April 2005

PM takes pole position in tungsten heavy alloy production

The scene-setter for discussion of tungsten was the keynote paper of Professor Kira Povarova, of the Baikov Institute of Metallurgy and Material Science, Russian Academy of Sciences, Moscow, on the Powder Metallurgy of Tungsten Alloys. This was almost a tour de force, complementing a data-dense printed contribution with a colourful, application-biased presentation that held even knowledgeable delegates' interest from start to finish.
Tungsten made up around 10-4 per cent (one part in a million) of the Earth's mantle and tungsten deposits were found in more than 50 countries. China, of course, was the leading commercial source. The author's chart (Figure 1) showed that production grew steadily from 1930 to a peak of about 50kT around 1990, then fell away sharply. These figures strongly reflect demand from the hardmetals industry: according to the author, 58 per cent of tungsten production was employed in cemented carbides, 28 per cent in steels and superalloys, 4 per cent in chemicals and 14 per cent in mill products (surprisingly, a total of 104 per cent).

Figure 1. It was claimed in Vienna that 70-80 per cent of tungsten materials are made by powder metallurgy.

Povarova explained that worldwide, most (70-80 per cent) of tungsten-based materials, including all tungsten heavy alloys, were produced by powder metallurgy. Methods had been developed to stabilise the mixed carbide-oxide strengthening in PM-based alloys and to realise high mechanical properties at temperatures 0.4 - 0.6 and 0.7 Tm (melting temperature) due to carbide precipitates and dispersed oxide particles respectively. Uniformity of distribution of alloying elements and phases and the diffusion-controlled processes during sintering were improved by mechanical activation of powder mixtures in high-energy ball mills. The high-temperature mechanical properties of the new PM tungsten alloys were as good as those of the best vacuum-melted material. The main properties of tungsten heavy alloys ( 90% W, 10%Ni-Fe-Co) could be improved by alloying the starting powders and by special thermomechanical treatment (Figure 2).

Tungsten of commercial purity (Tm 3410°C) was characterised by very strong interatomic bonding, high density ( = 19.23 g/cm3), Young's modulus E = 410 GPa, shear modulus G = 177 GPa, torsion modulus T = 91 GPa, heat conductivity 124 W/m K at 1000°C, low thermal-expansion coefficient (4.67x106/K from 0 to 1000°C), and high corrosion resistance in some media. Tungsten was also a strong carbide former. These characteristics defined the application of tungsten as the base of heavy and hard alloys and super-high-temperature alloys and composites for service in reducing and neutral media or in vacuum at temperatures that were too high for less refractory metals (Table 1).

Interestingly, in addition to illustrating familiar examples of tungsten components made by powder metallurgy, the author showed a variety of products made by alternative techniques, notably high-temperature vacuum casting and gas-phase deposition from tungsten fluoride. The latter, developed at Russia's
VNIITS, suggests possibilities as a high-tech coating operation for refractory powders.

Figure 2. Effect of thermomechanical treatment on the mechanical properties and microstructure of tungsten and tungsten-based alloys.

The disadvantage of tungsten as a bcc (body-centred cubic structure) transition metal of IV group was its low plasticity at temperatures below the ductile-brittle transition temperature (Td/b). The low-temperature plasticity, high-temperature strength and creep strength of tungsten and its alloys were conventionally improved by alloying and structural changes during production, whether by powder metallurgy or vacuum melting techniques, as well as by deformation and heat treatment.

Cheap and simple

The PM method was relatively cheap and simple, allowing the production of non-sag wire from alloys with potassium aluminosilicate dopants, electrode materials alloyed with refractory oxides (such as ThO2, CeO2, Y2O3 and La2O3), which could not be produced by melting, and W-Ni-Fe heavy alloys with special structures. The disadvantages of PM tungsten compared with vacuum-melted include lower product density (92-98 per cent of theoretical density), especially of large-size articles, higher impurity levels and non-uniform rhenium distribution in high-Re alloys, increasing the tendency for rhenium-rich -phase regions to appear. The thermodynamic stability of carbides, the main strengthening phase in vacuum-melted structural W alloys, was less in PM tungsten alloys than that of oxides.

Dr Povarova explained that the selection of alloying systems for tungsten was based on the study of phase diagrams and composition-property dependences.
All alloying elements decreased the melting point or solidus temperature of tungsten.

