
In summary, RPLD and reactive sputtering complement each other in the comprehension about finding the adequate processing conditions to obtain hard coatings of transition metal nitrides with optimal mechanical properties. These properties for coatings grown by both deposition techniques depend on the nitrogen content in the films, reaching a threshold value to obtain the best properties. The WNx coatings processed by reactive sputtering are superior to those synthesized by RPLD, although the surface roughness is higher. The films grown by RPLD and reactive sputtering with the highest hardness resulted to be 27.7 and 38.7 GPa, respectively. The films grown by RPLD and reactive sputtering with the highest plastic deformation ratio resulted to be 0.34 and 0.72 GPa. Appliances manufactured by stainless steel might be protected from corrosion utilizing these hard coatings. Acknowledgements We gratefully acknowledge the technical assistance of Victor García, Pedro Casillas, Juan Antonio Peralta, Margot Sainz and Carlos González. CONACyT provided financial support. References [1] Chung YW, Sproul WD. Superhard coating materials. MRS Bull 2003;28(3):164 [and articles therein]. [2] Veprêk S, Veprêk MJG. Industrial applications of superhard nanocomposite coatings. Surf Coat Technol 2008;202:5063e73. [3] Veprêk S, Veprêk-Heijman MGJ, Kárvanková P, Procházka J. Different approaches to superhard coatings and nanocomposites. Thin Solid Films 2005;476:1e29. [4] Ma S, Procházka J, Kárvanková P, Ma Q, Niu X, Wang X, et al. Comparative study of the tribological behaviour of superhard nanocomposite coatings ncTiN/a-Si3N4 with TiN. Surf Coat Technol 2005;194:143e8. [5] Kauffmann F, Ji B, Dehm G, Gao H, Arzt E. A quantitative study of the hardness of a superhard nanocrystalline titanium nitride/silicon nitride coating. Scr Mater 2005;52:1269e74. [6] Shar PL. Refractory metal gate processes for VLSI applications. IEEE Trans Electron Devices 1979;ED-26:631e40. [7] Kim YT, Min SK. New method to suppress encroachment by plasma-deposited b-phase tungsten nitride thin films. Appl Phys Lett 1991;59:929e31. [8] Chiu HT, Chuang SH. Tungsten nitride thin films prepared by MOCVD. J Mater Res 1993;8:1353e60. [9] Veprêk S, Haussmann M, Reiprich S. Superhard nanocrystalline W2N/amorphous Si3N4 composite materials. J Vac Sci Technol A 1996;14:46e51. [10] Shen YG, Mai YW, McBride WE, Zhang QC, McKenzie DR. Structural properties and nitrogen-loss characteristics in sputtered tungsten nitride films. Thin Solid Films 2000;372:257e64. [11] Shen YG, Mai YW. Structural studies of amorphous and crystallized tungsten nitride thin films by EFED, XRD and TEM. Appl Surf Sci 2000;167:59e68. [12] Parreira NMG, Carvalho NJM, Vaz F, Calveiro A. Mechanical evaluation of unbiased WeOeN coatings deposited by d.c. reactive magnetron sputtering. Surf Coat Technol 2006;200:6511e6. [13] Polcar T, Parreira NMG, Calveiro A. Tribological characterization of tungsten nitride coatings deposited by reactive magnetron sputtering. Wear 2007;262:655e65. [14] Klaus JW, Ferro SJ, George SM. Atomic layer deposition of tungsten nitride films using sequential surface reactions. J Electrochem Soc 2000;147:1175e81. [15] Migita T, Kamei R, Tanaka T, Kawabata K. Effect of dc bias on the compositional ratio of WNx thin films prepared by rfedc coupled magnetron sputtering. Appl Surf Sci 2001;169e170:362e5. [16] Hones P, Martin N, Regula M, Lévy F. Structural and mechanical properties of chromium nitride, molybdenum nitride and tungsten nitride thin films. J Phys D Appl Phys 2003;36:1023e9. [17] Abadias G, Dub S, Shmegra R. Nanoindentation hardness and structure of ion beam sputtered TiNx, W and TiN/W multilayer hard coatings. Surf Coat Technol 2006;200:6538e43.
E.C. Samano et al. / Vacuum 85 (2010) 69e77 [18] Soto G, De la Cruz W, Castillón FF, Díaz JA, Machorro R, Farías MH. Tungsten nitride films grown via pulsed laser deposition studied in situ by electron spectroscopies. Appl Surf Sci 2003;214:58e67. [19] Bereznai M, Tóth Z, Caricato AP, Fernández M, Luches A, Majni G, et al. Reactive pulsed laser deposition of thin molybdenum- and tungsten-nitride films. Thin Solid Films 2005;473:16e23. [20] Yamamoto T, Kawate M, Hasegawa H, Suzuki T. Effects of nitrogen concentration on microstructures of WNx films synthesized by cathodic arc method. Surf Coat Technol 2005;193:372e4. [21] Hones P, Consiglio R, Randall N, Lévy F. Mechanical properties of hard chromium tungsten nitride coatings. Surf Coat Technol 2000;125:179e84. [22] Silva PN, Dias JP, Cavaleiro A. Tribological behaviour of WeTieN sputtered thin films. Surf Coat Technol 2005;200:186e91. [23] Musil J, Daniel R, Soldán J, Zeman P. Properties of reactively sputtered WeSieN films. Surf Coat Technol 2006;200:3886e95. [24] Samano EC, Machorro R, Soto G, Cota L. In situ ellipsometric characterization of SiNx thin grown by laser ablation. J Appl Phys 1998;84: 5296e305. [25] Garcia V. Inhomogeneous film growth and ambient gas supervision in laser ablation. MSc Thesis. Ensenada, Mexico: CICESE; 2003. [26] Fischer-Cripps AC. Nanoindentation. 1st ed. New York, USA: Springer; 2002. [27] Fischer-Cripps AC. Critical review of analysis and interpretation of nanoindentation test data. Surf Coat Technol 2006;200:4153e65.
