3.3. X-ray Photoelectron Spectroscopy
XPS analysis allows information to be obtained about the chemical structure of the elements on the surface of samples. The samples were exposed to air prior to the XPS measurements. Therefore, the initial surface of all samples was oxidized and strongly contaminated. Survey spectra revealed Cr, Al, C and O peaks, and a carbon content of more than 50% was measured on the surface (Figure 3
a). Cr as well as Al concentrations on the untreated surface were observed at about 5%. The coating deposited at 90 V was measured after 30, 60, 90 and 120 min of argon ion etching (Ar+
etching rate: 2 Å/min) in order to study the chemical composition of the surface layers. Figure 3
b shows the atomic concentration of the elements versus etching time.
Argon etching led to a decrease of C and O and to an increase of Cr and Al concentration due to the removal of contaminated and oxidized layers on the surface. The composition reached a constant value after 60–90 min of argon etching. It is noticed that after 120 min of etching, oxygen still remained at approximately 20%. It is assumed that this amount of oxygen was incorporated into the coatings during deposition. The analysis of the deposition process reveals that the most likely source of oxygen contamination may be residual oxygen in the deposition chamber. The oxygen incorporation in the structure of M2
AlC phases is well known [31
]; however in our case the concentration is higher than expected.
Note that the initial surface was depleted with chromium; the Al/Cr ratio was about 1. The chromium concentration sharply increased in the first 30 min of etching and slightly changed during further etching, while the aluminum concentration remained almost constant. Thus, the inner layers (6–24 nm depth) were depleted with aluminum. The upper limit of the Al/Cr ratio (Figure 3
b) was estimated to be less than 0.35, which was substantially lower than the specified value of 0.5. The origin of such a low Al/Cr ratio obtained from XPS is discussed below (see the description of Al2p + Cr3s lines presented on Figure 4
). Thus, all samples were substantially depleted with aluminum in the subsurface layers. The amount of oxygen on the cleaned surfaces increased with the decrease in substrate bias from 120 to 60 V.
Typical high resolution XPS spectra of Cr2p, Al2p, and C1s core levels of coatings prepared at different bias voltages are presented on Figure 4
a–c. Etching of the surface for 120 min led to almost complete removal of oxidized Cr3+
species and carbon contaminants.
The carbon valence state represented mainly by the peak at 282.8 is typical for carbide-like carbon in all cases, i.e., typical for the Cr–C bond [38
]. Higher than typical values for Al4
(282.2 eV) indicate the absence of a discrete aluminum carbide phase [39
]. Only minor additional components were observed at 285.0 and 286.6 eV, which corresponds to elementary carbon (C–C bonds) and oxidized carbon species (C–O). They were probably formed due to the decomposition of carbide phases into pure elements under ion beam influence (Ar ion damage [40
]) and the subsequent removal of the decomposition products [41
]. Thus, a steady-state concentration of elementary carbon species was observed.
Chromium was represented by two states with binding energies; (BE
) = 574.3 and 577.3 eV. Minor components at 577.3 eV corresponding to Cr2
were observed [43
]. The main asymmetric component at 574.3 eV corresponds to chromium in carbide phase [44
] or in metallic Cr [38
]. Similar XPS results were presented by Zamulaeva et al. for Cr2
AlC prepared using pulsed electrospark deposition (PED) [45
]. Furthermore, both peaks are situated in the same position for all coatings, and no chemical shift could be discerned.
The minor Cr3s photoelectron line overlaps with the Al2p main photoelectron line of aluminum. The estimation of Al2p contribution to this spectrum is possible via two ways. Theoretical Al2p contribution can be estimated using the Cr2p photoelectron spectra and Scofield cross-sections [46
] of the 3s and 2p core-levels of chromium for the calculation of the Cr3s line intensity. Thus, the theoretical Al/Cr ratio could be obtained. The most probable experimental Al/Cr ratio can be estimated via the curve of best fit of the experimental spectrum from the total areas of the peaks at 72.7 and 74.4 eV in the Al2p + Cr3s on Figure 4
using binding energy and the shape of Cr3s taken from [43
]. The Al/Cr ratios obtained each way are presented in Figure 3
b. The Al/Cr ratio was calculated to be about 0.3 using the theoretical approach and about 0.2 using curve fitting of the experimental spectra, which were both lower than the specified value 0.5.