Three directions for reaerch

Strain hardening of tungsten and the elevation of recrystallisation temperature Trec, due to solid-solution alloying , were greater with increasing differences between atomic radii of tungsten and alloying elements, in the order Mo, Re, Ta, Nb, Hf and Zr. Increasing content of alloying elements above 5 atomic per cent caused an abrupt decrease in Tm,Td/b and Trec, and intense softening at temperatures above 0.5 Tm.
The speaker devoted the final part of her data-laden contribution to some of the key factors affecting the production of kinetic penetrators and alloy cones for shaped charges from tungsten heavy alloys. Three research directions were considered urgent: (1) new processes of consolidation by liquid-phase sintering, to obtain long rods with both lengthwise and crosswise uniform fine-grained structure; (2) optimisation or development of new methods of strengthening by thermomechanical treatment, to ensure high levels of mechanical and service properties; and (3) the development of new matrix structures capable of localised adiabatic shear. This would be to improve the ballistic data of tungsten-based kinetic energy penetrators to that of penetrators produced from depleted uranium alloy (U/0.75Ti). Current W-based penetrators resist shear and produce a large mushroom-shaped head on impact, increasing the cavity diameter but reducing penetration depth. By contrast, the depleted uranium fails by adiabatic shear, discarding the deformed material and improving armour penetration.
The traditional system of heavy-alloy compositions, in which the contents of basic elements are, in weight per cent, 89-96 W/4-11(Ni+Fe), with a Ni:Fe ratio of 7:3, was developed by partial substitution of Co for Fe, to give a Ni:Fe:Co ratio of 8:2:1. The revised alloy was two-phase, consisting of the W-based bcc a solid solution and the Ni/Fe/Co-based fcc g solid solution.
To summarise, the full paper, from which we have taken these extracts more than fulfilled its duty as a keynote contribution, not only as an introduction to its subject but also as a state-of-art reference in its own right.
Progressing from the general to the specific, Roland Taillard, of the Laboratoire de Métallurgie Physique & Génie des Matériaux, Université des Sciences & Technologies de Lille, France, spoke on the Effect of microstructure of W-Y based ball-milled powders on compaction and sintering. Since tungsten metal is frequently strengthened with particles of refractory oxides, this was a useful exercise in establishing the relative importance of some of the parameters.
Starting materials were of two kinds. The first of these consisted of 5µm tungsten containing 620ppm oxygen and 130ppm carbon, mixed with 99.9 per cent pure yttrium powder of <250µm particle size and no detectable oxygen content. The idea was that metallic yttrium powder would react with all the oxygen in the tungsten and simultaneously produce yttria powder. The second mix blended the tungsten with yttria (yttrium oxide Y2O3 ) containing 0.28 per cent carbon and having platelike particles with a large scatter of sizes but mean length of about 3µm. In each case the proportions tested were either 1 or 17 per cent of Y or Y2O3.

Figure 3. Effect of total oxygen content on green density.

Figure 4. Comparison of green and sintered densities of W/1Y samples.

Stainless steel balls

After mixing, all powders were milled at room temperature in a Fritsch Pulverisette 6 high-energy planetary ball-mill, those containing Y in argon and those with Y2O3 in air. Conditions included 250ml container, 400 rev/min plate rotation speed and WC/Co 1cm balls, with 16:1 ball-to-powder weight ratio. For comparison, W/Y batches were also milled with stainless steel balls.

The milled powders were compacted for 1 minute into 1cm diameter pellets at 1.2 GPa uniaxial pressure. These were sintered, under hydrogen or in <10-2Pa vacuo, by heating at 300 K/h, followed by four hours at 1800°C and cooling at 500 K/h. Microstructures of both powder particles and sintered samples were studied, identification of phase changes and their temperatures during vacuum sintering being further aided by differential thermal analysis and dilatometry. Under the experimental conditions, both elemental tungsten and blends behaved in a ductile manner during ball-milling. Aggregation, attrition, deformation, work hardening, fracture and welding caused particles to become lamellar or even sandwich-like, or possess a swaged aspect. Nevertheless, the powder became more consistent at the longest milling durations, with improved distribution of the second phase. Powder morphology seemed also to be rather insensitive to the composition of the milling media. Average particle size diminished with longer milling times but the grain-size distribution became slightly broader. The "most frequent particle size" fell below 1µm only with milling times longer than 20 hours, though crystallite size within the particles (as estimated by X-ray diffraction) could be much smaller.

Changes in chemical composition of the powder resulted from both mechanical milling (mechanical alloying) and contamination. Yttrium oxide second phases were just detectable in the W/1vol%Y blends milled up to about 20 hours and could be observed by thin-foil TEM at longer milling time. Powder contamination is apparent at long milling times, where it entails a noticeable increase in powder weight as a result of adhesive wear of the grinding media. This interaction is promoted by both the ductile behaviour of the blends during milling and the tendency of the Co and WC (from the grinding balls) and Fe and W (from the alloy particles) to form solid solutions.

Cobalt is found either as isolated second phases with variable oxygen content and crystallite size that steadily decreases down to a few tens of nanometres during milling, or as an oversaturated solid solution in tungsten. A cobalt content of 4.3at% was measured by XRD within the matrix of the W/1Y mixture after milling for three days.

