- Email: [email protected]
Examination of wear damage to rock-mining hardmetal drill bits
Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Contents lists available at ScienceDirect
Int. Journal of Refractory Metals and Hard Materials journal homepage: www.elsevier.com/locate/IJRMHM
Examination of wear damage to rock-mining hardmetal drill bits H.G. Jones a,⁎, S.M. Norgren b,c, M. Kritikos b, K.P. Mingard a, M.G. Gee a a b c
National Physical Laboratory, Hampton Road, Teddington, Middlesex TW11 0LW, UK Sandvik Rock Tools, Sandvik Coromant, Lerkrogsvägen 13-19, 12680 Stockholm, Sweden Ångström Tribomaterials Group, Applied Materials Science, Uppsala University, 75121 Uppsala, Sweden
a r t i c l e
i n f o
Article history: Received 12 July 2016 Received in revised form 24 November 2016 Accepted 30 January 2017 Available online 02 February 2017 Keywords: Rock drilling Impact wear Hardmetals Tribochemistry X-ray diffraction FIBSEM
a b s t r a c t WC/Co mining bits from a drill head used for drilling holes for roof support bolts in a mine were examined using a focused ion beam scanning electron microscope (FIB-SEM). This was combined with energy dispersive X-ray spectroscopy (EDX) and X-ray diffraction (XRD) analyses to study the chemical interaction between the drill bit and the rock. It was found that at the surface of the buttons there was depletion of cobalt, change in chemistry of the remaining binder regions, and changes to the morphology of the WC grains. Tribochemistry calculations were done to understand the possible formation of silicides at the surface of the drill bits, and thus emphasise the importance of quartz content in rock on wear. The evidence of mechanical damage combined with chemical reactions is another step towards understanding the complete wear process in hardmetal mining tools. © 2017 Elsevier Ltd. All rights reserved.
1. Introduction In the mining and tunnelling sector it is common practice among mining engineers, geologist and tool manufacturers to predict tool wear by to relating it to the equivalent amount of quartz in the rock [1,2]. In the case of the steel tools, this is understandable since quartz is the hardest mineral and will thus contribute most to abrasive wear. However, for hardmetal tools composed of WC/Co, this is somewhat surprising. Although the quartz is harder than the metallic cobalt binder phase, this is only a minor fraction (~ 10 vol%) of the composite. The major fraction is the hard WC phase with a hardness of HV1300 for the prismatic plane and HV2300 for the basal plane [3], compared to ~ HV1000 for quartz [1]. It was also shown by Gant et al. [4], comparing abrasion from SiO2 and Al2O3 particles, that SiO2 do not significantly abrade or fracture WC crystallites in the WC/Co hardmetals. In this work, possible reasons are described why quartz or silica content is so important in predicting the tool wear, including thermodynamic calculations for the tribochemical interactions that could occur between the tool and the rock. Few papers study actual wear mechanisms of drill bit inserts directly taken from the mine [5,6] or application [7], rather they attempt to simulate the conditions in laboratory tests [4, 8,9,10,15,16,17] Typical degradation mechanisms of hardmetals described in the literature [7–17] include loss of binder phase, crushing and fragmentation of tungsten carbide (WC) grains, as well as removal of fragments and unsupported grains. Beste [5,8,9,10] found rock material adhered to ⁎ Corresponding author. E-mail address: [email protected] (H.G. Jones).
http://dx.doi.org/10.1016/j.ijrmhm.2017.01.013 0263-4368/© 2017 Elsevier Ltd. All rights reserved.
the binder phase, which formed a binder-rock mixture that locally was found to penetrate several hundred micrometres below the surface. He also found oxidation of WC grains and subsequent scraping away of the binder-rock layers. Stjernberg et al. [14] found cracks reaching ~100 μm below the worn surface, caused by large thermal fluctuations which had caused tensile stresses in the surface region that opened cracks perpendicular to the surface. In this work, the aim was to gain a better understanding of the tribochemical wear mechanisms that lead to damage during top hammer drilling (where many of the cited works focused mainly on the mechanical wear mechanisms), and to provide a contribution to understand the chemical driving forces to the wear that contribute further to that previously reported. Improved understanding will help to determine limits on the maximum life of the bit during drilling. A forensic examination was conducted on several of the WC/Co buttons. A focused ion beam and scanning electron microscope (FIB-SEM), energy dispersive X-ray spectroscopy (EDX) and X-ray diffraction (XRD) were used to study the mechanical damage and chemical interactions in the WC/Co buttons. The observations were compared to thermodynamic calculations of the silica (SiO2)–Co-WC system, where investigation of these reactions gave a clearer insight into the cause of the wear mechanisms observed in the buttons. It is shown that the Co in the hardmetal reacts with the quartz and forms CoSi2, enhancing the wear of the bit. 2. Experimental method The 43 mm drill head from Sandvik Mining Rock Tools used in the analysis is shown in Fig. 1. The button geometry was GT7S100A-XT48
2
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
mounted with adhesive tape to the sample holder and XRD patterns were measured at several positions. 3. Results Multiple wear mechanisms took place at the surface and just beneath it in the regions examined by FIB cross-sectioning and EDX. Each type of phenomenon is described in the following sections. 3.1. Surface composition
Fig. 1. The top 10 mm of the steel drill head with embedded WC/Co buttons numbered for reference. The rotating direction of the drill is anti-clockwise, as seen from this view.
thus fully spherical and 10 mm in diameter. The button composition was 10 vol% cobalt WC hardmetal with a hardness (HV30) of 1450, grain size of 1.2 μm [18], coercivity of 11.5 kAm−1 and relative weight specific magnetic saturation (compared to pure cobalt) of 0.90. The drill head was used for drilling holes for roof support bolts in a zinc and lead mine. To examine the buttons, the top 10 mm of the drill head with the seven buttons was cut off (Fig. 1). It was mounted on a suitable holder with carbon glue for examination in the microscope. The buttons were numbered, where 2 and 5 are known as front buttons and the rest as gauge buttons. A Carl Zeiss Auriga 60 FIB-SEM equipped with an Oxford Instruments X-Max EDX detector was used for microstructural characterisation. Due to the large depth of field of the electron beam, it was straightforward to locate and look at features on the domed buttons. Cross-sections were made on the buttons using the FIB with 30 kV: 1– 4 nA beam conditions, and imaged at high resolution with the SEM at 3 kV accelerating voltage. Elemental mapping and line scans of the surface and some of the cross sections were carried out with EDX at higher accelerating voltages to gain understanding of the chemistry of the wear processes. The cross-sections were made randomly on the worn surface (“wear flats”) of the buttons, but as the sample was tilted in the FIB-SEM for analysis and the geometry of the buttons is domed, it was only possible to view and mill sections on certain portions of each button wear flat. As the position of the head remained the same throughout the course of the experiment (tilted along the line of symmetry through button 7, with buttons 1 to 3 uppermost), points on the surfaces nearer to the right of the buttons as viewed in Fig. 1 were most likely to have been analysed (as indicated approximately by arrows in Fig. 1). In some cases, e.g. Fig. 5, where the FIB curtaining is seen at an angle, then the cross-section was further down the side of the button. The FIB-SEM and EDX results shown in this paper are on buttons 1 and 4, which are gauge buttons, but similar surface characteristics and mechanisms were observed on the other buttons, including button 2, a front button. XRD measurements were performed on a Bruker Discover D8 diffractometer with Davinci design equipped with a IμS Microfocus Source (CuKα radiation, λ = 1.5418 Å), an Eulerian cradle and a Våntec-500 2D area detector. A laser-video positioning system was used for alignment of the sample. The XRD patterns were analysed with software DIFFRAC EVA (Bruker) and High Score Plus (PANalytical). XRD data were typically collected in the angular range 10° b 2θ b 140°. The drill bit was
Low magnification SEM imaging of the surface of the top of a gauge button (labelled 1 in Fig. 1) showed the surface texture to be rough (compared to a polished or even fine-ground surface) with evidence of rock debris adhered to the surface (Fig. 2a). EDX analysis at 15 kV accelerating voltage of the area in Fig. 2a revealed the surface composition on a larger scale and produced a spectrum showing the presence of rock constituents including carbon, oxygen, calcium, iron, aluminium, silicon and iron, as well as zinc and lead (Fig. 2b). Oxygen and tungsten were the most abundant elements on the surface. By contrast, little cobalt was detected, despite this being a 10 vol% cobalt grade hardmetal. Distinct regions of lead and zinc, as well as oxygen and silicon from rock debris, were identified in elemental EDX maps, which were dragged across and embedded in the button surface during rotation of the drill head and this directionality is observed in Fig. 2c. Other buttons with similar surface composition and mapped at a similar magnification did not show this smearing as defined as seen in Fig. 2c. At higher magnifications however, the button surfaces vary considerably in roughness and composition, as described in the sections below. 3.2. Carbide wear and morphology Fig. 3a shows the position of the FIB-section on the wear flat of button 1 and Fig. 3b–d show it in higher magnification. Fig. 4 shows the same but on another position on button 1. Fig. 5 shows a milled section of button 4. In Figs. 3 and 4 there was surprisingly little cracking of the carbide grains below the wear surface (Figs. 3b and 4b), however some intergranular fracture was observed just below the top layer of carbide grains (Fig. 3c). Transgranular fracture was observed in several carbide grains where there were fine cracks through crushed and re-adhered carbide fragments at the surface (Fig. 4c) and next to binder regions where there was depletion and porosity in the binder phase (Fig. 5). Fragmentation of the carbide grains also occurred at the surface and the crushed carbide particles were re-embedded in the surface layer and mixed with rock fragments and extruded cobalt (Figs. 3b and 4c). There was deformation of the top layer of carbide grains such that they had lost their facets and the boundaries between carbide grains were curved. This is shown in Fig. 4c where larger fragments have been compressed together and the boundaries are deformed, and in Fig. 3b and d down to ~10 μm from the surface, the original boundaries are barely visible. The vertical streaks and obvious “waviness” in the images are unfortunate artefacts of the FIB milling, exacerbated by the uneven surface due to omission of using a protective layer in these experiments, but the features described above are real features in the material. 3.3. Binder morphology and composition There was significant compression of the cobalt regions between the surface and up to 10 μm in depth as revealed by FIB (Figs. 3 and 4), where cobalt binder regions of the expected size have disappeared and large number of crushed WC grains and forced-together WC-WC boundaries are present. The long thin binder region in Fig. 4d has at the top a dark region indicated by the arrow containing adhered
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
3
Fig. 2. a) Secondary electron image of the surface on the top of a gauge button (labelled number 1), b) EDX spectra with elemental peaks labelled, and c) EDX at% elemental maps of oxygen, tungsten, silicon, aluminium, lead and zinc of the whole area. White and black pixels indicate 0 at% and 100 at% respectively.