Aluminum was much more oxidized compared to chromium due to its more metallic chemical properties, as obvious from the intensity of the peak at 74.3–74.4 corresponding to Al2
]. The accuracy of the Al2p + Cr3s photoelectron line fitting is confirmed by the quotient of the total oxygen amount and the oxidized Al + Cr species (doublet peak at 577.3 eV in Cr2p and peak at 74.4 in Al2p + Cr3s), which is close to 3 and presented in Figure 5
b. Thus, these peaks in the Cr2p and Al2p + Cr3s lines correspond to Cr2
The peak at 72.8–72.5 eV formally corresponds to metallic aluminum, which has a binding energy of about 72.6–72.8 eV [39
]. Aluminum carbide is characterized by a higher binding energy of Al2p core level 73.4 eV [39
]. Thus, taking into account the rather high reactivity of aluminum towards carbon [39
], we attribute this state to Cr2
AlC. The negative shift 0.6–0.9 eV relative to pure aluminum carbide can be explained with more metallic Al–C bonds in Cr2
AlC phase compared to Al4
. Zhang et al. observed a similar negative shift of the Al2p core level for Ti2
AlN phase compared to pure AlN. Thus, such a shift could be used as indicator of MAX phase formation [51
]. In the current study, the XPS results could confirm the formation of a Cr2
AlC MAX phase, and these results are in good agreement with the XRD measurements.
The Cr and Al carbide states standardized on carbide-like C reached a constant value after 30 min of ion etching, while total carbide-like C concentration reached a constant value only after 60 min of ion etching (Figure 5
a). Thus, despite the removal of carbon contaminants and oxidized species, the carbide phase composition is constant in the topmost ~24 nm; however, this phase is deplete of aluminum and corresponds to Cr2:Al0.25:C1 stoichiometry in the case of the Cr-Al-C coating at 90 V (Figure 5
a). Taking into account the oxidation of the surfaces with oxygen traces during ion etching, the initial “non-oxidized” surface composition was estimated to correspond to the Cr2:Al0.6–0.7:C1 stoichiometry under this assumption.
Moreover, the Cr carbide to carbide-like C state ratio was very stable, independent of the bias voltage (Figure 5
b), and equal to 2.03–2.1, while the Al non-oxidized carbide state to carbide-like C state varied from 0.12 to 0.22.
3.4. Atomic Force Microscopy
AFM gives information about the morphological properties of films. The main characteristics of Cr-Al-C coatings like thickness, surface roughness, and granular size are listed in Table 2
. Thickness, as a function of bias voltage, shows an inverse relationship with the bias. The film thickness reduces from 8.95 to 6.98 µm, which can be related to the notable sputtering off of the film (re-sputtering) [53
During increasing bias voltage, the incident ion energy rises and, therefore, more atoms from the growing coatings will be re-sputtered [55
]. These results are in good correlation with the literature [53
]. However, Jiang et al. showed that, at a low bias voltage (till 50 V), the thickness can increase with rising bias until the ion current density is saturated [54
The surface morphologies of the coated silicon substrates were examined by the AFM method and shown in Figure 6
. All films have a granular structure with visible agglomerated grains. The Cr-Al-C coating deposited at 60 V exhibits the morphology with the highest RMS (root-mean-square). As the substrate bias changes from 60 to 90 V, the structure changes from irregular to rectangular particulates with random alignment and the roughness sharply decreases to 33 nm, indicating a surface-smoothing phenomenon.
The lateral force map for the sample deposited at 90 V with the pronounced superstructure of the films is presented in Figure 6
d. A step-line structure is evident, and its edges are visible. At 120 V, the film became again rougher with a roughness of about 60 nm (Figure 6
c). The increase in roughness with rising bias voltage can be explained by higher ion energy [60
The average granular size was obtained using atomic force microscopy. Similar to the roughness, the granular size significantly decreases as the bias increases from 60 V to 90 V and then rises during a further increase to 120 V.
