Structural, optical and electrical properties of reactively sputtered CrxNy films: Nitrogen influence on the phase formation

The properties of various CrxNy films grown by direct current (DC) reactive sputtering process with different values of nitrogen partial pressures (0, 2×10, 3.5×10 and 5×10 mbar) were studied. The structural analysis of the samples was performed by using X-ray diffraction and transmission electron microscopy (TEM), while an elemental analysis was realized by means of Rutherford backscattering spectrometry. By varying nitrogen partial pressure the pure Cr layer, mixture of Cr, Cr2N and CrN phases, or single-phase CrN was produced. TEM analysis showed that at pN2 = 2×10 -4 mbar the layer has dense microstructure. On the other hand, the layer deposited at the highest nitrogen partial pressure exhibits pronounced columnar structure. The optical properties of CrxNy films were evaluated from spectroscopic ellipsometry data by the Drude or combined Drude and Tauc-Lorentz model. It was found that both refractive index and extinction coefficient are strongly dependent on the dominant phase formation (Cr, Cr2N, CrN) during the deposition process. Finally, the electrical studies indicated the metallic character of Cr2N phase and semiconducting behaviour of CrN.


I. Introduction
Transition metal nitrides exhibit highly covalent bonds in simple, usually cubic structures, which give them an extreme hardness, high corrosion and oxidation resistance and excellent mechanical and hightemperature stability [1][2][3][4].Due to their excellent tribological characteristics, they have become important materials for cutting tools and wear applications, as diffusion barriers in microelectronics, and as corrosion and abrasion-resistant coatings on optical and mechanical components [5].TiN and CrN are the most extensively investigated hard coatings.Besides its good mechanical properties TiN is not always corrosion resistant due to the presence of micro defects in the layers, while dense microstructure of CrN provide high wear and corrosion resistance [6].Much research is being conducted to study growth and properties of CrN film deposited by CVD and PVD methods [7].Nevertheless PVD methods, such as unbalanced magnetron sputtering, metal vapour vacuum arc, RF reactive sputtering, pulsed DC magnetron sputtering, arc discharge etc., are quite common for the deposition of CrN films, whereas CVD methods are not as popular.
In physical vapour deposition chromium-nitride films can grow in form of cubic CrN or hexagonal Cr 2 N phases, with Cr 2 N phase exhibiting a higher hardness.On the other hand, the CrN phase is also interesting due to its magnetic, optical and electronic properties, in the recent studies being reported as a semiconductor material [8,9].For sputtered thin films, process parameters such as gas partial pressure, bias voltage, temperature, and growth rate strongly influence the film properties.By controlling the partial pressure of nitrogen during reactive sputtering it is possible to produce films ranging in composition from the pure Cr through Cr 2 N-CrN mixtures to pure CrN as the amount of nitrogen is increased [10].The composition and growth parameters influence the microstructure, optical and electrical properties of the resulting materials.Although the microstructural features of CrN and Cr 2 N have been studied previously [10][11][12] low attention was given to their optical properties and electronic structure and their correlation to the process parameters during the deposition.
In this work the microstructure, optical and electrical properties of various Cr x N y coatings were studied.We investigate the effect of the nitrogen partial pressure on the phase formation (Cr, Cr 2 N, CrN) and, subsequently, on the properties of the produced coatings.Chemical composition of the layers was identified by Rutherford backscattering spectrometry (RBS), while X-ray diffraction (XRD) and transmission electron microscopy (TEM) were used for the structural characterization and the phase identification.The optical properties were determined by spectroscopic ellipsometry (SE), SE data being analysed with the Bruggeman Effective Medium Theory (BEMT) [13].We showed that spectroscopic ellipsometry can be used as an alternative technique for phase identification of single-phases (Cr, Cr 2 N and CrN) and their volume fractions in the layers.Using the Drude and combined Drude-Tauc Lorentz model [13][14][15], we defined the optical constants (refractive index n and extinction coefficient k) of the layers.We found that both optical and electrical properties of Cr x N y layers are strongly determined by the growing phase.