material from the rock which was probably forced into a hole created by micro-cracking of WC grains. However, larger cobalt binder regions were still present near the surface in many regions as seen in Figs. 4b and 5a. In this case, there was a surface oxide layer present, rather than exposed WC grains, as seen in Fig. 3b. The variation in thickness and composition of the surface oxide layer was likely to have caused the difference in wear mechanisms observed underneath. There was often a gradient or image contrast in the cobalt regions. In Fig. 4, the binder regions appear darker further from the surface, and were very pale between fragmented carbides (Fig. 3c), indicating possible composition variation within single binder regions. This contrast is exaggerated by the shadowing caused by the geometry of the FIB section, however is likely a real effect as it is quite dramatic in the image such as Fig. 4c underneath the carbide fragments. The W solubility in Co increases with temperature so the brighter contrast of the binder regions closer to the surface maybe an indication a temperature gradient. In Fig. 5, the cobalt binder regions showed image contrast where a dark border b100 nm in thickness was observed at the binder-carbide boundary (Fig. 5b). In these regions, there was also delamination of the cobalt at the carbide-binder interfaces, leaving “ligaments” connecting the binder phase to adjacent carbide grains. 3.4. Oxide surface layer and delamination As observed with EDX in Fig. 2 the surface of the buttons showed abundant oxygen, and the cross-sections often showed a surface layer above the hardmetal structure. This layer consisted of mostly oxygen, silicon, calcium and aluminium, but varied in elemental proportions in
different positions, with carbide fragments and cobalt often embedded in these features. EDX showed that the surface layer in Figs. 3 and 4 had a surface layer consisting of more silica than calcium oxide and alumina, and in Fig. 5, contained more calcium oxide than silica and alumina. The structure of the oxide layer also varied, where Figs. 3 and 4 show a “glossy” surface (silica) and Figs. 5 and 6 show a thicker more structured oxide (calcium oxide) that also exhibited some porosity and fracture. Approximately 20 μm from the site of Fig. 5 a layer of oxide was observed to be spalling from the surface; this is shown in Fig. 7. On the outer part of the wear flat of the button in Fig. 6 (pointing away from the centre of the drill head in Fig. 1), there was a large raised feature, measuring approximately 1 mm in length (Fig. 6a). At higher magnification, a relief feature was identified and an EDX map was acquired to include it (Fig. 6c). The area contained a high proportion of calcium oxide, with silica as the next abundant oxide and small amounts of other elements (Fig. 6b). On closer examination of the relief feature, an elevated “plate” or spall was observed ~100 μm in width, which appeared to stand proud of the rest of the surface by several micrometres (Fig. 7). A cross-section was made on the edge to reveal its internal structure. There was a top layer of darker material between 2 and 4 μm in thickness, and a higher brightness banded layer underneath with a thickness of 1–2 μm (Fig. 7c). At higher magnification, even thinner bands were revealed, where the thinnest layers were b50 nm in thickness (Fig. 7d). EDX mapping and a line scan of the cross-section showed the upper part of the section contained mostly calcium (oxide) and the lower layered part was tungsten-rich, but also contained cobalt and oxygen (Figs. 8 and 9). The upper layer had a gradient of tungsten and other elements
4
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 3. Secondary electron images of a region, on gauge button labelled 1, where the surface appeared roughened with exposed carbide grains embedded in a low contrast material, which was a mixture of crushed rock, crushed carbide grains and cobalt. a) Whole FIB-milled section and surface topography, b) full width of milled cross-section, c) fracture under a carbide grain at the surface, d) deformation of carbide boundaries and compression of cobalt regions.
Fig. 4. Secondary electron images of a region, on gauge button labelled 1, where the surface appears smoother and non-metallic due to charging of the electron beam. a) Whole FIB-milled section and surface topography, b) full width of milled cross-section, c) fragmentation of carbide grains, d) further milled section revealing compressed binder regions with variation in imaging contrast in the cobalt.
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
5
Fig. 5. Secondary electron images of a cross-section, on gauge button labelled 4, in a region with surface oxide layer of ~1 μm. a) Whole milled cross-section, b) binder regions with dark contrast at the edges and porosity between them and the carbide phase, c) porosity in oxide layer, transgranular fracture in the carbide and rock material in the binder phase.
shown in Fig. 9. Interestingly, cobalt was only detected in the layered structure, however there were a few discrete regions underneath the plate (Fig. 8), presumably where they were protected by the plate above. The line scan gave quantitative results of the variation of composition across the cross-section (Fig. 9). There were small amounts of other elements, such as lead, zinc, sulphur and potassium, but were not plotted due to little variation along the whole line. There were high amounts of silicon, calcium and oxygen on the surface (below
2 μm on graph in Fig. 9). At the edge of the cross-section (~2 μm), the calcium content increased, oxygen decreased, and silicon and tungsten increased slightly. At the boundary with the bright layers (~6 μm), tungsten increased quickly, as well as a smaller rise in cobalt. There was a peak in these two elements at 7 μm, which was the brightest layer in the secondary electron images (Fig. 7d). Gallium was also detected, particularly in the layers, which was an ion milling artefact. Although similar spalls were not observed on the other buttons, there were other regions on the other buttons where the oxide layer
Fig. 6. Analysis of a relief feature on the worn side of a gauge button (labelled 4). a) Low magnification secondary electron image of button, b) EDX spectra of the area boxed in Panel a, c) EDX at% elemental maps of tungsten, calcium, silicon and oxygen present in the area boxed in Panel a; white and black pixels indicate 0 at% and 100 at% respectively.
6
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 7. Secondary electron images of a relief feature on button 4 seen in Fig. 6. a) The whole lifted “plate” or spall, measuring ~100 μm in width, b) underneath and edge of spall, where crosssection was subsequently made, c) cross-section on the edge of the plate, positioned in the middle of what is shown in Panel b; the arrow shows the direction of the EDX line scan, d) area shown in the box in Panel c of layered structure on the underside of the spall (vertical streaks are an artefact of the ion beam milling).
had begun to lift and a similar layered structure had started to form underneath, as in Fig. 7d; including a front button (specifically, button labelled 2 in Fig. 1). The oxygen to metal ratios in Fig. 9b show that above the layered structure, there are high oxygen levels compared with the metallic phases that reduces with increasing depth. At the boundary to the layered structure, the oxygen and tungsten content is equal and then tungsten becomes more abundant throughout the layers. The oxygen to cobalt ratio remains above 1 but is much lower than in the bulk of the cross section. The oxygen to oxide-constituents ratio also decreases with increasing depth, however there is 10× more oxygen rather than 100× more compared to the metals. At the boundary to the layers, the calcium becomes more abundant and then both silicon and calcium reduce. This confirms that the layers contain mostly metallic elements and the bulk cross-section contain oxides from the quartz. However, higher resolution techniques than EDX in the SEM would be required to reveal the exact change in composition. 3.5. XRD The surface of drill bit number 3 was investigated by 2D XRD at several positions. The integrated 1D diffraction patterns showed strong reflections from the WC substrate, however at other positions on the drill
bit, other crystalline phases were present. The peak intensities of these phases were very low compared to substrate peak intensities, thus indicating that small amounts were present. The positions of the reflections (Fig. 10) agree well with the peak positions calculated for crystalline αSiO2 and disilicide CoSi2, respectively. Several peaks belonging to the SiO2 phase were identified in the experimental XRD patterns. For CoSi2 only (111) was observed since the (220) peak at 48.0° 2θ is completely shadowed by the strong (101) reflection of WC. Both Si-containing phases could be observed at several places on the drill bit. 4. Discussion 4.1. Analysis Focused ion beam and electron imaging characterisation of drill buttons from a real application provided a new insight into the interaction of rock material with the hardmetal during use in mining. A mixture of abrasion and oxidation took place, where evidence of chemical reactions, with the likely presence of water vapour, was observed in addition to mechanical wear. In many areas on the buttons, the rock material or rock-hardmetal product had formed a layer over large areas of the surface. At low magnification, the surface showed compositional variance indicating rock
Fig. 8. EDX at% element maps of tungsten, cobalt and calcium of the entire width of the cross-section on the edge of the spall in Fig. 7 (the centre portion is shown in Fig. 7c); white and black pixels indicate 0 at% and 100 at% respectively.
