Electronic structure and chemical bonding of amorphous chromium carbide thin films

Metal carbide coatings have many interesting electrical, mechanical and tribological properties. Examples of multifunctional applications for these materials are combinations of oxidation, wear and corrosion resistance, decorative purposes and electrical contacts [1, 2]. Magnetron sputtering of metal carbide films can give different microstructures ranging from epitaxial single-crystal materials to nanocomposites with crystallites of the carbide in an amorphous matrix [3, 4]. It is also possible to deposit completely amorphous metal carbide coatings in both binary and ternary systems such as Cr–C, Fe–C and W–Fe–C [1]. The explanation for the formation of non-crystalline phases during sputtering in these carbide systems remains unclear, but it is known that amorphous films often are found in carbide systems where the crystalline phases exhibit structural units that have carbon atoms in both octahedral and prismatic sites [5].

Magnetron sputtering of chromium carbides including the formation of amorphous Cr–C films has been reported (see, e.g., [6–9]). In a recent study, Andersson et al [10] related the crystallinity of Cr–C coatings to the deposition process used. A survey of the literature showed that reactive deposition tends to give more crystalline films while non-reactive sputtering from elemental targets favors amorphous growth. An observation in [10] is that elemental target sputtering can give completely amorphous films with two separate phases: (i) an amorphous chromium carbide, CrC_x and (ii) an amorphous C-rich phase dominated by C–C bonds. The amorphous metal carbide phase contained 20–30 at.% of carbon, giving a composition between the crystalline Cr₂₃C₆ and Cr₇C₃ phases while the amorphous carbon phase contained mainly pure carbon with a high amount of sp²-hybridized carbon [10]. However, it was also concluded that more studies than x-ray photoelectron spectroscopy (XPS) analysis are required to determine structure and bonding of the two amorphous phases in these films.

In this work, we have studied the detailed chemical bonding and structure of magnetron-sputtered amorphous Cr–C films using a combination of high-resolution transmission electron microscopy (HRTEM), electron energy loss spectroscopy (EELS), soft x-ray absorption (SXA) and soft x-ray emission spectroscopy (SXE). An important part of the study is to investigate short- and long-range ordering in the materials and how this is influenced by composition and charge transfer between the elements. We confirm that the films consist of two amorphous phases with the exception of composition at 47 at.% C, where nanocrystallites are formed. The importance of these structures and their interfaces for the film properties is also discussed. These short- to medium-range structures and the charge transfer between the elements are known to affect properties such as hardness, elasticity and electrical resistance.

Four chromium carbide (Cr_1−xC_x) samples were deposited in Ar atmosphere at 0.4 Pa using non-reactive DC magnetron sputtering in an UHV chamber with base pressure of 1×10⁻⁷ Pa. Two separate 2 in targets with a purity of 99.999% for C and 99.95% for Cr were used. The films were deposited onto SiO₂ substrates at 300 °C and with a substrate bias of −50 V. The distance between targets and substrate was about 14 cm. By tuning the Cr magnetron while keeping the C magnetron constant, compositions between 25 and 85 at.% were achieved at a deposition rate of 12–45 Å min⁻¹ depending on composition. The compositions of the Cr_1−xC_x films were determined using XPS as described in [10]. The deposition of the amorphous carbon (a-C) film is described in [11].

The Cr_1−xC_x samples were characterized with an analytical high-resolution (HR)-TEM FEI field emission gun instrument operated at 200 kV with 1.95 Å point resolution. The microstructure was analyzed by selected-area electron diffraction (SAED) in the TEM equipped with an electron energy loss (EELS) spectrometer with an energy resolution of 0.7 eV. The Cr_1−xC_x films were also measured with electron energy loss spectroscopy (EELS) at the C 1s edge. The radial distribution functions and bond distances were extracted using the ATHENA software package [12] in figure 2. The samples were prepared in a conventional way including grinding, polishing and finally ion milling.

