Metallurgical and Materials Transactions A

, Volume 38, Issue 10, pp 2464–2478

Oxidation Behavior and Mechanisms of TiAlN/VN Coatings


  • Z. Zhou
    • Department of Engineering MaterialsUniversity of Sheffield
    • Department of Engineering MaterialsUniversity of Sheffield
  • C. Rodenburg
    • Department of Engineering MaterialsUniversity of Sheffield
  • N.C. Hyatt
    • Department of Engineering MaterialsUniversity of Sheffield
  • D.B. Lewis
    • Materials Engineering Research InstituteSheffield Hallam University
  • P.E. Hovsepian
    • Materials Engineering Research InstituteSheffield Hallam University

DOI: 10.1007/s11661-007-9293-4

Cite this article as:
Zhou, Z., Rainforth, W., Rodenburg, C. et al. Metall and Mat Trans A (2007) 38: 2464. doi:10.1007/s11661-007-9293-4


Hard wear-resistant coatings require excellent oxidation resistance for high-speed machining operations. Moreover, the oxide formed is integral to the frictional behavior and therefore the success of the coating. The oxidation behavior of TiAlN/VN nanoscale multilayer coatings was investigated using high-resolution techniques and was compared with TiN and TiAlN coatings. Static oxidation of TiAlN/VN films was studied in the range 550 °C to 700 °C, and characterized by high-temperature in-situ X-ray diffraction (XRD) and scanning transmission electron microscopy/energy-dispersive X-ray/electron energy loss spectroscopy (STEM/EDX/EELS) of selected surface cross sections. The oxidation resistance of TiAlN/VN was found to be controlled by the VN layers, and consequently, oxidation was initiated at a lower temperature than TiN and TiAlN coatings. The onset of oxidation of the TiAlN/VN coating was found to be ≥550 °C with the VN being the first component to oxidize. At temperatures >600 °C, a duplex oxide structure was formed; the inner layer comprised a porous region of Ti-rich and V-rich nanocrystallites, while several phases were observed in the outer region, including V2O5, TiO2, and AlVO4. V2O5 was the dominant oxide at the outer layer at ≥638 °C. The outward diffusion of V depended on the species present; in the inner layer, V was present as V3+, V4+, whereas a significant V5+ was dominant in the outer layer of oxide at ≥638 °C. An Au marker study suggested roughly equal diffusivity of cations outward, and oxygen inward diffusion occurred during oxidation.

1 Introduction

Titanium nitride (TiN) with the B1 NaCl structure has been widely used as a hard wear-protective coating since the 1980s. However, a major limitation of TiN coatings for high-speed machining applications is that they oxidize rapidly to rutile, TiO2, at temperatures above 500 °C. Initial oxidation is parabolic, with O diffusion through TiO2 to the oxide/nitride interface as the rate-limiting step.[1] The large difference in molar volumes of TiO2 and TiN results in spallation of the oxide scale, resulting in a transition from parabolic to pseudo-linear kinetics.[2,3]

TiAlN coatings provide much improved high-temperature oxidation resistance, up to 750 °C to 900 °C, and have consequently been used extensively for high-temperature cutting operations with minimum use of lubricant or dry machining.[4] Titanium aluminum nitride (Ti1-xAlxN) has the B1 NaCl structure with 0 ≤ x ≤ 0.52, but as x is increased, phase decomposition occurs into B1 NaCl and wurtzite AlN structures.[5,6] Ti0.5Al0.5N exhibits stable parabolic oxidation kinetics, associated with an oxide comprised of a passive double layer oxide, which is a result of outward diffusion of Al to form Al-rich oxide at the topmost surface and inward diffusion of O to form Ti-rich oxide at the interface to TiAlN.[7] However, during sliding wear tests, the friction coefficient of TiAlN is usually high, typically μ = 0.8 for TiAlN measured from the pin-on-disc test against alumina ball.[8]

Both the friction and wear performance of TiAlN can be improved through the quaternary addition of V, as monolithic Ti-Al-V-N coatings or as TiAlN/VN multilayers. The VN, which also has the B1 NaCl structure, exhibits continuous solid solubility with a number of metal nitrides and carbides, such as TiN, TiC, TiAlN, VC, and NbN.[9] Knotek et al.[10] found that the addition of V to TiN and Ti-Al-N enhanced wear performance, with the best wear resistance obtained in a Ti-V-N coating with 29 at. pct V. TiAlN/VN multilayers, which incorporate VN into TiAlN as a nanoscale multilayer (also referred to as a “superlattice”) coating, have exhibited excellent sliding wear resistance (1.26 × 10−17 m3·N−1 m−1) with a lower friction coefficient at room temperature (μ = 0.4, pin-on-disc test, Al2O3 ball counterpart) in comparison to other wear-resistant coatings,[8] e.g., TiAlN/CrN (sliding wear coefficient 2.4 × 10−16 m3·N−1 m−1, μ = 0.7 to 0.9).

The oxidation of VN has been investigated by several authors.[1113] Oxidation is complex, with columnar V2O5 being the dominant scale, but also suboxides V6O13, VO2, and the oxygen-deficient Magnéli phases, VnO2n−1, VnO3n−1, or VnO3n−2, which have the rutile structure (TinO2n−1) formed at the interface with a lamellar structure. The limiting step was found to be the diffusion of both anionic and cationic species through the V2O5. Not surprisingly, the reaction rate accelerated dramatically at the melting point of the V2O5 (674 °C), where oxidation became catastrophic.[12]

Frictional heating during sliding often results in the formation of surface oxides. These oxides can have either a beneficial or detrimental effect on wear and friction behavior.[14,15] In the case of TiAlN/VN, it is believed that the products of oxidation can substantially reduce friction. For example, high-temperature wear tests of TiAlN/VN coatings[16] and Ti-Al-V-N films with various contents of V (2 to 25 at. pct in targets)[17] against an Al2O3 ball showed a dramatic decrease of friction coefficient from >0.8 at 500 °C to 0.2 at 700 °C. Gassner et al.[18] provided some evidence that this is associated with the formation of a variety of Magnéli phases, while V2O5 has also been observed in room-temperature tests using Raman spectroscopy.[19] The V2O5 has a melting point of 674 °C[20] and inherently low friction, leading to suggestions that the formation of this oxide is the key to understanding the low frictional behavior of these coatings. Moreover, the Magnéli phases, VnO2n−1, VnO3n−1, or VnO3n−2, all contain crystallographic shear (CS) planes, which are known to promote low friction.[21]