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
7
Fig. 9. EDX elemental line scans of surface and milled cross-section on relief feature (arrow in Fig. 7c). a) The most abundant elements are shown in atomic percent along the full length of the line. b) Oxygen to metal ratios along cross-section and c) oxygen to oxide-constituents ratios, not including data from top surface (first ~2 μm of line scan).
material was dragged across the surface and embedded into the surface, as well as showing lead and zinc from the ore. This is in agreement with the findings of Beste and Jacobsson [9] but this does not explain why wear is related to the amount of quartz in the rock. At high magnifications, distinct regions with a certain surface characteristics and composition were observed, where the oxide layer had formed with different elemental proportions and at different thicknesses. This large variation indicates that the temperature and abrasive conditions the drill underwent greatly varied over single buttons as well as between individual buttons. The carbide and binder phases both exhibited wear damage, but with characteristics which appeared to differ according to the surface oxide layer properties. Firstly, where there was intergranular fracture between carbide grains, or binder regions that had been compressed or removed, and deformation of carbide boundaries, the surface oxide layer was thinner and silicon rich. Some carbide grains were also exposed at the surface, where the surface layer was continually being removed during drilling. Here carbide grains fractured where they were always exposed to high temperatures and pressures. In addition, as there was little protection by surface oxides to absorb some of the impact force, hence causing the binder regions to compress and carbide grains to become rounded due to fracture of facets at high temperature, as seen in Fig. 4c. The absence of the binder phase in the outermost surface layers of the inserts has also been reported by Beste [5,8,9,10] and Larssen-Basse [12]. The carbide grains were squeezed towards each other, causing extrusion of the binder from between them. Where the carbide grains were forced together under compression, intergranular cracks formed. The compressed regions that appeared reached 2–6 μm depth in some places. In the second case, where binder regions were present closer the surface, greater transgranular fracture in the carbide took place, with separation between phases at the carbide-binder boundaries; here the surface oxide layer was much thicker and calcium rich. It was more likely to have acted as a protecting layer, resulting in less compression of the binder regions than without the thicker oxide layer. In particular, the delamination of cobalt from carbide suggests compression followed by expansion, possibly creating a channel for surface material to mix with the binder phase, causing the lower contrast borders in Fig. 5b. This in turn created voids at the boundary where the binder phase was left only partially attached to neighbouring carbide grains. There were more fragments of WC grains near the surface of the button, which embedded in the binder phase, getting less dense further into the button. Hence the cobalt regions appear lighter (high electron
imaging contrast) near the surface and darker below. It is mostly, if not only, seen in cobalt regions that have a channel to the surface where fragments can travel down from the surface, or at the edges of cracked WC grains. Being open to the surface, these channels could also be means through which the cobalt escaped. Cobalt was often detected with EDX at the surface of the hardmetal underneath the oxide layer (most abundantly at the left side of the cross-section in Fig. 5a). This also corroborates the theory that the cobalt is squeezed out from between the carbide grains, gets trapped beneath the oxide layer, but is eventually removed from the surface, also suggested by Beste et al. [5,8,9,10]. The lateral fracture in the oxide layer occurred above this cobalt-rich region and underneath the oxide layer itself (Fig. 5a); it is possible that this type of feature lead to the delamination of the surface seen in Fig. 6. Another mechanism that contributed to the binder-free outer surface is corrosion, as cobalt alloys easily corrode in hot water and/or acidic conditions. In the work by Beste [5,8,9,10], the cobalt binder is absent at much larger depths; however, they used a different sampling method in which the buttons were cracked to find the weakest region, whereas in this work the wear surface was affected very little, by using the FIB to make sections at local areas on the button surface. Montgomery [11] proposed different wear mechanisms for the gauge buttons compared to the front buttons (labelled 2 and 5 in Fig. 1). He stated sliding-abrasive wear was the predominant wear on the gauge buttons and that the face buttons wear by micro-chipping of small wear fragments, typically of 20 μm in size. Montgomery defined the predominant wear mechanisms on the face buttons as microspalling of the surface caused by stresses related to the blows of the bit on the rock. The fragments in Fig. 7 were observed on the wear flat and resemble the size and appearance of the microspalls despite being taken from a gauge button. This indicates that this type of wear was also present on the gauge buttons. The spall in Fig. 7 could have formed due to specific localised heating or a particular wear motion, due to chemical interaction between the quartz from the rock and the hardmetal. It is possible that after the surface oxide layer had reached a certain thickness, its adherence to the hardmetal button beneath weakened and subsequently loosened from the surface. The layers on the underside of the spall may have stuck to the surface at different stages of the drilling. This region underneath the adhered rock is thus no longer in contact with the air and the chemical interactions are discussed in Section 4.3. It is likely that this would be more common in areas that undergo slower, gradual wear rather than a sudden, high rate wear, which could account for the layering of
8
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 10. XRD patterns, and positions on button indicated by the laser point at the centre of each photo. a) CoSi2 (111) peak at 29.3° 2θ is indicated, all unmarked peaks are from the substrate. b) SiO2 (101),(011) peaks at 26.7° and the CoSi2 (111) peak at 28.9° 2θ. c) Several SiO2 peaks: (100) at 20.9°, (101),(011) at 26.7°, (102),(012) at 39.5° and (111),(11−1) at 40.3°; some smaller peaks remained unidentified.
material with varying composition, and that a similar phenomenon was observed on another button, which could be the same process but at an earlier stage. 4.2. Tribochemistry It is verified by many authors that adhered rock material is detected on the bit surface after drilling. However, no studies to date have considered the tribochemical interaction between the adhered rock and the hard metal bit. Drilling with air or water flushing means the hardmetal is in an oxidising atmosphere; thus the reaction: WC + Co + O2 → WyOx + CoyOx will occur at the unprotected surface. In the areas where the rock adheres to the surface, the following process will also occur at the interface between the quartz in the adhered rock material (here simplified to silica) and the hardmetal button: WC + Co + SiO2 (in rock) → WC + SiO2 + SiC + W-phases + Wand Co-silicides. Since there is an abundance of carbon from the WC, the silica will be reduced to and react with the Co binder, forming Co-silicides. According
to Lavoie et al. [19], CoSi2 forms readily at 620 °C and WSi2 at 100 °C higher temperature. Consulting the binary Co-Si phase diagram (Fig. 11), the first silicide to form on the Si side is the CoSi2, which was also detected by XRD. The W-phases formed could be tungsten oxides as also observed by Beste [5,8,9,10] or the W-Co intermetallic phases such as μ-phase (Co7W6) if there is enough Co left, or W-silicides if there is abundance of Si, although none of these additional phases were detected by XRD here. To further investigate the chemical driving forces a calculation was carried out at 800 °C on the WC-Co-SiO2 system, starting from a molar fraction, of X(Si) = 0.3, with increasing cobalt and decreasing silicon (Fig. 12). It is always difficult to know the exact temperature in rock drilling, but according to Heinö [1] it is between 600 and 2000 °C. The temperature of 800 °C was selected here based on this work and the work by Östberg [21], which correlated plastic deformation of the WC grains in the temperature range of 800–1200 °C. This is to be regarded as a guide since the system is not at equilibrium; despite this, there is a driving force to form silicon carbide and Co-silicides, where Si and C are abundant. CoSi2 would form where there was Co close to the
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 11. The Si-Co binary system - temperature versus at% Co [20].
