February 2004

New ways to make moly as it enters nano-phase production

Molybdenum production methods have been improved as attention swings back towards this important refractory metal, reports Ken Brookes...

Perhaps because of its relative lack of success as a raw material for hardmetals, production processes for molybdenum have never received the attention applied to its near cousin tungsten. Thus the inclusion of two important and complementary papers on the subject was a notable achievement for the MPIF PM2 TEC 2003 conference in Las Vegas.

In the first of these, Fenglin Yang of Harper International Inc described an Advanced reactor system for the fine control of properties of molybdenum powder. His intention was not to replace, but to investigate, refine and exercise greater control over the long-established, conventional two-stage reaction for producing fine molybdenum powder.

In the first stage, ammonium dimolybdate (ADM) or molybdenum trioxide MoO3 is reduced by hydrogen in either a rotary reduction or a pusher-type furnace, to molybdenum dioxide MoO2. Reaction via ADM is simpler but the final product is less desirable. In the second stage, the MoO2 is further hydrogen-reduced to Mo metal powder in a single-tunnel or multi-tube pusher furnace.

The first-stage reaction is highly exothermic and therefore difficult to control. Without this control, localised hot spots of 1000ºC or more can develop, followed by a runaway reaction. Heat of reaction at 600ºC is 155kcal/kg MoO3.

The rotary reactor's advantages over the older pusher furnace are substantial. They include diminished likelihood of hot spots, lower energy consumption and lower capital cost due to higher efficiency. This means the physical size of the equipment and reaction time can be reduced due to better gas-solid contact. Automated materials handling, improved and simplified control of grain size and morphology, lower maintenance and labour costs are additional pluses. The overall result is superior product quality.

Though not mentioned in the simplified equations, Mo4O11 is a distinctive and important intermediate product with needle-shaped grains that affect the morphology of the final powder and facilitate its control. Temperature profile, hydrogen dewpoint, hydrogen flow rate and residence time are the main process parameters.

Second-stage reduction to Mo metal is endothermic and is normally carried out in a multi-tube furnace to achieve optimal grain size and grain-size distribution, with minimal residual oxygen. Solid material bed depth is an additional parameter here.

The investigation described in the paper was carried out in a lab-scale pilot plant. Its major components were a rotary calciner for ADM to MoO3, 1500mm long and 150mm in diameter, a rotary reactor for stage one reduction and an 18-tube pusher furnace for stage two reduction, the latter taking alloy boats about 100mm wide, 50mm high and 480mm long. The plant also included a hydrogen dryer for recycling, dust filter and a scrubber for waste gas and dust collection. Angle and rotation speed could be varied and powder feed rate was controlled by a screw feeder.

Figures 1 and 2, showing equilibrium gas partial pressures for each of the reduction stages, shows that stage one reduction

occurs with extremely low (wet) H22 concentration, whereas stage two reduction

requires extremely high (dry) H2 concentration.

Figure 1. Chemical equilibrium calculations - stage one reduction.

Figure 2. Chemical equilibrium calculations - stage two reduction.

In fact the reactions are considerably more complicated than those implied by the simple equations, involving chemical vapour transport and almost certainly the volatile molybdenum oxide monohydrate MoO3(OH)2. The mechanism, beginning with nucleation of the product phase, is illustrated in Figure 3. Typical changes in crystal structure for each stage, which include the needle-like Mo4O11, are shown in Figures 4 and 5.

Figure 3. Grain morphology transformation mechanism.

Figure 4. Stage one reduction - grain morphology transformation.

Figure 5. Stage two reduction - grain morphology transformation.

As might be expected, there are similarities between the reactions and results achieved with molybdenum precursors and those related to tungsten reduction. For example, stage two reduction with wetter hydrogen (higher dewpoint) results in larger Mo grain size. Mo grain morphology is inherited from that of the MoO2.

In Figure 6, the author demonstrated the superiority of the "rotary (stage one) plus pusher" system over the traditional "pusher plus pusher" arrangement, in terms of grain size and shape. In each case the starting material was MoO3. A similar comparison showed that MoO3 was better than ADM as input.

Figure 6. Grain morphologies of materials processed from different reactor systems.

Since the investigation, the "rotary plus pusher" pilot plant has been successfully scaled up by 35 times, to a production rate of around 2.4 tonnes per day.

In a further contribution devoted to this important refractory metal, Sunil Jha of the Climax Molybdenum Company, Sahuarita, Arizona, introduced a significantly different process for making Nano-sized molybdenum powder.

Now being produced in commercial quantities, the high specific surface area of this product enables fully dense parts to be obtained at lower sintering temperatures. It also makes processing of Mo/Cu and Mo/Ag more cost-effective and the alloys easier to fabricate into electronic components, such as heat sinks and electrical contacts. Agglomerated to improve handling, the ultrafine powder can also be used in MIM feedstocks.

In principle, the production sequence for nanomolybdenum is much the same as that described in the previous paper, but in practice it is noticeably different. The patented process involves quenching MoO3 vapours to form nanomolybdenum trioxide (NTO) particles of 50-100nm grain size (Figure 7). The surface area of NTO approaches 60m2/g and its apparent density is in the range 0.11-0.14 g/cm3.

Figure 7. SEM micrograph x50k of nano MoO3 (NTO).