77
[28] Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564e83. [29] Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology. J Mater Res 2004;19:3e20. [30] Yao CD, Polycarpou AA. A correction to the nanoindentation technique for ultrashallow indenting depths. J Mater Res 2007;22:2359e62. [31] Soto G, Aparicio E, Avalos-Borja M. Correlation functions between specific volume and stoichiometry for transition metal nitrides. J Alloys Compd 2005;389:42e6. [32] Johnson KL. Contact mechanics. 1st ed. Cambridge, UK: Cambridge University Press; 1985. [33] Tsui TY, Pharr GM, Oliver WC, Bhatia CS, White RL, Anders S, et al. Nanoindentation and nanoscratching of hard carbon coatings for magnetic disks. Mater Res Soc Symp Proc 1995;383:447e52. [34] Musil J. Hard and superhard nanocomposite coatings. Surf Coat Technol 2000;125:322e30. [35] Perrone A. State-of-the-art reactive pulsed laser deposition of nitrides. Jpn J Appl Phys 2002;41:2163e70. [36] Berg S, Nyberg T. Fundamental understanding and modeling of reactive sputtering processes. Thin Solid Films 2005;476:215e30. [37] Polcar T, Cavaleiro A. Structure, mechanical properties and tribology of WeN and WeO coatings. Int J Refract Metals Hard Mater 2010;28:15e22.
Contents lists available at ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Mechanical properties optimization of tungsten nitride thin films grown by reactive sputtering and laser ablation E.C. Samano*, A. Clemente, J.A. Díaz, G. Soto CNyN-Universidad Nacional Autónoma de México, A. Postal 356, Ensenada, BC, Mexico
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 July 2009 Received in revised form 22 March 2010 Accepted 3 April 2010
Transition metal nitrides coatings are used as protective coatings against wear and corrosion. Their mechanical properties can be tailored by tuning the nitrogen content during film synthesis. The relationship between thin film preparation conditions and mechanical properties for tungsten nitride films is not as well understood as other transition metal nitrides, like titanium nitride. We report the synthesis of tungsten nitride films grown by reactive sputtering and laser ablation in the ambient of N2 or N2/Ar mixture at various pressures on stainless steel substrates at 400 C. The composition of the films was determined by XPS. The optimal mechanical properties were found by nanoindentation based on the determination of the proper deposition conditions. As nitrogen pressure was increased during processing, the stoichiometry and hardness changed from W9N to W4N and 30.8e38.7 GPa, respectively, for films deposited by reactive sputtering, and from W6N to W2N and 19.5e27.7 GPa, respectively, for those deposited by laser ablation. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: Hard coatings Nanoindentation Tungsten nitride Reactive sputtering RPLD
1. Introduction The term of hard coatings is applied to structures which improve wear resistance and extend the lifetime of the structure. Hard coatings are industrially used onto cutting tools, automotive engine parts, turbine blades, structural components, etc [1,2]. Traditionally, materials utilized as hard coatings had a single composition, crystalline phase, and microstructure. Some of the most common hard coatings used as thin films are diamond-like carbon (DLC), BN, B4C, SiC, Al2O3, Si3N4 and WC [1]. A growing interest in the synthesis of transition metal nitrides has risen due to their chemical inertness and high hardness to be used as wearresistant and hard protective coatings. In particular, TiN thin films and nanocomposites using TiN satisfy these requirements and have been extensively investigated by a variety of deposition techniques in the recent literature [1e5]. For instance, there is a close relationship between film composition and its mechanical properties for TiN films [3]. However, there are not as many reports as expected on other transition metal nitrides. Tungsten nitride films have generated considerable interest in the manufacturing of semiconductor devices because they are an excellent barrier for Cu diffusion into Si at high temperatures [6]. The deposition of diffusion barriers motivated the synthesis of WNx
* Corresponding author. Tel.: þ52 646 1744602; fax: þ52 646 1744603. E-mail address: [email protected] (E.C. Samano). 0042-207X/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.vacuum.2010.04.004
thin films by several techniques like CVD [7e9], dc reactive sputtering [10e13], ALD [14], rf reactive sputtering [15,16], ion beam sputtering [17], reactive laser ablation [18,19] and cathodic arc [20]. There are reports concerning to the mechanical properties of alloys containing tungsten nitride sputtered coatings, like CreWeN [21], TieWeN [22] and SieWeN [23]. WNx thin films are also a good candidate to be used as a hard coating [9,13,16,19,20]. The composition, mass density, electrical and optical properties of WNx thin films on silicon wafers at room temperature grown by reactive laser ablation have been already published by our group but the study of the mechanical properties was missing [18]. This study investigates the best possible mechanical properties as a function of film composition and microstructure for WNx coatings on stainless steel substrates synthesized by two different deposition techniques: reactive sputtering and laser ablation. 2. Experiment As previously mentioned, the WNx thin films were grown by two different deposition methods: RPLD and dc-sputtering. The experiment was carried out in a laser ablation system described elsewhere [24]. The system consists of three UHV stainless steel chambers: sample introduction, growth and analysis, as shown in Fig. 1. The base pressure of the chambers is approximately 107 Pa. A KrF excimer laser is focused onto a high purity (99.9% at.) tungsten disc of 5 cm in diameter at an angle of 50 off the surface normal in the growth or deposition chamber. The values of laser
70
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 1. Cross sectional view of the reactive pulsed laser deposition system. The gas control system is also displayed.
