- Email: [email protected]
Microstructure investigation of textured CVD alumina coatings
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
Contents lists available at ScienceDirect
International Journal of Refractory Metals & Hard Materials journal homepage: www.elsevier.com/locate/IJRMHM
Microstructure investigation of textured CVD alumina coatings ⁎
S. Shojaa,b, , N. Mortazavia,c, E. Lindahlb, S. Norgrenb, O. Bäckea, M. Halvarssona a b c
Department of Physics, Chalmers University of Technology, Gothenburg, Sweden Sandvik Coromant, Stockholm, Sweden School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: Coating Alpha alumina CVD Texture development H2S TKD
This study concerns the interfacial microstructure and texture development in three textured CVD α-Al2O3 coatings using X-ray diffraction, transmission Kikuchi diffraction, scanning transmission electron microscopy and energy dispersive X-ray analysis. It is well known that the performance of these types of coatings relies on the degree and type of texture. The aim of this work is to study the microstructure of three different CVD α-Al2O3 layers when deposited onto a Ti(C,N,O) bonding layer. The coatings were deposited onto cemented carbide/ cobalt substrates (WC/Co). It was observed that grain boundary diffusion of W and Co occurred through the TiN and Ti(C,N) layers to the bonding layer/α-Al2O3 interface. This may disturb the alumina layer nucleation and early growth. Interfacial porosity was observed at the bonding layer/α-Al2O3 interface. The number of voids that were detected in the (0001) and (011¯0) -textured coatings was similar when H2S was not used during the initial deposition step. When H2S was present during the nucleation step deposition of the α-Al2O3 more voids were introduced at the interface for the (0001)-textured samples. The alumina grain morphology developed from small (~100 nm) equiaxed grains at the start of the alumina coating to larger (several microns) columnar grains at the top of the coating. The inner part of the alumina coatings had a more random orientation. The texture changed either: (i) gradually over several grains, or (ii) more abruptly from one grain to another; from more random orientations to the desired texture. The desired texture develops earlier when using H2S at the start for the (0001)-textured coatings, while the (011¯0) -texture development benefits from the absence of H2S. Thus, in this study, H2S promotes (0001) texture and interfacial void formation.
1. Introduction
The properties of α-Al2O3 coatings using CVD processes have been significantly improved through optimizing bonding layers for phase control as well as performing microstructural refinements in the coating layers, e.g. via lowering the fraction of porosities at interfaces, see for example [10–17]. As a rather new development step, highly textured alumina coatings [14] have attracted industrial and scientific attention as a solution for further improvement of the CVD Al2O3 coatings, resulting in increased cutting tool life, as well as higher metal removal rates. During the last decade, the properties of the different texture components of α-Al2O3 coatings have been widely characterized [14,18–20]. In this regard, Ruppi [18] and M'Saoubi et al. [20] studied wear properties of CVD α-Al2O3 layers with different growth textures and reported significant texture effects on wear performance of the αAl2O3 layers. Suggesting plastic deformation as the main degradation mechanism, they reported that the best wear resistance is achieved for (0001)-textured coating. Ruppi et al. [18,19] examined the ability to
The technology of machining of metallic materials demands cutting tools with longer lifetime and better performance, striving to accomplish the optimal combination of productivity and tool life. Cutting tools must withstand rigorous operating conditions as they are subjected to high temperatures, large temperature gradients, thermal chocks, fatigue, abrasion, attrition and diffusion wear [1–3]. It is well known that alumina (Al2O3) is suitable to be used in a coating system on metal cutting tools owing to its chemical inertness and high hardness at elevated temperatures, i.e. 650 °C to 1000 °C, which is the typical temperature range in the metal cutting applications [4–6]. The Al2O3 coatings are usually applied to the tools using chemical vapor deposition (CVD). The CVD technique has been used for many years for industrial deposition of thick and uniform Al2O3 coatings on cemented carbide (WC/Co) substrates, first as κ-Al2O3 and later also as α-Al2O3 [7–9].
⁎
Corresponding author at: Department of Physics, Chalmers University of Technology, Gothenburg, Sweden. E-mail address: [email protected] (S. Shoja).
https://doi.org/10.1016/j.ijrmhm.2019.105125 Received 5 July 2019; Received in revised form 29 September 2019; Accepted 3 October 2019 0263-4368/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: S. Shoja, et al., International Journal of Refractory Metals & Hard Materials, https://doi.org/10.1016/j.ijrmhm.2019.105125
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
undergo uniform plastic deformation of α-Al2O3 coatings with a number of textures, namely (101¯2) , (101¯4), (0001) and (101¯0). They reported that the (0001)-textured coating exhibits better wear performance as compared to the other textured coatings. These conclusions were also confirmed by M'Saoubi et al. [21] in their work. They used controlled machining tests to investigate the wear properties of the αAl2O3 for three textures, (0001), (011¯2) and (101¯0) and showed that the (0001) textured alumina is the preferred one. However, a fundamental description of texture development during growth and the effect of catalyzing gases on the coating's texture were not discussed. This study focuses on grain development as a function of height during texture evolution in differently textured CVD α-Al2O3 coatings. The microstructural differences of the different CVD α-alumina coatings were investigated by means of Scanning Transmission Electron Microscopy (STEM). Micro-textural analyses were also performed through acquiring grain orientation mapping using Transmission Kikuchi Diffraction (TKD). Thus, the aim of the present investigation is to further elucidate the microstructural differences in CVD alumina coatings with different textures.
layers was analyzed in order to understand the coating properties and also to gain insight about the reasons for the formation of interfacial defects and their influence on alumina nucleation and early growth. The texture of the upper and lower parts of the α-Al2O3 coatings was analyzed using pole figure analysis and inverse pole figure (IPF) mapping based on TKD data acquired from thin foil samples. Details of specimen preparation, and the imaging and analytical techniques are described below and the results for the different samples are presented and discussed in Section 3. 2.2. Focused ion beam/scanning electron microscopy (FIB/SEM) An FEI Versa 3D combined Focused Ion Beam/Scanning Electron Microscope (FIB/SEM) workstation was used to produce cross-section thin foils from the coatings and subjacent substrate. The thin foils were later used for STEM/EDX analysis in a TEM/STEM and for TKD in a scanning electron microscope (SEM), see below. In order to protect the surface from ion beam damage, a thin Pt-layer with a thickness of around 100 nm was first deposited using electrons onto the surface (25 μm long and 5 μm wide) and then a thicker Pt layer (~2 μm) was deposited on top with the assistance of Ga-ions. Low accelerating voltages, i.e. 5 kV and 2 kV with ion currents of 49 pA and 27 pA, respectively, were used in the final thinning process to minimize the ion beam induced damage to the milled surfaces.
2. Materials and methods 2.1. Materials The studied coatings were comprised of several sub-layers obtained from different processing steps. All coatings were deposited on a WC/ Co substrate using thermally activated CVD. They consisted of an inner TiN layer, ~0.3 μm thick deposited at 930 °C, followed by a medium temperature CVD Ti(C,N) layer of ~1 μm thick, deposited at 885 °C. Then a bonding layer (Ti(C,N,O)) with the thickness of ~0.5 μm was deposited at 1000 °C followed by an α-Al2O3 layer of ~5 μm which was deposited at 1000 °C with two different textures, (0001) and (011¯0) . We investigated the gradual texture development of the alumina layers in three different samples. Two of the specimens had identical texture, (0001), but different deposition starting conditions of with and without the catalyzing gas H2S. This gas is generally used in industrial production of the coatings to enhance the alumina growth rate and to produce coatings with even thicknesses. Thus, these two samples were used for comparing the initial alumina layer growth, and the influence of the catalyzing gas (H2S) on the microstructure at the bonding layer/ alumina interface. These coatings are referred to as Sample (001) and Sample (001)+, where the “+” denotes that the H2S was added during the whole process, while the (001) sample was produced without H2S in the initial step which is the standard case in industry. This initial step corresponds to around 100 nm in thickness. The third sample, referred to as Sample (010) in this paper, had a (011¯0) texture and was produced without H2S in the initial step. The individual layer thicknesses and presence of an initial step including the catalyzing gas H2S are summarized in Table 1 for the three coatings. The texture and microstructure of the different coatings were investigated using STEM, Energy Dispersive X-ray analysis (EDX) and TKD. Special attention was paid to the fraction of voids and defects at interphases. The chemical composition of the underlying Ti-based
2.3. X-Ray diffraction (XRD) The X-Ray Diffraction (XRD) measurements were done with a Bruker D8 Advance instrument using CuKα radiation. The XRD was used with a θ-2θ setup. Using the XRD measurements, textures for the alumina layers were evaluated using texture coefficients, TCs, which were calculated using Eq. (1), see below. The equation uses the intensity ratios of diffraction peaks (hkl) from possibly textured coatings over non-textured coatings, in order to calculate the degree of texture.