It is clear that the oxidation behavior, both under static and sliding conditions, strongly affects the performance of these coatings. To date, very little knowledge is available on the static oxidation of the Ti-Al-V-N system. Preliminary investigations by the current authors into the thermal behavior of TiAlN/VN coatings[22,23] showed that oxidation initiated at ≥550 °C with V2O5, TiO2, and AlVO4 identified at 600 °C. Small changes in (Ti + Al)/V ratio changed the exact temperature for the onset of oxidation, with an increase in (Ti + Al)/V increasing the onset temperature, but these changes did not apparently change the mechanism.[2325] The current work provides a comprehensive understanding of the oxidation mechanisms in these coatings for approximately constant (Ti + Al)/V. Although not the focus of the study, oxidation kinetic data are included for TiN and TiAlN coatings as well, to provide a baseline.

2 Experimental

TiAlN/VN multilayer coatings were grown on stainless steel substrates by reactive unbalanced magnetron (UBM) sputtering in an industrial scale physical vapour deposition coating machine (HTC-1000 ABS, manufactured by Hauzer Techno Coating BV, Venlo, The Netherlands). This system comprises four vertically mounted targets that can be run in either cathodic arc or UBM mode, 2 TiAl (50:50) and 2 V (99.99 pct) targets in this case. Prior to deposition, a V metal ion etch was undertaken in cathodic arc mode, which was used to enhance adhesion between coating and substrate. Following metal ion etching, a thin (∼0.25-μm) VN base layer was deposited by UBM sputtering to further improve adhesion and also to reduce the residual stress. Coating composition can be altered by changing the bias voltage applied to substrates during processing or changing power dissipated to the targets, as shown in Table I. All the coatings were deposited at a constant temperature of 450 °C in a common Ar + N2 atmosphere at a total pressure of 4.5 × 10−3 mbar. A more detailed description of the process parameters is provided elsewhere.[8] Coatings were deposited with threefold planetary substrate rotation. Because there was no mechanical shielding/shutters during deposition, intermixing between layers occurred, as detailed elsewhere.[26]
Table I

Details of Coating Deposition Conditions, Composition, Dominant Crystallographic Texture, and Residual Stress


Bias Voltage

Al (At. Pct)

Ti (At. Pct)

V (At. Pct)

(Ti + Al)/V


Residual Stress (GPa)

















In order to compare the oxidation kinetics of the TiAlN/VN multilayer coatings, TiN and TiAlN coatings were prepared. The coatings were deposited on the same machine using bias voltages of 75 V. Full details of the deposition and structure of these coatings are provided elsewhere.[27,28]

A Cahn TG 131 microbalance was used for the thermogravimetry work, with the following specifications: temperature drift stability of 10 μg/°C, temperature repeatability of ±3 °C, and mass sensitivity of 1 μg. The TG allows the determination of the onset point of oxidation and the extent of weight gain. All coatings were deposited on stainless steel 304 coupons (dimensions 50 × 15 × 0.75 mm). The coupons were completely covered with coating material, including the hole drilled for fixing the sample in the TG furnace, such that the oxidation would only occur in the coating and the substrate would only become oxidized when the coating failed. Oxidation of TiAlN/VN was assessed dynamically by thermogravimetric analysis (TGA) in a linear-temperature-ramp (400 °C to 1000 °C at 1 °C/min) mode. Isothermal heat treatment at 550 °C and 600 °C was also conducted. Coatings were subsequently heat treated at 550 °C, 600 °C, 638 °C, and 672 °C in a conventional ambient air atmosphere furnace in order to correlate the TG traces with phase transformations on the coating surfaces, in particular, the formation of V2O5. These temperatures were selected based on the TG trace.

The X-ray diffraction (XRD) of the as-deposited and oxidized coatings was undertaken on a PHILIPS MRD-Xpert, Cu Kα radiation with 2θ from 10 to 140 deg using standard Bragg Brentano diffraction conditions. Glancing angle geometry with 1 deg incidence was used when the oxide scale was thin, i.e., at 550 °C.

A coating deposited with a bias voltage of −85 V was examined using high-temperature XRD. The sample was heated in a Bruker D5005 diffractometer with a high-temperature stage from room temperature to 700 °C at 25 °C/min. The sample was held at the chosen temperature for 40 minutes to complete a scan. The θ to 2θ scan was performed from 10 to 55 deg, at 0.014-deg step size using Cu Kα radiation.

The as-deposited and oxidized surfaces were characterized by a scanning electron microscope (SEM, PHILIPS1 XL30) equipped with Link-Isis EDX spectrometer. Cross sections of the as-deposited coatings and oxidized surfaces were prepared for TEM observation in the usual manner. Two slices of oxidized surface were bonded face to face using epoxy glue, with the composite bar subsequently sectioned, ground, polished, and ion milled. Ion milling was performed by a Gatan2 precision ion pump system with ion modulation control (which prevents Ar ions from striking from the surface to substrate direction, which helps to retain the original outmost surface; Reference 29). Samples were characterized using a JEOL3 2010F field emission gun TEM equipped with a scanning transmission unit and an EDX detector with Link-Isis interface. The scanning transmission electron microscopy/energy-dispersive X-ray (STEM/EDX) elemental maps were obtained using Al Kα, Ti Kα, and V Kβ peaks. Energy-filtered TEM (EFTEM) images were obtained using a Gatan GIF200 attached to the JEOL 2010F using the conditions given in Table II.
Table II

Experimental Conditions for the Acquisition of Elemental and Jump Ratio Images on the Gatan GIF200


N-K (401 eV)

Ti-L2,3 (456 eV)

V-L2,3 (513 eV)

O-K (532 eV)





N/A (jump ratio)











Slit width (eV)





The Au marker studies were undertaken to provide information of the relative diffusion rates of the cations and anions. A TiAlN/VN coating was covered by a piece of net fabric and then sputter coated with Au for 12 minutes. The Au islands were left on the coating surface once the net fabric was removed. The coating was then heat treated in air at 600 °C for 0.5 hours. It was assumed that Au does not react with the coating and the oxides up to this temperature. The oxidized coating was then Ni plated to preserve oxide scale prior to cross-sectional metallographic preparation.