energy, number of pulses, and repetition rate were optimized in a previous experiment and were kept fixed at 260 mJ, 18,000 and 5 Hz, respectively [18]. The film composition was studied by introducing N2 gas in the 0e10.0 Pa range. The ablated species were deposited on stainless steel ANSI304 substrates kept at 400 C. Every film was in situ characterized by XPS at the end of the film processing. The films were also grown in a reactive dc magnetron sputtering system, as shown in Fig. 2. A current-regulated dc power supply was used to provide a discharge with an input power of 100 W and a target-substrate separation of 3 cm. The target diameter and thickness are 5 cm and 0.6 cm, respectively. The power density is approximately 5 W/cm2 with an input current of 170 mA. The WNx films were grown on stainless steel ANSI304 substrates at 400 C by using the same tungsten target described above. Prior to deposition, the system was pumped down to a base pressure of 103 Pa. Afterwards, a flow of an Ar/N2 gas mixture, independently regulated by mass-flow controllers, fed the deposition chamber. The partial Ar pressure was kept constant at 0.53 Pa for the whole experiment, while the N2 partial pressure was varied from 0 to 1.6 Pa. Before film processing, the target was sputter cleaned for 5 min, with a shutter shielding the substrate. A deposition time of 10 min was used for every experimental run, the same for all depositions. The total pressure, Pt, as read from a Pirani-gauge was led to an analog input in a computer. This signal is compared to the preset pressure; the error signal is supplied to a proportional, integral, derivative (PID) algorithm. The PID output is used to compensate the gas flow controlled by the mass-flow controllers. The partial pressure of the reactive gas in the deposition chamber was estimated by knowing the total pressure, chamber volume, pumping speed, elapsed time, and gas flow of both gases [25]. This scheme is a closed-loop partial pressure regulator and it can be implemented to produce gas mixtures in a controlled way. For RPLD the same scheme was applied, even if nitrogen was the only gas. A Dektak profilemeter is utilized to measure the thickness of all samples. The thickness was between 0.5 and 0.7 mm for films grown by RPLD, and between 1.5 and 1.8 mm by reactive sputtering. A Philips difractometer, model X’pert, was employed to determine the crystallinity of the samples by X-ray diffraction (XRD) using the Cu Ka line with a 0.154 nm wavelength. The film morphology and roughness were determined by atomic force
microscopy (AFM) using a NanoScope III, Veeco Instruments, in contact mode. Several images of the samples surface were taken at scales of 10, 5, 2, 1 mm and 500 nm. The mechanical properties of the samples were measured using a TriboScope nanoindenter, Hysitron. The nanoindenter is used in conjunction with the AFM so that the area of indentation can be visualized with the same probe. The probe is a commercial Berkovich diamond tip attached to a transducer. This unit replaces the conventional optical head with a cantilever of the NanoScope III AFM. The tip penetrates into the sample from zero up to a maximum predetermined force (loading region); then, the tip returns to its original position (unloading region). The “golden rule” in nanoindentation to be used in the study of the mechanical properties of thin films is that the maximum depth of penetration in a test must be no more than 10% of film thickness [26,27]. The graph showing the relationship between force or load and penetration or displacement is called the indentation curve. The hardness, H, and reduced elastic modulus, Er, were found from the unloading region of the indentation curve by means of the theory developed by Oliver and Pharr [28,29], and corrections to this theory for very shallow indentations [30]. After choosing a region on the sample, 4 indentations (for 4 different loads) were performed. These loads were 200, 400, 800 and 1500 mN; and 800, 1500, 2000 and 3000 mN for films processed by RPLD and reactive sputtering, respectively. A number of 4 different regions, corresponding to a total of 16 indentations, were analyzed. 3. Results The WNx thin films were deposited on stainless steel disks kept at 400 C on both deposition methods. The nitrogen partial pressure was varied from 0 to 10.0 Pa in the RPLD system and from 0 to 1.6 Pa in the sputtering system. The analysis was in situ performed immediately after each deposition for the films grown by RPLD, while it was done ex situ for those grown by reactive sputtering. The only elements present in a typical XPS spectrum of a WNx film grown by RPLD were W, N, and O. Meanwhile, W and N were just observed in the XPS spectrum for the film synthesized by reactive sputtering. Fig. 3 shows the relative atomic concentrations of tungsten, nitrogen and oxygen obtained by XPS as a function of nitrogen pressure for WNx films grown by RPLD. The relative atomic concentration of nitrogen steadily grows in steps from 0 to
E.C. Samano et al. / Vacuum 85 (2010) 69e77
71
Fig. 2. Cross sectional view of the reactive dc magnetron sputtering system. The gas control system is also shown.