interface between the hardmetal and the adhered rock, with greatest concentration closest to the surface, but penetrating below the surface with the gradient in Si concentration diffusing below the hardmetal surface. The thermodynamics of the system shows that a reaction is to be expected between the lower surface of the adhered rock material where it meets the underlying hardmetal. Hence, the bright layers in Fig. 7 are probably reaction products between the adhered rock and the hardmetal consisting of Co- and W- silicides. If there is an abundance of SiO2 originating from the quartz in the rock and carbon from the WC, quartz will be reduced locally in these surface regions: SiO2 (excess) + WC + Co → SiO2 + SiC + WSi2 + CoSi2. Cobalt is only available through patches of binder layers between the WC grains but this is sufficient to form CoSi2, confirmed by XRD at several places on the wear flat. These reactions suggest tool wear is partly a tribochemical process due to the interactions between the quartz (SiO2) and the rock and it explains why the tool wear can be related to the quartz content in the rock, despite that SiO2 being softer than the WC phase. Approximately 1 μm below the surface of the oxygen-rich tribolayer in Fig. 5a, a cobalt pocket, shown in Fig. 5b, was observed. The Co was close to the surface and had a dark rim towards the outer edge, adjacent to WC grains. This could not result from alloying with W as that would give a whiter contrast; a plausible explanation is that this zone is enriched by a lighter element such as O or Si. However, the solubility of O in Co is low, but Si in Co is high, thus most likely this is a Si enriched zone where Si is supplied by grain boundary diffusion from the outer
9
more Si rich part. This needs to further investigation with transmission electron microscopy (TEM) for confirmation. Figs. 4d and 5c suggested increasing cobalt content in the binder and decreasing content of other elements, i.e. tungsten, Ca/Si surface oxides, with increasing depth beneath the surface. Where the binder was exposed at the surface, the oxide layer/silicon diffused into the binder but stopped at the carbide interface. In an attempt to understand this, the binary Co-Si phase diagram in Fig. 12 shows silicon alloys are still present in the binder containing N20 at% cobalt, decreasing the melting point drastically. Hence the exposed regions might be in fact Si-rich cobalt binder. This may also explain the regions in Fig. 5 where the binder appeared to melt and re-solidify, giving rise to voids at the phase/interphases, which no longer support the adjacent WC grains as they wear away. This relates to the observation of Beste et al. that the WC grains are removed one by one from the surface and their observation of a melted structure. 4.3. Experimental limitations The use of the FIB technique in the work where very specific regions were targeted, enabled viewing of the top surface and cross-section simultaneously. This meant the FIB milling could be stopped precisely to view an exact feature in cross-section. In order for this to be possible, a deposited protective layer was omitted from the experiment. This unfortunately meant more milling artefacts were introduced in the crosssections. Further work would include sections being made with a protective layer to ensure better-controlled milling conditions, and compared with the existing results to remove the ambiguity regarding such artefacts. A limitation of the EDX technique, especially on this material, is the X-ray peak overlap between tungsten Mα and silicon Kα peaks. However, the analysis software was set to interpret the concentrations using different tungsten peaks to distinguish between the elements and represented the final results in atomic percentage to overcome this ambiguity. When EDX maps were taken of a cross-section, the interaction volume limited the resolution of the element identification at given point. The interaction volume would have been of the order of 1 μm in depth beneath the visible surface, therefore detecting other elements than those on the surface. In the case of identifying the chemistry of the darker areas in the binder regions (Fig. 5) using EDX, compositional variance due to high temperature and/or pressure in these regions was
Fig. 12. Mole fraction of phases (NPM) versus increasing cobalt/decreasing silicon content, starting from the composition (at%)X(C) = 0.1, X(Co) = 0.725, X(O) = 5E-2, X(Si) = 2.5E-2, X(W) = 0.1,. Thus WC-Co-SiO2 (left side) and exchanging Co for Si, calculated at 800 °C and 1 bar calculated using the Thermocalc software SSUB5 database [22].
10
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
not possible to detect with confidence due to overlap with carbide grains behind. A lamella containing this region would need to be removed from the bulk sample and analysed separately to get more accurate EDX data without the interaction volume. Ideally transmission electron microscopy (TEM) would also be done on the lamella to confirm the phases with greater certainty. Following on from the limitations of the techniques used in the work in Section 4.2, a number of further experiments would complement and corroborate the existing results. Sectioning the buttons transversely and polishing would confirm if there was deeper rock penetration into the hardmetal. Using the lamella lift-out technique, as previously mentioned, with electron backscatter diffraction (EBSD) analysis could be done to examine plastic deformation at the surface, at compressed carbide boundaries, and surrounding fractures. 5. Conclusion A study was conducted on WC/Co rock drill bits. FIB-SEM microscopy and XRD with consideration of metallurgical thermodynamics was used to lead to the following conclusions in regard to the wear processes taking place. Large variation in subsurface damage was observed on single buttons as well as between different buttons, where there was either exposed carbide grains or a protective oxide layer of varying thickness and composition on the surface. The high temperatures and pressures during drilling had different effects on the hardmetal microstructure depending on these surface characteristics. EDX analysis showed the oxide layer usually consisted mostly of silicon or calcium, with some tungsten (from exposed grains), and no cobalt. The cobalt tended to get squeezed out at the surface (where there was not an oxide protective layer) which would close up any obvious channels. The depth of damage found was only within the top few microns of the impact surface; despite the FIB-SEM cross-sectioning technique only being able to analysis ~15 μm, there was no evidence of channels that rock could implant into and reach much lower depths. By using a 2D-XRD detector and a microfocus X-ray source it was possible to analyse small surface regions of the drill bits. The X-ray powder patterns showed the presence of crystalline α-SiO2 and cobaltdisillicide CoSi2. These results agree with thermodynamic calculations which show that quartz (SiO2) in the rock can be reduced by the presence of C from the WC in the hardmetal and leads to the formation of Co silicides. The formation of CoSi2 by diffusion through Co below the drill bit surface and local melting of this phase removes support from the surrounding WC grains and accelerates wear by fracture and removal of WC fragments. Thus the determination of the wear rate by the equivalent amount of quartz in a rock may be related to formation of Co silicides rather than the abrasive nature of the rock itself. Formation of crystalline α-SiO2 and disilicide CoSi2. These results indicate that the outcome of tribochemical reactions under non-model conditions can be studied by X-ray diffraction methods.
Acknowledgements The authors would like to thank Sandvik for the supply of the worn drill bit and for the UK Government Department for Business, Innovation and Skills for partial funding of this work under the Materials Programme of the National Measurement System. References [1] M. Heiniö, Rock Excavation Handbook, Editor in Chief, Sandvik-Tamrock, 1999. [2] K. Thuro, Drillability prediction - geological influences in hard rock drill and blast tunnelling, Int. J. Earth Sci. 86 (2) (1997) 426–438. [3] D.N. French, D.A. Thomas, in: F.V. Vahldiek, J. Mersol (Eds.), Anisotropy in Single Crystal Refractory Compounds, 1, Plenum New York, 1968. [4] A.J. Gant, M.G. Gee, Abrasion of tungsten carbide hardmetals using hard counterfaces, Int. J. Refract. Met. Hard Mater. 24 (2006) 189–198. [5] U. Beste, T. Hartzell, H. Engqvist, N. Axén, Surface damage on cemented carbide rock-drill buttons, Wear 249 (2001) 324–329. [6] M. Olsson, K. Yvell, J. Heinrichs, M. Bengtsson, S. Jacobson, Surface degradation mechanisms of cemented carbide drill buttons exposed to iron ore rock drilling, The 17th Nordic Symposium on Tribology June 2016, pp. 14–17 (Hämeenlinna, Finland). [7] S. Momeni, S. Moseley, M. Ante, J. Allaart, The Wear of WC-Co Drill Bits During Rotary-Percussive Drilling of Reinforced Concrete, Submitted to International Journal of Refractory Metals and Hard Materials, 2016. [8] U. Beste, E. Coronel, S. Jacobson, Wear induced material modifications of cemented carbide rock drill buttons, Int. J. Refract. Met. Hard Mater. 24 (2006) 168–176. [9] U. Beste, S. Jacobson, A new view of the deterioration and wear of WC/Co cemented carbide rock drill buttons, Wear 264 (11–12) (2008) 1129–1141. [10] U. Beste, S. Jacobson, S. Hogmark, Rock penetration into cemented carbide drill buttons during rock drilling, Wear 264 (11–12) (2008) 1142–1151. [11] R.S. Montgomery, The mechanism of percussive wear on tungsten carbide composites, Wear 12 (1968) 309–329. [12] J. Larsen-Basse, Binder extrusion in sliding wear of WC-Co alloys, Wear 105 (1985) 247–256. [13] I. Konyashin, B. Ries, Wear damage of cemented carbides with different combinations of WC mean grain size on Co content. Part II: laboratory performance tests on rock cutting and drilling, Int. J. Refract. Met. Hard Mater. 45 (2014) 230–237. [14] K.G. Stjernberg, U. Fisher, N.I. Hugoson, Wear mechanisms due to different rock drilling conditions, Powder Metall. 18 (1975) 89–106. [15] H.G. Jones, K.P. Mingard, J. Zunega, M.G. Gee, B. Roebuck, Test methods for high rate impact loading of hardmetals, European Congress and Exhibition on Powder Metallurgy. European PM Conference Proceedings, 1–6, 2012. [16] M. Antonov, R. Veinthal, D.-L. Yung, D. Katušin, I. Hussainova, Mapping of inpactabrasive wear performance of WC-Co cemented carbides, Wear 332–333 (2015) 971–978. [17] J. Angseryd, A. From, J. Wallin, S. Jacobson, S. Norgren, On a wear test for rock drill inserts, Wear 301 (2013) 109–115. [18] B. Roebuck, M. Gee, E.G. Bennett, R. Morrell, Good Practice Guide No. 20: Mechanical Tests for Hardmetals, National Physical Laboratory, 1999. [19] C. Lavoie, C. Cabral Jr., F.M. d'Heurle, J. Jordan-Sweet, J.M.E. Harper, T.J. IBM, Watson Research Center, Yorktown Heights, NY, NSLS Activity Report, 2001 2–20. [20] D. Choi, The Co-Si system, Calphad (1992) 151–159. [21] G. Ostberg, K. Buss, M. Christensen, S. Norgren, H.O. Andren, D. Mari, G. Wahnstrom, I. Reineck, Mechanisms of plastic deformation of WC–Co and Ti(C, N)–WC–Co, Int. J. Refract. Met. Hard Mater. 24 (2006) 135–144. [22] Scientific group Thermodata Europe SSUB5 v. 5.1database available from www. thermocalc.se.