3 Results

3.1 Composition and Structure of the As-Deposited Coatings

The coatings deposited are shown in Table I. Relative contents of (Ti + Al)/V were determined by SEM/EDX at the plan-view coating surface, using a standardless routine (error ±0.5 pct), and therefore the results should be regarded as semiquantitative. Coatings deposited at −75 and −85 V had approximately the same (Ti + Al)/V ratio. Low-angle XRD revealed multilayer repeat periods of ∼3 nm for all coatings. The coating deposited with a substrate bias voltage of −85 V possessed a residual compressive stress of −7.9 GPa and (111) texture, whereas the coatings deposited at −75 V bias had a lower residual stress of −3.9 GPa and exhibited a (110) texture. More details about the influence of substrate bias voltage on the coating’s composition, microstructure, texture, residual stress, and hardness were provided in Reference 30.

Figures 1(a) and (b) show the typical microstructures of as-deposited coatings with −75 V bias. The coatings comprised three parts: stainless steel substrate (including V ion-etched surface), base layer (250 nm), and multilayer coating (3.2 μm). The V ion-etched region cannot be seen, but the etching performed using the cathodic arc process usually leads to occasional growth defects, which result from droplets caused by localized melting and vaporization of the cathode.[31] The coating that grows on top of a droplet subsequently gives rise to a growth defect.[32] The coatings generally had a columnar grain structure. Regions of low density are occasionally present at the columnar grain boundaries (Figure 1(b)). The population of low-density regions depends on growth bias voltage, with more low density regions present for the −75 V bias voltage than the −85 V bias voltage. The plan view of the coatings at low bias voltage, i.e., −75 V also showed voids at columnar grain boundaries.[30]
Fig. 1

TEM bright-field images showing (a) typical microstructures of as-deposited coatings with −75 V substrate bias voltage and (b) voids at columnar grain boundaries in coatings with −75 V substrate bias voltage

3.2 Continuous and Isothermal TGA

Figure 2(a) plots TGA oxidation rate measurements for TiN, TiAlN (3 at. pct Cr), and TiAlN/VN coatings on stainless steel coupons using the linear-temperature ramp (400 °C to 1000 °C at 1 °C/min). The differential thermal gain is also plotted as a function of temperature at the bottom half in Figure 2(a), used to determine the onset of oxidation and the oxidation rate. The TGA of a bare stainless steel coupon was acquired to exclude any mass gains resulting from the substrate contribution to the coating TGA curve, showing a negligible amount of mass gain up to 1000 °C (Figure 2(a)). The TiAlN (3 at. pct Cr) coating exhibited detectable steady mass gains at ≥880 °C, with a significant mass increase at 930 °C (onset for rapid oxidation). The TiN showed a detectable steady mass gain at ≥600 °C. In contrast, TiAlN/VN coatings showed detectable steady mass gain at ≥550 °C with a significant mass increase at 638 °C ± 5 °C (onset for rapid oxidation) up to 672 °C. The mass gain did not increase appreciably with temperature between 650 °C and 820 °C, but exhibited a catastrophic increase from >820 °C.
Fig. 2

(a) TGA of the oxidation rate for TiN, TiAlN (3 at. pct Cr), and TiAlN/VN on stainless steel coupons using the linear-temperature ramp (400 °C to 1000 °C at 1 °C/min). The bottom part of the graph is the differential TG by which the onset for rapid oxidation was determined. (b) Isothermal TG traces obtained by heating the TiAlN/VN coatings, −85 V bias, at 550 °C and 600 °C, respectively. The temperature ramp is also shown

Figure 2(b) gives the isothermal TG traces obtained by heating the TiAlN/VN coating (−85 V bias) for 1.5 hours at 550 °C and 600 °C, respectively. The temperature profile is also shown, from which it is clear that oxidation may have started during the heating phase for the 600 °C sample. The oxidation rates at 550 °C initially appeared logarithmic, but changed to parabolic after ∼0.4 hours, and became linear after ∼0.7 hours with a much slower rate of mass gain. The behavior of the 600 °C sample was similar, except the parabolic oxidation rate was more rapid, and the oxidation rate >0.7 hours was slightly greater (0.08 g/m2 h) compared to the 550 °C sample (0.06 g/m2 h). The oxidation kinetics of the TiAlN/VN coatings at 550 °C and 600 °C closely resembles that of the Nb and Ta metals below 600 °C.[33]

3.3 Surface Structure and Phase Evolution during Oxidation

Figure 3 gives high-temperature in-situ XRD traces showing the phase evolution for various temperatures for the coating with a bias voltage of −85 V. The XRD of the coating deposited with a bias voltage of −75 V revealed near identical results, and therefore, the difference in deposition conditions did not appear to affect the oxidation mechanisms. The V2O5 phase was first detected through the appearance of the (001) peak at 597 °C, which increased in intensity to 628 °C. At T ≥ 638 °C, V2O5 (001) decreased and V2O5 (110) increased and exceeded the (001) peak significantly. The (110) preferred orientation was a maximum at 648 °C but decreased as the temperature was raised further. Thus, there was not only a change in phase proportion, but also a change in the orientation of the V2O5 crystals with respect to the coating surface. At temperatures ≥678 °C, peaks associated with the V2O5 phase all disappeared, which is consistent with the melting temperature of V2O5 at 674 °C. AlVO4 was found at ≥608 °C and TiO2 at ≥638 °C, and were the only peaks present for temperatures ≥678 °C, at which temperatures V2O5 phase is molten.
Fig. 3

High-temperature XRD showing phase evolution upon heating on the coating (−85 V bias). Each scan took 40 min