TC (hkl) =
Sample (001)
Sample (001)+
Sample (010)
5.2 No 0.5 1.2 0.3
5.3 Yes 0.5 1.2 0.33
6.5 No 0.65 1.5 0.45
I (hkl) ⎫ I0 (hkl) ⎬ ⎭
(1)
2.4. Transmission kikuchi diffraction (TKD) Grain orientation mapping of the alumina layers was performed by means of TKD. This enabled us to carry out high resolution microstructural (grain size and shape) and micro-textural (preferred orientation of α-alumina grains) analyses on samples. The TKD analyses with high indexing rates were performed using a special thin foil holder, designed earlier, see [22,23]. The analyses were performed using an HKL Channel 5 Electron Backscattered Diffraction (EBSD) system with a Nordlys II detector, mounted on a Zeiss Ultra Field Emission Gun (FEG) SEM. The TKD data were collected from the FIB/ SEM-prepared thin foils. An accelerating voltage of 30 kV in high current mode and an aperture size of 120 μm were used. The analysis grid size for most of the cases was 160 × 100 with a step size of 10–20 nm and SEM magnifications of around 35,000 X.
Coating layer (μm) Al2O3 Initial catalyzing step Bonding layer Ti(C,N) TiN
−1
∑
I(hkl) is the XRD peak intensity of the textured coating, I0(hkl) is the non-textured XRD peak intensity from a reference material, and n is the number of peaks considered. The TC was calculated based on eight reflections (011¯2) , (101¯4) , (112¯0) , (112¯3) , (022¯4) , (112¯6) , (033¯0) and (0 0 0 12), which makes the maximum value of the texture coefficient equal to 8. This value corresponds to a perfectly oriented material. In contrast, the value of 1 corresponds to a randomly oriented sample.
Table 1 Measured coating layer thicknesses for the three samples, including an indication if there was an initial catalyzing step with H2S or not. Sample ID
I (hkl) ⎧ 1 I0 (hkl) ⎨ ⎩n
2.5. Scanning transmission electron microscopy/energy dispersive X-ray analysis (STEM/EDX) The thin foils were also investigated by STEM in an FEI Titan 2
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 1. SEM secondary electron micrographs showing the surface morphologies of the α-alumina coatings. (a) and (b) The basal plane (0001) is parallel to the surface. (c) The prismatic plane (011¯0) is parallel to the surface.
Fig. 2. (a) STEM BF micrograph showing a cross-section overview of the coating microstructure for Sample (001). There are small and equiaxed grains at the bottom of α-alumina layer. The shape and size of the grains develop to larger, columnar grains from the bottom to the top. STEM (b) BF and (c) HAADF images showing some of the porosities (arrowed) at the interface of α-Al2O3 and the bonding layer in higher magnification. Fig. 3. (a) STEM BF image showing the cross-section overview of Sample (001)+; H2S was used during the whole alumina deposition. The coating exhibits small and equiaxed grains at the bottom of the αalumina layer. The grains develop to larger and columnar grains at the outer parts. STEM (b) BF and (c) HAADF images showing the interfacial porosity (circled) at the α-Al2O3/bonding layer at higher magnification.
general microstructure, such as grain morphology and local composition of differently grown layers, and also to characterize the processinduced pores in the multilayer coating.
80–300 TEM/STEM equipped with a FEG operated at an accelerating voltage of 300 kV. The microscope is equipped with an Oxford Inca EDX detector. STEM micrographs were acquired using high angle annular dark field (HAADF) and bright field (BF) detectors in combination with EDX analysis. The STEM analyses were conducted to examine the 3
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 4. (a) STEM BF micrograph showing the cross-section overview of Sample (010); H2S was not used during the initial alumina deposition. The coating exhibits small and equiaxed grains at the bottom of the α-alumina layer, while there are larger and columnar grains at the outer parts. STEM (b) BF and (c) HAADF images showing the needle shaped morphology of the bonding layer and some porosity (circled) at the interface of α-Al2O3 and the bonding layer.
Fig. 5. (a). STEM HAADF image at the region of the coating/WC-Co substrate interface showing the migration of heavy elements (white contrast) from the substrate to the coating along grains boundaries. (b) STEM HAADF micrograph at the interface of alumina and the bonding layer showing that heavy elements have migrated all the way to this interface. (c) STEM/EDX maps from the boxed region in (b) showing the presence of W and Co in this region. Table 2 Calculated texture coefficients for eight reflections show that all three samples are highly textured. Sample (001) and Sample (001)+ have high texture coefficients for (0 0 0 12), which corresponds to an (0001) texture and Sample (010) has a high texture coefficient for (033¯0) , which corresponds to an (011¯0) texture. TC
(0 1 1¯ 2)
(1 0 1¯ 4)
(1 1 2¯ 0)
(1 1 2¯ 3)
(0 2 2¯ 4)
(1 1 2¯ 6)
(0 3 3¯ 0)
(0 0 0 12)
0.30 0.02 0.22
0.59 0.46 0.07
0.06 0.14 0.13
0.05 0.00 0.03
0.09 0.00 0.38
0.33 0.06 0.06
0.16 0.11 7.11
6.69 7.21 0.00
Sample ID Sample (001) Sample (001)+ Sample (010)
3. Results
are presented and discussed in this section. SEM micrographs of the surface of the coatings for all three α-Al2O3 layers are shown in Fig. 1. The surface morphology of the coatings exhibits pyramidal-shaped grains with the size of 1–2 μm. To get a better understanding of the coatings, cross-section thin-foils were prepared by FIB/SEM and investigated using STEM/EDX. Fig. 2a shows a STEM BF cross-section overview of the
3.1. General microstructure As mentioned earlier, different aspects of the coatings were investigated using the SEM, STEM, EDX and TKD techniques. The results for all three samples, Sample (001), Sample (001)+ and Sample (010),
4
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 6. (a). Band contrast (top) and corresponding IPF map (bottom) for Sample (001). There are highly aligned (red) (0001) alumina grains in the top part of the coating and more randomly oriented (blue/green) small grains at the bottom of the coating. 3D view (left) of the crystals are shown for some grains demonstrating an abrupt texture development (b). Calculated pole figures {0001} and {101¯0 } of the top and bottom parts of the coating. (TKD data). (c). Inverse pole figures for Sample (001) as a function of slice height. The texture gradually develops towards (0001). The grain orientations in slice 1 and 2 are relatively random, and start to be aligned along (0001) from slice 3. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
higher number of pores compared to Sample (001) can be seen right at the interface between the α-Al2O3 and the bonding layer. Fig. 3b and c show STEM BF and HAADF images of the voids at the interface in higher magnifications. The average number of pores per unit of length for this sample was measured to 2.41 ± 0.53 pore/μm. Fig. 4a shows the cross-section overview for Sample (010) by STEM BF imaging. The α-Al2O3 layer was in this case deposited without H2S during the initial step of deposition. The bonding layer has a needleshaped morphology, as can be seen in Fig. 4b. Some porosity is observed at the interface of α-Al2O3 and the bonding layer, which is shown in the STEM HAADF image in Fig. 4c. The average number of interfacial pores per unit of length were determined to 1.35 ± 0.53 pore/μm. The extent of pores is roughly half the amount for Sample (001)+, and almost the same for Sample (001). However, with a small sample size i.e. TEM thin-foils, caution must be applied as the statistics is limited for these measurements. Turning to the Ti-containing layers below the alumina layer, these layers are the same in all three samples and can be seen in Figs. 2–4. The innermost TiN layer is comprised of very small grains, in the order of 10–100 nm, especially closest to the WC/Co substrate. The Ti(C,N) layer has columnar, larger grains, while the bonding layer contains small, sometimes columnar grains. A closer inspection of the inner parts
microstructure of Sample (001), which was produced without introducing H2S at the initial step of alumina deposition. The micrograph provides information about the thickness, grain size, morphology and defects of different layers in the coating. The outer Pt-layer was deposited during specimen preparation in the FIB/SEM to protect the surface from ion beam damage. In Fig. 2a there are small and equiaxed grains with a size of about 100 nm visible at the start of the α-Al2O3 layer. The shape and size of the grains develop to columnar forms with the width of 1–2 μm and column lengths of several microns, almost equal to the layer thickness. Some interfacial pores, marked by arrows, can be observed at the interface of α-Al2O3 and the bonding layer. These porosities are shown in higher magnification in STEM BF and HAADF in Fig. 2b and c, respectively. The number of interfacial pores per unit of length was measured in a few TEM micrographs taken from different parts of the interface. The average number measured for this sample was 1.05 ± 0.21 pore/μm. A STEM cross-sectional overview micrograph of Sample (001)+, i.e. when H2S was introduced from the beginning of the alumina deposition, is shown in Fig. 3a. Almost the entire α-Al2O3 layer consists of large and columnar grains, several microns in height and widths of 1–2 μm. At the start of the α-Al2O3 layer, small and equiaxed grains with a grain size of around 100 nm is also observed in this coating. A 5
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 7. (a). Band contrast (top) and corresponding IPF map (bottom) for coating Sample (001)+. There are highly aligned grains towards (0001) at the top of the coating and more randomly aligned grains at the bottom of the coating. 3D view (left) of some crystals demonstrating a gradual texture development (b). Calculated pole figures {0001} and {101¯0 } of the top and bottom parts of the coating for Sample (001)+. (c). Inverse pole figures for sample (001)+ as a function of slice height. The texture gradually develops towards (0001). The grain orientation in slice 1 is relatively random, and start to be aligned along (0001) from slice 2.
