Nanostructuring and coherency strain in multicomponent hard coatings Open Access

Cubic (c)-TiAlN is one of the most commonly used coating material systems for metal cutting tools in high abrasive wear applications. To increase the productivity and to maintain a high precision during the cutting operation, it is necessary for protective coatings to provide a high hardness at high temperatures around 1000 °C.^1,2 Upon exposure to elevated temperatures, c-TiAlN coatings undergo iso-structural spinodal decomposition into coherent Ti- and Al-enriched cubic phases.^3,4 This decomposition process is beneficial since the generated coherency strain in combination with the variation in elastic properties⁵ between the phases increases the hardness. However, since c-AlN is unstable, the Al-enriched phases eventually transform into their stable hexagonal structure. When the hexagonal transformation occurs, the coherency strain is relaxed and the hardness decreases significantly.⁶ During the last 10 years, numerous studies and concepts have been applied attempting to understand and to enhance different mechanisms responsible for stabilizing the metastable cubic structure of AlN.^3,4,7–15 Some explicit examples are multilayer nanostructures where the cubic structure may be stabilized by strain or by altering the diffusion during the coarsening.^16–19 Other examples are through addition of different alloying elements altering the mixing enthalpy and driving force for spinodal decomposition and the hexagonal transformation.^20–26 A common focus of these investigations and efforts is to delay the detrimental transformation of c-AlN to hexagonal (h)-AlN to higher temperatures. This delay is considered essential for the formulation of new and improved AlN-based functional hard coatings.

Recent investigations have shown that quaternary TiCrAlN coatings exhibit higher hardness^25–27 and improved wear behavior^20,28 at elevated temperatures compared to the TiAlN system. When adding Cr to c-TiAlN, the driving force for spinodal decomposition into c-TiN and c-AlN is reduced.²⁵ It results in a spinodal decomposition at higher temperatures which consequently also pushes the subsequent formation of h-AlN to higher temperatures. In order to fully understand the effect of multicomponent alloying on the decomposition process, the local structure and metallic concentrations have to be determined on an atomic level. Here, we utilize the combination of a double aberration corrected FEI Titan³ 60-300 scanning transmission electron microscope ((S)TEM) equipped with a Super-X energy dispersive x-ray spectrometer (EDX) together with a LEAP atom probe tomography microscope (APT). The combination allows for quantitative determination of the metallic concentrations and the lattice structure of the nanometer-sized phases and their interfaces formed during the decomposition process. The results provide a detailed understanding of how the thermal stability of c-AlN is affected when alloying TiAlN with Cr.

We present results from the investigation of the Ti_0.31Cr_0.17Al_0.52N coating exhibiting almost 20% higher hardness after annealing at 1000 °C and 1100 °C compared to Ti_0.33Al_0.67N.²⁶ Detailed descriptions regarding the coating deposition parameters, (S)TEM sample preparation, hardness, and x-ray diffractograms can be found in Ref. 26. In short, the coating was deposited using a Sulzer Metaplas industrial scale cathodic arc evaporation system onto WC-Co substrates. The annealing was carried out in an inert atmosphere, and the temperature was kept constant for 2 h before cooling down to room temperature. The (S)TEM samples were prepared by mechanical grinding and ion milling to electron transparency ending with a low energy polish to minimize Ar-ion induced defects. To overcome the mass to charge ratio overlaps between the thermal tails of some isotopes of Ti and Cr in the APT tip reconstructions, the shank angles of the tips were optimized to minimize the thermal tails in the charge to mass ratio spectra. This allows for a precise quantification based on the natural abundance of the isotopes (Ti/Cr ratio error <2.5%).

Figures 1(a) and 1(b) show an overview of Z-contrast micrograph of the Ti_0.31Cr_0.17Al_0.52N coating annealed at 1100 °C and the corresponding color coded EDX map (Ti-red, Cr-blue, and Al-green). According to the map, the darker phases in the Z-contrast image are Al-rich (green), while the brighter are TiCr-rich (pink). The map further identifies an Al rich region in the middle of the micrograph extending in the growth direction which is identified as a grain boundary (partly damaged by sample preparation). In addition, an ∼2 nm thick Cr-enriched phase (blue) is separating the TiCr- (pink) and the Al-rich phase (green) at the interfaces inside the grain.

Figure 1(c) shows color coded APT reconstructions of tips from the coating annealed at 1000 °C and 1100 °C. Reconstruction parameters and the tip shapes were optimized to minimize artifacts that may arise from anomalies in the evaporation field at the interfaces by qualitatively verifying the shapes and the dimensions of the phases with the (S)TEM results. The tips shown here contain the same microstructure as identified by the EDX map. The domains are larger at 1100 °C compared to 1000 °C due to the more evolved coarsening.

