April 2004

Valuable nuggets for those in the right place at the right time

It was by no means a vintage conference for hardmetals interests, says Ken Brookes, but there was gold dust to be found for those who panned the waters of the conference...

After the opening plenary session, Julie Russias of Groupe d'Etudes de Métallurgie Physique et de Physique des Matériaux, Institut National des Sciences Appliquées de Lyon, Villeurbanne, France, kicked off the hard materials part of the proceedings by describing the Synthesis of carbide and nitride materials obtained from ceramics SHS starting powders.

Titanium carbide and titanium nitride are widely used in hardmetals, both as constituents of carbide alloys or as wear-resistant coatings. They have the advantage of being electroconductive and therefore machinable by EDM processes.

This research programme was targeted at the production of pure bulk compacts by hot pressing. Powders were prepared by the SHS (self-propagating high- temperature synthesis) method, which permits economic production of particulate materials. The starting powders are described in Table 1.

Although the average grain sizes look good, examination of grain-size distribution indicates a substantial proportion of larger grains up to 6µm (Figures 1 and 2).

Figure 1. Particle size distribution in TiC starting powders, measured by a Horiba Capa 700.

Figure 2. Observation of TiN powder by transmission electron microscopy.

Test samples were 37mm diameter and 5mm high. Powders were cold compacted in a graphite die and the die inserted into the hot press. Degassing occurred at 600°C under vacuum. The final hot pressing was carried out in an argon atmosphere at 50 or 100 MPa, the temperature being increased at 10K/min up to the desired temperature (1400°C to 1900°C), which was maintained for one hour.

Mechanical properties of the sintered materials are noted in Tables 2 and 3. Optimum hot pressing temperature for the TiN at 50MPa was around 1600°C, with improved results from powders that had been initially thermally treated. Grain growth occurs at higher temperatures. Maximum relative density was 98.3 per cent.

Best results were obtained from the TiC hot-pressed at 100MPa and 1900°C. Hardness reached 2800HV. However, because the microstructure did not appear to be homogeneous, it was proposed to attritor-mill the powders for future experiments. Sinter-HIP and hot isostatic pressing will also be investigated.

A contribution of fundamental interest was that of Anders Petersson of the Royal Institute of Technology, Stockholm, Sweden, on the Densification of cemented carbide alloys during
sintering. However, somewhat less comprehensive than its title implied, its scope was confined to the simplest WC/Co alloy system. Even this, though, had to cope with the extensive ramifications of cobalt and carbon contents, nominal or mean carbide grain size, grain size distribution, powder preparation method and sintering conditions.

Figures 3 and 4, of three compositions with 10 weight per cent (about 16 volume per cent) cobalt and differing grain sizes, demonstrate that, though they start sintering at similar temperatures, coarser materials require higher temperatures to achieve substantial densification. The two overlapping peaks in the densification curves relate successively to solid-state and liquid-phase sintering.

Figure 3. Sintering shrinkage of three WC/10Co materials with different carbide grain sizes.

Figure 4. Densification rates of three WC/10Co materials with different carbide grain sizes.

Similar pairs of diagrams showed the effects of different cobalt and carbon contents. In the latter case (Figures 5 and 6), the compacts were of 2 µm - 3 µm nominal grain size, 10 per cent cobalt content and 5.42 per cent, 5.55 per cent and 5.69 per cent carbon. These compositions are respectively sub-stoichiometric, approximately stoichiometric and above stoichiometric. In the first and last of these one would expect to see eta phase and free carbon respectively in the final products. As can be seen from the densification-rate diagram, those with the low and near-standard carbon contents shrink similarly, but there is a massive difference with the high-carbon variant.

Figure 5. Sintering shrinkage of three WC-Co materials with different carbon content.

Figure 6. Densification rate of three WC-Co materials with different carbon contents.

Further experiments were carried out, sintering WC/10Co compacts with two different nominal grain sizes (2µm and 6µm) under uniaxial loads. Equations were derived for free sintering strain rates and compared with experimental dilatometric data.

The author concluded that much of the sintering process took place in the solid state (Figure 7a). There was a marked difference at a temperature 1380°C just above the liquidus (Figure 7b) but, after a hold time at 1430°C (Figure 7c), the cobalt binder was more homogeneously distributed and the prismatic shape of the carbide crystals was more pronounced. The latter was due to the solution-reprecipitation recognised as the second stage of liquid-phase sintering.

Figure 7. Microstructure of WC/10Co, 2 µm samples sintered to (a) 1280°C, (b) 1380°C and (c) 1430°C.

It seems a pity that, at some conferences, there is no official record of the discussions that follow each paper. On this occasion, the first questioner (Bryan Roebuck of Britain's National Physical Laboratory), having elicited that the same equipment could be used to investigate binder phase changes on cooling as well as on heating, suggested that it might well show changes in the ratio between fcc and hcp. Answering another question, the author stated that a global model was being developed, with the aid of finite element analysis. Finally, it was reported that sintering stress had been measured at about 10MPa.

Turning now to functionally graded hardmetals, we come to the latest chapter in the fruitful collaboration between sponsor Kennametal Widia and TU Vienna, Austria. Entitled Gas/solid interactions in the evolution of graded microstructures in hardmetals, it was presented by researcher Alexander Eder. Interested readers will find it rewarding to read the data-rich original paper.

Gas/solid interactions are in no sense a side issue in the formation of graded microstructures. They are the modus operandi of the process, generally involving highly complex hardmetal alloys. Those examined in this investigation - simplified in Table 3 - were of the general type WC+(Ti,W)(C,N)+Co, with the addition of some TaC, NbC and carbon black. The object of adding carbon was to maintain near-stoichiometry whilst outgassing the adsorbed or combined oxygen as CO. The ultimate object of the research was to produce high-grade performance-enhancing coatings during a modified sintering schedule, eliminating the costly separate coating operation of conventional manufacture.