measurements are summarized in Table 2. The highest TC for Sample (001) and (001)+ is around 7 and belongs to (0 0 0 12) reflection. This reflection represents the (0001) orientation. These high TC values mean that both Sample (001) and Sample (001)+ are highly (0001)-textured coatings. The highest TC value for Sample (010) is around 7 for the (011¯0) peak. Thus, the coating is highly (011¯0) -textured, as expected. Transmission Kikuchi Diffraction analysis on cross-section thin foils was used to understand the development of the textures as a function of height in the coatings. The results for Sample (001) are shown in Fig. 6. The band contrast and IPF maps (Fig. 6a) of the α-Al2O3 cross-section show large columnar grains in the top part of the coating with lattice plane normals (0001) along the growth direction, shown as red grains in the IPF map (the IPF color legend is also shown in Fig. 6a). Fig. 6b shows the {0001} and {101¯0 } pole figures extracted from the TKD data. The two pole figures on top show the results from the top part of the coating, above the dashed line in Fig. 6a. They have a high intensity in the center for {0001} and a ring at 90° (i.e. at the periphery) in {101¯0 }, which indicates that almost all the grains in the top part of the alumina layer have (0001) orientation. However, the pole figures for the bottom
of the Ti(C,N) and TiN layers reveals a network of bright contrast in the STEM HAADF micrographs (see Fig. 5a for Sample (001)). This implies the migration of heavy elements from the WC/Co substrate along the TiN and Ti(C,N) grain boundaries. These elements have diffused all the way up through the (Ti;C,N) and bonding layer to the alumina interface, as can be seen in Fig. 5b. STEM/EDX was used to analyze the elemental distribution in the top part of the Ti(C,N) layer, the bonding layer and the interface towards αAl2O3. EDX maps from the region indicated by a box in Fig. 5b are presented in Fig. 5c. W is present as vague elongated areas, probably along grain boundaries, while Co is present as small (~10 nm) round spheres. This shows that W and Co have migrated to this sensitive region where they may affect the nucleation and growth of the α-Al2O3 layer. It can also be observed from the EDX map that these elements probably have not diffused further into the α-Al2O3 layer. 3.2. Texture analysis The
calculated
texture
coefficients
(TCs)
from
the
XRD 6
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 8. IPF maps showing grains with the c-axis tilted 70° to 90°away from the surface normal. The number of grains are around 70% of all grains for Sample (001) and 35% for Sample (001)+.
orientation. The IPFs for horizontal subsets (Fig. 7c) show that the start/bottom slice has a relatively random texture. However, to a lesser extent compared to the same section for Sample (001). It can also be seen that the texture starts to be relatively orientated towards (0001) already from Section 2 to the top. To investigate the small grains at the bottom of the α-Al2O3 layers in more detail, the grains with their c-axis tilted 70° to 90° away from the surface normal, which are the grains with non-preferred orientations, were selected. Fig. 8 shows these grains for both Sample (001) and Sample (001)+. There are significant differences in the shape and number of selected grains for the two samples. In Sample (001), there are many grains, ~70% of the total, within this angular range, i.e. with the c-axis almost lying down. These grains are agglomerated and extended to 0.5 to 1 μm height. In Sample (001)+ these types of grains are fewer (~35%) with a height of only one grain. The texture analysis of Sample (010) obtained by TKD is shown in Fig. 9. The band contrast (Fig. 9a top) is in agreement with the microstructure of this sample observed by TEM, see Fig. 4. The IPF map (Fig. 9a bottom) shows that almost all the grains have either (101¯0) or (011¯0) orientations (green or blue); thus, Sample (010) is highly textured. This also agrees with the XRD and texture coefficient results as described above. According to the IPF map the majority of the grains nucleate and grow with the preferred orientation, i.e. they are not colored red. For the other (red) grains the growth orientation often changes abruptly from the bottom to the top. The three-dimensional view of three crystals, next to the IPF map in Fig. 9a, illustrates one example of an abrupt orientation change. The IPF map texture analysis results can be verified by viewing the {011¯0 } and {112¯0 } pole figures for the top and start/bottom part of the coating (Fig. 9b). The pole figures indicate that both top and bottom parts of the α-Al2O3 layer are highly textured and very close to ideal {011¯0 } texture, corresponding to intensity at the center and at a ring 60° away for {011¯0 }, and two rings at 30° and 90° for {112¯0 }. The IPFs for horizontal subsets (Fig. 9c) all exhibit strong intensities at the (101¯0) and (011¯0) corners, which refers to highly aligned grains from bottom to the top. Only some low intensity can be observed near the (0001) corner, which disappears gradually after the distance corresponding to the third slice.
part, (start), show that the intensities are not as concentrated as for the top part, which gives that fewer grains have (0001) orientations. Thus, the start/bottom part of the coating is not as textured as the top part. These pole figure results are in agreement with the IPF map grain orientations, where many small, equiaxed grains at the start/bottom of αAl2O3 layer have the “wrong” orientation, i.e. not (0001), and have blue or green color combinations instead of being red. Three-dimensional views of the crystal orientation for a column of grains are shown next to the IPF map in Fig. 6a. The inner grain has the c-axis nearly parallel to the interface (lying down), while the outer three have their c-axis (almost) parallel to the growth direction. This is an example of abrupt grain orientation change from non-preferred to the preferred orientation. However, most of the grains in Sample (001) show a gradual change in their orientations from start/bottom to the top in such a way that the c-axis is gradually aligned along the growth direction. To study texture development in more detail, seven horizontal subsets with heights of around 700 nm were created (see Fig. 6a) and IPFs for each subset were made (Fig. 6c). Thus, Fig. 6c shows the texture development as a function of height. It is apparent that the grains start with more random orientations, and gradually develop an (0001) texture. The IPFs support that the grains start to be relatively oriented towards (0001) from Section 3 and upward. Fig. 7 gives the texture analysis of the Sample (001)+ obtained by TKD. The band contrast of the α-Al2O3 layer (Fig. 7a top) shows the shape and size of the grains reflecting the small and equiaxed grains at the start/bottom close to the α-Al2O3/bonding interface and larger and columnar grains on top, as seen earlier in the STEM micrographs in Fig. 3. The IPF map (Fig. 7a bottom) for this sample shows that most of the grains have the preferred orientation of (0001), already from the start. However, there are a few grains with colors (blue/green) further away from (0001) corner, which means that they don't have the desired (0001) orientation. This can also be verified by viewing the {0001} and {101¯0 } pole figures (Fig. 7b top) for the top part of the coating, above the dashed line in Fig. 7a. The orientations are mainly at the center for {0001} pole figure; however, there are some dispersed and scattered intensities around the center, demonstrating that a few grains deviate from the (0001) orientation. The pole figures for the start/bottom part of Sample (001)+ (Fig. 7b bottom) show slightly more (0001)-aligned grains, as compared to Sample (001) (Fig. 6b bottom). The 3D view of the crystals in the IPF map (Fig. 7a) exhibits one example of a gradual texture development from the bottom to the top part of the α-Al2O3 layer. Nevertheless, as the IPF map indicates, most of the grains in this sample either nucleate and start growing with the preferred orientation or sharply shift from non-preferred to preferred
4. Discussion Previous studies have noted the importance of understanding and improving characteristics of CVD alumina coatings such as enhanced bonding [10,12–14,17] and texture [18–20] on wear properties of the cutting tools. However, very little is found in the literature on the understanding of texture development during growth and the effect of 7
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 9. (a). Band contrast (top) and corresponding IPF map (bottom) for Sample (010). There are highly aligned grains towards both the (101¯0) and (011¯0) corners in the top and bottom parts of the coating. Some misaligned grains are observed at the bottom of coating. A 3D view is shown for three grains, demonstrating an example of abrupt texture development (b). Calculated pole figures {011¯0 } and {112¯0 } for the top and bottom parts of Sample (010). (c). IPFs for Sample (010) as a function of slice height. The contouring demonstrates highly aligned grains even at the bottom of the coating, located at both (101¯0) and (011¯0) corners, which are the preferred orientations. Some low intensities referring to grains with non-preferred orientations can be observed towards (0001) in slice 1 which disappears gradually after three slices.