Reduction of granular size can also be related to a modification of the superficial morphology by argon ion bombardment, increasing the energy associated with the atoms on the substrate surface and/or the growing coating surface [62
]. In addition, Lee et al. [63
] reported that applied DC bias may cause an increase in the point defects in the film’s structure, increasing compressive residual stress and reducing the grain size. However, as mentioned above, when the bias voltage increases, the ion energy increases, resulting in higher adatom mobility. High energy promotes diffusion as well as grain boundary migration that might cause increasing granular size [60
]. Similar results were also reported by Gangopadhyay et al. [53
] and Zhang et al. [65
]. As available from Table 2
, the first increase in bias voltage results in a decreasing granular size, while a further increase in bias voltage leads to an increase in granular size. Therefore, in the literature, both effects were discussed contradictorily. Thus, a superposition of different effects takes place. The problem of whether higher adatom mobility, higher ion energy, and more defects result in smaller or larger granular size has still not been solved [53
3.5. Transmission Electron Microscopy
Cross-sectional TEM samples of the Cr2
AlC film deposited at 90 V were analyzed by means of TEM to understand the coating growth. The coating layer exhibits a columnar polycrystalline growth structure consisting of 120–250 nm width columns (Figure 7
a). The selective area electron diffraction taken from a 3-µm diameter circle shows a textured coating with preferential columnar orientation. Cr2
AlC phase was recognized and the Cr2
AlC planes (012) (d012
= 0, 2311 nm), (013) (d013
= 0, 2143 nm), (016) (d016
= 0, 1619 nm), and (008) (d008
= 0, 1346 nm) were detected. The coating is visibly textured. The angle measurement of the most varied diffraction spots in the same interplanar distance showed that most of the grains are oriented in ±15° angle to the main growth axis.
The substrate-coating interface region shows the presence of two layers marked with the numbers 1 and 2 in Figure 8
. An amorphous or nanocrystalline layer marked with number 1 has about a 50–60 nm thickness. Between the amorphous region and the columnar crystal, a 100–120 nm nanocrystalline layer exists. In this region, the beam is strongly dispersed but has not created a circular polycrystalline pattern. It is possible to observe individual crystalline reflections, but the pattern has the attributes of an amorphous or nanocrystalline structure.
The diffraction taken in a single grain confirms the presence of Cr2
AlC phase (Figure 9
a). The selected area electron diffraction reveals blurred fuzzy streaks and extra diffraction spots along the direction  (Figure 9
b). They can be caused by the formation of a superlattice structure inside the Cr2
AlC grain, ordered stacking faults, or another long-period ordered crystal arrangement. This theory corresponds to the bright-field image, showing an oriented substructure inside the Cr2
AlC grain (Figure 9
a). Further results of the TEM analysis of the crystal growth will be published in the near future.
The hardness (H
), elastic modulus (E
), and the H
ratios of the coatings were obtained using a nanoindenter. The results had shown that the mechanical properties of the coating were influenced by the bias values (Table 3
). The hardness and E-modulus values rapidly increased from 8.8 to 15.8 GPa and from 223.4 to 307.7 GPa, respectively, when bias rose from 60 V to 90 V. By further increasing the bias voltage up to 120 V, H
were decreased. The H
values possessed maximums of 0.042 and 0.052, obtained at a bias voltage of 90 V. High H
values mean larger elastic strain to failure and higher fracture toughness [66
]. The high hardness at 90 V is probably due to the reduction in granular size caused by the ionic bombardment [62
]. Consequently, this enhancement of the mechanical properties with the increase in the applied bias voltage is caused by a blockage for the displacement of the cracks, and thus the energy necessary for the motion of cracks through the coating increases [67
]. As previously reported in the literature, an increase in substrate bias raises the level of compressive residual stresses due to an increasing number of impinging ions onto the substrate as well as higher ion current density [68
]. The XPS results show a change in the ratio of Al carbide state to carbide-like C state, determining the mechanical properties of the films. This statement needs further investigation and more detailed analysis. The nanoindentation results are consistent with the previous work of the authors [17
], with hardness values in the range between 11 and 14 GPa at various sputtering powers as well as with further investigations about Cr2
]. Schneider et al. [6
] measured the hardness value of Cr2
AlC as 13 ± 2 GPa. Zamulaeva et al. [25
] reported a hardness of 15.4 GPa and an elastic modulus of 288 GPa for coatings produced using pulsed electrospark deposition (PED). These hardness values are rather low in comparison to the nanoindentation data of Ti2
] and Ti3
] and are similar to V2
AlC coatings [72
] prepared using magnetron sputtering. Furthermore, the hardness values presented above are significantly higher than those for bulk MAX phases, which are typically 2–4 GPa [1
]. This increase could be explained by the Hall-Petch’s effect [74
], wherein the strength of materials is inversely proportional to the square of the crystalline size before a decreased threshold value. Due to the rapid cooling during PVD vapor condensation, the coatings developed nanocrystalline structure, which can be attributed to the better mechanical properties of coatings, compared to bulk material, and confirmed for different MAX phase coatings [6