0.15, from 0.15 to 0.2, from 0.2 to 0.25, and from 0.25 to 0.38 when the nitrogen pressure changes, respectively, from 0 to 2, from 2 to 3.33, from 3.33 to 4, and from 4 to 6.66 Pa. Meanwhile, the relative atomic concentration of tungsten gradually drops from 0.98 to 0.83, from 0.83 to 0.79, from 0.79 to 0.74, and from 0.74 to 0.61 when the nitrogen pressure changes from 0 to 2, from 2 to 3.33,
Fig. 3. Relative atomic concentration of each element in the WNx films processed by RPLD as a function of N2 background pressure.
from 3.33 to 4, and from 4 to 6.66 Pa, respectively. The relative atomic concentration of any of the elements did not change for pressures above 6.66 Pa. The atomic oxygen amount in average never exceeds 0.02 in the whole pressure range from 0 to 6.66 Pa. Because the amount of oxygen is quite low, it is ignored in the estimate of the stoichiometry of the films. Following this reasoning, the stoichiometry of the films could be given approximately by W6N, W4N, W3N and W2N, at the pressure range between 0.66 and 2, 2 and 3.33, 3.33 and 4, 5.33 and 6.66 Pa, respectively, ignoring spurious oxygen. The average deposition rates for the films grown by RPLD vary from 0.4 A/pulse to 0.28 A/ pulse when the nitrogen partial pressure changes from 0 to 6.66 Pa. These results are shown in Table 1. Fig. 4 shows the relative atomic concentrations obtained by XPS of tungsten and nitrogen as a function of nitrogen pressure for WNx films grown by reactive sputtering. Similarly to Fig. 3, the relative atomic concentration of nitrogen slowly grows from 0 to 0.07 when the nitrogen pressure changes from 0 to 0.53 Pa, then it suddenly jumps to 0.2 when the nitrogen pressure changes from 0.53 to 0.66 Pa. The amount of nitrogen remains almost constant for nitrogen pressures above 0.66 Pa. Meanwhile, the relative atomic concentration of tungsten diminishes from 1 to 0.93 when the nitrogen pressure changes from 0 to 0.53 Pa; afterwards it abruptly drops to 0.8 when the nitrogen pressure changes from 0.53 to 0.66 Pa. The amount of tungsten keeps basically unchanged for nitrogen pressures above 0.66 Pa. According to these values, the stoichiometry of the films could be represented approximately by W9N and W4N in the 0 < pN2 0.53 Pa and pN2 0.66 Pa nitrogen pressure ranges, respectively, keeping in mind that the nitrogen
72
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Table 1 Average numerical values of deposition rate (D. R.), roughness (R), hardness (H), reduced elastic modulus (Er) and plastic resistance parameter (H3/E2r ) as a function of N2 nitrogen pressure (film composition) for the most representative samples processed by reactive pulsed laser deposition and reactive sputtering. pN2 (Pa)
0 0.53 0.8 1.06 1.33 1.6 2.66 4 5.33 6.66 8
Reactive pulsed laser deposition
Reactive sputtering
Stoichiometry
D. R. ( A/p.)
R (nm)
W e e e W6N e W4N W3N W2N W2N W2N
0.40 e e e 0.36 e 0.34 0.33 0.30 0.28 0.28
0.1 e e e 0.1 e 0.2 0.2 0.3 0.4 0.5
0.02
0.02
0.03 0.02 0.02 0.03 0.03
H (GPa)
H3/E2r (GPa)
Stoichiometry
D. R. ( A/s)
R (nm)
19.2 e e e 19.5 e 20.3 21.5 27.0 27.7 22.2
0.104 e e e 0.095 e 0.114 0.156 0.331 0.338 0.218
W W9N W4N W4N W4N W4N e e e e e
30.0 29.0 27.0 27.0 26.0 25.0 e e e e e
2.1 2.3 2.8 3.2 3.3 3.5 e e e e e
1.2
1.3 1.6 1.8 2.1 2.2 2.4
content is underestimated due to argon ion sputtering during processing. The average deposition rates for the films grown by reactive sputtering vary from 30 A/s to 25 A/s when the nitrogen partial pressure changes from 0 to 1.6 Pa. The results are shown in Table 1. The films processed by sputtering required a cleaning procedure to remove physisorbed species with an argon gun prior to surface characterization by XPS. The argon ions preferentially sputter nitrogen atoms which might cause an underestimation of the nitrogen content in the WNx films. Oxygen contamination might be expected to occur for the films grown by magnetron sputtering but oxygen atoms were also removed due to the sputter cleaning effect. XRD was used to examine the crystallinity of WNx films deposited at various nitrogen pressures by both film processing techniques. XRD spectra using the BraggeBrentano geometry of representative WNx films by RPLD with a substrate at 400 C are shown in Fig. 5. The corresponding diffractogram of the stainless steel substrate has been added as a comparison, Fig. 5(a). A peak corresponding to a-W(110) planes [12,17,19,20] can be observed for the film grown in high vacuum, 0 Pa. Its intensity and width is a clear sign of its crystalline structure, as shown in Fig. 5(b). However, this peak slightly shifts to lower diffraction angles by increasing its width and decreasing its intensity as nitrogen is incorporated into the film, as shown in Fig. 5(c)e(g), until it completely vanishes at pN2 ¼ 8 Pa, as shown in Fig. 5(h). The shift to lower diffraction angles is interpreted as an increase in the
Fig. 4. Relative atomic concentration of each element in the WNx films processed by reactive sputtering as a function of N2 background pressure.