observed at the grain boundaries originating from the WC/Co substrate. These elements migrate through the grain boundaries of the TiN, Ti (C,N) and bonding layer all the way to the upper interface towards the alumina (Fig. 5). W has high solubility in Ti(C,N) so there is a thermodynamic driving force to form (Ti,W)(C,N). However, bulk diffusion is not possible at these deposition temperatures, i.e. 885–1000 °C [24] so the only way for W to enter Ti(C,N) is by grain boundary diffusion. It is also known that there is a driving force for Co to segregate to all types of carbide grain boundaries [24,25]. It is therefore not surprising to find also Co at the grain boundaries, and since it can be expected to have a higher mobility compared to W, faster transport of Co at the grain boundaries can be expected. It is surprising however, to find spherical Co particles in the bonding layer. It is known that the Co/Ti(C,N) interface has a high interfacial energy per unit area [26] (that manifests itself in the well-known inadequate wetting of Ti(C,N) by Co); thus, the total energy is minimized by lowering the surface energy, which leads to a spherical shape of the Co particles if they manage to nucleate. In order to have ~10 nm spheres of Co, it can be hypothesized that
using catalyzing gases on the early stage grain nucleation, which ultimately leads to the texture of the coating. This work aimed to study the development of (0001) and (011¯0) textured alumina coatings and the influence of H2S as catalyzing gas. The general microstructure of all three samples is very similar. Almost the entire α-Al2O3 layers consist of large, columnar grains with a thin inner region of small, equiaxed grains. It appears as having different textures, (0001) or (011¯0) , has no major effect on the microstructure, such as the size and morphology of alumina grains. However, it was observed that the degree of porosity at the interface of α-Al2O3 and the bonding layer interface is higher (almost double the amount) in Sample (001)+, compared to Sample (001) and Sample (010), see Figs. 2–4. This suggests that the number of voids is related to the presence of H2S during the nucleation step of the α-Al2O3 layer deposition. It is possible to hypothesize that the H2S, which is an aggressive gas, etches the surface of the bonding layer that later transforms to voids when alumina is deposited. In the underlying, Ti-containing, layers diffusion of W and Co is 8
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
• The desired texture develops early in the (011¯0)-textured coating (without H S in the beginning). • Thus, in this study, H S promotes (0001) texture and interfacial void
possible small pores in the bonding layer can act as nucleation sites. These pores can later be fed by fast transport of Co through grain boundaries. It may be the case, therefore, that these diffusions affect the nucleation and growth of alumina at early stages and may be the cause of the porosities observed at the interface of the alumina and bonding layer. The texture analysis of the coatings, presented as IPF maps and pole figures (Figs. 6–9) using TKD data, shows that all three samples are highly textured. This is also in accordance with the XRD results, which showed that all three samples have high texture coefficients (Table 2). Nevertheless, by comparing the grains at the top and bottom of the alumina layer, it is revealed that the bottom part, which contains the small and equiaxed grains, has a more random orientation. Using H2S in the beginning increases the probability for growing grains with an (0001) texture. This can be seen by comparing the IPFs for the bottom subsets (slice 1) of Sample (001) and Sample (001)+ (Figs. 6c and 7c). This means that the desired texture develops earlier for (0001)-textured alumina when H2S is present from the start of deposition. This is consistent with results observed in Fig. 8, which signifies that introducing H2S reduces the number of grains with the c-axis laying down, and limit them to a very thin (one grain high) region at the start/bottom of alumina layer. It can be observed in Fig. 9c that in (011¯0) -textured alumina the desired texture develops earlier without H2S being present from the beginning. The grains either grow with the desired orientation from the start or are modified to the desired orientation at early stages of deposition. In general, the texture develops in two modes, either gradually over several grains or more abruptly from one grain to another. The combination of the methods and results in this work provides support for the conclusion that H2S promotes (0001) texture and interfacial void formation.
2
2
formation.
Acknowledgements The authors would like to thank Swedish Foundation for Strategic Research (SSF) for their financial support via the contract ID16-0048. This work was performed in part at the Chalmers Materials Analysis Laboratory, CMAL. We would like to acknowledge Prof. Hans-Olof Andrén for his valuable input in the work and Dr. Linus von Fieandt from Sandvik Coromant for the support and discussions. References [1] A.G. King, W.M. Wheildon, Ceramics in Machining Processes, vol. 1966, Academic Press, NEW YORK, N. Y., 1966, p. 327 111 FIFTH AVE. [2] B.M. Kramer, P.K. Judd, Computational design of wear coatings, J. Vac. Sci. Technol. A 3 (6) (1985) 2439–2444. [3] O. Knotek, F. Löffler, G. Krämer, Cutting performance of multicomponent and multilayer coatings on cemented carbides and cermets for interrupted cut machining, Int. J. Refract. Met. Hard Mater. 14 (1–3) (1996) 195–202. [4] G. Brandt, Thermal shock resistance of ceramic cutting tools, Surf. Eng. 2 (2) (1986) 121–132. [5] G.K.L. Goh, et al., Transitions in wear mechanisms of alumina cutting tools, Wear 201 (1) (1996) 199–208. [6] S. Ruppi, M. Halvarsson, TEM investigation of wear mechanisms during metal machining, Thin Solid Films 353 (1) (1999) 182–188. [7] C.J. Chatfield, N.M. Lindström, E. Sjöstrand, Microstructure of CVD - Al2O3, J. Phys. Colloques 50 (C5) (1989) C5–377-C5-387. [8] M. Halvarsson, S. Vuorinen, Epitaxy in multilayer coatings of k-Al2O3, Surf. Coat. Technol. 80 (1) (1996) 80–88. [9] A. Larsson, M. Halvarsson, S. Vuorinen, Microstructural investigation of as-deposited and heat-treated CVD Al2O3, Surf. Coat. Technol. 94–95 (1997) 76–81. [10] S. Vuorinen, L. Karlsson, Phase transformation in chemically vapour-deposited κalumina, Thin Solid Films 214 (2) (1992) 132–143. [11] M. Halvarsson, S. Vuorinen, The influence of the nucleation surface on the growth of CVD α-Al2O3 and κ-Al2O3, Surf. Coat. Technol. 76–77 (1995) 287–296. [12] S. Ruppi, Advances in chemically vapour deposited wear resistant coatings, Le J. de Phys. IV 11 (PR3) (2001) Pr3-847–Pr3-859. [13] S. Söderberg, M. Sjöstrand, B. Ljungberg, Advances in coating technology for metal cutting tools, Met. Powder Rep. 56 (4) (2001) 24–30. [14] S. Ruppi, Deposition, microstructure and properties of texture-controlled CVD αAl2O3 coatings, Int. J. Refract. Met. Hard Mater. 23 (4–6) (2005) 306–316. [15] S. Canovic, et al., CVD TiC/alumina and TiN/alumina multilayer coatings grown on sapphire single crystals, Int. J. Refract. Met. Hard Mater. 28 (2) (2010) 163–173. [16] S. Canovic, B. Ljungberg, M. Halvarsson, CVD TiC/alumina multilayer coatings grown on sapphire single crystals, Micron 42 (8) (2011) 808–818. [17] P.W. Trimby, Orientation mapping of nanostructured materials using transmission Kikuchi diffraction in the scanning electron microscope, Ultramicroscopy 120 (2012) 16–24. [18] S. Ruppi, Enhanced performance of α-Al2O3 coatings by control of crystal orientation, Surf. Coat. Technol. 202 (17) (2008) 4257–4269. [19] S. Ruppi, A. Larsson, A. Flink, Nanoindentation hardness, texture and microstructure of α-Al2O3 and κ-Al2O3 coatings, Thin Solid Films 516 (18) (2008) 5959–5966. [20] R. M'Saoubi, S. Ruppi, Wear and thermal behaviour of CVD α-Al2O3 and MTCVD Ti (C,N) coatings during machining, CIRP Ann. 58 (1) (2009) 57–60. [21] R. M’Saoubi, et al., Microstructure and wear mechanisms of texture-controlled CVD α-Al2O3 coatings, Wear 376–377 (2017) 1766–1778. [22] N. Mortazavi, M. Esmaily, M. Halvarsson, The capability of transmission Kikuchi diffraction technique for characterizing nano-grained oxide scales formed on a FeCrAl stainless steel, Mater. Lett. 147 (2015) 42–45. [23] N. Mortazavi, et al., Interplay of water and reactive elements in oxidation of alumina-forming alloys, Nat. Mater. 17 (7) (2018) 610–617. [24] U. Rolander, H.-O. Andrén, Atom probe microanalysis of the γ phase in cemented carbide materials, Mater. Sci. Eng. A 105-106 (1988) 283–287. [25] A. Henjered, et al., Quantitative microanalysis of carbide/carbide interfaces in WC–co-base cemented carbides, Mater. Sci. Technol. 2 (8) (1986) 847–855. [26] M. Christensen, S. Dudiy, G. Wahnström, First-principles simulations of metalceramic interface adhesion: Co/WC versus Co/TiC, Phys. Rev. B 65 (4) (2002) 45408.
5. Conclusions This study investigated the interfacial microstructure and texture development in three textured CVD α-Al2O3 coatings using XRD, STEM/EDX and SEM/TKD. The following conclusions were made using the presented results:
• Diffusion of W and Co occurred through the TiN and Ti(C,N) layers • • • • •
all the way to the bonding layer/α-Al2O3 interface. It was suggested that this might disturb the nucleation and early growth of the alumina layer. Interfacial porosity at the bonding layer/α-Al2O3 interface was observed. A similar number of voids was present in the (0001) and (011¯0) -textured coatings when H2S was not used during the initial deposition step. When H2S was present during the nucleation step of the α-Al2O3 in the deposition process, a higher fraction of voids (almost double the amount) was introduced at the interface of the (0001)-textured sample. The alumina grain morphology developed from small (~100 nm) equiaxed grains at the start/bottom of the coating to larger (several microns) columnar grains at the top of the coating. The inner part of the alumina coatings exhibited grains with a more random orientation. The texture changed either: (i) gradually over several grains, or (ii) more abruptly from one grain to another; from more random orientations to the desired texture. The desired texture develops earlier with H2S for the (0001)-textured coatings.