TiAlN/VN coatings deposited with a bias voltage of −85 V were heat treated at 550 °C, 600 °C, 638 °C, and 672 °C, respectively, for 0.5 hours, cooled, and examined by XRD (Figure 4). The heat-treatment temperatures were chosen on the basis of the continuous TG measurements (Figure 2(a)) and represent the stages of the onset of oxidation, the onset of rapid oxidation, and the point at which spallation occurred. The time was chosen on the basis of the isothermal TG measurements (Figure 2(b)) corresponding to the parabolic oxidation. In addition, the coating was heated at 600 °C for 1 hour in order to examine the oxide evolution as a function of time at temperature. The room-temperature XRD data from these isothermal oxidation studies exhibited only small differences to the high-temperature XRD presented in Figure 3. The differences were believed to be attributable to the total time at temperature (which was clearly much longer for the high-temperature XRD) rather than any difference in mechanism. The as-deposited coating was single phase, with a (111) texture. The coating oxidized at 550 °C/0.5 h exhibited little change overall, although some weak additional peaks were present indicating new phases had formed, but there was insufficient intensity to allow identification. After 600 °C/0.5 h, V2O5 with small amounts of AlVO4 and TiO2 phases were identified at the surface of coating. After 600 °C for 1 hour, the AlVO4 peaks had increased in intensity, whereas there was comparatively little change in the V2O5 and TiO2 peaks compared to the 0.5-hour sample. The V2O5 phase appeared to have a (001) preferred orientation. After heating at 638 °C/0.5 h, a substantial amount of V2O5 with preferred orientation (001) dominated the phases at the surface of coating, although again smaller quantities of AlVO4 and TiO2 were also present. After heating at 672 °C/0.5 h, the amount of V2O5 reduced, with slightly more TiO2 and AlVO4. In addition, a stronger stainless steel substrate peak was present, but peaks from the coating had disappeared, indicating that coating was completely oxidized and spallation had probably occurred.
Fig. 4

XRD patterns of the as-deposited and oxidized TiAlN/VN coatings, −85 V bias, taken at room temperature

Figure 5 gives SEM micrographs of the starting and oxidized TiAlN/VN coating (deposited with a −85 V bias voltage). Figure 5(a) shows the as-deposited coating, which exhibited a largely featureless surface apart from growth defects. Growth defects are dome shape in plan view and have dense columnar grains extending outward nearly perpendicular to the local surface of the droplet in the cross section.[34] Figure 5(b) is the coating heated at 550 °C/0.5 h, which shows the surface was uniformly covered by a thin layer of oxide with needle morphology, consistent with the weak peaks in the XRD spectra. The growth defects are still visible after this treatment, indicating that the surface oxide thickness is less than the height of growth defects, ∼0.17 μm, derived from the surface roughness of the coating. At 600 °C, the surface was similar to that at 550 °C, being dominated by needle-shaped crystals, but the size and density of the needles had increased (Figure 5(c)). The growth defects were less obvious, indicating that the oxide thickness was greater than the height of the growth defect. Figure 5(d) shows the surface microstructure of the coating after extended heating at 600 °C for 1 hour. Large crystals were observed extending approximately parallel to the surface, which EDX analysis indicated were predominantly V and O, indicating, with the XRD data, that they were probably V2O5 crystals. The finer angular particles that covered the rest of the oxidized surface were Al/V rich, and, according to the XRD data and the TEM investigation (discussed subsequently), were believed to be predominantly AlVO4. A further increase in temperature to 638 °C led to a substantial increase in the V2O5 crystals growing parallel to the surface, which occupied approximately 90 pct of the surface area (region A in Figure 5(e)). The remaining area (region B) was found to be Al and Ti rich, with a certain level of V, suggesting the presence of TiO2 and AlVO4 phases, but these regions were recessed below the V2O5, which essentially formed the outer layer. Figure 5(f) shows the coating heated at 672 °C. The large V2O5 crystals present at 638 °C had disappeared. Blisters appeared on the surface with diameters of ∼40 μm, and there was general breakup of the integrity of the oxide layer. The EDX analysis on the surface of blisters suggested they were Ti deficient with respect to the adjacent flat area on the surface.
Fig. 5

Scanning electron micrographs showing the surface microstructures of the oxidized TiAlN/VN coatings (−85 V): (a) as deposited, (b) 550 °C/0.5 h, (c) 600 °C/0.5 h, (d) 600 °C/1 h, (e) 638 °C/0.5 h, and (f) 672 °C/0.5 h

3.4 Cross Sections of Oxidized Surface by TEM

On the basis of the TGA, SEM, and XRD data, selected samples were examined in detail in the TEM. Figure 6 shows the bright-field STEM image and associated EDX maps from cross sections of the TiAlN/VN coating (−75 V bias) heat treated at 550 °C for 0.5 hours. Ti Kα (4.51 keV) and V Kβ (5.43 keV) were selected to form the EDX maps, because Ti Kβ (4.93 keV) overlaps with V Kα (4.95 keV). The coating had an oxide layer of ∼300-nm thickness, consistent with SEM observations. The oxide exhibited a rudimentary duplex oxide structure, with the outermost surface comprised of needle-/plate-shaped crystals, but the inner layer was porous with the greatest pore density adjacent to the nonoxidized coating. The elemental maps suggested the outermost surface was V rich, while the region adjacent to the coating was dominated by Ti and Al. Phase identification of the oxide was difficult owing to the small dimensions of the particles.
Fig. 6

(a) Cross-sectional bright-field TEM micrograph of the surface oxide and pores in TiAlN/VN coating (−75 V bias) heat treated at 550 °C for 0.5 h. (b) STEM/EDX map group of the oxidized coating

Figure 7 shows a cross section of the TiAlN/VN coating (−75 V bias) oxidized at 600 °C for 0.5 hours. A well-defined duplex oxide structure was now present, comprising an outer layer of well-defined crystals and a much finer nanocrystalline porous inner. The EDX indicated that the majority of the V was located in the outer layer, but that the outerlayer also contained crystals rich in Ti (which sometimes extended into the inner layer) and those that contained both Al and V. The inner layer was mainly Ti and Al, although with some V.
Fig. 7

Bright-field STEM cross section of TiAlN/VN coating (−75 V bias) oxidized at 600 °C for 0.5 h