Soft x-ray absorption (SXA) and soft x-ray emission (SXE) measurements were performed at the undulator beamline I511-3 at MAX II (MAX IV Laboratory, Lund, Sweden), comprising a 49-pole undulator and a modified SX-700 plane grating monochromator [13]. The SXA spectra were measured at 5° grazing-incidence angle that was found to be necessary for amorphous samples in total fluorescence yield (TFY) mode with 0.15 and 0.1 eV resolution at the Cr 2p and C 1s absorption edges, respectively. The SXE spectra were measured with a high-resolution Rowland-mount grazing-incidence grating spectrometer with a two-dimensional detector. The Cr L and C K SXE spectra were recorded using a spherical grating with 1200 lines mm⁻¹ of 5 m radius in the first order of diffraction. During the Cr L and C K SXE measurements, the resolutions of the beamline monochromator were 0.68, and 0.2 eV, respectively. The SXE spectra were recorded with spectrometer resolutions of 0.7 and 0.2 eV, respectively. The measurements were performed with a base pressure lower than 6.7×10⁻⁷ Pa. In order to minimize self-absorption effects [14], the angle of incidence was 15° from the surface plane during the SXE measurements. The x-ray photons were detected parallel to the polarization vector of the incoming beam in order to minimize elastic scattering. The Cr L_2,3 and C K SXE spectra were normalized to the main peak heights and were plotted on a photon energy scale (bottom) and an energy scale relative to the E_F (top) in figures 3 and 4.

A series of Cr–C samples with different compositions were deposited at 300 °C. The compositions were determined by XPS to Cr_0.75C_0.25,Cr_0.53C_0.47,Cr_0.33C_0.67 and Cr_0.15C_0.85. The compositions are similar to those used in [10].

3.1. Transmission electron microscopy

Figure 1 shows HRTEM images at low to high resolutions with corresponding SAED patterns of the Cr_1−xC_x samples. As observed, the structural ordering largely depends on the carbon content. The most Cr-rich film in the top panel (x = 0.25) is fully amorphous with no indication of (crystalline) ordering. Its SAED pattern shows only weak and broad rings typical for amorphous samples [15]. In contrast, an increase of the carbon content to x = 0.47 shows short- and medium-range order within defect-rich nanocrystalline grains with lattice fringe spacings between 1.9 and 2.1 Å in different directions for each crystallite. The SAED pattern from this sample has sharper rings typical for the formation of a crystalline structure with a slightly larger structural distance of 2.21 Å in comparison to the other films. While it is difficult to identify the phase compositions of the randomly distributed grains, these are likely cubic CrC possibly mixed with Cr₃C₂. The small white spots in the SAED originate from the substrate.

Zoom In Zoom Out Reset image size

An increase in carbon content to x = 0.67 and 0.85 changes the structure again. No indication of crystalline phases can be seen in these two C-rich samples. However, the HRTEM images in the left panels show dark and light areas. This can be explained by elemental contrast where the darker areas are amorphous domains richer in Cr while the lighter areas can be attributed to amorphous domains more rich in carbon. Higher magnification of the two most C-rich films shows small, embedded carbon clusters with a graphene-like structure with atomic distances between 3.5 and 4.0 Å, clearly observed along the extension of the thin white lines that are guides for the eyes. The amount of graphene layers is not large enough and the small size of these clusters (typically 1–10 nm in length and 2–4 atomic layers thick) explains why the intensity is too low in the background of the central beam to be observed in SAED.

3.2. Radial distribution functions

In order to determine the different bond lengths in the samples, atomic radial distribution functions (RDFs) were extracted by the standard EXAFS procedure [12] from extended electron energy loss fine structure (EXELFS) measurements and the results are shown in figure 2. The first peak (1) at the position of ∼1.5 Å corresponds to the C–C bond length. For comparison, the C–C bond length is 1.44 Å in amorphous carbon, 1.43 Å in graphite and 1.54 Å in diamond [15–17]. The results qualitatively agree well with the relative distribution of carbon in the carbidic, CrC_x phase and the a-C phase as reported in [10] and summarized in table 1. We estimate the error bar for the bond length determination within half a standard deviation ± 0.15 Å. However, quantitatively, the ratio between the C–C (1) and C–Cr (2) peaks in figure 2 is lower in the RDFs than for XPS. As observed, the first peak (1) gradually decreases in intensity with reduced C content. This signifies that the C–C bonds in free carbon in the films disappear for x = 0.25 while XPS shows 11% of the amorphous carbon phase. The second peak (2) at ∼2.2 Å is due to the C–Cr bond distance consistent with Cr 1s EXAFS studies (2.18 Å) [18]. The third (3) and fourth (4) peaks are related to the second and third coordination shell of C–C bonding. The experimentally observed C–C (∼1.5 Å) and C–Cr (∼2.2 Å) bond lengths are shorter than the calculated bond lengths for the CrC crystallites; Cr₃C₂ (C–C=2.11 Å, C–Cr=2.63 Å), Cr₇C₃ (C–C=2.14 Å, C–Cr=2.58 Å) and Cr₂₃C₆ (C–C=2.10 Å, C–Cr=2.56 Å) [19]. The RDFs also show structures around ∼3.5 Å that could correspond to graphene-like structures but the uncertainties with overlapping high order coordination shells are too large to draw any definite conclusions.