Contents lists available at ScienceDirect
Int. Journal of Refractory Metals and Hard Materials journal homepage: www.elsevier.com/locate/IJRMHM
Examination of wear damage to rock-mining hardmetal drill bits H.G. Jones a,⁎, S.M. Norgren b,c, M. Kritikos b, K.P. Mingard a, M.G. Gee a a b c
National Physical Laboratory, Hampton Road, Teddington, Middlesex TW11 0LW, UK Sandvik Rock Tools, Sandvik Coromant, Lerkrogsvägen 13-19, 12680 Stockholm, Sweden Ångström Tribomaterials Group, Applied Materials Science, Uppsala University, 75121 Uppsala, Sweden
a r t i c l e
i n f o
Article history: Received 12 July 2016 Received in revised form 24 November 2016 Accepted 30 January 2017 Available online 02 February 2017 Keywords: Rock drilling Impact wear Hardmetals Tribochemistry X-ray diffraction FIBSEM
a b s t r a c t WC/Co mining bits from a drill head used for drilling holes for roof support bolts in a mine were examined using a focused ion beam scanning electron microscope (FIB-SEM). This was combined with energy dispersive X-ray spectroscopy (EDX) and X-ray diffraction (XRD) analyses to study the chemical interaction between the drill bit and the rock. It was found that at the surface of the buttons there was depletion of cobalt, change in chemistry of the remaining binder regions, and changes to the morphology of the WC grains. Tribochemistry calculations were done to understand the possible formation of silicides at the surface of the drill bits, and thus emphasise the importance of quartz content in rock on wear. The evidence of mechanical damage combined with chemical reactions is another step towards understanding the complete wear process in hardmetal mining tools. © 2017 Elsevier Ltd. All rights reserved.
1. Introduction In the mining and tunnelling sector it is common practice among mining engineers, geologist and tool manufacturers to predict tool wear by to relating it to the equivalent amount of quartz in the rock [1,2]. In the case of the steel tools, this is understandable since quartz is the hardest mineral and will thus contribute most to abrasive wear. However, for hardmetal tools composed of WC/Co, this is somewhat surprising. Although the quartz is harder than the metallic cobalt binder phase, this is only a minor fraction (~ 10 vol%) of the composite. The major fraction is the hard WC phase with a hardness of HV1300 for the prismatic plane and HV2300 for the basal plane [3], compared to ~ HV1000 for quartz [1]. It was also shown by Gant et al. [4], comparing abrasion from SiO2 and Al2O3 particles, that SiO2 do not significantly abrade or fracture WC crystallites in the WC/Co hardmetals. In this work, possible reasons are described why quartz or silica content is so important in predicting the tool wear, including thermodynamic calculations for the tribochemical interactions that could occur between the tool and the rock. Few papers study actual wear mechanisms of drill bit inserts directly taken from the mine [5,6] or application [7], rather they attempt to simulate the conditions in laboratory tests [4, 8,9,10,15,16,17] Typical degradation mechanisms of hardmetals described in the literature [7–17] include loss of binder phase, crushing and fragmentation of tungsten carbide (WC) grains, as well as removal of fragments and unsupported grains. Beste [5,8,9,10] found rock material adhered to ⁎ Corresponding author. E-mail address: [email protected] (H.G. Jones).
http://dx.doi.org/10.1016/j.ijrmhm.2017.01.013 0263-4368/© 2017 Elsevier Ltd. All rights reserved.
the binder phase, which formed a binder-rock mixture that locally was found to penetrate several hundred micrometres below the surface. He also found oxidation of WC grains and subsequent scraping away of the binder-rock layers. Stjernberg et al. [14] found cracks reaching ~100 μm below the worn surface, caused by large thermal fluctuations which had caused tensile stresses in the surface region that opened cracks perpendicular to the surface. In this work, the aim was to gain a better understanding of the tribochemical wear mechanisms that lead to damage during top hammer drilling (where many of the cited works focused mainly on the mechanical wear mechanisms), and to provide a contribution to understand the chemical driving forces to the wear that contribute further to that previously reported. Improved understanding will help to determine limits on the maximum life of the bit during drilling. A forensic examination was conducted on several of the WC/Co buttons. A focused ion beam and scanning electron microscope (FIB-SEM), energy dispersive X-ray spectroscopy (EDX) and X-ray diffraction (XRD) were used to study the mechanical damage and chemical interactions in the WC/Co buttons. The observations were compared to thermodynamic calculations of the silica (SiO2)–Co-WC system, where investigation of these reactions gave a clearer insight into the cause of the wear mechanisms observed in the buttons. It is shown that the Co in the hardmetal reacts with the quartz and forms CoSi2, enhancing the wear of the bit. 2. Experimental method The 43 mm drill head from Sandvik Mining Rock Tools used in the analysis is shown in Fig. 1. The button geometry was GT7S100A-XT48
2
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
mounted with adhesive tape to the sample holder and XRD patterns were measured at several positions. 3. Results Multiple wear mechanisms took place at the surface and just beneath it in the regions examined by FIB cross-sectioning and EDX. Each type of phenomenon is described in the following sections. 3.1. Surface composition
Fig. 1. The top 10 mm of the steel drill head with embedded WC/Co buttons numbered for reference. The rotating direction of the drill is anti-clockwise, as seen from this view.
thus fully spherical and 10 mm in diameter. The button composition was 10 vol% cobalt WC hardmetal with a hardness (HV30) of 1450, grain size of 1.2 μm [18], coercivity of 11.5 kAm−1 and relative weight specific magnetic saturation (compared to pure cobalt) of 0.90. The drill head was used for drilling holes for roof support bolts in a zinc and lead mine. To examine the buttons, the top 10 mm of the drill head with the seven buttons was cut off (Fig. 1). It was mounted on a suitable holder with carbon glue for examination in the microscope. The buttons were numbered, where 2 and 5 are known as front buttons and the rest as gauge buttons. A Carl Zeiss Auriga 60 FIB-SEM equipped with an Oxford Instruments X-Max EDX detector was used for microstructural characterisation. Due to the large depth of field of the electron beam, it was straightforward to locate and look at features on the domed buttons. Cross-sections were made on the buttons using the FIB with 30 kV: 1– 4 nA beam conditions, and imaged at high resolution with the SEM at 3 kV accelerating voltage. Elemental mapping and line scans of the surface and some of the cross sections were carried out with EDX at higher accelerating voltages to gain understanding of the chemistry of the wear processes. The cross-sections were made randomly on the worn surface (“wear flats”) of the buttons, but as the sample was tilted in the FIB-SEM for analysis and the geometry of the buttons is domed, it was only possible to view and mill sections on certain portions of each button wear flat. As the position of the head remained the same throughout the course of the experiment (tilted along the line of symmetry through button 7, with buttons 1 to 3 uppermost), points on the surfaces nearer to the right of the buttons as viewed in Fig. 1 were most likely to have been analysed (as indicated approximately by arrows in Fig. 1). In some cases, e.g. Fig. 5, where the FIB curtaining is seen at an angle, then the cross-section was further down the side of the button. The FIB-SEM and EDX results shown in this paper are on buttons 1 and 4, which are gauge buttons, but similar surface characteristics and mechanisms were observed on the other buttons, including button 2, a front button. XRD measurements were performed on a Bruker Discover D8 diffractometer with Davinci design equipped with a IμS Microfocus Source (CuKα radiation, λ = 1.5418 Å), an Eulerian cradle and a Våntec-500 2D area detector. A laser-video positioning system was used for alignment of the sample. The XRD patterns were analysed with software DIFFRAC EVA (Bruker) and High Score Plus (PANalytical). XRD data were typically collected in the angular range 10° b 2θ b 140°. The drill bit was
Low magnification SEM imaging of the surface of the top of a gauge button (labelled 1 in Fig. 1) showed the surface texture to be rough (compared to a polished or even fine-ground surface) with evidence of rock debris adhered to the surface (Fig. 2a). EDX analysis at 15 kV accelerating voltage of the area in Fig. 2a revealed the surface composition on a larger scale and produced a spectrum showing the presence of rock constituents including carbon, oxygen, calcium, iron, aluminium, silicon and iron, as well as zinc and lead (Fig. 2b). Oxygen and tungsten were the most abundant elements on the surface. By contrast, little cobalt was detected, despite this being a 10 vol% cobalt grade hardmetal. Distinct regions of lead and zinc, as well as oxygen and silicon from rock debris, were identified in elemental EDX maps, which were dragged across and embedded in the button surface during rotation of the drill head and this directionality is observed in Fig. 2c. Other buttons with similar surface composition and mapped at a similar magnification did not show this smearing as defined as seen in Fig. 2c. At higher magnifications however, the button surfaces vary considerably in roughness and composition, as described in the sections below. 3.2. Carbide wear and morphology Fig. 3a shows the position of the FIB-section on the wear flat of button 1 and Fig. 3b–d show it in higher magnification. Fig. 4 shows the same but on another position on button 1. Fig. 5 shows a milled section of button 4. In Figs. 3 and 4 there was surprisingly little cracking of the carbide grains below the wear surface (Figs. 3b and 4b), however some intergranular fracture was observed just below the top layer of carbide grains (Fig. 3c). Transgranular fracture was observed in several carbide grains where there were fine cracks through crushed and re-adhered carbide fragments at the surface (Fig. 4c) and next to binder regions where there was depletion and porosity in the binder phase (Fig. 5). Fragmentation of the carbide grains also occurred at the surface and the crushed carbide particles were re-embedded in the surface layer and mixed with rock fragments and extruded cobalt (Figs. 3b and 4c). There was deformation of the top layer of carbide grains such that they had lost their facets and the boundaries between carbide grains were curved. This is shown in Fig. 4c where larger fragments have been compressed together and the boundaries are deformed, and in Fig. 3b and d down to ~10 μm from the surface, the original boundaries are barely visible. The vertical streaks and obvious “waviness” in the images are unfortunate artefacts of the FIB milling, exacerbated by the uneven surface due to omission of using a protective layer in these experiments, but the features described above are real features in the material. 3.3. Binder morphology and composition There was significant compression of the cobalt regions between the surface and up to 10 μm in depth as revealed by FIB (Figs. 3 and 4), where cobalt binder regions of the expected size have disappeared and large number of crushed WC grains and forced-together WC-WC boundaries are present. The long thin binder region in Fig. 4d has at the top a dark region indicated by the arrow containing adhered
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
3
Fig. 2. a) Secondary electron image of the surface on the top of a gauge button (labelled number 1), b) EDX spectra with elemental peaks labelled, and c) EDX at% elemental maps of oxygen, tungsten, silicon, aluminium, lead and zinc of the whole area. White and black pixels indicate 0 at% and 100 at% respectively.