Figure 8 gives diffraction patterns from tilting studies of the arrowed crystal at the outermost surface of the cross section in Figure 7. The only possible crystals that match with the EDX data are AlVO4 (triclinic)[35] and AlV2O4 (cubic, spinel).[36] The diffraction patterns shown in Figure 8 are consistent with AlVO4 (known to be present from the XRD), but excluded the possibility of AlV2O4. The AlVO4 was found to be present as discrete, approximately equiaxed crystals (such as that arrowed in Figure 7), bridging almost entirely the outer layer of the duplex oxide layer and distributed homogenously between other phases across the surface layer.
Fig. 8

Selected area diffraction patterns for the angular particle (arrowed) in Fig. 7, identified as AlVO4 (triclinic). Coating was heat treated at 600 °C for 0.5 h

Figure 9 gives a cross section of the oxidized surface, where TiO2 (rutile) was confirmed by selected area electron diffraction patterns (the directions of the patterns shown were experimentally tilted by 19.5 deg, which agrees well with the theoretical angle of 19.4 deg). Further, TiO2 (rutile) particles were found situated in both outer and inner layers, the distribution of which can be seen in the Ti map in Figure 7. The rutile was present in the outer layer, but was also found in the inner layer, generally as finer crystals than in the outer oxide layer.
Fig. 9

Bright-field TEM image of a cross section of the oxidized surface, as shown in Fig. 7, where TiO2 (rutile) was confirmed by two selected area electron diffraction patterns and EDX. Coating was heat treated at 600 °C for 0.5 h

Figure 10 shows a further STEM bright-field image and associated EDX maps of outer layer oxide, taken from a region of the oxidized surface where more needles were present. Apart from V-rich only needles (later identified as V2O5), additional needle-shaped crystals were present, which were Al/Ti rich. Systematic tilting studies were not successful because of the fine dimensions being finer than the V2O5 needles. The limited electron diffraction data obtained was consistent with Al2TiO5, which is a probable phase from the phase diagram, but no absolute identification could be made.
Fig. 10

Bright-field STEM/EDX maps of another area at the surface where more needles were present with Al and Ti rich, possibly Al2TiO5. Coating was heat treated at 600 °C for 0.5 h

Figure 11 shows an EDX spectrum and selected area diffraction patterns taken from the needlelike regions, which confirmed that some, but not all, of the needle-shaped crystals to be V2O5 (residual Ti and Al in the spectra were from surrounding particles). The streaking effect of diffraction spots is consistent with the “layered”[37] or “lamellar”[38] structure of V2O5. These V2O5 needles tended to grow out in-between the angular AlVO4 in the outer oxide layer. However, some needles were also found in the inner layer where TiO2 was located.
Fig. 11

TEM bright-field image, EDX spectrum, and selected area diffraction patterns by which V2Oneedle-shaped particles are identified. Coating was heat treated at 600 °C for 0.5 h

Figure 12 shows a typical area at the interface between the oxide and the coating in the TiAlN/VN (−75 V bias) sample heated at 600 °C for 0.5 hours. The oxidation front followed the columnar structure rather than the multilayer structure. Oxidation was most rapid along the columnar boundaries, as expected. The oxide in this inner layer was porous and nanocrystalline. The electron diffraction pattern from this region, given in Figure 7, was complex. Indexing could be made to V2O5 and TiO2, but additional spots were also present in the diffraction pattern. No unique identification of these additional spots could be made, although a good fit was found with (100) and (220) planes of VO2 and (230) and (006) planes of Al2TiO5.
Fig. 12

Bright-field TEM image showing a typical area at the interface of the oxide/coating in TiAlN/VN (−75 V bias) heated at 600 °C, which grows preferentially along columnar grain boundaries

In order to explore the structure of the interface and inner layer further, energy-filtered TEM and associated EELS spectra were acquired. Figure 13 shows a zero loss bright-field TEM image along with elemental maps of the same area for N, O, V, Ti, and Al. A thickness map is also provided to aid in interpretation and to show where the voids were present. The thickness map (t/λ typically 0.2 to 0.3 in the oxide; brighter contrast corresponds to thicker regions) also showed that there was preferential ion beam thinning of the oxide with respect to the coating. Figure 14 gives a similar series of energy-filtered maps taken from a similar area, but including a columnar boundary in the coating. The maps show a complex variation in composition, of a scale typically 10 to 100 nm in size. Figure 15 shows a series of EEL spectra (background subtracted) obtained by point analysis from the oxide, 3nm above oxide/coating interface, at the interface, 3nm below the interface in the coating and the bulk coating well away from the interface. Note the overlap of the OK edge and the V L2,3 edge. Thus, it is probable that the V energy-filtered maps in Figures 13 and 14 contained some of the O signal, and vice versa.
Fig. 13

Zero loss bright-field TEM image, thickness map, and elemental maps of the same area for N, O, V, Ti, and Al at the interface and inner layer of oxide for the sample heated at 600 °C for 0.5 h
Fig. 14

Bright-field TEM image of the oxide/coating interface containing a grain boundary and elemental maps of the same area for V, O, Ti, and N at interface and inner layer of oxide for the sample heated at 600 °C for 0.5 h
Fig. 15

Background subtracted EELS spectra across the interface for the sample oxidized at 600 °C

The first notable point is that the N content in the oxide film appears to be very small, suggesting that most of the N must have been liberated as gas, discussed later. However, the EEL spectra show that the structure of the N K edge is substantially different at the interface compared to the coating. The shape of the edge (near edge structure) reflects the bonding environment of the N and indicates a significant change in bonding at the oxide interface.

The energy-filtered maps indicate an abrupt interface between the oxide and the coating. The O K edge is distorted by the overlying near edge structure from the V L2,3 edge. However, it would appear that while O is clearly present at the interface, it is below detectable quantities 3 nm into the coating. The V L2,3 intensity significantly reduced across the interface and was generally lower within this oxide layer, although the exact intensity changed from point to point depended on the local phase constitution, as shown by the V energy filtered map in Figures 13 and 14. The reduction in V intensity was coincident with the higher porosity in the coating, but this did not reduce the intensity in the Ti L2,3 or Al L2,3 maps. The intensity in the Ti L2,3 maps also varied substantially from point to point, although locally in the oxide film the edge intensity was as high as in the coating, as in the spectrum shown in Figure 15.

