October 2005

Manufacturers 'need better quality titanium PM powders'

As delegates at Euro PM 2005 in Prague discuss the future for powder titanium and the FFC process at a special seminar, a plea from a US research group underlines the need for higher purity powder while reporting success in injection moulding titanium using a binder based on naphthalene...

The use of titanium in biomedical applications continues to gain attention because of its unique properties, including high specific strength, low density and lightweight feel, excellent corrosion resistance, and biocompatibility.
These translate into tremendous clinical advantages in terms of reduced recovery time and rehabilitation and improved patient comfort, making it a nearly ideal material for the development of medical bone reinforcement and replacement products. At present however, the fabrication of titanium-based implants is limited to a costly, multi-step process of vacuum arc melting, hot rolling, scale removal, vacuum annealing, machining, and surface treatment. There is an overwhelming need in the industry for an alternative method of titanium manufacture, one that simultaneously reduces the cost and complexity of processing, enhances design flexibility, and improves the biological activity of the implant surface without compromising its structural integrity, i.e. its strength, fatigue resistance, and biocompatibility. Metal injection moulding (MIM) is a leading candidate.

The use of easily removed fugitive phases in the feedstock mixture and control of the de-binding and sintering treatments allow the porosity of the component to be optimised, affording potential advantages in a number of specialised applications including the design of self-lubricating parts and biomedical implants, say the researchers.

However, while MIM has been used with great success in manufacturing a wide variety of metal products, including those made from stainless steel, nickel-based superalloys, and copper alloys, it has found far less application with reactive metals, such as titanium, primarily because of problems with interstitial impurities. In particular, high levels of oxygen are often found in inexpensive titanium powders.

Also, given the high binder loadings generally employed in powder injection moulding, carbon can be readily introduced due to incomplete binder removal prior to sintering and/or to deleterious reactions between the decomposing binder, the debinding atmosphere, and the metal phase. Even at low concentration, these impurities can cause severe degradation in the mechanical properties of titanium and its alloys.

Earlier investigations on titanium injection moulding employed the thermoplastic and/or thermoset binder systems typically used in other MIM processes, with these materials constituting some 40 - 50 vol% of the feedstock, and downstream binder removal taking place via thermal pyrolysis. More recently, binders have been designed that are soluble in water or organic solvents and/or contain a catalyst that facilitates in situ
decomposition. However, feedstocks that employ these materials still require high binder loadings.

An alternate binder system has been developed at Pacific Northwest National Laboratory (PNNL) in Richland, Washington that is aimed specifically at injection moulding reactive metals such as titanium. It uses naphthalene as the primary binder component. The advantages of this compound are that it melts at relatively low temperature, 81°C, and can be removed via sublimation at temperatures below its melting point. Consequently naphthalene can act both as the binder backbone and as a solvent for additional binder constituents used to improve the mixing properties of the feedstock and the green strength of the moulded parts.

A research team from PNNL set up a series of experiments in co-operation with the nearby Battelle Institute to test the theory further, using titanium hydride powder to minimise concerns over oxygen contamination.
Several reports suggest that TiH2 self-reduces oxygen impurities during thermal decomposition (around 350°C or higher, depending on the ambient atmosphere) to form metallic titanium, although the mechanisms for this are not well understood. The characteristics of this powder are given in Table 1, and its distinctive angular morphology is shown in Figure 1. The binder system contained 93 vol% naphthalene, 6vol% ethylene vinyl acetate (EVA) to improve green strength, and 1 vol% stearic acid as a powder surfactant and lubricant. A combination of mould release agents was applied to the cavities and feedstock formulations of TiH2 and the premixed binder were prepared by shear mixing at 85°C.

Figure 1. The distinctive andgular morphology of TiH2 powder can be seen here.

The naphthalene was removed by sublimation under about 2 X10-2 Torr of vacuum at a 48-hour isothermal hold of 75°C. Final removal of the remaining binder and subsequent sintering were conducted in a single-step heat treatment run.

Specimens were placed onto a pre-reduced zirconia setter plate, and heat treated in a refractory metal furnace under a set of conditions determined from previously reported thermal analysis experiments: heat in Ar/2.75% H2 at 1°C/min to 375°C; hold at 375°C for three hours; heat at 1°C/min to 750°C, turn off the gas flow, and begin pulling a vacuum on the heating chamber while holding at 750°C for three hours; continue heating in 10-6 Torr vacuum at 5°C/min to a final soak temperature of 1100°C and hold for four hours; followed by argon-assisted convective cooling to room temperature.
Density measurements of the sintered specimens were conducted by simple geometric and weight measurements, as well as by the Archimedes technique using distilled water as the displacement medium. The microstructures of the sintered samples were characterised on polished cross sections using an optical light microscope and stereo microscope as well as a scanning electron
microscope quipped with an energy dispersive X-ray spectrometer. X-ray diffraction analysis was also carried out. Quantitative combustion analyses using the American Society for Testing and Materials standard ASTM D5373 were performed on the raw TiH2 powders and the as-sintered titanium MIM components to determine the nitrogen, carbon, and oxygen contents prior to and after MIM processing. Tensile testing was performed according to the conditions specified by ASTM E8.