material from the rock which was probably forced into a hole created by micro-cracking of WC grains. However, larger cobalt binder regions were still present near the surface in many regions as seen in Figs. 4b and 5a. In this case, there was a surface oxide layer present, rather than exposed WC grains, as seen in Fig. 3b. The variation in thickness and composition of the surface oxide layer was likely to have caused the difference in wear mechanisms observed underneath. There was often a gradient or image contrast in the cobalt regions. In Fig. 4, the binder regions appear darker further from the surface, and were very pale between fragmented carbides (Fig. 3c), indicating possible composition variation within single binder regions. This contrast is exaggerated by the shadowing caused by the geometry of the FIB section, however is likely a real effect as it is quite dramatic in the image such as Fig. 4c underneath the carbide fragments. The W solubility in Co increases with temperature so the brighter contrast of the binder regions closer to the surface maybe an indication a temperature gradient. In Fig. 5, the cobalt binder regions showed image contrast where a dark border b100 nm in thickness was observed at the binder-carbide boundary (Fig. 5b). In these regions, there was also delamination of the cobalt at the carbide-binder interfaces, leaving “ligaments” connecting the binder phase to adjacent carbide grains. 3.4. Oxide surface layer and delamination As observed with EDX in Fig. 2 the surface of the buttons showed abundant oxygen, and the cross-sections often showed a surface layer above the hardmetal structure. This layer consisted of mostly oxygen, silicon, calcium and aluminium, but varied in elemental proportions in
different positions, with carbide fragments and cobalt often embedded in these features. EDX showed that the surface layer in Figs. 3 and 4 had a surface layer consisting of more silica than calcium oxide and alumina, and in Fig. 5, contained more calcium oxide than silica and alumina. The structure of the oxide layer also varied, where Figs. 3 and 4 show a “glossy” surface (silica) and Figs. 5 and 6 show a thicker more structured oxide (calcium oxide) that also exhibited some porosity and fracture. Approximately 20 μm from the site of Fig. 5 a layer of oxide was observed to be spalling from the surface; this is shown in Fig. 7. On the outer part of the wear flat of the button in Fig. 6 (pointing away from the centre of the drill head in Fig. 1), there was a large raised feature, measuring approximately 1 mm in length (Fig. 6a). At higher magnification, a relief feature was identified and an EDX map was acquired to include it (Fig. 6c). The area contained a high proportion of calcium oxide, with silica as the next abundant oxide and small amounts of other elements (Fig. 6b). On closer examination of the relief feature, an elevated “plate” or spall was observed ~100 μm in width, which appeared to stand proud of the rest of the surface by several micrometres (Fig. 7). A cross-section was made on the edge to reveal its internal structure. There was a top layer of darker material between 2 and 4 μm in thickness, and a higher brightness banded layer underneath with a thickness of 1–2 μm (Fig. 7c). At higher magnification, even thinner bands were revealed, where the thinnest layers were b50 nm in thickness (Fig. 7d). EDX mapping and a line scan of the cross-section showed the upper part of the section contained mostly calcium (oxide) and the lower layered part was tungsten-rich, but also contained cobalt and oxygen (Figs. 8 and 9). The upper layer had a gradient of tungsten and other elements
4
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 3. Secondary electron images of a region, on gauge button labelled 1, where the surface appeared roughened with exposed carbide grains embedded in a low contrast material, which was a mixture of crushed rock, crushed carbide grains and cobalt. a) Whole FIB-milled section and surface topography, b) full width of milled cross-section, c) fracture under a carbide grain at the surface, d) deformation of carbide boundaries and compression of cobalt regions.
Fig. 4. Secondary electron images of a region, on gauge button labelled 1, where the surface appears smoother and non-metallic due to charging of the electron beam. a) Whole FIB-milled section and surface topography, b) full width of milled cross-section, c) fragmentation of carbide grains, d) further milled section revealing compressed binder regions with variation in imaging contrast in the cobalt.
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
5
Fig. 5. Secondary electron images of a cross-section, on gauge button labelled 4, in a region with surface oxide layer of ~1 μm. a) Whole milled cross-section, b) binder regions with dark contrast at the edges and porosity between them and the carbide phase, c) porosity in oxide layer, transgranular fracture in the carbide and rock material in the binder phase.
shown in Fig. 9. Interestingly, cobalt was only detected in the layered structure, however there were a few discrete regions underneath the plate (Fig. 8), presumably where they were protected by the plate above. The line scan gave quantitative results of the variation of composition across the cross-section (Fig. 9). There were small amounts of other elements, such as lead, zinc, sulphur and potassium, but were not plotted due to little variation along the whole line. There were high amounts of silicon, calcium and oxygen on the surface (below
2 μm on graph in Fig. 9). At the edge of the cross-section (~2 μm), the calcium content increased, oxygen decreased, and silicon and tungsten increased slightly. At the boundary with the bright layers (~6 μm), tungsten increased quickly, as well as a smaller rise in cobalt. There was a peak in these two elements at 7 μm, which was the brightest layer in the secondary electron images (Fig. 7d). Gallium was also detected, particularly in the layers, which was an ion milling artefact. Although similar spalls were not observed on the other buttons, there were other regions on the other buttons where the oxide layer
Fig. 6. Analysis of a relief feature on the worn side of a gauge button (labelled 4). a) Low magnification secondary electron image of button, b) EDX spectra of the area boxed in Panel a, c) EDX at% elemental maps of tungsten, calcium, silicon and oxygen present in the area boxed in Panel a; white and black pixels indicate 0 at% and 100 at% respectively.
6
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 7. Secondary electron images of a relief feature on button 4 seen in Fig. 6. a) The whole lifted “plate” or spall, measuring ~100 μm in width, b) underneath and edge of spall, where crosssection was subsequently made, c) cross-section on the edge of the plate, positioned in the middle of what is shown in Panel b; the arrow shows the direction of the EDX line scan, d) area shown in the box in Panel c of layered structure on the underside of the spall (vertical streaks are an artefact of the ion beam milling).
had begun to lift and a similar layered structure had started to form underneath, as in Fig. 7d; including a front button (specifically, button labelled 2 in Fig. 1). The oxygen to metal ratios in Fig. 9b show that above the layered structure, there are high oxygen levels compared with the metallic phases that reduces with increasing depth. At the boundary to the layered structure, the oxygen and tungsten content is equal and then tungsten becomes more abundant throughout the layers. The oxygen to cobalt ratio remains above 1 but is much lower than in the bulk of the cross section. The oxygen to oxide-constituents ratio also decreases with increasing depth, however there is 10× more oxygen rather than 100× more compared to the metals. At the boundary to the layers, the calcium becomes more abundant and then both silicon and calcium reduce. This confirms that the layers contain mostly metallic elements and the bulk cross-section contain oxides from the quartz. However, higher resolution techniques than EDX in the SEM would be required to reveal the exact change in composition. 3.5. XRD The surface of drill bit number 3 was investigated by 2D XRD at several positions. The integrated 1D diffraction patterns showed strong reflections from the WC substrate, however at other positions on the drill
bit, other crystalline phases were present. The peak intensities of these phases were very low compared to substrate peak intensities, thus indicating that small amounts were present. The positions of the reflections (Fig. 10) agree well with the peak positions calculated for crystalline αSiO2 and disilicide CoSi2, respectively. Several peaks belonging to the SiO2 phase were identified in the experimental XRD patterns. For CoSi2 only (111) was observed since the (220) peak at 48.0° 2θ is completely shadowed by the strong (101) reflection of WC. Both Si-containing phases could be observed at several places on the drill bit. 4. Discussion 4.1. Analysis Focused ion beam and electron imaging characterisation of drill buttons from a real application provided a new insight into the interaction of rock material with the hardmetal during use in mining. A mixture of abrasion and oxidation took place, where evidence of chemical reactions, with the likely presence of water vapour, was observed in addition to mechanical wear. In many areas on the buttons, the rock material or rock-hardmetal product had formed a layer over large areas of the surface. At low magnification, the surface showed compositional variance indicating rock
Fig. 8. EDX at% element maps of tungsten, cobalt and calcium of the entire width of the cross-section on the edge of the spall in Fig. 7 (the centre portion is shown in Fig. 7c); white and black pixels indicate 0 at% and 100 at% respectively.
