At any top-class conference, there’s generally one “stand-out” paper. On the technical level, my chosen example from MPIF’s 2007 PowderMet conference in Denver is about “nanoscale” powders and “nanostructured” materials, but its lessons should be required reading, not only for every research director but also for those who need to evaluate the commercial importance of any new development.
Thus, for this review, I’ve chosen the Economics of processing nanoscale powders by John L Johnson, R&D Director of ATI Alldyne, Huntsville, Alabama. The author defined nanoscale particles as those with dimensions less than 100nm, but the impeccable logic of his excellent contribution is unaffected by his choice of units. For convenience, I’ll stick to the author’s terminology.
As the presenter made clear, the industry has failed to live up to early predictions of widespread use of nanoscale powders. In spite of undoubtedly useful properties, their implementation in press-and-sinter processing remains extremely limited, almost to niche (though important) markets such as tungsten carbide microtools and circuit-board drills. Barriers to their use included powder cost, contamination, handling difficulties, low packing densities and rapid grain growth during sintering.
A typical new processing path devised for nanoscale tungsten employed compaction pressures above 1 GPa and low sintering temperatures to preserve nanosized grains. In today’s competitive world, innovations such as this must show a substantial cost or performance benefit in order to displace established press-and-sinter processes. Discussing this problem, the author extended sintering and property models for nanoscale tungsten to other metals and ceramics, processing economics being established by comparing compaction and sintering costs with performance gains.
A sintering model was employed to calculate apparent density, green density, sintered density and grain size as functions of particle size. These were then used to predict further properties. The relevant equations can be found in the full paper, in the Conference Proceedings CD-ROM** published by MPIF.
Derived from the model predictions, green density, sintered density and grain size were plotted as functions of particle size for two different systems. In the first, tungsten powder was pressed at 250MPa and sintered at 2000°C for 60 minutes. The low apparent density of nanoscale powders resulted in low green density, which could not be sintered to high density. Density also decreased with large particles, due to their slower sintering rate, leaving an optimal particle size range of 1µm to 3µm for near full density. The grain size of nanoscale powders sintered at 2000°C was substantially larger than the initial particle size. In the second system, the compacting pressure was 1280MPa. Much higher green density was attained and nanoscale powders were predicted to sinter to near full density at only 1000°C. In both instances, once particle size became too large to press and sinter to full density under the specified conditions, grain growth became negligible and grain size was determined by the initial particle size.
In general, the properties of refractory metals such as tungsten and molybdenum were far more sensitive to particle size than those of other metals. Properties of ceramics were even more sensitive to particle size, since their high hardness and brittle nature prevented pressing to high green densities.
Properties are to a great extent governed by grain size and density. For this paper, properties for nanograined materials were predicted by extrapolating conventional models, but they sometimes provided unique benefits not predicted by the models. For example, while hardness was predicted to increase with decreasing grain size, and fracture toughness to decrease, nanograined materials provided some exceptional combinations of hardness and toughness. Conventional models established baseline properties expected of nanoscale powders (tabulated in the paper) and enabled unpredicted property improvements found in experimental work to be quantified.
As examples of the model predictions, properties of tungsten were plotted as functions of particle size for the two systems referred to above. For tungsten powder pressed at 250 MPa and sintered at 2000°C, properties followed density values and peaked at approximately 1µm to 3µm. Higher strengths and hardnesses were possible with nanoscale powders and ultra-high compaction pressures.
Property maps, constructed by plotting properties as functions of each other, were employed to identify unique property combinations. Plots of modulus, hardness, ductility, electrical conductivity, thermal conductivity and wear resistance as functions of strength were given for 10 different metals. Most properties either correlated or inversely correlated with one another. For example, modulus, hardness and abrasion resistance correlated directly with strength, whilst ductility was inversely related. More interesting combinations were identified for electrical resistivity or thermal conductivity with strength. Lower-strength copper and aluminium tended to be the best conductors, but the unique combination of high strength and high conductivity of tungsten was particularly significant.