0.05 0.05 0.07 0.06 0.04 0.08
H (GPa) 30.3 30.8 31.6 32.6 38.2 38.7 e e e e e
3.2 3.5 4.1 3.3 3.9 3.1
H3/E2r (GPa) 0.214 0.298 0.373 0.451 0.726 0.719 e e e e e
tungsten specific volume, or the average WeW distance, as expected for nitrogen incorporation in a tungsten matrix [31]. The width and intensity loss of the peaks in Fig. 5(c)e(g) suggest that the nitrogen atoms were incorporated into the films in a disordered manner and, therefore, their initial long-range order has been lost. This is confirmed by observing a blunted peak in the curve given in Fig. 5(c). This assumption was based on the WNx film densities already reported in a previous work [18]. A small but clear feature corresponding b-W2N (200) plane is found for the film grown at pN2 ¼ 8 Pa, Fig. 5(g) and (h) [13,19,20]. Tiny crystallites oriented along the (200)-planes start to grow in films with high nitrogen content, W2N stoichiometry. This might affect the mechanical properties of the WNx films grown by RPLD. Fig. 6 shows representative qe2q diffractogram for films grown by reactive sputtering at a 400 C substrate temperature. Again, a peak corresponding to a-W (110) planes can be observed for the film grown in the absence of N2 gas, only Ar, as a clear sign of the crystalline structure, as shown in Fig. 6(a). Nevertheless, this peak disappears as a small amount of nitrogen is incorporated into the film and two new peaks corresponding to b-W2N (111) and b-W2N (200) appear, as shown in Fig. 6(b), at pN2 ¼ 0.66 Pa [13,19,20]. As the nitrogen pressure
Fig. 5. XRD qe2q scans spectra showing the evolution of the main peaks at various pressures: (a) stainless steel substrate, (b) pN2 ¼ 0 Pa, (c) pN2 ¼ 1.33 Pa, (d) pN2 ¼ 2.66 Pa, (e) pN2 ¼ 4.0 Pa, (f) pN2 ¼ 5.33 Pa, (g) pN2 ¼ 6.66 Pa and (h) pN2 ¼ 8.0 Pa for WNx films grown by RPLD.
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 6. XRD qe2q scans spectra showing the evolution of the main peaks at various pressures: (a) pN2 ¼ 0 Pa, (b) pN2 ¼ 0.66 Pa, (c) pN2 ¼ 1.07 Pa and (d) pN2 ¼ 1.6 Pa, for WNx films grown by reactive sputtering.
increases, this peak remains unchanged in position but the intensity gradually decreases, as shown in Fig. 6(c)e(d). This means that the films tend to lose long-range crystallinity as the nitrogen content increases, as it occurs in the films grown by RPLD, with a possible influence in the mechanical properties. Fig. 7(a) shows a typical 2 2 mm scan of a 0.5 mm thick of a WNx film grown by RPLD at pN2 ¼ 6.66 Pa before indentation. The smoothness and uniformity of the coating can be observed; in fact, the average roughness resulted to be 0.35 nm. This smooth surface topography with a roughness comparable to the initial surface (stainless steel) indicates that the film is growing with negligible roughening. Fig. 7(b) shows the same spot area as in Fig. 7(a) just after indentation, a clear view of the Berkovich tip geometry is viewed in the impression. The corresponding indentation curve with the measured mechanical properties is also shown in Fig. 7(b). Fig. 8(a) shows a typical 1 1 mm scan of a 1.5 mm thick WNx film deposited by reactive sputtering pN2 ¼ 1.6 Pa before indentation. This surface is rougher, average roughness of 3.5 nm, than the WNx thin film grown by RPLD. The corresponding indentation curve with the measured mechanical properties is shown in Fig. 8(b); a very shallow impression on the WNx film can also be viewed. The mechanical properties for the films grown by both deposition methods were determined by means of the corresponding loadingeunloading curves, similar to those observed in Figs. 7 and 8. Figs. 9 and 10 show the hardness, H, and H3/E2r ratio as a function of nitrogen gas pressure for films grown by RPLD and reactive sputtering, respectively. As mentioned on the previous section, these data points were obtained after averaging the indentations on four different regions on each sample. Then, the quantitative analysis shown in Figs. 9 and 10 correspond to the mean statistical values of the measurements for a given deposition condition per sample. If the hardness values for hard coatings are plotted as a function of penetration depth, as obtained from the indentation curves, the hardness increases at low penetration depths. This is due to the fact that the probe-sample contact produces an elastic indentation due to the roundness of the tip [27]. As the penetration depth increases, a plateau region is found and the hardness values
73