9
Contents lists available at ScienceDirect
International Journal of Refractory Metals & Hard Materials journal homepage: www.elsevier.com/locate/IJRMHM
Microstructure investigation of textured CVD alumina coatings ⁎
S. Shojaa,b, , N. Mortazavia,c, E. Lindahlb, S. Norgrenb, O. Bäckea, M. Halvarssona a b c
Department of Physics, Chalmers University of Technology, Gothenburg, Sweden Sandvik Coromant, Stockholm, Sweden School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, United States
A R T I C LE I N FO
A B S T R A C T
Keywords: Coating Alpha alumina CVD Texture development H2S TKD
This study concerns the interfacial microstructure and texture development in three textured CVD α-Al2O3 coatings using X-ray diffraction, transmission Kikuchi diffraction, scanning transmission electron microscopy and energy dispersive X-ray analysis. It is well known that the performance of these types of coatings relies on the degree and type of texture. The aim of this work is to study the microstructure of three different CVD α-Al2O3 layers when deposited onto a Ti(C,N,O) bonding layer. The coatings were deposited onto cemented carbide/ cobalt substrates (WC/Co). It was observed that grain boundary diffusion of W and Co occurred through the TiN and Ti(C,N) layers to the bonding layer/α-Al2O3 interface. This may disturb the alumina layer nucleation and early growth. Interfacial porosity was observed at the bonding layer/α-Al2O3 interface. The number of voids that were detected in the (0001) and (011¯0) -textured coatings was similar when H2S was not used during the initial deposition step. When H2S was present during the nucleation step deposition of the α-Al2O3 more voids were introduced at the interface for the (0001)-textured samples. The alumina grain morphology developed from small (~100 nm) equiaxed grains at the start of the alumina coating to larger (several microns) columnar grains at the top of the coating. The inner part of the alumina coatings had a more random orientation. The texture changed either: (i) gradually over several grains, or (ii) more abruptly from one grain to another; from more random orientations to the desired texture. The desired texture develops earlier when using H2S at the start for the (0001)-textured coatings, while the (011¯0) -texture development benefits from the absence of H2S. Thus, in this study, H2S promotes (0001) texture and interfacial void formation.
1. Introduction
The properties of α-Al2O3 coatings using CVD processes have been significantly improved through optimizing bonding layers for phase control as well as performing microstructural refinements in the coating layers, e.g. via lowering the fraction of porosities at interfaces, see for example [10–17]. As a rather new development step, highly textured alumina coatings [14] have attracted industrial and scientific attention as a solution for further improvement of the CVD Al2O3 coatings, resulting in increased cutting tool life, as well as higher metal removal rates. During the last decade, the properties of the different texture components of α-Al2O3 coatings have been widely characterized [14,18–20]. In this regard, Ruppi [18] and M'Saoubi et al. [20] studied wear properties of CVD α-Al2O3 layers with different growth textures and reported significant texture effects on wear performance of the αAl2O3 layers. Suggesting plastic deformation as the main degradation mechanism, they reported that the best wear resistance is achieved for (0001)-textured coating. Ruppi et al. [18,19] examined the ability to
The technology of machining of metallic materials demands cutting tools with longer lifetime and better performance, striving to accomplish the optimal combination of productivity and tool life. Cutting tools must withstand rigorous operating conditions as they are subjected to high temperatures, large temperature gradients, thermal chocks, fatigue, abrasion, attrition and diffusion wear [1–3]. It is well known that alumina (Al2O3) is suitable to be used in a coating system on metal cutting tools owing to its chemical inertness and high hardness at elevated temperatures, i.e. 650 °C to 1000 °C, which is the typical temperature range in the metal cutting applications [4–6]. The Al2O3 coatings are usually applied to the tools using chemical vapor deposition (CVD). The CVD technique has been used for many years for industrial deposition of thick and uniform Al2O3 coatings on cemented carbide (WC/Co) substrates, first as κ-Al2O3 and later also as α-Al2O3 [7–9].
⁎
Corresponding author at: Department of Physics, Chalmers University of Technology, Gothenburg, Sweden. E-mail address: [email protected] (S. Shoja).
https://doi.org/10.1016/j.ijrmhm.2019.105125 Received 5 July 2019; Received in revised form 29 September 2019; Accepted 3 October 2019 0263-4368/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: S. Shoja, et al., International Journal of Refractory Metals & Hard Materials, https://doi.org/10.1016/j.ijrmhm.2019.105125
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
undergo uniform plastic deformation of α-Al2O3 coatings with a number of textures, namely (101¯2) , (101¯4), (0001) and (101¯0). They reported that the (0001)-textured coating exhibits better wear performance as compared to the other textured coatings. These conclusions were also confirmed by M'Saoubi et al. [21] in their work. They used controlled machining tests to investigate the wear properties of the αAl2O3 for three textures, (0001), (011¯2) and (101¯0) and showed that the (0001) textured alumina is the preferred one. However, a fundamental description of texture development during growth and the effect of catalyzing gases on the coating's texture were not discussed. This study focuses on grain development as a function of height during texture evolution in differently textured CVD α-Al2O3 coatings. The microstructural differences of the different CVD α-alumina coatings were investigated by means of Scanning Transmission Electron Microscopy (STEM). Micro-textural analyses were also performed through acquiring grain orientation mapping using Transmission Kikuchi Diffraction (TKD). Thus, the aim of the present investigation is to further elucidate the microstructural differences in CVD alumina coatings with different textures.
layers was analyzed in order to understand the coating properties and also to gain insight about the reasons for the formation of interfacial defects and their influence on alumina nucleation and early growth. The texture of the upper and lower parts of the α-Al2O3 coatings was analyzed using pole figure analysis and inverse pole figure (IPF) mapping based on TKD data acquired from thin foil samples. Details of specimen preparation, and the imaging and analytical techniques are described below and the results for the different samples are presented and discussed in Section 3. 2.2. Focused ion beam/scanning electron microscopy (FIB/SEM) An FEI Versa 3D combined Focused Ion Beam/Scanning Electron Microscope (FIB/SEM) workstation was used to produce cross-section thin foils from the coatings and subjacent substrate. The thin foils were later used for STEM/EDX analysis in a TEM/STEM and for TKD in a scanning electron microscope (SEM), see below. In order to protect the surface from ion beam damage, a thin Pt-layer with a thickness of around 100 nm was first deposited using electrons onto the surface (25 μm long and 5 μm wide) and then a thicker Pt layer (~2 μm) was deposited on top with the assistance of Ga-ions. Low accelerating voltages, i.e. 5 kV and 2 kV with ion currents of 49 pA and 27 pA, respectively, were used in the final thinning process to minimize the ion beam induced damage to the milled surfaces.
2. Materials and methods 2.1. Materials The studied coatings were comprised of several sub-layers obtained from different processing steps. All coatings were deposited on a WC/ Co substrate using thermally activated CVD. They consisted of an inner TiN layer, ~0.3 μm thick deposited at 930 °C, followed by a medium temperature CVD Ti(C,N) layer of ~1 μm thick, deposited at 885 °C. Then a bonding layer (Ti(C,N,O)) with the thickness of ~0.5 μm was deposited at 1000 °C followed by an α-Al2O3 layer of ~5 μm which was deposited at 1000 °C with two different textures, (0001) and (011¯0) . We investigated the gradual texture development of the alumina layers in three different samples. Two of the specimens had identical texture, (0001), but different deposition starting conditions of with and without the catalyzing gas H2S. This gas is generally used in industrial production of the coatings to enhance the alumina growth rate and to produce coatings with even thicknesses. Thus, these two samples were used for comparing the initial alumina layer growth, and the influence of the catalyzing gas (H2S) on the microstructure at the bonding layer/ alumina interface. These coatings are referred to as Sample (001) and Sample (001)+, where the “+” denotes that the H2S was added during the whole process, while the (001) sample was produced without H2S in the initial step which is the standard case in industry. This initial step corresponds to around 100 nm in thickness. The third sample, referred to as Sample (010) in this paper, had a (011¯0) texture and was produced without H2S in the initial step. The individual layer thicknesses and presence of an initial step including the catalyzing gas H2S are summarized in Table 1 for the three coatings. The texture and microstructure of the different coatings were investigated using STEM, Energy Dispersive X-ray analysis (EDX) and TKD. Special attention was paid to the fraction of voids and defects at interphases. The chemical composition of the underlying Ti-based
2.3. X-Ray diffraction (XRD) The X-Ray Diffraction (XRD) measurements were done with a Bruker D8 Advance instrument using CuKα radiation. The XRD was used with a θ-2θ setup. Using the XRD measurements, textures for the alumina layers were evaluated using texture coefficients, TCs, which were calculated using Eq. (1), see below. The equation uses the intensity ratios of diffraction peaks (hkl) from possibly textured coatings over non-textured coatings, in order to calculate the degree of texture.