Zoom In Zoom Out Reset image size

Table 1.  Composition of the Cr_1−xC_x films for x = 0.25,0.47,0.67 and 0.85. The C–C and C–Cr chemical bond contributions were determined by integrating the areas under the corresponding peak structures in C 1s XPS spectra [10]. The Cr L₃ SXE peak positions and L₃/L₂ ratios in figure 3 are also listed.

Composition (XPS)	Cr0.75C0.25	Cr0.53C0.47	Cr0.33 C0.67	Cr0.15 C0.85
% a-C phase	11	61	85	98
% CrCx phase	89	39	15	2
Cr L3 (eV ) (SXE)	576.38	575.98	576.48	575.79
Cr L3/L2 (SXE)	3.46	3.02	2.88	2.40
π*/[π* + σ*] (C 1s SXA)	0.074	0.093	0.079	0.095

Zoom In Zoom Out Reset image size

3.3. Cr 2p x-ray absorption spectra

3.4. Cr L_2,3 x-ray emission spectra

Figure 3 (bottom) shows Cr L_2,3 SXE spectra of Cr_1−xC_x in comparison to Cr metal, nonresonantly excited at 620 eV. The SXE spectra probes the occupied Cr 3d states following the Cr 3d→2p_3/2,1/2 dipole transitions. Starting with the Cr L_2,3 SXE spectra of Cr metal (x = 0, black curve), the main L₃ emission line due to metal Cr 3d–Cr 3d interactions is observed at −1.5 eV on the energy scale at the top of figure 3. The broader and less intense L₂ emission line is observed at +7 eV for Cr metal and at +6 eV for the carbide films. For the Cr_1−xC_x films there are three interesting observations; (i) the L₃ peak is shifted 0.5–1.0 eV to lower energies, (ii) the L₃ peak is significantly broader compared to Cr metal and (iii) the L₃/L₂ branching ratio decreases with increasing carbon content (table 1). For the Cr L_2,3 SXE spectra of the Cr_1−xC_x films, the main L₃ peak is due to Cr 3d–C 2p hybridization similar to the case of nc-TiC/a-C nanocomposites [11]. The low-energy shifts away from the Fermi level and the increased broadening of the L₃ and L₂ peaks for increasing carbon content can be attributed to a stronger Cr 3d–C 2p interaction compared to the Cr 3d–Cr 3d interactions in Cr metal [21]. The trend in the L₃/L₂ branching ratio in transition metal compounds is a signature of the degree of ionicity in the systems. For conducting systems, the L₃/L₂ ratio is usually significantly higher than the statistical ratio 2:1 due to the additional Coster–Kronig decay process [21]. The observed L₃/L₂ ratio systematically decreases as the carbon content increases (table 1), and is a sign of decreased metallicity [22]. Consequently, the Cr atoms in the most C-rich Cr_1−xC_x films appear to be more ionic than those with lower carbon content. For Cr metal, the L₃/L₂ ratio higher than 3:1 is a metallic signature.

Zoom In Zoom Out Reset image size

3.5. C 1s x-ray absorption spectra

3.6. C K x-ray emission spectra

The HRTEM images in figure 1 show that the Cr_1−xC_x films are mainly amorphous. The most chromium-rich film Cr_0.75C_0.25, as well as the most C-rich films Cr_0.33C_0.67 and Cr_0.15C_0.85, show no indication of crystallinity. This confirms the observations in [10] based on x-ray diffraction (XRD) and XPS that the films consist of two amorphous phases; (i) a chromium-rich carbide CrC_x phase and (ii) a C-rich phase. The presence of two amorphous phases is also supported by the C–C and C–Cr RDF peaks in figure 2 with characteristic atomic distances. The SAED and RDF from EELS provide complementary information due to the very high sensitivity of the loss energies in EELS for light elements such as C compared to SAED which is more sensitive to heavier elements such as Cr. The observation of areas with graphene-like structures in the most C-rich films suggests a considerable sp² hybridization of carbon. This is also supported by Raman data on similar films reported in [10] that indicate that the sp² contribution in the a-C phase is very high (≥80%). The graphene-like structures in figure 1 consist of several layers (2–4) of carbon with a layer distance of 3–4 Å. It is likely that these graphene-like structures influence some of the properties but are too rare to be observed in spectroscopy. The presence of two phases is also supported in figure 2 which shows C–C bonds and C–Cr bonds. The variations of the intensities are qualitatively similar with the variations observed in XPS.