II. Experimental
Cr x N y films used in the present experiments were deposited by means of DC reactive ion sputtering in a Balzers Sputtron II system.The layers were grown on commercial Si(100) wafers.The substrates were cleaned by standard HF etching and dip in deionized water before being mounted into the deposition chamber.Then, they were ion etched for 2 minutes in Ar atmosphere at a pressure of 1×10 -3 mbar and a negative bias voltage of 1 kV applied to the substrate holder.The base pressure prior to sputtering was approximately 5×10 -6 mbar.Sputter deposition was performed using a Cr target (99.9%) in a mixed Ar (99.999%) and N 2 (99.999%) discharge.During deposition, the argon partial pressure was initially set to 1×10 -3 mbar and the reactive gas, i.e.N 2 , was subsequently added to obtain the desired gas composition.The nitrogen partial pressure was set either to 2×10 -4 , 3.5×10 -4 , or 5×10 -4 mbar.The layers were grown at room temperature (RT), at a rate of ∼10 nm/min, to a total thickness of 240-280 nm.The pure Cr film (p N 2 = 0 mbar), with the thickness of ∼220 nm was also deposited.The thicknesses of the deposited structures were measured with a profilometer and confirmed by TEM.
For RBS (Rutherford backscattering spectrometry) analysis we used a 900 keV He 2+ ion beam, generated by the IONAS facility in Göttingen [16].We collected random RBS spectra at normal incidence to the sample surface with two detectors, positioned at 165°scattering angle in ibm geometry.The experimental spectra were fitted by using the Data Furnace code [17].Transmission electron microscopy was done on a Philips EM 400T microscope operated at 120 keV, the samples being prepared for cross-sectional analysis by standard technique of ion beam thinning.Bright-field contrast imaging was done, and we also used micro-diffraction (MD) analysis to study the crystalline structure of the samples.XRD measurements were carried out at normal and grazing incidence on a standard Bruker D8 Diffractometer with parallel beam optics using Cu K α diffraction patterns.Angle 2θ was scanned in the range from 30°to 70°with step of 0.02°, in time sequence of 10 s.SE data were obtained using HORIBA-Jobin Yvon variable angle spectroscopic ellipsometer (model UVISEL 5) equipped with DeltaPsi 2 data analysis software [18].The ellipsometer consisted of a light source, a monochromator, a photoelastic modulator, collimating optics, polarizing elements, a sample holder, and a detector.Simultaneously, the system acquired a spectrum ranging from 0.6 to 4.8 eV with 0.1 eV intervals.Acquisition time per each point was 200 ms.SE measurements were taken using 1 mm spot size and at an angle of incidence of 70°.We also measured sheet resistance of the samples with a four point probe, the values being calculated in specific resistivity to compare the results.