TC (hkl) =
Sample (001)
Sample (001)+
Sample (010)
5.2 No 0.5 1.2 0.3
5.3 Yes 0.5 1.2 0.33
6.5 No 0.65 1.5 0.45
I (hkl) ⎫ I0 (hkl) ⎬ ⎭
(1)
2.4. Transmission kikuchi diffraction (TKD) Grain orientation mapping of the alumina layers was performed by means of TKD. This enabled us to carry out high resolution microstructural (grain size and shape) and micro-textural (preferred orientation of α-alumina grains) analyses on samples. The TKD analyses with high indexing rates were performed using a special thin foil holder, designed earlier, see [22,23]. The analyses were performed using an HKL Channel 5 Electron Backscattered Diffraction (EBSD) system with a Nordlys II detector, mounted on a Zeiss Ultra Field Emission Gun (FEG) SEM. The TKD data were collected from the FIB/ SEM-prepared thin foils. An accelerating voltage of 30 kV in high current mode and an aperture size of 120 μm were used. The analysis grid size for most of the cases was 160 × 100 with a step size of 10–20 nm and SEM magnifications of around 35,000 X.
Coating layer (μm) Al2O3 Initial catalyzing step Bonding layer Ti(C,N) TiN
−1
∑
I(hkl) is the XRD peak intensity of the textured coating, I0(hkl) is the non-textured XRD peak intensity from a reference material, and n is the number of peaks considered. The TC was calculated based on eight reflections (011¯2) , (101¯4) , (112¯0) , (112¯3) , (022¯4) , (112¯6) , (033¯0) and (0 0 0 12), which makes the maximum value of the texture coefficient equal to 8. This value corresponds to a perfectly oriented material. In contrast, the value of 1 corresponds to a randomly oriented sample.
Table 1 Measured coating layer thicknesses for the three samples, including an indication if there was an initial catalyzing step with H2S or not. Sample ID
I (hkl) ⎧ 1 I0 (hkl) ⎨ ⎩n
2.5. Scanning transmission electron microscopy/energy dispersive X-ray analysis (STEM/EDX) The thin foils were also investigated by STEM in an FEI Titan 2
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 1. SEM secondary electron micrographs showing the surface morphologies of the α-alumina coatings. (a) and (b) The basal plane (0001) is parallel to the surface. (c) The prismatic plane (011¯0) is parallel to the surface.
Fig. 2. (a) STEM BF micrograph showing a cross-section overview of the coating microstructure for Sample (001). There are small and equiaxed grains at the bottom of α-alumina layer. The shape and size of the grains develop to larger, columnar grains from the bottom to the top. STEM (b) BF and (c) HAADF images showing some of the porosities (arrowed) at the interface of α-Al2O3 and the bonding layer in higher magnification. Fig. 3. (a) STEM BF image showing the cross-section overview of Sample (001)+; H2S was used during the whole alumina deposition. The coating exhibits small and equiaxed grains at the bottom of the αalumina layer. The grains develop to larger and columnar grains at the outer parts. STEM (b) BF and (c) HAADF images showing the interfacial porosity (circled) at the α-Al2O3/bonding layer at higher magnification.
general microstructure, such as grain morphology and local composition of differently grown layers, and also to characterize the processinduced pores in the multilayer coating.
80–300 TEM/STEM equipped with a FEG operated at an accelerating voltage of 300 kV. The microscope is equipped with an Oxford Inca EDX detector. STEM micrographs were acquired using high angle annular dark field (HAADF) and bright field (BF) detectors in combination with EDX analysis. The STEM analyses were conducted to examine the 3
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 4. (a) STEM BF micrograph showing the cross-section overview of Sample (010); H2S was not used during the initial alumina deposition. The coating exhibits small and equiaxed grains at the bottom of the α-alumina layer, while there are larger and columnar grains at the outer parts. STEM (b) BF and (c) HAADF images showing the needle shaped morphology of the bonding layer and some porosity (circled) at the interface of α-Al2O3 and the bonding layer.
Fig. 5. (a). STEM HAADF image at the region of the coating/WC-Co substrate interface showing the migration of heavy elements (white contrast) from the substrate to the coating along grains boundaries. (b) STEM HAADF micrograph at the interface of alumina and the bonding layer showing that heavy elements have migrated all the way to this interface. (c) STEM/EDX maps from the boxed region in (b) showing the presence of W and Co in this region. Table 2 Calculated texture coefficients for eight reflections show that all three samples are highly textured. Sample (001) and Sample (001)+ have high texture coefficients for (0 0 0 12), which corresponds to an (0001) texture and Sample (010) has a high texture coefficient for (033¯0) , which corresponds to an (011¯0) texture. TC
(0 1 1¯ 2)
(1 0 1¯ 4)
(1 1 2¯ 0)
(1 1 2¯ 3)
(0 2 2¯ 4)
(1 1 2¯ 6)
(0 3 3¯ 0)
(0 0 0 12)
0.30 0.02 0.22
0.59 0.46 0.07
0.06 0.14 0.13
0.05 0.00 0.03
0.09 0.00 0.38
0.33 0.06 0.06
0.16 0.11 7.11
6.69 7.21 0.00
Sample ID Sample (001) Sample (001)+ Sample (010)
3. Results
are presented and discussed in this section. SEM micrographs of the surface of the coatings for all three α-Al2O3 layers are shown in Fig. 1. The surface morphology of the coatings exhibits pyramidal-shaped grains with the size of 1–2 μm. To get a better understanding of the coatings, cross-section thin-foils were prepared by FIB/SEM and investigated using STEM/EDX. Fig. 2a shows a STEM BF cross-section overview of the
3.1. General microstructure As mentioned earlier, different aspects of the coatings were investigated using the SEM, STEM, EDX and TKD techniques. The results for all three samples, Sample (001), Sample (001)+ and Sample (010),
4
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 6. (a). Band contrast (top) and corresponding IPF map (bottom) for Sample (001). There are highly aligned (red) (0001) alumina grains in the top part of the coating and more randomly oriented (blue/green) small grains at the bottom of the coating. 3D view (left) of the crystals are shown for some grains demonstrating an abrupt texture development (b). Calculated pole figures {0001} and {101¯0 } of the top and bottom parts of the coating. (TKD data). (c). Inverse pole figures for Sample (001) as a function of slice height. The texture gradually develops towards (0001). The grain orientations in slice 1 and 2 are relatively random, and start to be aligned along (0001) from slice 3. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
higher number of pores compared to Sample (001) can be seen right at the interface between the α-Al2O3 and the bonding layer. Fig. 3b and c show STEM BF and HAADF images of the voids at the interface in higher magnifications. The average number of pores per unit of length for this sample was measured to 2.41 ± 0.53 pore/μm. Fig. 4a shows the cross-section overview for Sample (010) by STEM BF imaging. The α-Al2O3 layer was in this case deposited without H2S during the initial step of deposition. The bonding layer has a needleshaped morphology, as can be seen in Fig. 4b. Some porosity is observed at the interface of α-Al2O3 and the bonding layer, which is shown in the STEM HAADF image in Fig. 4c. The average number of interfacial pores per unit of length were determined to 1.35 ± 0.53 pore/μm. The extent of pores is roughly half the amount for Sample (001)+, and almost the same for Sample (001). However, with a small sample size i.e. TEM thin-foils, caution must be applied as the statistics is limited for these measurements. Turning to the Ti-containing layers below the alumina layer, these layers are the same in all three samples and can be seen in Figs. 2–4. The innermost TiN layer is comprised of very small grains, in the order of 10–100 nm, especially closest to the WC/Co substrate. The Ti(C,N) layer has columnar, larger grains, while the bonding layer contains small, sometimes columnar grains. A closer inspection of the inner parts
microstructure of Sample (001), which was produced without introducing H2S at the initial step of alumina deposition. The micrograph provides information about the thickness, grain size, morphology and defects of different layers in the coating. The outer Pt-layer was deposited during specimen preparation in the FIB/SEM to protect the surface from ion beam damage. In Fig. 2a there are small and equiaxed grains with a size of about 100 nm visible at the start of the α-Al2O3 layer. The shape and size of the grains develop to columnar forms with the width of 1–2 μm and column lengths of several microns, almost equal to the layer thickness. Some interfacial pores, marked by arrows, can be observed at the interface of α-Al2O3 and the bonding layer. These porosities are shown in higher magnification in STEM BF and HAADF in Fig. 2b and c, respectively. The number of interfacial pores per unit of length was measured in a few TEM micrographs taken from different parts of the interface. The average number measured for this sample was 1.05 ± 0.21 pore/μm. A STEM cross-sectional overview micrograph of Sample (001)+, i.e. when H2S was introduced from the beginning of the alumina deposition, is shown in Fig. 3a. Almost the entire α-Al2O3 layer consists of large and columnar grains, several microns in height and widths of 1–2 μm. At the start of the α-Al2O3 layer, small and equiaxed grains with a grain size of around 100 nm is also observed in this coating. A 5
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 7. (a). Band contrast (top) and corresponding IPF map (bottom) for coating Sample (001)+. There are highly aligned grains towards (0001) at the top of the coating and more randomly aligned grains at the bottom of the coating. 3D view (left) of some crystals demonstrating a gradual texture development (b). Calculated pole figures {0001} and {101¯0 } of the top and bottom parts of the coating for Sample (001)+. (c). Inverse pole figures for sample (001)+ as a function of slice height. The texture gradually develops towards (0001). The grain orientation in slice 1 is relatively random, and start to be aligned along (0001) from slice 2.