The NTO precursor is reduced, again via a patented process, to nanosize MoO2 and then Mo metal powder. Figures 8a and 8b compare the nano Mo with conventional Mo metal powder, and figures 9a and 9b show the nanomaterial at higher magnification. Crystallite sizes of individual nanomolybdenum particles are in the range 100 to 500nm, but they are generally loosely agglomerated or sintered together.

Figures 8a and 8b. SEM micrographs x5000 of nano and conventional Mo metal powders.

Figures 9a and 9b. SEM micrographs of nano Mo at x30k and x100k magnification.

There is a school of thought that would describe material with average grain size as large as 0.1-0.5µm as “sub-micron” rather than “nano”, saving the “nano” expression for much finer powders when available (even the MoO3 precursor is much finer). The author's expression is used here for convenience, but with reservations. This problem should disappear when the ISO and/or EPMA come to an agreement on nomenclature for fine and ultrafine grain sizes.

Typical physical properties of nano and conventional Mo powders are compared in Table 1. The chemical purity of nano Mo is 99.95 per cent, excluding gases. Table 2 compares the green strength of Charpy test bars of nano material pressed with and without lubricant. High green strengths are attributed to the irregular nature of the particle agglomerates.

The nanopowder showed high sintering activity even at temperatures that would normally be considered low for refractory metal processing (Figure 10). With its high surface area and fine particle size, target applications include electronic substrates, electrical contacts, ink for thick films, alloy blends, MIM feedstocks and thermal spray powders.

Figure 10. Sintering of nano Mo in hydrogen atmosphere, as a function of time and temperature.

Tungsten's refractory or high-temperature properties are relatively unimportant in the heavy alloys application. They would come into play during manufacture, but the addition of nickel and/or other alloying elements make it possible to carry out liquid-phase sintering at relatively low temperatures.

Because of their very high density and other advantageous properties, tungsten heavy alloys are massively employed in ammunition and armaments of every kind, conferring unusual topicality on our next choice, A study of the microstructural evolution of tungsten heavy alloys during liquid phase sintering. The paper was presented by Nicholas B. Erhardt of Pennsylvania State University.

Two W/NiFe alloys were chosen for this investigation, with 83 and 93 weight per cent of tungsten and the remainder NiFe in a 7:3 ratio. The ratio between nickel and iron was chosen to avoid intermetallics. The tungsten powder was initially rod milled for one hour, using pure tungsten rods and an inert argon atmosphere to prevent oxidation. Elemental powders were combined to give the desired composition, with 0.5 per cent Acrawax pressing lubricant, then blended in a Turbula mixer for 20 minutes to ensure homogeneity.

The blended powders were hand pressed, at a pressure of 175MPa, into cylindrical compacts 12.54mm in diameter and heights between 10mm and 12mm. Density of the green compacts was near 60 per cent of theoretical. Dewaxing and presintering were carried out in a single step, initially at 500ºC, then heating at 10K/min to 1000ºC for a 1 hour hold. The process gas was dry hydrogen with a dewpoint of -67ºC.

Samples were sintered in a molybdenum crucible suspended in the hot zone of a quench furnace by a tungsten wire. When the desired quench temperature was reached, the furnace was purged with nitrogen (not mentioned in the printed paper but mentioned in the presentation) and the entire crucible dropped into a bucket of water, to capture the in situ microstructure. Temperatures and isothermal hold times are listed in Table 3.

Microstructures were examined on a central vertical cross-section. Princeton Gamma-Tech Imagist II software was employed to examine optical micrographs and to measure and classify the tungsten grains. Connectivity was estimated by counting the number of other grains in contact with each grain selected. A minimum of 400 grains were taken for each microstructural measurement.

Figure 11 plots grain size against the various sintering conditions. For the 83W alloy, it shows continuous grain growth as either time or temperature is increased but, for the 93W alloy, a decrease was noted after two and five minutes hold time at 1500ºC. A suggested mechanism for this is rapid solution of W in the newly formed NiFe alloy, which has a liquidus temperature around 1465ºC. Solubility of W in NiFe at this temperature is about 19 per cent.

Figure 11. Plot of average grain size with sintering temperature and isothermal hold time at 1500°C.

With regard to density, both materials approached full theoretical density soon after the formation of NiFe liquid phase. Soon afterwards, however, distortion occurred in the 83W alloy and its apparent density fell dramatically. Structural examination disclosed "gaping" internal pores reaching up to 1-2mm across.

The contiguity check also showed major differences. In each case, the results were attributed to the number, location and curvature of the necks formed between tungsten particles during solid-state sintering, and their solution in the presence of NiFe liquid phase. The tungsten was said to dissolve preferentially at the necks, the driving force increasing with curvature. This was a gradual effect in the 83W samples but more rapid with the 93W.

There is a sharp increase in contiguity in the 83W alloy after more than a brief hold at 1500ºC. This is due to settling under gravity, during which the grains can move position and the compact takes up a so-called "elephant's foot" shape.

During the question period, the presenter was asked why the two alloys (83W and 93W) were chosen, since there was no explanation in the paper. He explained that it was known that 93W would not distort, whereas 83W would do so. With regard to heating rate, a faster rate would give smaller necks between particles during solid-state sintering. Finally, answering my own question, he agreed that the macroporosity was almost certainly due to an oxygen-hydrogen reaction to form bubbles of water vapour.