remain almost constant. However, as the contact depths or loads increase to larger values, the hardness decreases due to the substrate influence on the film mechanical properties [27]. The data shown in Figs. 9 and 10 correspond to the average values on the plateau region; the corresponding reduced elastic modulus is also obtained from this graph. The plateau corresponds to a fully developed plastic zone for intermediate contact depth values [27]. The values of the plastic resistance parameter, H3/E2r , have been included in both figures for having a better understanding of the mechanical characteristics of the WNx thin films. K.L. Johnson found the required contact load to initiate plastic deformation when two solid surfaces touch each other [32]. Tsui et al. [33] and Musil [34] used the analysis developed by Johnson to show that this contact load is proportional to the H3/E2r ratio; i.e., plastic deformation might be impeded in materials with high H and low Er. Therefore, the plastic resistance parameter determines the most favorable combination of hardness and reduced elastic modulus for obtaining an optimum protective coating, which is the ultimate goal of this investigation. As observed from Fig. 9, the hardness and plastic deformation ratio for films grown by RPLD are almost constant at pN2 2.66 Pa. Suddenly, these values increase in the 2.66 Pa pN2 5.33 Pa pressure range, and reach maximum values at pN2 z 6.5 Pa. The maximum hardness and plastic resistance ratio values resulted to be 27.7 GPa and 0.34, respectively. By another hand, as observed from Fig. 10, the hardness and plastic deformation ratio for films grown by reactive sputtering are almost constant in the pN2 0.8 Pa pressure range. Suddenly, these values increase for pN2 > 0.8 Pa. The films with the maximum hardness and plastic resistance ratio values resulted to be the films grown at pN2 z 1.6 Pa with values of 38.7 and 0.72, respectively. The mechanical properties apparently do not change at pN2 1.6 Pa, as observed from Fig. 10. Both, the hardness and the plastic resistance ratio are found to be higher for films grown by sputtering compared to those grown by RPLD under the same substrate temperature and nitrogen pressure. A summary of statistical averages of surface roughness, hardness, and reduced elastic modulus, as a function of nitrogen pressure is given in Table 1 for the WNx films. 4. Discussion RPLD and reactive sputtering are non-equilibrium thermodynamic processes. The different tungsten ablated species in RPLD collide with the N2 molecules in the plume expansion process and react with them, after dissociation, to form a WNx compound as a thin film on the substrate [35]. The species are released from their initial bonds in ground state in the reactive gas and free radicals, only in the ablation plume, and they rearrange themselves somewhere in the network of the growing layers during thin film deposition. On the other hand, the sputtered species of WNx, from a “poisoned” tungsten target, in the reactive sputtering process are preferentially transferred to the substrate surface to grow a thin film [36]. A saturation of the “poisoning” of the tungsten target is an indication that the nitrogen content in the films does not change. The species in both deposition methods tend to be mobile and, then, enhance nucleation homogeneously when the substrate is at 400 C. XPS data show that tungsten and nitrogen are the only elements present in the films, except for a negligible oxygen contamination only for the films grown by RPLD. Although the atomic concentration curves look similar, as observed from Figs. 3 and 4, these two different synthesis routes might explain the difference in stoichiometry in the WNx films. As a matter of fact, the saturation pressure value in Fig. 4, 0.66 Pa, is lower than the saturation pressure value in Fig. 3, 5.33 Pa. This is why the nitrogen content in the films synthesized by RPLD is higher than those
74
E.C. Samano et al. / Vacuum 85 (2010) 69e77
Fig. 7. AFM images of a 0.35 mm thick WNx film deposited by RPLD in a N2 ambient (pN2 ¼ 6.66 Pa) while the substrate is at 400 C (a) before indentation and (b) after indentation showing the corresponding loadingeunloading curve with H ¼ 31.34 GPa and Er ¼ 345.6 GPa for that particular spot.
synthesized by reactive sputtering when the substrate temperature is 400 C. However, the nitrogen content in the films processed by RPLD at a substrate temperature of 400 C in this report is lower than those previously reported at room temperature [18]. Therefore, the substrate temperature affects on the film composition. Besides, there is an influence of the film composition on the mechanical properties, as observed in Table 1, which will be discussed below in this section. The film surface roughness also depends on the deposition method, as shown in Table 1. The films processed by RPLD are smoother than those prepared by reactive sputtering. In part, this
discrepancy is due to difference in kinetic energy of the species; the lower the kinetic energy, the coarser the grain size. The substrate temperature of the WNx films grown was kept at 400 C while film was processed by both methods. The average roughness of the films synthesized by RPLD in this report is approximately an order of magnitude smaller than those reported in the literature using the same deposition technique [19]. As observed from Figs. 5 and 6, in addition to the substrate temperature, the nitrogen pressure during synthesis is mainly responsible for film crystallinity. M. Bereznai et al. have already noticed this effect in the deposition of WNx films by RPLD at room
E.C. Samano et al. / Vacuum 85 (2010) 69e77
75
Fig. 8. AFM images of a 1.5 mm thick WNx film deposited by reactive sputtering in a N2 ambient (pN2 ¼ 1.6 Pa) while the substrate is at 400 C (a) before indentation and (b) after indentation showing the corresponding loadingeunloading curve with H ¼ 42.46 GPa and Er ¼ 318.3 GPa for that particular spot.