Figure 16 shows the cross section of TiAlN/VN coating (−85 V bias) oxidized at 638 °C for 0.5 hours. The oxide was broadly similar to that observed at 600 °C. However, this sample also included a section through a large V2O5 crystal at the outermost surface of thickness 300 to 400 nm, consistent with the structure observed in the SEM micrograph in Figure 5(e). The two-layer structure below this outer crystal was similar to that found at 600 °C, and as far as was practically possible, it would appear to be comprised of the same phases. However, as is clear in the EDX maps in Figure 16, the TiO2 appeared to be more concentrated in the inner layer, while the AlVO4 comprised much of the layer below the V2O5. This is consistent with the XRD spectra in Figure 4, which clearly shows an enhancement of AlVO4 at 638 °C compared with 600 °C.
Fig. 16

Bright-field STEM cross section of TiAlN/VN coating (−85 V) oxidized at 638 °C for 0.5 h

Figure 17 shows the interface of oxide/coating in TiAlN/VN (−85 V) heated at 638 °C. The Fresnel contrast from the multilayer structure was retained right up to the reaction interface, suggesting that interdiffusion of species between individual TiAlN and VN layers was not sufficient to remove the multilayer structure prior to oxidation and that one layer was not oxidized before the other. Interestingly, occasional fragments of the coating remained within the oxide region (arrow Figure 17(b)), implying that oxygen supply may have been a rate-limiting step.
Fig. 17

Bright-field TEM images from the interfaces of oxide/coating in TiAlN/VN (−85 V) heated at 638 °C/0.5 h, showing (a) preferential oxidation at the columnar grain boundaries and (b) a fragment of nonoxidized coating buried in the inner layer of oxide

Thus, the TEM section indicates that at 638 °C, the oxide comprises three layers: two layers with essentially the same phase constitution as at 600 °C, but with a change in proportion of phases, and with the V2O5 forming a near continuous outer layer, whereas this layer was occasional at 600 °C.

3.5 Au Marker Study

As the diffusion rates of the reacting species through the scale are largely unknown, Au marker studies were undertaken on the TiAlN/VN coating deposited with a bias voltage of −85 V. Figure 18(a) shows a SEM image of the cross section of the oxidized coating. The oxide structure was essentially identical to that observed in the TEM studies. An X-ray linescan was conducted, as indicated on the image, and Figure 18(b) shows the profiles of elements. The presence of Ni defined the surface of the oxide. The V, Al, and Ti were all found located above the Au marker, with Al associated with V at the outermost surface; Ti followed closer to the Au marker. This indicates V, Al, and Ti diffused outward to the surface, but with V and Al diffusing faster than Ti. The results also suggest that some nitrogen was present in the coating, and, as expected, there was an oxygen gradient from outer to inner surface.
Fig. 18

(a) SEM image of the cross section of the oxidized coating (−75 V) containing an Au marker. (b) X-ray line scans, as indicated in (a), show profiles of the elements

4 Discussion

4.1 Oxidation Kinetics and the Temperature for the Onset of Oxidation

There was no evidence that the oxidation kinetics or mechanism of the TiAlN/VN was significantly modified by the substrate bias voltage during deposition, even though the bias voltage did appreciably change the crystallographic texture and residual stresses (Table I). Thus, the observations of oxidation mechanism from the bias voltages of −75 and −85 V can be combined.

The TGA traces of TiN and TiAlN coatings were believed to be, to a good approximation, representative of the mass gain of the coatings, because there was no significant diffusion between TiAlN coatings and stainless steel substrate was expected up to 600 °C. However, minor Fe has been found at columnar grain boundaries of TiAlCrN at 700 °C resulting from the diffusion of Fe from the substrate.[39] In the case with TiAlN/VN, the mass gain in TGA probably included a contribution from the steel substrate at 600 °C, because V2O5 forms eutectics FeVO4 with Fe2O3 at 635 °C and 840 °C corresponding to phase diagram V2O5/Fe2O3.[40] However, the TEM examination presented here indicated no significant ingress of Fe into the coating at the temperatures studied here. The other onset of rapid oxidation at 840 °C was possibly associated with oxidation of substrate, but at that temperature, the coating had completely failed and was therefore of no interest.

The onset of rapid oxidation of the TiAlN/VN coating was around 635 °C (Figure 2(a)), which compares to ∼700 °C for TiN and ∼900 °C for TiAlN. Thus, as expected, the addition of VN to TiAlN reduced the oxidation resistance. However, as noted earlier, the product oxides are known to be beneficial during sliding contact,[14] and therefore, this should not necessarily be regarded as a disadvantage for tribological applications. Although no data were obtained for VN in the current study, Borgianni et al.[11] have shown that VN oxidizes at just below 400 °C in flowing oxygen, clearly indicating that the oxidation kinetics of TiAlN/VN was dictated by a combination of the TiAlN and VN, not just the VN.

The oxidation of the TiAlN/VN appeared to have parabolic stages at both 550 °C and 600 °C (Figure 2(b)), although the mass gain after 0.7 hours decreased abruptly to a much lower rate. Note that this transition was not believed to be associated with exposure of the substrate, because TEM sections through the coating indicated that oxidation was wholly contained within the coating well after this change in kinetics. The kinetics could not be adequately evaluated at 638 °C and 670 °C because of rapid failure of the coating through oxidation. The oxidation kinetics of TiAlN have been found to be parabolic (e.g., Reference 7) also with a change in oxidation kinetics to lower growth rates after a certain incubation time, which was believed to be due to the formation of a passive aluminum oxide layer. In contrast, the isothermal oxidation of VN at 640 °C has been found to follow linear kinetics,[11] rather than the parabolic kinetics observed here, indicating that although the addition of VN to TiAlN had significantly adjusted the oxidation kinetics, it had not prevented the formation of a passivating oxide layer.

4.2 Oxide/Coating Interface

The interface between the oxide and the substrate was abrupt (Figures 12 and 17). The oxide front did not follow the multilayer structure, as shown by the orientation of the Fresnel fringes (which delineate the position of the individual layers), which could be imaged right up to the oxide/coating interface.