Figure 2(a) shows an EDX line scan obtained from the coating annealed at 1100 °C. The line scan indicates two Al-enriched phases from 5 to 20 nm and 50 to 70 nm along the axis, where the phase in the beginning of the scan contains little or no Cr, while the one at the end of the scan has a notable Cr-content. From the atomically resolved (S)TEM micrographs of these phases (see Figure 3), it is found that the phases containing Cr are cubic while the phase without Cr is hexagonal. This result is the key to understand the improved thermal stability of TiCrAlN coatings over TiAlN coatings. Through the formation of the intermediate decomposition product, c-CrAlN, the Cr atoms stabilize the cubic structure of AlN.²⁵ Figure 2(b) shows line scans obtained at 1000 °C and 1100 °C in the reconstructed tip data depicting the same phases as identified with EDX. The aforementioned Cr-enriched phase (blue) at the interfaces of the TiCr and Al-enriched phases is also seen in the APT results between the dashed lines where the Ti-concentration gradient is sharper than the Cr gradient resulting in a Cr-enriched phase and a more blue color at the phase interfaces in the EDX maps. The metallic concentration of each phase was determined by constructing proximity histograms across 50% Al iso-surfaces utilizing the entire tip reconstruction. The obtained local concentrations are presented in Figure 2(c). At 1000 °C, the cubic Al-enriched phase contains 14% Cr and at 1100 °C 9% Cr.

Figure 3(a) shows the atomically resolved Z-contrast micrograph of the two Cr-containing phases in the coating annealed at 1100 °C. The structure is entirely cubic with coherent interfaces between the TiCr- and the Al-enriched phases, which means that the lattice is strained resulting in a significant hardness enhancement.^3,9,25,26 The stability of the c-AlN phase is not only affected by the Cr-concentration but also by the coherency strain since transformation to h-AlN yields a volume increase.^10,29 It is seen that in the temperature range of 1000 °C and 1100 °C, the size of the c-AlN domains grows to 20-30 nm while the structure is cubic containing 9% Cr- and 2% Ti-content. This is a much larger size of metastable c-AlN than what has been reported in TiAlN coatings (<5 nm).^6,29 This means that the Cr-concentration of 9% (effectively lowering the lattice mismatch between the TiCr- and the Al-rich domains by 1%)²⁵ plays a key role for the stability of the cubic structure.

Figure 3(b) shows the atomically resolved micrograph of the interface between the cubic TiCr-enriched phase and the hexagonal Al-enriched phase. The cubic and the closed packed hexagonal interfaces are semi-coherent with approximately one atomic layer thick concentration gradient at the interface. The semi-coherent crystallographic relationship between the cubic and the hexagonal phases is revealed when comparing the FFT patterns shown in the insets in Figure 3(b), i.e., c-TiCrN 〈110〉 ∥ h-AlN 〈100〉 and c-TiCrN 〈111〉 ∥ h-AlN 〈001〉. The crystallographic relationship is also given as a schematic sketch in Figure 3(c). The lattice mismatch along the interface in the image plane (c-TiCrN $〈 1 ̄ 01〉∥h$-AlN $〈0 1 ̄ 0〉)$ and perpendicular to the image plane (c-TiCrN 〈110〉 ∥ h-AlN 〈100〉) is 4.1% (measured) or 4.4% (calculated by assuming a linear combination of the equilibrium lattice parameters of 4.1 Å for c-AlN (theoretical³⁰), 4.24 Å for c-TiN, and 4.14 Å for c-CrN with the phase compositions taken from the APT results). As suggested for Ti_0.5Al_0.5N,^31,32 semi-coherency between cubic and hexagonal phases in the as deposited state results in strain beneficial for the hardness.³³ However, normally, it is not considered beneficial to have hexagonal AlN in the as deposited state of hard coatings since the degradation of the hardness is accelerated when the hexagonal phases coarsen upon exposure to elevated temperatures. The microstructure behind the improved thermal stability and hardness for the TiCrAlN coating investigated here relies on the delayed spinodal decomposition and the improved stability of the cubic structure as compared to TiAlN.²⁵ The hexagonal structure in these coatings is thus formed at a much higher temperatures, and due to the semi-coherency, the strain does not completely relax. This explains why the hardness of the Ti_0.31Cr_0.17Al_0.52N (Ref. 26) coating only decreases by ∼10% at the studied annealing temperature of 1100 °C, where the phases are semi-coherent. These results highlight the importance of considering the coherency strain even for the hexagonal phase which is important to consider in future strategies for theoretical studies and experimental design of hard coatings.

This work was supported by the SSF-project Designed Multicomponent Coatings, MultiFilms, and VINNOVA Strategic Faculty Grant VINNMER. The Knut and Alice Wallenbergs Stiftelse was acknowledged for the electron microscopy laboratory in Linköping and Dr. M. P. Johansson-Jõesaar at Seco Tools AB is acknowledged for the support regarding the depositions.