The Cr_0.53C_0.47 film exhibits a different microstructure compared to the other films. In this case, we observe randomly oriented nanocrystallites embedded in a matrix of both amorphous carbon as well as non-crystallized amorphous CrC_x. Thus, this sample consists of three phases; two amorphous phases and one nanocrystalline chromium carbide phase. As observed in figure 1, the nanocrystallites have a typical size of less than 3 nm with different shapes and orientation. XRD data published in [10] showed no sign of crystallinity for this composition that can be explained by the small grain size, making the films x-ray amorphous. The nanocrystallites in the Cr_0.53C_0.47 sample have crystalline plane distances in the range 1.9–2.1 Å. Several chromium carbides such as Cr₃C₂ and metastable CrC have plane distances in this range and it is therefore difficult to identify the exact crystal structure of the embedded nanocrystallites. The crystallization kinetics of amorphous chromium carbides as a function of composition has not yet been studied and it is therefore not clear why only the Cr_0.53C_0.47 film has a partially crystallized carbide phase. However, it should be noted that this composition is close to the metastable cubic CrC phase observed by Gassner et al [26] and is close to the composition of Cr₃C₂. It is reasonable that total film compositions close to those of the crystallization products favor nucleation. However, further studies of the crystallization kinetics are required to confirm this.

Further details of the chemical bonding and the structure in the chromium carbide films are obtained from the SXA and SXE spectra in figures 3 and 4. The intensity of the Cr 2p SXA peaks in figure 3 is clearly increased with increasing carbon content. This reflects a change in the density of unoccupied states and is attributed to a reduced Cr 3d electron density at higher carbon contents, as will be discussed in more detail below. Notably, a characteristic observation in the Cr 2p SXA spectra is that the intensity and position of the (1)–(3) sub-peaks at the 2p_3/2 absorption threshold are strongly dependent on the carbon content. The total intensity trend for increasing carbon content in the Cr SXA spectra agrees with the Ti 2p SXA spectra in an earlier study [11]. In contrast to the Cr_1−xC_x films, the Ti 2p_3/2 and 2p_1/2 SXA peaks in nanocomposite nc-TiC/a-C films were strongly split into two sub-peaks due to the t_2g–e_g symmetry in the cubic TiC phase. The energy positions of the t_2g–e_g sub-peaks in the Ti SXA spectra were independent of the total carbon content since all films contained cubic TiC with a well-defined nanocrystalline structure. This suggests that the large variations in intensities and energy positions of the (1)–(3) sub-peaks in figure 3 reflect changes in the bonding and coordination of the Cr atoms in the amorphous CrC_x phase. Peaks (2)–(5) are similar to the t_2g–e_g splitting observed in the cubic TiC crystallites [11] but considerably weaker. Although we have not identified the specific origin of peak (1), it likely reflects a different coordination and bonding where the specific structure is unknown. The exact coordination and bonding of Cr in the amorphous CrC_x domains is also unknown, but the results in [10] show that the total carbon content is about 20–30 at.%. This is in the same range as the crystalline Cr₂₃C₆ and Cr₇C₃ phases. In these crystalline phases, the C atoms are placed in a mixture of octahedral and trigonal prismatic sites. It is likely that a mixture of such coordination polyhedra exists in the amorphous CrC_x domains and the distribution is dependent on the total carbon content. Additional experimental and theoretical studies are needed to fully correlate the observed features (1)–(3) in the Cr 2p_3/2 peak with structure. The intensity and position of the peaks are changed as a function of composition. For x = 0.47 and 0.67, the spectra exhibit a significant intensity increase of peak (1) that may be due to a change of coordination, while for x = 0.85, an additional increase of valency of the Cr atoms likely accounts for a shift of peak (1) towards peak (2). Similar spectral changes and shifts have been observed in hard x-ray SXA at the Cr 1s-edge when going from octahedral to tetrahedral systems [27, 28].

For the Cr L_2,3 SXE spectra in figure 3, we observe a broadening of the spectral shape with increasing carbon content in a similar way as for the Ti L_2,3 SXE peaks in the nanocomposite nc-TiC/a-C films [11]. The Cr L₃ peak exhibits a 0.5–1.0 eV low-energy shift for the Cr_1−xC_x films in comparison to pure Cr metal. The chemical shift of the Cr L₃ peak towards lower energy systematically increases for increasing carbon content. Simultaneously, the L₃/L₂ ratio systematically decreases as the carbon content increases (table 1), signifying that the Cr atoms in the most C-rich Cr_1−xC_x films are less metallic than those with lower carbon content. This is in agreement with the observations of the trend of the L₃/L₂ ratio of Ti in nanocomposite nc-TiC/a-C films [11] and the increasing resistivity of the Cr_1−xC_x samples with increasing carbon content [10].