Tungsten carbide stands out

Modulus, hardness, fracture toughness, electrical conductivity, thermal conductivity and wear resistance were plotted as functions of strength for eight different ceramics. The correlations were similar to those of metals. In general, ceramics provided high modulus, high hardness, good wear resistance and a wide range of conductivities, but with poor fracture toughness. The high strength of tungsten carbide set it apart from the other selected ceramics.
The wide range of properties for the different materials described were partly the result of the range of sintered densities achieved by different particle sizes. While some industrial users required maximum values for certain properties, for which they were willing to pay a premium, for the most common applications users would pay only for materials providing the best price/performance ratios.
As a first estimate of the value provided by nanoscale powders, predicted properties were divided by powder prices to “normalise” them. Powder costs were analysed over a wide range of particle sizes. Differences in impurity levels had a large effect on powder costs, as did fluctuations in raw material prices. Plots shown for metals and ceramics were mostly related to the greater cost of creating higher surface area, though some exceptions existed if a low-cost chemical process route were feasible. For example, coarse tungsten powders (above 10µm) were no cheaper than 1µm powders and could even be more expensive because of longer processing times or additives to enhance grain growth. The costs of refractory metal powders produced from chemical processes did not tend to fall as particle size increased above a few micrometres. On the other hand, costs of metals such as copper, titanium and iron continued to decrease until they neared the cost of wrought metal.
At the other end of the scale, making nanoscale powders usually required alternative and much more expensive techniques, such as gas-phase synthesis or exploding wires. Some of this extra cost was related to the lower volumes of powder produced and the inefficiencies of new processes, but nanoscale powders were unlikely ever to match the economics of conventional powders.
Dividing predicted properties by powder costs enabled cost-normalised
properties to be plotted. For powders pressed at 250 MPa and sintered at 2000°C, peak values occurred at about 1µm for strength, ductility and thermal conductivity, indicating the best value proposition. As particle size decreased below 1µm, the value dropped. This finding agreed with the current market place, in which the most commonly used powders had particle sizes between 0.8µm and 3µm. The higher strengths and hardnesses possible with nanoscale powders pressed with ultra-high compaction pressures were offset by higher powder costs. For nanoscale tungsten powders to be economically useful on a large scale, they needed to be produced at lower cost or provide much greater improvement in property performance than predicted.
The cost-normalised properties of iron were shown but, unlike tungsten, no peaks were seen, implying that coarse powders provided the best value. Even the improvements in properties attained with ultra-high compaction pressures provided negligible incremental value compared with coarser particle sizes. This simple economic analysis showed why conventional press-and-sinter ferrous powder metallurgy used coarse powders and indicated the hurdles that nanoscale powders needed to overcome in order to provide superior value.
As with absolute properties, cost-normalised property maps were produced by plotting cost-normalised properties as functions of one another. The plots showed the high strength-to-cost ratio of iron and why it was so widely used for structural applications. The value of copper for thermal conductivity was also obvious. The plots for ceramics showed that tungsten carbide and alumina provided the best value for high volume, cost-sensitive applications. Other metals and ceramics only had niche applications where premium cost would provide the highest absolute performance.
Powder costs were part of the cost of producing powder metallurgy components, as were compaction and sintering costs. The use of ultra-high compaction pressures to consolidate nanoscale powders required larger presses with higher tonnage, resulting in higher compaction costs. The low apparent density of nanoscale powders also resulted in a longer stroke, which slowed the production rate. Nanoscale powders had the potential to sinter at lower temperatures, reducing energy consumption and possibly allowing for lower-cost furnace construction.
For full and detailed analyses, a cost model contained over 150 variables, requiring realistic assumptions for overhead costs such as rent, maintenance, interest rates, facility use, utilities, depreciation, load balance and so on. Input values were obtained from current databases and supplemented as needed from external sources. Each operation was independently audited and sensitivity analyses were conducted to determine the most significant factors. Part geometry was a key attribute that affected costs in several ways.
To illustrate the concepts, a ring geometry with 2.54mm outer diameter, 1.27mm inner diameter and height of 0.85mm was assumed (Figure 1). Compaction costs were calculated as a function of material, particle size, pressure and part size. A press was selected from a six-press database, based on the tonnage required to achieve desired pressures. Hourly operating cost was calculated based on the costs of capital, floor space, utilities, maintenance and labour.
Cycle time was based on stroke rate and distance, which depended on the apparent density of the powder and the compaction pressure. Green density was calculated as an output variable for input into the sintering cost model.
A plot of compaction cost as a function of particle size for different compaction pressures for the ring geometry is shown in Figure 2. This plot was not material specific, but the green density for a given combination of particle size and compaction pressure was a function of material hardness. Compaction cost depended on particle size only at ultra-high pressures. Ultra-high pressures provided little benefit for ceramics, which could not be pressed to high densities because of their lack of ductility. In any case, the cost of compacting metals and ceramics was small compared to their raw material cost.
Sintering costs were calculated based on the sintering temperature required to achieve 96 per cent of theoretical density. This temperature depended on material, particle size and green density. Sintering costs were normalised per unit volume and the number of ring parts that could be sintered per unit volume was calculated. Hourly operating cost was calculated from the costs of capital, floor space, utilities, maintenance and labour. A 12-hour sintering cycle was assumed.
The sintering temperature required to achieve full density increased with coarser powders and lower compaction pressures. Not surprisingly, sintering costs increased substantially at particle sizes which could not be sintered in conventional furnaces, but in some cases slightly smaller particle sizes allowed the use of conventional furnaces. The effect of particle size on sintered part cost for various metals is shown in Figure 3. Powder costs dominated sintered part costs. Except in special cases, lower sintering temperatures for finer powders did not offset higher powder costs.

Refined microstructure

Ultra-high pressure compaction and low-temperature sintering of nanoscale powders produced refined microstructures with improved hardness and strength. This was more costly than employing conventional pressures, but the compaction cost was small relative to powder cost. The reduction in sintering costs due to lower sintering temperatures of nanoscale powders did not offset their higher powder costs.
The key result of this investigation, from analysis of cost-normalised properties, was that the incremental improvement that nanoscale powders provided in performance did not compensate for their higher powder and manufacturing costs. For nanoscale powders to become commonplace in press-and-sinter powder metallurgy, they needed to be produced at much lower cost or to provide a much larger improvement in property performance than was currently predicted by conventional models.