Composition and microstructure
In general, elemental composition measurements carried out by RBS on chromium-nitride samples revealed homogeneous Cr and N concentrations over the whole depth of the layers, which was shown in our previous paper [19].For illustration, the Cr concentration profiles in Cr x N y layers, as deduced by means of the WiNDF code from the experimental RBS spectra were presented in Fig. 1.The layers have different thickness, but we can see uniform Cr depth profiles.The presented spectra clearly show that the nitrogen partial pressure determines the composition of the deposited layers, which is manifested in the decrease of the Cr concentration with the increase of the N 2 partial pressure.More precise insight into the layers composition can be revealed  by observing the extracted Cr/N ratios for the deposited layers, presented in Table 1.Sample S-1 corresponds to the pure Cr coating, having 100 at.% of chromium, since it was deposited with no flow of nitrogen in the chamber.The extracted Cr/N ratios for the lower nitrogen partial pressures of 2×10 -4 mbar and 3.5×10 -4 mbar, clearly showed that the metal-to-nitrogen ratio is ∼73/27 and ∼61/39, respectively, hinting the formation of nonstoichiometric Cr 2 N phase.The sample S-4 deposited at the highest nitrogen pressure (5×10 -4 mbar) lead to near stoichiometric composition of Cr/N, with the mean atomic composition of Cr 51 N 49 .This is an indication that under these conditions the pure CrN phase is formed.
Figure 2 shows XRD spectra taken from the samples S-2 and S-3, deposited at nitrogen partial pressure of 2×10 -4 and 3.5×10 -4 mbar, respectively.These data indicate that different phases are present in the layers.At low nitrogen partial pressure of 2×10 -4 mbar the film S-2 contains a mixture of Cr, Cr 2 N and CrN phases.When the nitrogen partial pressure is increased to 3.5×10 -4 mbar (the sample S-3) again the mixture of Cr + Cr 2 N + CrN was formed, with larger amount of Cr 2 N phase, which is visible through the appearance of new peaks at around 68°.In both spectra the peaks are overlapped due to their large width, indicating very finegrained structure of the layers.Figure 3 shows XRD spectrum corresponding to the sample S-4 deposited at 5×10 -4 mbar of nitrogen.Well defined diffraction pattern of the pure fcc CrN phase was obtained, with strong (111) preferred orientation and weaker peaks  220) planes, in accordance with minimization of the overall energy of the film.This result is in good agreement with Lu et al. [20] who found that the deposited CrN films exhibit a (111) texture.The XRD reflections of CrN film are shifted to lower angular values than expected, which is typical for in-plane compressed d-metal nitrides [13,14].The deduced lattice parameter of 0.4183 nm found to be in the range of the values characteristic for polycrystalline CrN films: a 0 = 0.4133-0.4185nm [3,21].The average grain size of the CrN films, estimated from the FWHM of the XRD lines using the Scherrer's formula, was found to be 14 ±2 nm.
Figure 4 shows typical cross-section TEM images of Cr x N y thin films deposited under different nitrogen partial pressures.One sees that nitrogen pressure during deposition have significant influence on the microstructure of the layers.Bright-field image of the S-2 layer deposited at the lowest nitrogen partial pressure of 2×10 -4 mbar exhibit a dense morphology, as shown in Fig. 4a.With increasing p N 2 to 3.5×10 -4 mbar (Fig. 4b) the microstructure is less compact with an indication of columnar growth.Diffraction rings on the corresponding MD patterns for both low-nitrogen-pressure layers are blurred and reveal no obvious phases.This is the consequence of two effects: i) the presence of more than one phase and ii) the formation of very small crystalline grains.This is supported by interpretation of corresponding XRD spectra, which showed fine-grained mixture of Cr, Cr 2 N and CrN phases.When the nitrogen partial pressure was increased to 5×10 -4 mbar (Fig. 4c) a pronounced columnar structure of the layer is obvious.Columnar growth is present throughout the whole film thickness, with the columns width of few tens of nanometers.The corresponding MD pattern reveals obvious CrN phase.The first three rings belong to (111), ( 200) and (220) CrN phase, while the outer rings correspond to higher indexed planes of this phase.The changes in the morphology of the growing film during deposition were investigated by Hones et al. [22]  on different nitride thin films deposited by RF reactive magnetron sputtering.They revealed that a significant fraction of the Cr + ions exhibits a high flux and kinetic energy if the nitrogen partial pressure p N 2 is low.These high-energy ions effectively bombard the growing film and a densely packed morphology results.In contrast, in an absence of a significant amount of high-energy ions at higher p N 2 , the authors observed columnar crystal morphology.