measurements are summarized in Table 2. The highest TC for Sample (001) and (001)+ is around 7 and belongs to (0 0 0 12) reflection. This reflection represents the (0001) orientation. These high TC values mean that both Sample (001) and Sample (001)+ are highly (0001)-textured coatings. The highest TC value for Sample (010) is around 7 for the (011¯0) peak. Thus, the coating is highly (011¯0) -textured, as expected. Transmission Kikuchi Diffraction analysis on cross-section thin foils was used to understand the development of the textures as a function of height in the coatings. The results for Sample (001) are shown in Fig. 6. The band contrast and IPF maps (Fig. 6a) of the α-Al2O3 cross-section show large columnar grains in the top part of the coating with lattice plane normals (0001) along the growth direction, shown as red grains in the IPF map (the IPF color legend is also shown in Fig. 6a). Fig. 6b shows the {0001} and {101¯0 } pole figures extracted from the TKD data. The two pole figures on top show the results from the top part of the coating, above the dashed line in Fig. 6a. They have a high intensity in the center for {0001} and a ring at 90° (i.e. at the periphery) in {101¯0 }, which indicates that almost all the grains in the top part of the alumina layer have (0001) orientation. However, the pole figures for the bottom
of the Ti(C,N) and TiN layers reveals a network of bright contrast in the STEM HAADF micrographs (see Fig. 5a for Sample (001)). This implies the migration of heavy elements from the WC/Co substrate along the TiN and Ti(C,N) grain boundaries. These elements have diffused all the way up through the (Ti;C,N) and bonding layer to the alumina interface, as can be seen in Fig. 5b. STEM/EDX was used to analyze the elemental distribution in the top part of the Ti(C,N) layer, the bonding layer and the interface towards αAl2O3. EDX maps from the region indicated by a box in Fig. 5b are presented in Fig. 5c. W is present as vague elongated areas, probably along grain boundaries, while Co is present as small (~10 nm) round spheres. This shows that W and Co have migrated to this sensitive region where they may affect the nucleation and growth of the α-Al2O3 layer. It can also be observed from the EDX map that these elements probably have not diffused further into the α-Al2O3 layer. 3.2. Texture analysis The
calculated
texture
coefficients
(TCs)
from
the
XRD 6
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 8. IPF maps showing grains with the c-axis tilted 70° to 90°away from the surface normal. The number of grains are around 70% of all grains for Sample (001) and 35% for Sample (001)+.
orientation. The IPFs for horizontal subsets (Fig. 7c) show that the start/bottom slice has a relatively random texture. However, to a lesser extent compared to the same section for Sample (001). It can also be seen that the texture starts to be relatively orientated towards (0001) already from Section 2 to the top. To investigate the small grains at the bottom of the α-Al2O3 layers in more detail, the grains with their c-axis tilted 70° to 90° away from the surface normal, which are the grains with non-preferred orientations, were selected. Fig. 8 shows these grains for both Sample (001) and Sample (001)+. There are significant differences in the shape and number of selected grains for the two samples. In Sample (001), there are many grains, ~70% of the total, within this angular range, i.e. with the c-axis almost lying down. These grains are agglomerated and extended to 0.5 to 1 μm height. In Sample (001)+ these types of grains are fewer (~35%) with a height of only one grain. The texture analysis of Sample (010) obtained by TKD is shown in Fig. 9. The band contrast (Fig. 9a top) is in agreement with the microstructure of this sample observed by TEM, see Fig. 4. The IPF map (Fig. 9a bottom) shows that almost all the grains have either (101¯0) or (011¯0) orientations (green or blue); thus, Sample (010) is highly textured. This also agrees with the XRD and texture coefficient results as described above. According to the IPF map the majority of the grains nucleate and grow with the preferred orientation, i.e. they are not colored red. For the other (red) grains the growth orientation often changes abruptly from the bottom to the top. The three-dimensional view of three crystals, next to the IPF map in Fig. 9a, illustrates one example of an abrupt orientation change. The IPF map texture analysis results can be verified by viewing the {011¯0 } and {112¯0 } pole figures for the top and start/bottom part of the coating (Fig. 9b). The pole figures indicate that both top and bottom parts of the α-Al2O3 layer are highly textured and very close to ideal {011¯0 } texture, corresponding to intensity at the center and at a ring 60° away for {011¯0 }, and two rings at 30° and 90° for {112¯0 }. The IPFs for horizontal subsets (Fig. 9c) all exhibit strong intensities at the (101¯0) and (011¯0) corners, which refers to highly aligned grains from bottom to the top. Only some low intensity can be observed near the (0001) corner, which disappears gradually after the distance corresponding to the third slice.
part, (start), show that the intensities are not as concentrated as for the top part, which gives that fewer grains have (0001) orientations. Thus, the start/bottom part of the coating is not as textured as the top part. These pole figure results are in agreement with the IPF map grain orientations, where many small, equiaxed grains at the start/bottom of αAl2O3 layer have the “wrong” orientation, i.e. not (0001), and have blue or green color combinations instead of being red. Three-dimensional views of the crystal orientation for a column of grains are shown next to the IPF map in Fig. 6a. The inner grain has the c-axis nearly parallel to the interface (lying down), while the outer three have their c-axis (almost) parallel to the growth direction. This is an example of abrupt grain orientation change from non-preferred to the preferred orientation. However, most of the grains in Sample (001) show a gradual change in their orientations from start/bottom to the top in such a way that the c-axis is gradually aligned along the growth direction. To study texture development in more detail, seven horizontal subsets with heights of around 700 nm were created (see Fig. 6a) and IPFs for each subset were made (Fig. 6c). Thus, Fig. 6c shows the texture development as a function of height. It is apparent that the grains start with more random orientations, and gradually develop an (0001) texture. The IPFs support that the grains start to be relatively oriented towards (0001) from Section 3 and upward. Fig. 7 gives the texture analysis of the Sample (001)+ obtained by TKD. The band contrast of the α-Al2O3 layer (Fig. 7a top) shows the shape and size of the grains reflecting the small and equiaxed grains at the start/bottom close to the α-Al2O3/bonding interface and larger and columnar grains on top, as seen earlier in the STEM micrographs in Fig. 3. The IPF map (Fig. 7a bottom) for this sample shows that most of the grains have the preferred orientation of (0001), already from the start. However, there are a few grains with colors (blue/green) further away from (0001) corner, which means that they don't have the desired (0001) orientation. This can also be verified by viewing the {0001} and {101¯0 } pole figures (Fig. 7b top) for the top part of the coating, above the dashed line in Fig. 7a. The orientations are mainly at the center for {0001} pole figure; however, there are some dispersed and scattered intensities around the center, demonstrating that a few grains deviate from the (0001) orientation. The pole figures for the start/bottom part of Sample (001)+ (Fig. 7b bottom) show slightly more (0001)-aligned grains, as compared to Sample (001) (Fig. 6b bottom). The 3D view of the crystals in the IPF map (Fig. 7a) exhibits one example of a gradual texture development from the bottom to the top part of the α-Al2O3 layer. Nevertheless, as the IPF map indicates, most of the grains in this sample either nucleate and start growing with the preferred orientation or sharply shift from non-preferred to preferred
4. Discussion Previous studies have noted the importance of understanding and improving characteristics of CVD alumina coatings such as enhanced bonding [10,12–14,17] and texture [18–20] on wear properties of the cutting tools. However, very little is found in the literature on the understanding of texture development during growth and the effect of 7
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
Fig. 9. (a). Band contrast (top) and corresponding IPF map (bottom) for Sample (010). There are highly aligned grains towards both the (101¯0) and (011¯0) corners in the top and bottom parts of the coating. Some misaligned grains are observed at the bottom of coating. A 3D view is shown for three grains, demonstrating an example of abrupt texture development (b). Calculated pole figures {011¯0 } and {112¯0 } for the top and bottom parts of Sample (010). (c). IPFs for Sample (010) as a function of slice height. The contouring demonstrates highly aligned grains even at the bottom of the coating, located at both (101¯0) and (011¯0) corners, which are the preferred orientations. Some low intensities referring to grains with non-preferred orientations can be observed towards (0001) in slice 1 which disappears gradually after three slices.