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
7
Fig. 9. EDX elemental line scans of surface and milled cross-section on relief feature (arrow in Fig. 7c). a) The most abundant elements are shown in atomic percent along the full length of the line. b) Oxygen to metal ratios along cross-section and c) oxygen to oxide-constituents ratios, not including data from top surface (first ~2 μm of line scan).
material was dragged across the surface and embedded into the surface, as well as showing lead and zinc from the ore. This is in agreement with the findings of Beste and Jacobsson [9] but this does not explain why wear is related to the amount of quartz in the rock. At high magnifications, distinct regions with a certain surface characteristics and composition were observed, where the oxide layer had formed with different elemental proportions and at different thicknesses. This large variation indicates that the temperature and abrasive conditions the drill underwent greatly varied over single buttons as well as between individual buttons. The carbide and binder phases both exhibited wear damage, but with characteristics which appeared to differ according to the surface oxide layer properties. Firstly, where there was intergranular fracture between carbide grains, or binder regions that had been compressed or removed, and deformation of carbide boundaries, the surface oxide layer was thinner and silicon rich. Some carbide grains were also exposed at the surface, where the surface layer was continually being removed during drilling. Here carbide grains fractured where they were always exposed to high temperatures and pressures. In addition, as there was little protection by surface oxides to absorb some of the impact force, hence causing the binder regions to compress and carbide grains to become rounded due to fracture of facets at high temperature, as seen in Fig. 4c. The absence of the binder phase in the outermost surface layers of the inserts has also been reported by Beste [5,8,9,10] and Larssen-Basse [12]. The carbide grains were squeezed towards each other, causing extrusion of the binder from between them. Where the carbide grains were forced together under compression, intergranular cracks formed. The compressed regions that appeared reached 2–6 μm depth in some places. In the second case, where binder regions were present closer the surface, greater transgranular fracture in the carbide took place, with separation between phases at the carbide-binder boundaries; here the surface oxide layer was much thicker and calcium rich. It was more likely to have acted as a protecting layer, resulting in less compression of the binder regions than without the thicker oxide layer. In particular, the delamination of cobalt from carbide suggests compression followed by expansion, possibly creating a channel for surface material to mix with the binder phase, causing the lower contrast borders in Fig. 5b. This in turn created voids at the boundary where the binder phase was left only partially attached to neighbouring carbide grains. There were more fragments of WC grains near the surface of the button, which embedded in the binder phase, getting less dense further into the button. Hence the cobalt regions appear lighter (high electron
imaging contrast) near the surface and darker below. It is mostly, if not only, seen in cobalt regions that have a channel to the surface where fragments can travel down from the surface, or at the edges of cracked WC grains. Being open to the surface, these channels could also be means through which the cobalt escaped. Cobalt was often detected with EDX at the surface of the hardmetal underneath the oxide layer (most abundantly at the left side of the cross-section in Fig. 5a). This also corroborates the theory that the cobalt is squeezed out from between the carbide grains, gets trapped beneath the oxide layer, but is eventually removed from the surface, also suggested by Beste et al. [5,8,9,10]. The lateral fracture in the oxide layer occurred above this cobalt-rich region and underneath the oxide layer itself (Fig. 5a); it is possible that this type of feature lead to the delamination of the surface seen in Fig. 6. Another mechanism that contributed to the binder-free outer surface is corrosion, as cobalt alloys easily corrode in hot water and/or acidic conditions. In the work by Beste [5,8,9,10], the cobalt binder is absent at much larger depths; however, they used a different sampling method in which the buttons were cracked to find the weakest region, whereas in this work the wear surface was affected very little, by using the FIB to make sections at local areas on the button surface. Montgomery [11] proposed different wear mechanisms for the gauge buttons compared to the front buttons (labelled 2 and 5 in Fig. 1). He stated sliding-abrasive wear was the predominant wear on the gauge buttons and that the face buttons wear by micro-chipping of small wear fragments, typically of 20 μm in size. Montgomery defined the predominant wear mechanisms on the face buttons as microspalling of the surface caused by stresses related to the blows of the bit on the rock. The fragments in Fig. 7 were observed on the wear flat and resemble the size and appearance of the microspalls despite being taken from a gauge button. This indicates that this type of wear was also present on the gauge buttons. The spall in Fig. 7 could have formed due to specific localised heating or a particular wear motion, due to chemical interaction between the quartz from the rock and the hardmetal. It is possible that after the surface oxide layer had reached a certain thickness, its adherence to the hardmetal button beneath weakened and subsequently loosened from the surface. The layers on the underside of the spall may have stuck to the surface at different stages of the drilling. This region underneath the adhered rock is thus no longer in contact with the air and the chemical interactions are discussed in Section 4.3. It is likely that this would be more common in areas that undergo slower, gradual wear rather than a sudden, high rate wear, which could account for the layering of
8
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 10. XRD patterns, and positions on button indicated by the laser point at the centre of each photo. a) CoSi2 (111) peak at 29.3° 2θ is indicated, all unmarked peaks are from the substrate. b) SiO2 (101),(011) peaks at 26.7° and the CoSi2 (111) peak at 28.9° 2θ. c) Several SiO2 peaks: (100) at 20.9°, (101),(011) at 26.7°, (102),(012) at 39.5° and (111),(11−1) at 40.3°; some smaller peaks remained unidentified.
material with varying composition, and that a similar phenomenon was observed on another button, which could be the same process but at an earlier stage. 4.2. Tribochemistry It is verified by many authors that adhered rock material is detected on the bit surface after drilling. However, no studies to date have considered the tribochemical interaction between the adhered rock and the hard metal bit. Drilling with air or water flushing means the hardmetal is in an oxidising atmosphere; thus the reaction: WC + Co + O2 → WyOx + CoyOx will occur at the unprotected surface. In the areas where the rock adheres to the surface, the following process will also occur at the interface between the quartz in the adhered rock material (here simplified to silica) and the hardmetal button: WC + Co + SiO2 (in rock) → WC + SiO2 + SiC + W-phases + Wand Co-silicides. Since there is an abundance of carbon from the WC, the silica will be reduced to and react with the Co binder, forming Co-silicides. According
to Lavoie et al. [19], CoSi2 forms readily at 620 °C and WSi2 at 100 °C higher temperature. Consulting the binary Co-Si phase diagram (Fig. 11), the first silicide to form on the Si side is the CoSi2, which was also detected by XRD. The W-phases formed could be tungsten oxides as also observed by Beste [5,8,9,10] or the W-Co intermetallic phases such as μ-phase (Co7W6) if there is enough Co left, or W-silicides if there is abundance of Si, although none of these additional phases were detected by XRD here. To further investigate the chemical driving forces a calculation was carried out at 800 °C on the WC-Co-SiO2 system, starting from a molar fraction, of X(Si) = 0.3, with increasing cobalt and decreasing silicon (Fig. 12). It is always difficult to know the exact temperature in rock drilling, but according to Heinö [1] it is between 600 and 2000 °C. The temperature of 800 °C was selected here based on this work and the work by Östberg [21], which correlated plastic deformation of the WC grains in the temperature range of 800–1200 °C. This is to be regarded as a guide since the system is not at equilibrium; despite this, there is a driving force to form silicon carbide and Co-silicides, where Si and C are abundant. CoSi2 would form where there was Co close to the
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
Fig. 11. The Si-Co binary system - temperature versus at% Co [20].