Optical analyses
Optical properties of Cr x N y films were studied by SE (spectroscopic ellipsometry), which is non-destructive, surface-sensitive technique determining the complex dielectric function ε(ω) of the materials.Figure 5 shows the real (ε r ) and the imaginary (ε i ) parts of dielectric functions obtained from Cr x N y films deposited under different values of nitrogen partial pressure.
The dielectric functions of Cr x N y films were analysed through appropriate modelling, using the following points: • One-layer model was used for each sample since these layers are optically thick.• The pure Cr layer was modelled with Cr, while the other layers were modelled as a mixture of Cr, Cr 2 N and CrN phases, by using the Bruggeman effective approximation.In BEMT any complex material is considered to consist of randomly mixed regions of pure constituent materials (Cr, Cr 2 N, CrN); then, the dielectric function of complex material can be described by the dielectric functions of the constituent materials and the corresponding volume fractions.• The dielectric function for the pure Cr layer was modelled using the Drude oscillator model, which is commonly used for metal systems and well describes metallic character due to the intraband transitions of free conduction electrons.The analysis of other samples was made by Drude term and Tauc-Lorentz oscillator (DTL model) described in literature [23], where Tauc-Lorentz model describes the interband transitions due to valence electrons.A special feature of materials with conduction electrons is the plasma frequency ω p , which is correlated with the conduction electron density.Therefore, ω p can be used to determine the metallic or semiconducting character of the Cr x N y films.Another important parameter is the damping factor Γ D .This parameter is proportional to the collision frequency which can be related to the col-lision processes between the free charge carriers and defects and hence give us information about the crystalline order in the system.The fitting results of different Cr x N y films, obtained by using above mentioned modelling, are shown in Fig. 5.The volume fractions of different phases obtained by BEMT analysis are listed in Table 2.The four presented layers exhibit different composition, crystal structure and optical properties and correspond to Cr (p N 2 = 0 mbar), mixed Cr + Cr 2 N + CrN (p N 2 = 2×10 -4 and 3.5×10 -4 mbar) and CrN (p N 2 = 5×10 -4 mbar).According to the raw spectra, the pure Cr have ε r < 0, indicating metallic character of the layer.Similar behaviour was observed for samples S-2 and S-3 deposited at nitrogen pressure of 2×10 -4 and 3.5×10 -4 mbar, respectively.On the other hand, the layer S-4 deposited at 5×10 -4 mbar of nitrogen shows quite different behaviour having ε r > 0. The results of the BEMT analysis show the variation of the constituent phases (Cr, Cr 2 N and CrN) with the increase of nitrogen partial pressure.We identify three distinct regions where different phases dominate.For p N 2 = 0 there is no nitrogen incorporation in the layer and the pure Cr phase is formed.In the range of 2×10 -4 mbar ≤ p N 2 ≤ 3.5×10 -4 mbar primarily formed phase is Cr 2 N, with different content of Cr and CrN.For p N 2 = 2×10 -4 mbar the film consists of 70% of Cr 2 N, 20% of Cr and 10% of CrN.When the nitrogen partial pressure is increased up to p N 2 = 3.5×10 -4 mbar, Cr 2 N becomes predominant phase in the layer with the volume fraction of 86%.Finally, for p N 2 = 5×10 -4 mbar a pure CrN is formed.These results are in a good agreement with RBS and XRD findings.
Based on the results for the best-fit DTL parameters we calculated the refractive index n and extinction coefficient k in the energy region of 0.6-4.8eV. Figure 6 shows the calculated n and k of the layer deposited at p N 2 = 5×10 -4 mbar (the pure CrN) and the layers deposited at p N 2 = 2×10 -4 and 3.5×10 -4 mbar, containing predominantly Cr 2 N phase.The solid and dashed lines represent the n, k values for the pure Cr 2 N and CrN, respectively, reported by Aouadi et al. [24].One can see that optical constants of our CrN layer quite well match with the one from the literature.Slightly smaller absolute values of refraction index in the whole energy range means that it is optically less dense as compared to the referent layer.As for the layers deposited at lower nitrogen partial pressures corresponding n and k versus photon energy shows different slopes, where the values for p N 2 = 3.5×10 -4 mbar is in a good agreement with the literature values [24].If one knows that these layers have different content of Cr 2 N phase, namely 70% and 86% for p N 2 = 2×10 -4 mbar and p N 2 = 3.5×10 -4 mbar, respectively, it is concluded that even small change in the layer's composition influences its optical constants.Thus, establishing of an optical database for the Cr-N system enables us to easily identify different chromiumnitride phases (Cr 2 N, CrN) and control their purity by monitoring the optical properties (based on SE data) of the layers.
The results of the DTL model fits can also determine the evolution of the transport properties of Cr x N y 's conduction electrons.Figure 7 shows the ω p and Γ D values calculated from the Drude term of the DTL model.The discrimination of the various phases is obvious in the evolution of ω p and Γ D .In all cases Γ D exhibits higher values with N incorporation as compared to the initial Cr layer.This parameter is closely related to the electrical resistivity and is influenced by the existence of grain boundaries and defects in the layers.Hence, the increase of Γ D is attributed to the lower free electron mobility due to the structural differences in the layers.Indeed, there are some differences between our layers, not only compositional but also structural.As shown by TEM analysis (Fig. 4), by increasing the nitrogen partial pressure the average energy per atom of the growing film converts the compact stacking of the film (p N 2 = 2×10 -4 mbar) over the less dense structure (p N 2 = 3.5×10 -4 mbar) to the more porous, columnar structure (p N 2 = 5×10 -4 mbar).Patsalas et al. [25] found that the Γ D parameter is directly related to the content of voids in the layer and that Γ D has the lowest value when the dense, compact structure is obtained, which is attributed to the elimination of voids.Figure 7 also shows an overall decrease of ω p with increasing nitrogen content in the layers.Changes of ω p could be directly related to the contribution of different phases and their metallic character.According to the results, Cr layer has pure metallic behaviour with the considerable density of conduction electrons (ω p = 24 eV).In addition, with increasing nitrogen partial pressure to 2×10 -4 mbar ω p becomes lower and stays almost constant with further increase of p N 2 to 3.5×10 -4 mbar.This is due to the incorporation of N into the layer, which is manifested in the formation of Cr 2 N phase, with some contribution of Cr and CrN phases.For the nitrogen partial pressure of 5×10 -4 mbar the layer exhibits no conduction electrons and consequently ω p = 0.This is an indication of semiconducting character of CrN, since at this p N 2 value a stoichiometric CrN phase is formed.