temperature and 500 C [19]. In the present study, the films are crystalline when they are grown in high vacuum, either processed by RPLD or reactive sputtering, as shown by the a-W (110) peak in both curves at the bottom of Figs. 5 and 6. However, the crystallinity of the films is different as nitrogen is incorporated during the deposition process. The WNx films grown by RPLD lose long-range order as the nitrogen pressure increases, as shown in Fig. 5. The only exception to this leaning is the appearance of a small feature associated to tiny crystallites of W2N oriented along the (200)planes for the films grown at pN2 z 8 Pa, Fig. 5(g) and (h). On the contrary, new crystalline phases corresponding to b-W2N (111), small peak, and b-W2N (200), large peak appear in the WNx films grown by reactive sputtering as soon as a small amount of nitrogen
is introduced during film deposition, as shown in Fig. 6. These peaks become smaller at pN2 > 0.66 Pa, and the peak related to a-W (110) suddenly disappears at any nitrogen pressure different to vacuum. Consequently, the WNx films have the tendency to lose long-range crystallinity as the nitrogen content increases for both deposition methods. Regarding the relationship between crystallinity and mechanical properties for the WNx films grown by RPLD and reactive sputtering, the films are composed by tiny crystalline grains embedded in an amorphous matrix. This can be observed from Fig. 5(g), with a W2N stoichiometry, and Fig. 6(d), with a W4N stoichiometry, for the films processed at pN2 ¼ 6.66 Pa and pN2 ¼ 1.6 Pa, respectively. This is in good agreement with Tsui [33]
76
E.C. Samano et al. / Vacuum 85 (2010) 69e77
WNx films reported in this work are comparable to those values published in the literature by reactive sputtering [12,16], RPLD [19]. Our results also agree well with recent reports for WNx coatings with thickness in the order of a few microns, hardness in the 28 and 40 GPa range [37]. 5. Conclusions
Fig. 9. Relationship of hardness and plastic deformation ratio as a function of N2 background gas pressure for WNx films grown on stainless steel by RPLD.
and Musil [34], a decrease of film crystallinity, characterized by Xray peaks with a reduced intensity, results in an improvement of hardness and plastic deformation ratio. Remarkable enhancement in mechanical properties of WNx films are induced by incorporation of nitrogen into interstitial sites of tungsten, producing changes in crystallographic orientation of grains in the material. However, an increment of the nitrogen content beyond a threshold value, pN2 > 6.66 Pa, for films grown by RPLD produces a decrement on hardness and elastic recovery probably due to voids from unreacted nitrogen. A similar effect in the mechanical properties might occur for films grown by reactive sputtering at pN2 > 1.6 Pa. In overall, the WNx films grown by reactive sputtering resulted to have hardness and plastic deformation parameter values higher than those grown by RPLD at any stoichiometry or any nitrogen pressure, as shown in Table 1. The WNx films with the most favorable mechanical properties depend on the deposition method. The film grown by RPLD having an approximated stoichiometry of W2N has H and H3/E2r values of 27.7 and 0.34 GPa, respectively, as observed from Fig. 9. Meanwhile, the film grown by reactive sputtering having an approximated stoichiometry of W4N has H and H3/E2r values of 38.7 and 0.72 GPa, respectively, as observed from Fig. 10. These are the coatings with the best mechanical properties for each film processing technique. The mechanical properties obtained by indentation methods of the
Fig. 10. Relationship of hardness and plastic deformation ratio as a function of N2 background gas pressure for WNx films grown on stainless steel by reactive sputtering.
In summary, RPLD and reactive sputtering complement each other in the comprehension about finding the adequate processing conditions to obtain hard coatings of transition metal nitrides with optimal mechanical properties. These properties for coatings grown by both deposition techniques depend on the nitrogen content in the films, reaching a threshold value to obtain the best properties. The WNx coatings processed by reactive sputtering are superior to those synthesized by RPLD, although the surface roughness is higher. The films grown by RPLD and reactive sputtering with the highest hardness resulted to be 27.7 and 38.7 GPa, respectively. The films grown by RPLD and reactive sputtering with the highest plastic deformation ratio resulted to be 0.34 and 0.72 GPa. Appliances manufactured by stainless steel might be protected from corrosion utilizing these hard coatings. Acknowledgements We gratefully acknowledge the technical assistance of Victor García, Pedro Casillas, Juan Antonio Peralta, Margot Sainz and Carlos González. CONACyT provided financial support. References [1] Chung YW, Sproul WD. Superhard coating materials. MRS Bull 2003;28(3):164 [and articles therein]. [2] Veprêk S, Veprêk MJG. Industrial applications of superhard nanocomposite coatings. Surf Coat Technol 2008;202:5063e73. [3] Veprêk S, Veprêk-Heijman