The oxide adjacent to the coating contained extensive porosity, with a near continuous layer at the interface itself (confirmed by the presence of Fresnel fringes around pores and by the abrupt changes in contrast in the thickness map in Figure 13). The rounded shape of the pores suggests they did not originate from microcracks arising from stresses induced by changes in the molecular volume on oxidation. The porosity is consistent with the reaction of the nitride coating with oxygen to liberate nitrogen gas. Point EEL spectra were obtained from this region (Figure 15). The shape of the N K edge was appreciably different to that found in the coating, with a single peak found at the interface, compared to a double peak in the coating. Moreover, the position of the first peak for the interface K edge was shifted by ∼1.4 eV with respect to the first peak for the K edge for the spectrum from the coating. The interface N K edge can be interpreted in two ways. First, the peak shape and first peak intensity fits that of V2N, as reported by Hofer et al.[41] However, they observed a shift of ∼1 eV between the first peak for VN and V2N, compared to the ∼1.4 eV observed in the present work. Alternatively, the N K edge could be interpreted as nitrogen gas, as shown by Trasobares et al.,[42] who examined nitrogen gas inside a carbon nanotube and observed a single peak N K edge with the peak located at 401 eV. In the present work, the peak shape and the increased intensity at the interface in the V L2,3 map (Figure 13) are more consistent with V2N, while the peak position is more appropriate to nitrogen gas. The porosity at the interface was most probably a result of the liberation of nitrogen gas by oxidation of the coating (as discussed later). Moreover, the porosity was typically 5 to 20 nm in diameter and could have been contained within the TEM sample, which was probably of the order of 30-nm thickness in this region. However, while the results do not uniquely identify the presence of nitrogen gas or V2N, it is clear that the bonding of the nitrogen in this region is completely different than that in the coating.

The Fresnel contrast present in the coating right up to the interface (Figures 12 and 17) demonstrated that there was no preferential oxidation of VN in preference to the TiAlN. In this respect, the TiAlN did not provide any “protection” to the VN layers. Moreover, although the TiAlN was present as a discrete phase, the onset of oxidation of this phase had been reduced by >100 °C from the ∼750 °C observed in monolithic TiAl(Cr)N coatings (Figure 2). This is probably because of the elemental intermixing between TiAl and V in the multilayers, such that all TiAlN layers contained appreciable quantities of V within them.[26] Equally, the onset of oxidation of the TiAlN/VN occurred at a much higher temperature than for VN (∼400 °C in flowing oxygen), suggesting that the Ti and Al content of the VN layers was also significant in raising the oxidation resistance.

As shown in Figures 12 and 17, the oxide grew preferentially along columnar grain boundaries. However, the EFTEM maps and associated point EELS analyses showed that the oxygen had not penetrated far into the coating. The structure of columnar grain boundaries in PVD coatings is a strong function of the deposition conditions. Boundaries can exhibit voiding, the volume fraction of which increases with increasing heat-treatment temperature.[43] However, in the current study, most of the columnar boundaries were sufficiently dense and the composition homogeneous, such that they did not appreciably affect the oxidation kinetics.

4.3 Reactions/Stabilities of Phases

At 550 °C, the oxide had started to form a duplex structure, but much of the surface contained a relatively uniform mixture of oxides (Figures 5(b) and 6). However, at 600 °C, the oxide had become sharply duplex, with an abrupt interface between the outer and inner oxide layers. However, at 638 °C, the oxide had become even more complex, with a third layer of V2O5 covering the same duplex layer observed at 600 °C. Figure 19 gives a schematic of the oxide structure at 600 °C and 638 °C.
Fig. 19

Schematic of the oxide structure illustrating the diffusion mechanism

At all three temperatures studied, the inner oxide layer was porous, comprising TiO2 and a phase most likely to be Al2TiO5, although not uniquely identified. Although dominated by Ti- and Al-based oxides, the inner layer did contain some V-based oxides, believed to be VO2 and V2O5, although the majority of the V was found in the outer oxide. The observation of VO2 and V2O5 is consistent with the work of Lugscheider et al.,[44] who found these two phases in coatings deposited at 603 °C. The presence of other oxides and an amorphous component could not be ruled out. The EEL spectra suggested that some nitrogen was retained 3 nm into the oxide, which could suggest the presence of an oxynitride (which was distinct from the structure observed at the interface (Figure 15)), but this had disappeared a few nanometers further into the oxide, suggesting that nitrogen transport in the oxide was not important, consistent with studies on the oxidation of TiAlN.[6] Thus, the principal initial oxidation reactions can be summarized:
$$ \begin{aligned}{} & 4{\text{TiAlN}} + 7{\text{O}}_{2} \to 2{\text{Al}}_{2} {\text{TiO}}_{5} + 2{\text{TiO}}_{2} + 2{\text{N}}_{2} \\ & 4{\text{VN }} + {\text{ }}5{\text{O}}_{2} \to 2{\text{V}}_{2} {\text{O}}_{5} + 2{\text{N}}_{2} \\ & 2{\text{VN}} + 2{\text{O}}_{2} \to 2{\text{VO}}_{2} + {\text{N}}_{2} \\ \end{aligned} $$
Other possible reactions may have occurred initially, but were followed by rapid transformation a few nanometers into the oxide, namely,
$$ \begin{aligned}{} & {\text{VN}} + {\text{O}}_{2} \to {\text{V}}({\text{O}}_{x} {\text{N}}_{y} ) \\ & 2{\text{VN}} + {\text{O}}_{2} \to 2{\text{VO}} + {\text{N}}_{2} \\ \end{aligned} $$

The increase in oxygen potential as the oxide grew would be expected to transform any VO or VO2 in the series VO → VO2 → V2O5.