Electrical resistivity
The transport properties of Cr x N y 's conduction electrons are better illustrated by the electrical resistivity measurements, based on the four point probe method.Figure 8 shows variation of resistivity of the Cr x N y films deposited under different nitrogen partial pressures.The resistivity of the pure Cr thin film corresponding to p N 2 = 0 mbar found to be about 58 µΩ cm, is in agreement with the value already reported by Högberg et al. [26].When the nitrogen partial pressure is increased to 2×10 -4 mbar (the sample S-2) the resistivity becomes more than two times higher than that of the Cr layer, reaching a value of 120 µΩ cm and stays almost constant when the nitrogen partial pressure is increased further up to 3.5×10 -4 mbar (the sample S-3).The increasing trend of the electrical resistivity is due to the increase of N concentration in the layers which has as a consequence the increase of impurity defects or it may also be attributed to a decrease of the carrier density due to the nitrization of Cr.Based on the optical measurements  2), combined by the XRD results (Fig. 2) the presence of different phases (Cr, Cr 2 N and CrN) is established in the layers, with the primarily formed Cr 2 N phase.Hence, electrical resistivity is higher as compared to the initial Cr film, but still metallic character is present in the layers [27].By further increasing the nitrogen partial pressure to 5×10 -4 mbar very high value of resistivity of 1353 µΩ cm is obtained.This clearly proves semiconducting character of CrN as only CrN phase grown under given nitrogen partial pressure.Similar findings were also reported by Subramanian et al. [28], who examined the influence of nitrogen flow rates on the properties of magnetron sputtered CrN x films.

IV. Conclusions
The microstructure, optical and electrical properties of various Cr x N y films grown by DC reactive sputtering were studied.The layers were deposited at several values of nitrogen partial pressure ranging from 0 to 5×10 -4 mbar.RBS and XRD were employed for the compositional analysis and phase identification.It was revealed that at different values of nitrogen partial pressure different phases are obtained: Cr (p N 2 = 0), mixed Cr + Cr 2 N + CrN (p N 2 = 2×10 -4 and 3.5×10 -4 mbar) or pure CrN (p N 2 = 5×10 -4 mbar).TEM has shown the changes in the morphology of the growing film: from compact and dense structure observed at p N 2 = 2×10 -4 mbar to the well defined columnar structure for the highest nitrogen pressure.SE analysis was employed for the optical characterization of the samples.The results showed that there are well-defined p N 2 regions, where specific phases dominate (Cr, Cr 2 N, CrN) and the optical constants n and k are directly determined by the presence of the growing phase.The layers with predominantly formed Cr 2 N phase exhibit metallic behaviour, while CrN is a semiconductor.to D. Purschke for skillfully running the IONAS accelerator.Ion beam analysis of samples was done by Rutherford backscattering technique at the University in Göttingen.

Figure 5 .
Figure 5. Real (ε 1 ) and imaginary (ε 2 ) part of the dielectric function of Cr x N y thin films deposited at different nitrogen partial pressures (continuous lines correspond to the fits obtained by models described in the text)

Table 2 .Figure 6 .
Figure 6.The calculated refractive index (n) and extinction coefficient (k) for Cr x N y thin films deposited at partial pressure of nitrogen of 2×10 -4 , 3.5×10 -4 and 5×10 -4 mbar compared to the corresponding n, k curves for the pure CrN and Cr 2 N [24]

Figure 7 .
Figure 7.The evolution of ω p and Γ D calculated from the Drude term of the Drude and Drude-Tauc Lorentz model for different Cr x N y layers

Figure 8 .
Figure 8. Variation of resistivity of Cr x N y layers as a function of nitrogen partial pressure during deposition (Table2), combined by the XRD results (Fig.2) the presence of different phases (Cr, Cr 2 N and CrN) is established in the layers, with the primarily formed Cr 2 N phase.Hence, electrical resistivity is higher as compared to the initial Cr film, but still metallic character is present in the layers[27].By further increasing the nitrogen partial pressure to 5×10 -4 mbar very high value of resistivity of 1353 µΩ cm is obtained.This clearly proves semiconducting character of CrN as only CrN phase grown under given nitrogen partial pressure.Similar findings were also reported by Subramanian et al.[28], who examined the influence of nitrogen flow rates on the properties of magnetron sputtered CrN x films.

Table 1 .
Chromium nitride thin film compositions obtained from RBS analyses (the films were deposited under the different values of nitrogen partial pressure, p N