MGJ, Kárvanková P, Procházka J. Different approaches to superhard coatings and nanocomposites. Thin Solid Films 2005;476:1e29. [4] Ma S, Procházka J, Kárvanková P, Ma Q, Niu X, Wang X, et al. Comparative study of the tribological behaviour of superhard nanocomposite coatings ncTiN/a-Si3N4 with TiN. Surf Coat Technol 2005;194:143e8. [5] Kauffmann F, Ji B, Dehm G, Gao H, Arzt E. A quantitative study of the hardness of a superhard nanocrystalline titanium nitride/silicon nitride coating. Scr Mater 2005;52:1269e74. [6] Shar PL. Refractory metal gate processes for VLSI applications. IEEE Trans Electron Devices 1979;ED-26:631e40. [7] Kim YT, Min SK. New method to suppress encroachment by plasma-deposited b-phase tungsten nitride thin films. Appl Phys Lett 1991;59:929e31. [8] Chiu HT, Chuang SH. Tungsten nitride thin films prepared by MOCVD. J Mater Res 1993;8:1353e60. [9] Veprêk S, Haussmann M, Reiprich S. Superhard nanocrystalline W2N/amorphous Si3N4 composite materials. J Vac Sci Technol A 1996;14:46e51. [10] Shen YG, Mai YW, McBride WE, Zhang QC, McKenzie DR. Structural properties and nitrogen-loss characteristics in sputtered tungsten nitride films. Thin Solid Films 2000;372:257e64. [11] Shen YG, Mai YW. Structural studies of amorphous and crystallized tungsten nitride thin films by EFED, XRD and TEM. Appl Surf Sci 2000;167:59e68. [12] Parreira NMG, Carvalho NJM, Vaz F, Calveiro A. Mechanical evaluation of unbiased WeOeN coatings deposited by d.c. reactive magnetron sputtering. Surf Coat Technol 2006;200:6511e6. [13] Polcar T, Parreira NMG, Calveiro A. Tribological characterization of tungsten nitride coatings deposited by reactive magnetron sputtering. Wear 2007;262:655e65. [14] Klaus JW, Ferro SJ, George SM. Atomic layer deposition of tungsten nitride films using sequential surface reactions. J Electrochem Soc 2000;147:1175e81. [15] Migita T, Kamei R, Tanaka T, Kawabata K. Effect of dc bias on the compositional ratio of WNx thin films prepared by rfedc coupled magnetron sputtering. Appl Surf Sci 2001;169e170:362e5. [16] Hones P, Martin N, Regula M, Lévy F. Structural and mechanical properties of chromium nitride, molybdenum nitride and tungsten nitride thin films. J Phys D Appl Phys 2003;36:1023e9. [17] Abadias G, Dub S, Shmegra R. Nanoindentation hardness and structure of ion beam sputtered TiNx, W and TiN/W multilayer hard coatings. Surf Coat Technol 2006;200:6538e43.
E.C. Samano et al. / Vacuum 85 (2010) 69e77 [18] Soto G, De la Cruz W, Castillón FF, Díaz JA, Machorro R, Farías MH. Tungsten nitride films grown via pulsed laser deposition studied in situ by electron spectroscopies. Appl Surf Sci 2003;214:58e67. [19] Bereznai M, Tóth Z, Caricato AP, Fernández M, Luches A, Majni G, et al. Reactive pulsed laser deposition of thin molybdenum- and tungsten-nitride films. Thin Solid Films 2005;473:16e23. [20] Yamamoto T, Kawate M, Hasegawa H, Suzuki T. Effects of nitrogen concentration on microstructures of WNx films synthesized by cathodic arc method. Surf Coat Technol 2005;193:372e4. [21] Hones P, Consiglio R, Randall N, Lévy F. Mechanical properties of hard chromium tungsten nitride coatings. Surf Coat Technol 2000;125:179e84. [22] Silva PN, Dias JP, Cavaleiro A. Tribological behaviour of WeTieN sputtered thin films. Surf Coat Technol 2005;200:186e91. [23] Musil J, Daniel R, Soldán J, Zeman P. Properties of reactively sputtered WeSieN films. Surf Coat Technol 2006;200:3886e95. [24] Samano EC, Machorro R, Soto G, Cota L. In situ ellipsometric characterization of SiNx thin grown by laser ablation. J Appl Phys 1998;84: 5296e305. [25] Garcia V. Inhomogeneous film growth and ambient gas supervision in laser ablation. MSc Thesis. Ensenada, Mexico: CICESE; 2003. [26] Fischer-Cripps AC. Nanoindentation. 1st ed. New York, USA: Springer; 2002. [27] Fischer-Cripps AC. Critical review of analysis and interpretation of nanoindentation test data. Surf Coat Technol 2006;200:4153e65.
77
[28] Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564e83. [29] Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology. J Mater Res 2004;19:3e20. [30] Yao CD, Polycarpou AA. A correction to the nanoindentation technique for ultrashallow indenting depths. J Mater Res 2007;22:2359e62. [31] Soto G, Aparicio E, Avalos-Borja M. Correlation functions between specific volume and stoichiometry for transition metal nitrides. J Alloys Compd 2005;389:42e6. [32] Johnson KL. Contact mechanics. 1st ed. Cambridge, UK: Cambridge University Press; 1985. [33] Tsui TY, Pharr GM, Oliver WC, Bhatia CS, White RL, Anders S, et al. Nanoindentation and nanoscratching of hard carbon coatings for magnetic disks. Mater Res Soc Symp Proc 1995;383:447e52. [34] Musil J. Hard and superhard nanocomposite coatings. Surf Coat Technol 2000;125:322e30. [35] Perrone A. State-of-the-art reactive pulsed laser deposition of nitrides. Jpn J Appl Phys 2002;41:2163e70. [36] Berg S, Nyberg T. Fundamental understanding and modeling of reactive sputtering processes. Thin Solid Films 2005;476:215e30. [37] Polcar T, Cavaleiro A. Structure, mechanical properties and tribology of WeN and WeO coatings. Int J Refract Metals Hard Mater 2010;28:15e22.