The phase constitution in the outer layer formed at 600 °C, and the middle layer with the same morphology formed at 638 °C comprised V2O5, TiO2, probably Al2TiO5, and AlVO4. The diffraction pattern reflections believed to be VO2 had disappeared. However, although the layer contained crystals approximately an order of magnitude larger than in the inner layer, some oxides could not be fully identified. The presence of AlVO4 suggests one possible reaction:
$$ {\text{Al}}_{2} {\text{TiO}}_{5} + {\text{V}}_{2} {\text{O}}_{5} \to 2{\text{AlVO}}_{4} + {\text{TiO}}_{2} $$
At 638 °C, much of the outer surface was found to comprise V2O5, with the layer below comprising some V2O5, Al2TiO5, and AlVO4 with no evidence of γ-Al2O3 (α-Al2O3 is only stable at much higher temperatures). The formation of V2O5 on the outer layer is perhaps not surprising given the combined strong thermodynamic driving force for formation (Table III) and the low melting temperature and therefore high vapor pressure. Although the diffusion rate of V in TiO2 or Al2TiO5 is not available, V5+ has a much smaller ionic radius than Al3+ or Ti4+ (Table III), which should allow more rapid diffusion of the V5+ species to the surface than the Al and Ti. The XRD also demonstrated that the change in temperature from 600 °C to 668 °C substantially changed the texture of the V2O5, with the (001) texture found at the lower temperatures replaced by a (011) texture, which was strongest at 648 °C. The strong (011) texture was believed to be associated with the formation of the large crystals that covered much of the outer oxide scale (e.g., Figure 5(e)).
Table III

Ionic Radii, Thermal Data, and Molecular Volume of Al2O3, TiO2, and Vanadium Oxides[45,46]








Gold–Schmidt Ionic Radius (nm)

Al3+: 0.057






ΔG1000, kJ, 727 °C







Melting point, °C







Molecular volume, cm3

25.6 (AlN:12.8)

18.8 (TiN: 11.8)

10.5/11.4 (VN:10.7)

29.2 (VN:10.7)

18 (VN:10.7)

54.0 (VN:10.7)








The Au marker studies were used to give some indication of the relative diffusion rates of the cations and anions, given the absence of reliable diffusion data. A comparison of the oxide with (Figure 18) and without the Au marker indicated that the marker had not altered the morphology and thickness of the oxide. The marker for the 600 °C, 0.5-hour sample was located at the interface between the inner and oxide scales. This suggests roughly equal fluxes of the outward diffusion of the cations to form the outer layer and the inward diffusion of oxygen. However, the marker study does not allow any further comment on the relative diffusivities of the individual cations.

Although it might be convenient to compare the duplex oxide structure observed here with the duplex oxide structure found for TiAlN, there inevitable differences since the TiAlN/VN oxides could not be studied above around 650 °C, while the oxides on TiAlN cannot be studied <750 °C, since oxidation is so slow. Moreover, the diffusion rate of Al is low below 700 °C, while in contrast, oxygen diffuses rapidly inward through the growing oxide film and forms both Al and Ti oxides at the oxide/nitride interface.[7] In the current study, the inner layer was TiO2, V2O5, probably Al2TiO5, and VO2, and the outer layer was Al2TiO5, V2O5, AlVO4, and TiO2. In TiAlN, a duplex oxide scale is found for temperatures of 750 °C to 900 °C,[6] with an outer layer of Al-rich oxide and an inner layer of Ti-rich oxide.[6,7] Reduced oxidation kinetics after a temperature-dependent incubation period was associated with the passivating nature of this outer Al-rich oxide layer. However, at 900 °C, the formation of TiO2 was positively identified, which grew at an accelerated rate through cracks in the oxide layer. Interestingly, for <850 °C, the diffusion of Al and O was found to be relatively equal and noninteractive, suggesting different primary diffusion paths for the two species.

One of the drivers behind this work was to explore whether static oxidation yields the lubricious oxides of V2O5 or related Magnéli phases. While this was indeed found to be the case, the V2O5 was only dominant over a small temperature range. At 600 °C, it was only one of several phases present at the surface, with a proportion of the V being lost to AlVO4. However, a small increase in temperature to 638 °C led to a substantial change with much of the surface being covered in V2O5 crystals. However, by 674 °C, the V2O5 will have melted, although such liquids can also provide low friction. In any event, the current work confirms the rather narrow temperature range in which frictional heating would produce potentially low friction Magnéli phases at the surface.

5 Conclusions

  1. 1.

    TiAlN/VN multilayer coatings exhibited more rapid oxidation kinetics compared to TiAlN due to prevention of a passive surface oxide layer by the presence of V.

  2. 2.

    Oxidation of the TiAlN/VN coating initiated at ≥550 °C with the VN the first to oxidize. Kinetics were logarithmic initially, parabolic after 0.4 hours, and linear after ∼0.7 hours with a much slower mass gain. The behavior of the 600 °C sample was similar, except the parabolic rate was more rapid, and the oxidation rate >0.7 hours was slightly greater (0.08 g/m2 h) compared to the 550 °C sample (0.06 g/m2 h).

  3. 3.

    Cross sections by STEM/EDX and marker study revealed that the total cation outward diffusion was approximately equal to the oxygen inward diffusion. However, the diffusion rate of the V cations was clearly the most rapid as it always formed the dominant outer oxide.

  4. 4.

    At 550 °C, the oxide exhibited a rudimentary duplex oxide structure, with the outermost surface comprised of needle-/plate-shaped crystals, but the inner layer was porous and nanocrystalline. Positive phase identification could not be made, but elemental maps suggested the outermost surface was V rich while the region adjacent to the coating was dominated by Ti and Al. Phase identification of the oxide was difficult owing to the small dimensions of the particles.

  5. 5.

    At temperatures >600 °C, a duplex oxide structure was formed; the inner layer comprised a porous region of Ti-rich and V-rich nanocrystallites, while several phases were observed in the outer region, including V2O5, TiO2, and AlVO4. V2O5 was the dominant oxide at the outer layer at ≥638 °C. The outward diffusion of V depended on the species present; in the inner layer, V was present as V3+, V4+, whereas a significant V5+ was dominant in the outer layer of oxide at ≥638 °C.


PHILIPS is a trademark of Philips Electronic Instruments Corp. Mahwah, NJ.


Gatan is a trademark of Gatan Inc., Pleasanton, CA.


JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.



We are grateful to EPSRC for funding through Grant No. GR/N23998/01.

Copyright information