observed at the grain boundaries originating from the WC/Co substrate. These elements migrate through the grain boundaries of the TiN, Ti (C,N) and bonding layer all the way to the upper interface towards the alumina (Fig. 5). W has high solubility in Ti(C,N) so there is a thermodynamic driving force to form (Ti,W)(C,N). However, bulk diffusion is not possible at these deposition temperatures, i.e. 885–1000 °C [24] so the only way for W to enter Ti(C,N) is by grain boundary diffusion. It is also known that there is a driving force for Co to segregate to all types of carbide grain boundaries [24,25]. It is therefore not surprising to find also Co at the grain boundaries, and since it can be expected to have a higher mobility compared to W, faster transport of Co at the grain boundaries can be expected. It is surprising however, to find spherical Co particles in the bonding layer. It is known that the Co/Ti(C,N) interface has a high interfacial energy per unit area [26] (that manifests itself in the well-known inadequate wetting of Ti(C,N) by Co); thus, the total energy is minimized by lowering the surface energy, which leads to a spherical shape of the Co particles if they manage to nucleate. In order to have ~10 nm spheres of Co, it can be hypothesized that
using catalyzing gases on the early stage grain nucleation, which ultimately leads to the texture of the coating. This work aimed to study the development of (0001) and (011¯0) textured alumina coatings and the influence of H2S as catalyzing gas. The general microstructure of all three samples is very similar. Almost the entire α-Al2O3 layers consist of large, columnar grains with a thin inner region of small, equiaxed grains. It appears as having different textures, (0001) or (011¯0) , has no major effect on the microstructure, such as the size and morphology of alumina grains. However, it was observed that the degree of porosity at the interface of α-Al2O3 and the bonding layer interface is higher (almost double the amount) in Sample (001)+, compared to Sample (001) and Sample (010), see Figs. 2–4. This suggests that the number of voids is related to the presence of H2S during the nucleation step of the α-Al2O3 layer deposition. It is possible to hypothesize that the H2S, which is an aggressive gas, etches the surface of the bonding layer that later transforms to voids when alumina is deposited. In the underlying, Ti-containing, layers diffusion of W and Co is 8
International Journal of Refractory Metals & Hard Materials xxx (xxxx) xxxx
S. Shoja, et al.
• The desired texture develops early in the (011¯0)-textured coating (without H S in the beginning). • Thus, in this study, H S promotes (0001) texture and interfacial void
possible small pores in the bonding layer can act as nucleation sites. These pores can later be fed by fast transport of Co through grain boundaries. It may be the case, therefore, that these diffusions affect the nucleation and growth of alumina at early stages and may be the cause of the porosities observed at the interface of the alumina and bonding layer. The texture analysis of the coatings, presented as IPF maps and pole figures (Figs. 6–9) using TKD data, shows that all three samples are highly textured. This is also in accordance with the XRD results, which showed that all three samples have high texture coefficients (Table 2). Nevertheless, by comparing the grains at the top and bottom of the alumina layer, it is revealed that the bottom part, which contains the small and equiaxed grains, has a more random orientation. Using H2S in the beginning increases the probability for growing grains with an (0001) texture. This can be seen by comparing the IPFs for the bottom subsets (slice 1) of Sample (001) and Sample (001)+ (Figs. 6c and 7c). This means that the desired texture develops earlier for (0001)-textured alumina when H2S is present from the start of deposition. This is consistent with results observed in Fig. 8, which signifies that introducing H2S reduces the number of grains with the c-axis laying down, and limit them to a very thin (one grain high) region at the start/bottom of alumina layer. It can be observed in Fig. 9c that in (011¯0) -textured alumina the desired texture develops earlier without H2S being present from the beginning. The grains either grow with the desired orientation from the start or are modified to the desired orientation at early stages of deposition. In general, the texture develops in two modes, either gradually over several grains or more abruptly from one grain to another. The combination of the methods and results in this work provides support for the conclusion that H2S promotes (0001) texture and interfacial void formation.
2
2
formation.
Acknowledgements The authors would like to thank Swedish Foundation for Strategic Research (SSF) for their financial support via the contract ID16-0048. This work was performed in part at the Chalmers Materials Analysis Laboratory, CMAL. We would like to acknowledge Prof. Hans-Olof Andrén for his valuable input in the work and Dr. Linus von Fieandt from Sandvik Coromant for the support and discussions. References [1] A.G. King, W.M. Wheildon, Ceramics in Machining Processes, vol. 1966, Academic Press, NEW YORK, N. Y., 1966, p. 327 111 FIFTH AVE. [2] B.M. Kramer, P.K. Judd, Computational design of wear coatings, J. Vac. Sci. Technol. A 3 (6) (1985) 2439–2444. [3] O. Knotek, F. Löffler, G. Krämer, Cutting performance of multicomponent and multilayer coatings on cemented carbides and cermets for interrupted cut machining, Int. J. Refract. Met. Hard Mater. 14 (1–3) (1996) 195–202. [4] G. Brandt, Thermal shock resistance of ceramic cutting tools, Surf. Eng. 2 (2) (1986) 121–132. [5] G.K.L. Goh, et al., Transitions in wear mechanisms of alumina cutting tools, Wear 201 (1) (1996) 199–208. [6] S. Ruppi, M. Halvarsson, TEM investigation of wear mechanisms during metal machining, Thin Solid Films 353 (1) (1999) 182–188. [7] C.J. Chatfield, N.M. Lindström, E. Sjöstrand, Microstructure of CVD - Al2O3, J. Phys. Colloques 50 (C5) (1989) C5–377-C5-387. [8] M. Halvarsson, S. Vuorinen, Epitaxy in multilayer coatings of k-Al2O3, Surf. Coat. Technol. 80 (1) (1996) 80–88. [9] A. Larsson, M. Halvarsson, S. Vuorinen, Microstructural investigation of as-deposited and heat-treated CVD Al2O3, Surf. Coat. Technol. 94–95 (1997) 76–81. [10] S. Vuorinen, L. Karlsson, Phase transformation in chemically vapour-deposited κalumina, Thin Solid Films 214 (2) (1992) 132–143. [11] M. Halvarsson, S. Vuorinen, The influence of the nucleation surface on the growth of CVD α-Al2O3 and κ-Al2O3, Surf. Coat. Technol. 76–77 (1995) 287–296. [12] S. Ruppi, Advances in chemically vapour deposited wear resistant coatings, Le J. de Phys. IV 11 (PR3) (2001) Pr3-847–Pr3-859. [13] S. Söderberg, M. Sjöstrand, B. Ljungberg, Advances in coating technology for metal cutting tools, Met. Powder Rep. 56 (4) (2001) 24–30. [14] S. Ruppi, Deposition, microstructure and properties of texture-controlled CVD αAl2O3 coatings, Int. J. Refract. Met. Hard Mater. 23 (4–6) (2005) 306–316. [15] S. Canovic, et al., CVD TiC/alumina and TiN/alumina multilayer coatings grown on sapphire single crystals, Int. J. Refract. Met. Hard Mater. 28 (2) (2010) 163–173. [16] S. Canovic, B. Ljungberg, M. Halvarsson, CVD TiC/alumina multilayer coatings grown on sapphire single crystals, Micron 42 (8) (2011) 808–818. [17] P.W. Trimby, Orientation mapping of nanostructured materials using transmission Kikuchi diffraction in the scanning electron microscope, Ultramicroscopy 120 (2012) 16–24. [18] S. Ruppi, Enhanced performance of α-Al2O3 coatings by control of crystal orientation, Surf. Coat. Technol. 202 (17) (2008) 4257–4269. [19] S. Ruppi, A. Larsson, A. Flink, Nanoindentation hardness, texture and microstructure of α-Al2O3 and κ-Al2O3 coatings, Thin Solid Films 516 (18) (2008) 5959–5966. [20] R. M'Saoubi, S. Ruppi, Wear and thermal behaviour of CVD α-Al2O3 and MTCVD Ti (C,N) coatings during machining, CIRP Ann. 58 (1) (2009) 57–60. [21] R. M’Saoubi, et al., Microstructure and wear mechanisms of texture-controlled CVD α-Al2O3 coatings, Wear 376–377 (2017) 1766–1778. [22] N. Mortazavi, M. Esmaily, M. Halvarsson, The capability of transmission Kikuchi diffraction technique for characterizing nano-grained oxide scales formed on a FeCrAl stainless steel, Mater. Lett. 147 (2015) 42–45. [23] N. Mortazavi, et al., Interplay of water and reactive elements in oxidation of alumina-forming alloys, Nat. Mater. 17 (7) (2018) 610–617. [24] U. Rolander, H.-O. Andrén, Atom probe microanalysis of the γ phase in cemented carbide materials, Mater. Sci. Eng. A 105-106 (1988) 283–287. [25] A. Henjered, et al., Quantitative microanalysis of carbide/carbide interfaces in WC–co-base cemented carbides, Mater. Sci. Technol. 2 (8) (1986) 847–855. [26] M. Christensen, S. Dudiy, G. Wahnström, First-principles simulations of metalceramic interface adhesion: Co/WC versus Co/TiC, Phys. Rev. B 65 (4) (2002) 45408.
5. Conclusions This study investigated the interfacial microstructure and texture development in three textured CVD α-Al2O3 coatings using XRD, STEM/EDX and SEM/TKD. The following conclusions were made using the presented results:
• Diffusion of W and Co occurred through the TiN and Ti(C,N) layers • • • • •
all the way to the bonding layer/α-Al2O3 interface. It was suggested that this might disturb the nucleation and early growth of the alumina layer. Interfacial porosity at the bonding layer/α-Al2O3 interface was observed. A similar number of voids was present in the (0001) and (011¯0) -textured coatings when H2S was not used during the initial deposition step. When H2S was present during the nucleation step of the α-Al2O3 in the deposition process, a higher fraction of voids (almost double the amount) was introduced at the interface of the (0001)-textured sample. The alumina grain morphology developed from small (~100 nm) equiaxed grains at the start/bottom of the coating to larger (several microns) columnar grains at the top of the coating. The inner part of the alumina coatings exhibited grains with a more random orientation. The texture changed either: (i) gradually over several grains, or (ii) more abruptly from one grain to another; from more random orientations to the desired texture. The desired texture develops earlier with H2S for the (0001)-textured coatings.
9