In an atmosphere of flowing argon, melting points of all alloys were found in the relatively narrow range of 1335°C - 1350°C. As examples, Figure 8 shows the DTA (differential thermal analysis) signal of alloy B in an argon atmosphere. This is compared with alloy E in an atmosphere of flowing nitrogen.

Measured eutectic temperatures of alloys A, B and C in both atmospheres were generally similar. By contrast, for the high-fcc alloys D and E, melting behaviour changed drastically in a flowing nitrogen atmosphere. The curve at the bottom of figure 8 shows a succession of eutectic temperatures and other overlapping reactions, replaced by a single composite peak on cooling.

Figure 8. DTA signals of alloy B (top) in Ar atmosphere showing a distinct melting point representative of the phase reaction in the system WC-fcc-Co and melting behaviour of a coarse sample of alloy E (bottom) in an atmosphere of flowing N2 with several peaks within a temperature range of 1270°C - 1350°C during heating.

Figure 9. Evolved gas analysis (EGA) results on investigated alloys.

Figure 9 shows CO and N2 outgassing data as a function of fcc/hex ratio. The results of EGA (evolved gas analysis) showed maximum outgassing of CO at a temperature of 1180°C, independently of composition. The variation of N2 onsets and maxima show no dependence with respect to the ratio fcc/WC or N2 content. Mass loss during measurements was higher with increasing N2 content. Since alloy A contained no N2, the mass loss resulted from CO reduction and Co evaporation. Owing to desorption, oxide reduction, outgassing and evaporation of Co binder, mass loss could be measured and is also shown in Figure 9. The higher the fcc/hex ratio, the higher the weight loss.

The influence of nitrogen on shrinkage behaviour was investigated for alloy D (Figure 10). For the experiments the dilatometer was filled with high-purity nitrogen of different pressures (10, 50 and 300mbar) after evacuating to a pressure of 10-3 mbar, after which heating was commenced. The resulting curves were divided into two types. Those at 10 mbar and 50 mbar showed an increase in shrinkage rate at higher temperatures, with a clear peak at temperatures above the liquid-phase formation temperature (termed type 1), whilst the other two cycles with 300mbar and dynamic vacuum show a broad distribution of shrinkage rate with temperature (type 2).

Figure 10. Shrinkage rates of alloy D in different atmospheres show two modes: for low nitrogen pressure a distinct maximum above the liquid phase formation temperature, whilst high nitrogen pressure and dynamic vacuum lead to broad shrinkage distribution with maxima below the liquidus.

Figure 11. Shrinkage behaviour of alloy D in a technical sintering cycle. After the second dwell at 1300°C, application of N2 led to a change in shrinkage mode. The melting point can be identified by a signal peak at 1345°C.

As a follow-up, the shrinkage behaviour of D was studied within a technical sintering cycle with dwell times at 1200°C, 1300°C and 1400°C (Figure 11) and nitrogen pressure of 50 mbar applied at the end of the 1300°C plateau. A change from type 1 to type 2 shrink age can be seen at the change of atmosphere.

I asked which of the alloys had been selected for commercial manufacture and was told that all were "near to production."

Although our next choice - Nanocrystalline (Ti,M)(CN)-based cermets for cutting-tool applications - was presented in the ìfunctionally graded materialsî session, this interesting paper dealt only with homogeneous hardmetals. It was presented by Shinhoo Kang of Seoul National University, Korea.

The mechanical properties of TiC- or Ti(C,N)-based hardmetals are typically poor, partly because the constituent powders are generally mixed rather than prealloyed. In this instance, oxide starter materials were high-energy milled to maintain homogeneous nanosized particles, then carbothermally reduced to ensure maximum intersolubility. Though different in detail, the method is in principle broadly similar to that we employed at Tungsten Electric Co for complex carbide hardmetals during the 1950s.

Starting materials were TiO2, NiO of average particle size 45µm and WO3 of average particle size 20µm. All were of purity greater than 99 per cent. They were mixed with carbon, then high-energy milled in a Fritsch Pulverisette 7 planetary mill. WC balls of 5mm diameter were used as milling media at a ball-to-powder ratio of 20:1, in a WC bowl for 20h in air at 250 rev/min. After milling, nickel oxide crystallites were said to have been reduced to around 40nm. Final (Ti,W)C-Ni and (Ti,W)(C,N)-Ni nanopowders were prepared by carbothermal reduction in flowing argon, hydrogen and nitrogen as appropriate, the nitrogen in the final powder being derived from the furnace atmosphere. Chemical analyses of the synthesised powders are given in Table 4, quoted from the paper. The apparent accuracy of these analyses seems meaningless unless the contents of W and Ti had been measured to corresponding precision, which in this instance did not occur. In any case, the oxygen was a retained impurity and categorised by the author as "somewhat high." When nitrides are present, the retained oxygen content is decreased, but the grain structure is noticeably coarsened.

Figure 12. XRD of pre-alloyed powders synthesised from oxide mixtures at 1300oC for 2h : (a) (Ti,W)C solid solution obtained from 15 wt % WC and (b), (c) (Ti,W)C-Ni and (Ti,W)(C,N) prealloyed powders obtained from 30 wt % WC, respectively.

Figure 2. Observation of TiN powder by transmission electron microscopy.

X-ray diffraction (XRD) results (Figure 12) confirmed that the synthesised particles were in the form of homogeneous solid solutions. Microstructures, with particle sizes from 200 to 800nm, were very consistent (Figure 13), but porosity was higher than in typical micron-sized structures (Figure 14). If full density could be attained, mechanical properties of the nano materials were expected to be significantly improved.