High contamination

Pollution by Co was associated with high contamination by WC, from the milling balls like the Co. Analyses demonstrated that C, as well as O, were essentially located close to the surfaces of the powder particles. Relative pickup of Fe and Cr similarly depended on the chemical composition of the stainless steel grinding balls.

The author discussed at length the effect of milling on the rate of oxidation or oxygen adsorption in the powders, noting that green density decreased with oxygen content (Figure 3). Adsorption is especially marked in the yttrium, cobalt and iron components, and there were changes in the size and shape of crystallites. The presence of WC second phases encouraged fracture in the W/Y2O3 blends milled in the WC/Co system, increasing with lower yttria content. The ductility of these powders, and thereby their volume fraction of WC wear fragments, was improved by reduced yttria content. Similarly, at 20 hours milling time, oxidation of the Y second phase seemed to diminish ductility and favour the fracture of powder particles.

Whatever the chemical composition of the powders, milling atmosphere and milling media, green density continuously decreased with milling time. For instance, relative density dropped from 78.1 to 51.3 per cent for the W/1Y blend milled for five minutes and three days respectively. However, this trend is not followed by sintered density (Figure 4).

With hydrogen sintering and relatively short milling times, gas evolution (reduction of oxides by the hydrogen) may result in sintered density even lower than green density.

With long-term milling and vacuum sintering, by comparison, sintered density may exceed 96 per cent, even at the relatively low sintering temperature of 1800ºC. The author considered that this was likely to be due to localised melting of one or more alloyed constituents or peritectics.

Following on from the Taillard paper, both logically and chronologically, was the contribution by Anish Upadhyaya, Department of Materials & Metallurgical Engineering, Indian Institute of Technology, Kanpur, on the Sintering of tungsten-based heavy alloys with NiB and FeB Additives.

As an introduction, the oral presentation led with a fairly detailed exposition of tungsten heavy-alloy compositions and structures, not only those based on tungsten-nickel-iron combinations but also those with additives of cobalt, copper, tin or bronze, nickel and iron aluminides, and combinations of metallic nickel or iron with their aluminides. Any of these alloys are consolidated by liquid-phase sintering.

The most popular binders for tungsten heavy alloys are nickel and iron, in the Ni:Fe ratio 7:3 and with W content 90 to 98 per cent by weight.

Figure 5. Sintered density variations in boride-modified tungsten alloys.

Figure 6. Microhardness variations in boride-modified tungsten alloys.

Irregular morphology

The Ni/Fe ratio is chosen to avoid the formation of brittle intermetallic phases. The 90W alloy was chosen as the basis for the author's investigation, substituting nickel or iron borides - melting points 1018ºC and 1650ºC respectively - to lower the solidus temperature and enhance the sintering characteristics.

The seven compositions investigated, all of which follow the basic 7:3 ratio, are listed in Table 2.

Supplied by Osram Sylvania, Towanda, the (M55 grade) tungsten was relatively coarse, with an average grain size of 5.5µm and irregular morphology. Details of the additive powders were not provided. Cylindrical test samples 12.7mm diameter and between 3 and 6mm thickness were pressed at 200MPa and sintered at 1500ºC in a dry hydrogen atmosphere with -35ºC dewpoint and 1 l/min flow rate.

Compacts were heated at 5K/min and held at sintering temperature for 60 minutes. After sintering, densities were calculated from dimension measurements (Figure 5), Vickers microhardness measurements made on a Leco V-100-C1 tester at 2kg load (Figure 6), and structures examined by optical microscopy at x1000 magnification.

The highest density, near theoretical, was achieved by the 90W/7Ni/3FeB composition, even better than the 97 per cent of the base (90W/7Ni/3Fe) composition when liquid-phase sintered. 90W/7NiB/3FeB gave the poorest results.

In spite of its low density, the 90W/7NiB/3FeB gave the highest hardness values, due to intermetallics formation, with high standard deviation because of the inhomogeneity. All the microstructures showed typical liquid-phase sintered microstructure except the 90W/7NiB/3FeB, which had less W solubility and in consequence less reprecipitation.

Upadhyaya summarised the results as follows:

• 90W/7Ni/3FeB alloy achieved the highest density during sintering and consequently resulted in appreciable hardness

• The hardness values of boride-modified tungsten alloys were similar and in some cases even higher when compared with the 90W/7Ni/3Fe alloy

• Both the boride-modified tungsten alloys and the tungsten-nickel-iron alloys have similar microstructure which is indicative of similar sintering behaviour

• About 20 to 25 weight per cent tungsten solubility is observed in the boride-modified matrix for the investigated alloys, comparable to that in the tungsten-nickel-iron alloys.

These useful results make it more than likely that commercial products will be developed from this boride-modified group of alloys