interface between the hardmetal and the adhered rock, with greatest concentration closest to the surface, but penetrating below the surface with the gradient in Si concentration diffusing below the hardmetal surface. The thermodynamics of the system shows that a reaction is to be expected between the lower surface of the adhered rock material where it meets the underlying hardmetal. Hence, the bright layers in Fig. 7 are probably reaction products between the adhered rock and the hardmetal consisting of Co- and W- silicides. If there is an abundance of SiO2 originating from the quartz in the rock and carbon from the WC, quartz will be reduced locally in these surface regions: SiO2 (excess) + WC + Co → SiO2 + SiC + WSi2 + CoSi2. Cobalt is only available through patches of binder layers between the WC grains but this is sufficient to form CoSi2, confirmed by XRD at several places on the wear flat. These reactions suggest tool wear is partly a tribochemical process due to the interactions between the quartz (SiO2) and the rock and it explains why the tool wear can be related to the quartz content in the rock, despite that SiO2 being softer than the WC phase. Approximately 1 μm below the surface of the oxygen-rich tribolayer in Fig. 5a, a cobalt pocket, shown in Fig. 5b, was observed. The Co was close to the surface and had a dark rim towards the outer edge, adjacent to WC grains. This could not result from alloying with W as that would give a whiter contrast; a plausible explanation is that this zone is enriched by a lighter element such as O or Si. However, the solubility of O in Co is low, but Si in Co is high, thus most likely this is a Si enriched zone where Si is supplied by grain boundary diffusion from the outer
9
more Si rich part. This needs to further investigation with transmission electron microscopy (TEM) for confirmation. Figs. 4d and 5c suggested increasing cobalt content in the binder and decreasing content of other elements, i.e. tungsten, Ca/Si surface oxides, with increasing depth beneath the surface. Where the binder was exposed at the surface, the oxide layer/silicon diffused into the binder but stopped at the carbide interface. In an attempt to understand this, the binary Co-Si phase diagram in Fig. 12 shows silicon alloys are still present in the binder containing N20 at% cobalt, decreasing the melting point drastically. Hence the exposed regions might be in fact Si-rich cobalt binder. This may also explain the regions in Fig. 5 where the binder appeared to melt and re-solidify, giving rise to voids at the phase/interphases, which no longer support the adjacent WC grains as they wear away. This relates to the observation of Beste et al. that the WC grains are removed one by one from the surface and their observation of a melted structure. 4.3. Experimental limitations The use of the FIB technique in the work where very specific regions were targeted, enabled viewing of the top surface and cross-section simultaneously. This meant the FIB milling could be stopped precisely to view an exact feature in cross-section. In order for this to be possible, a deposited protective layer was omitted from the experiment. This unfortunately meant more milling artefacts were introduced in the crosssections. Further work would include sections being made with a protective layer to ensure better-controlled milling conditions, and compared with the existing results to remove the ambiguity regarding such artefacts. A limitation of the EDX technique, especially on this material, is the X-ray peak overlap between tungsten Mα and silicon Kα peaks. However, the analysis software was set to interpret the concentrations using different tungsten peaks to distinguish between the elements and represented the final results in atomic percentage to overcome this ambiguity. When EDX maps were taken of a cross-section, the interaction volume limited the resolution of the element identification at given point. The interaction volume would have been of the order of 1 μm in depth beneath the visible surface, therefore detecting other elements than those on the surface. In the case of identifying the chemistry of the darker areas in the binder regions (Fig. 5) using EDX, compositional variance due to high temperature and/or pressure in these regions was
Fig. 12. Mole fraction of phases (NPM) versus increasing cobalt/decreasing silicon content, starting from the composition (at%)X(C) = 0.1, X(Co) = 0.725, X(O) = 5E-2, X(Si) = 2.5E-2, X(W) = 0.1,. Thus WC-Co-SiO2 (left side) and exchanging Co for Si, calculated at 800 °C and 1 bar calculated using the Thermocalc software SSUB5 database [22].
10
H.G. Jones et al. / Int. Journal of Refractory Metals and Hard Materials 66 (2017) 1–10
not possible to detect with confidence due to overlap with carbide grains behind. A lamella containing this region would need to be removed from the bulk sample and analysed separately to get more accurate EDX data without the interaction volume. Ideally transmission electron microscopy (TEM) would also be done on the lamella to confirm the phases with greater certainty. Following on from the limitations of the techniques used in the work in Section 4.2, a number of further experiments would complement and corroborate the existing results. Sectioning the buttons transversely and polishing would confirm if there was deeper rock penetration into the hardmetal. Using the lamella lift-out technique, as previously mentioned, with electron backscatter diffraction (EBSD) analysis could be done to examine plastic deformation at the surface, at compressed carbide boundaries, and surrounding fractures. 5. Conclusion A study was conducted on WC/Co rock drill bits. FIB-SEM microscopy and XRD with consideration of metallurgical thermodynamics was used to lead to the following conclusions in regard to the wear processes taking place. Large variation in subsurface damage was observed on single buttons as well as between different buttons, where there was either exposed carbide grains or a protective oxide layer of varying thickness and composition on the surface. The high temperatures and pressures during drilling had different effects on the hardmetal microstructure depending on these surface characteristics. EDX analysis showed the oxide layer usually consisted mostly of silicon or calcium, with some tungsten (from exposed grains), and no cobalt. The cobalt tended to get squeezed out at the surface (where there was not an oxide protective layer) which would close up any obvious channels. The depth of damage found was only within the top few microns of the impact surface; despite the FIB-SEM cross-sectioning technique only being able to analysis ~15 μm, there was no evidence of channels that rock could implant into and reach much lower depths. By using a 2D-XRD detector and a microfocus X-ray source it was possible to analyse small surface regions of the drill bits. The X-ray powder patterns showed the presence of crystalline α-SiO2 and cobaltdisillicide CoSi2. These results agree with thermodynamic calculations which show that quartz (SiO2) in the rock can be reduced by the presence of C from the WC in the hardmetal and leads to the formation of Co silicides. The formation of CoSi2 by diffusion through Co below the drill bit surface and local melting of this phase removes support from the surrounding WC grains and accelerates wear by fracture and removal of WC fragments. Thus the determination of the wear rate by the equivalent amount of quartz in a rock may be related to formation of Co silicides rather than the abrasive nature of the rock itself. Formation of crystalline α-SiO2 and disilicide CoSi2. These results indicate that the outcome of tribochemical reactions under non-model conditions can be studied by X-ray diffraction methods.
Acknowledgements The authors would like to thank Sandvik for the supply of the worn drill bit and for the UK Government Department for Business, Innovation and Skills for partial funding of this work under the Materials Programme of the National Measurement System. References [1] M. Heiniö, Rock Excavation Handbook, Editor in Chief, Sandvik-Tamrock, 1999. [2] K. Thuro, Drillability prediction - geological influences in hard rock drill and blast tunnelling, Int. J. Earth Sci. 86 (2) (1997) 426–438. [3] D.N. French, D.A. Thomas, in: F.V. Vahldiek, J. Mersol (Eds.), Anisotropy in Single Crystal Refractory Compounds, 1, Plenum New York, 1968. [4] A.J. Gant, M.G. Gee, Abrasion of tungsten carbide hardmetals using hard counterfaces, Int. J. Refract. Met. Hard Mater. 24 (2006) 189–198. [5] U. Beste, T. Hartzell, H. Engqvist, N. Axén, Surface damage on cemented carbide rock-drill buttons, Wear 249 (2001) 324–329. [6] M. Olsson, K. Yvell, J. Heinrichs, M. Bengtsson, S. Jacobson, Surface degradation mechanisms of cemented carbide drill buttons exposed to iron ore rock drilling, The 17th Nordic Symposium on Tribology June 2016, pp. 14–17 (Hämeenlinna, Finland). [7] S. Momeni, S. Moseley, M. Ante, J. Allaart, The Wear of WC-Co Drill Bits During Rotary-Percussive Drilling of Reinforced Concrete, Submitted to International Journal of Refractory Metals and Hard Materials, 2016. [8] U. Beste, E. Coronel, S. Jacobson, Wear induced material modifications of cemented carbide rock drill buttons, Int. J. Refract. Met. Hard Mater. 24 (2006) 168–176. [9] U. Beste, S. Jacobson, A new view of the deterioration and wear of WC/Co cemented carbide rock drill buttons, Wear 264 (11–12) (2008) 1129–1141. [10] U. Beste, S. Jacobson, S. Hogmark, Rock penetration into cemented carbide drill buttons during rock drilling, Wear 264 (11–12) (2008) 1142–1151. [11] R.S. Montgomery, The mechanism of percussive wear on tungsten carbide composites, Wear 12 (1968) 309–329. [12] J. Larsen-Basse, Binder extrusion in sliding wear of WC-Co alloys, Wear 105 (1985) 247–256. [13] I. Konyashin, B. Ries, Wear damage of cemented carbides with different combinations of WC mean grain size on Co content. Part II: laboratory performance tests on rock cutting and drilling, Int. J. Refract. Met. Hard Mater. 45 (2014) 230–237. [14] K.G. Stjernberg, U. Fisher, N.I. Hugoson, Wear mechanisms due to different rock drilling conditions, Powder Metall. 18 (1975) 89–106. [15] H.G. Jones, K.P. Mingard, J. Zunega, M.G. Gee, B. Roebuck, Test methods for high rate impact loading of hardmetals, European Congress and Exhibition on Powder Metallurgy. European PM Conference Proceedings, 1–6, 2012. [16] M. Antonov, R. Veinthal, D.-L. Yung, D. Katušin, I. Hussainova, Mapping of inpactabrasive wear performance of WC-Co cemented carbides, Wear 332–333 (2015) 971–978. [17] J. Angseryd, A. From, J. Wallin, S. Jacobson, S. Norgren, On a wear test for rock drill inserts, Wear 301 (2013) 109–115. [18] B. Roebuck, M. Gee, E.G. Bennett, R. Morrell, Good Practice Guide No. 20: Mechanical Tests for Hardmetals, National Physical Laboratory, 1999. [19] C. Lavoie, C. Cabral Jr., F.M. d'Heurle, J. Jordan-Sweet, J.M.E. Harper, T.J. IBM, Watson Research Center, Yorktown Heights, NY, NSLS Activity Report, 2001 2–20. [20] D. Choi, The Co-Si system, Calphad (1992) 151–159. [21] G. Ostberg, K. Buss, M. Christensen, S. Norgren, H.O. Andren, D. Mari, G. Wahnstrom, I. Reineck, Mechanisms of plastic deformation of WC–Co and Ti(C, N)–WC–Co, Int. J. Refract. Met. Hard Mater. 24 (2006) 135–144. [22] Scientific group Thermodata Europe SSUB5 v. 5.1database available from www. thermocalc.se.