The C K SXE spectra in figure 4 originate from several different types of carbon. The most important types are (i) carbon from the amorphous carbide, CrC_x, and from (ii) the a-C phase. However, interface bonding between amorphous carbide and a-C domains will likely contribute. For the Cr_0.53C_0.47 sample, an additional contribution comes from carbon in the nanocrystallites. Comparing the spectral shapes of the C K SXE spectra with calculated density of states, the weight of the low-energy shoulder depends on the specific structure [19, 29, 30]. Experimentally, the overlap between the low-energy shoulder and the main σ peak shows the complexity of the amorphous structure. The low-energy shoulder successively gains weight relative to the main σ peak as also previously observed in nanocomposite nc-TiC/a-C films [11]. In contrast to C K SXE spectra of graphite and diamond [31] where dispersive peaks are due to the symmetry-resolved band structure, the spectral shapes of the Cr_1−xC_x samples do not exhibit any significant excitation energy dependence.

If we compare the influence of carbon content on the C 1s SXA spectra, we observe a difference compared to nanocomposite nc-TiC/a-C films in [11] where the intensity of the unoccupied states was observed to systematically decrease for increasing carbon content. In table 1, no clear trend can be seen in the π*/[π* + σ*] contributions as a function of C content. Furthermore, the C 1s SXA spectra are a sum of contributions from the a-C and a-CrC_x phases, and since the relative amount of these two phases varies with C content, it is impossible to draw any quantitative conclusions regarding sp² content in the amorphous a-C phase.

Another important detail is the effect of charge transfer between and within the different structures. The relative intensities of the Cr 2p SXA spectra show that the Cr atoms in the most C-rich samples are most affected by charge transfer from the Cr 3d to the C 2p orbitals. The decrease of the occupied Cr 3d states and the SXE L₃/L₂ branching ratio show that the Cr atoms become more ionic for increased carbon content, an effect of charge transfer. For the C K emission spectra, the situation is opposite to that of the Cr spectra and the intensity and the occupation of the C 2p states systematically increase due to charge transfer for increasing carbon content. The trend in the C 1s SXA spectra is less obvious, but the least empty C 2p states are generally observed for the most C-rich samples. However, due to the smaller difference in electronegativity between Cr–C in comparison to Ti–C, we expect the charge transfer effect to be smaller in the former than in the latter system. The charge transfer from Cr to C may occur within the amorphous CrC_x domains, but also at the interfaces between the domains as an overlapping effect in the C 1s SXA spectra. The size of the charge transfer depends on the total surface area at the interfaces. It is difficult to estimate the interface area in the different samples as a function of C content, due to, for example, the formation of nanocrystallites at x = 0.47 and the nature of the a-C phase (formation of graphene-like structures). We can therefore not, in contrast to the nc-TiC/a-C case [11], quantify the charge transfer effect. This calls for further experimental and theoretical studies on the electronic structure of amorphous transition metal carbide materials.

We have shown that the structure, bonding and properties of chromium carbide (Cr_1−xC_x with x = 0.25–0.85) films deposited by magnetron sputtering are very sensitive to the carbon content. While at low carbon content x = 0.25, the samples are fully amorphous with no structural order, the amount of short-range ordering and charge transfer increase for higher carbon content. For x = 0.47, nanocrystalline grains are observed while for x = 0.67, and 0.85, domains of different carbon content with short- and medium-range structural order are identified. For these compositions, we demonstrate that the films consist of two-phase amorphous domain structures with an amorphous carbon-rich matrix phase and an amorphous chromium carbide phase. The presence of two amorphous phases likely affects the corrosion properties of this material. For the higher carbon contents, x = 0.67 and 0.85, the coordination of the Cr atoms appears to transform as the valency increases and the films also contain graphene-like layers between amorphous domains and clusters. The Cr 2p SXA spectra reveal features that are different from the simple octahedral coordination, suggesting a more complex coordination and bonding. Finally, charge transfer effects are expected in the system and an increasing ionicity of the Cr atoms is observed in Cr SXE for increasing carbon content. However, due to the complex nature of the C 1s SXA spectra with contributions from different types of carbon, this effect is difficult to quantify for each component in the samples.

We would like to thank the staff at the MAX IV Laboratory for experimental support. This work was supported by the Swedish Research Council, the Göran Gustafsson Foundation and the Swedish Foundation for Strategic Research (SSF) through the research programs ProViking and MS²E.