In forming the initial liquid binder, both the EVA and stearic acid appeared to dissolve completely in molten naphthalene, yielding a transparent solution after a few minutes of heating at 90°C. After injection moulding and heat treating, specimen density was evaluated as a function of TiH2 loading in the initial feedstock material. Sintered density displayed a monotonic increase with increased powder loading. These two parameters commonly display little or no correlation in MIM processing. Instead, a strong dependence of part shrinkage on powder loading is typically observed, with changes in specimen volume occurring during both binder removal and sintering. That is, with many thermoplastic and thermoset binder systems, the debound parts tend to achieve approximately the same pre-sintered density independent of the initial powder loading in the MIM feedstock. However, the amount of volumetric shrinkage is inversely related to powder loading. In the naphthalene based process, the opposite appears to occur. Even though the specimens lose a significant amount of weight during debinding due to naphthalene sublimation, their volumes remain essentially constant. The density of the debound part (ie green density) increases with increasing powder loading.

Figure 2. The cylindrical test pieces used in the experiments at PNNL.

A similar phenomenon is observed in supercritical drying. In fact, this drying technique is used to synthesise extremely low-density aerogel materials precisely because evaporation takes place with virtually no commensurate change in material volume. The phenomenon is typically attributed to the low surface energies intrinsic to supercritical fluids. Consequently, as the fluid is removed through pores in the drying body, it exerts very little capillary or collapsing force on the remaining solid, subsequently causing little change in the solid's bulk volume during drying. Similarly, sublimation involves low surface energies in the vaporising ''solvent''. While this means that part distortion can be mitigated during debinding, it also implies that one must be careful to maximise the initial powder loading in the feedstock to achieve good part sinterability and reduce void formation. That is, higher powder loadings in the feedstock, and therefore higher green densities and greater particle-to-particle contact within the specimen, will yield less porosity and higher density in the final sintered body; a finding that agrees with results typically obtained in powder pressing studies.

An optical image of a sintered Ti- MIM tensile bar fabricated from feedstock initially containing 55 vol% TiH2 is shown in Figure 3(a). XRD and EDAX analyses indicate that the material is essentially pure a-Ti, which is consistent with the slow cooling rate that the specimen experienced after sintering. Two types of voids are apparent in this micrograph. The first are a set of large pores, typically >50 mm in size. Based on their size and shape, they appear to be the result of air bubbles that remained entrapped in the feedstock either after mixing or upon injection moulding and can likely be eliminated with further process optimisation.

The observation of the second set of voids, much smaller in size (~10 mm) and more spherical in shape, suggested initially that sintering in this sample may have been prematurely interrupted. However, as shown in the etched image in Figure 3(b) the majority of these pores are isolated within the grains of the material, indicating that the final stage of sintering is nearly complete.
The micrograph shown in Figure 4(a) is of a sintered specimen fabricated from feedstock containing 67 vol% TiH2. Like the specimen in Figure 3, this sample is also composed solely of a -Ti, but far fewer large flaws. In addition, the number of fine residual sintering pores is somewhat lower and their average size is measurably smaller, on the order of 3 - 5 mm in size. The difference in these two microstructures reflects the densification results shown in Figure 4, which as discussed previously originates from the dependence of green density on powder loading.

Figure 3 left. The large voids are probably trapped air bubbles. The sample contained 55 per cent titanium hydride. Figure 4 right. At higher titanium loadings, void size is much reduced. This sample contained 67 per cent titanium hydride.

Figure 5 shows a typical load-displacement curve obtained from tensile testing, in this case of a 92.3 per cent dense specimen fabricated from a 65 vol% powder loaded feedstock. The curve displays linear elastic deformation followed by a small amount of plastic yielding and strain hardening up to a peak load. The subsequent drop in load is relatively rapid, indicating non-ductile behaviour. Approximately 1 - 3 per cent ductility was measured in the specimens, substantially lower than the 12 - 18 per cent or higher levels generally exhibited by wrought a-Ti. The elastic modulus of these specimens decreases linearly with increasing porosity, which agrees with results obtained by Oh et al and Ledbetter, and has potential significance in eventually using this process to designing an implant material with a stiffness matching that of natural bone. A lower titanium modulus may be desirable in that it allows greater load bearing contribution from the adjacent bone, which in turn helps to minimise atrophy and other undesirable interactions with the implant.

Figure 5. A load displacement curve from a 92.3 per cent dense sample. These characteristics could be useful in medical applications.

There appears to be several reasons for the reduced level of ductility observed in tensile testing. As indicated in Table 2, the original TiH2 powder contains a significant amount of oxygen, which may be due to handling and storage under ambient conditions. Although the concentration of oxygen in the MIM specimens does not increase above that of the starting powder, it is substantially higher than the amount typically allowed in commercially pure titanium. Interstitial impurities such as oxygen, carbon, and nitrogen are tightly controlled in wrought product specifically because of their deleterious effect on ductility. Additionally, we observed moulding flaws on the exterior surfaces of the specimens. These can act as stress risers, causing premature failure in specimens already embrittled by a -Ti formation and high oxygen content. To test this hypothesis, the gage section of several specimens was ground to a -10 mm finish on a lathe using sequentially finer grades of emery paper and the specimens were subsequently tensile tested. The removal of the exterior moulding flaws significantly improves the strength of the MIM material, which is slightly higher than wrought titanium. In addition, the ductility of these samples improves to 3 - 5 per cent, although this is still only half of the desired value. Again, the higher levels of oxygen in the MIM materials are the likely cause of both higher strength and lower ductility relative to wrought product.

The strength of the as-moulded specimens is substantially lower than the ground specimens, indicating that the exterior moulding flaws do play a significant role in these MIM-formed materials. It is anticipated that with process optimisation, particularly in the feedstock mixing and injection moulding steps, and the use of higher purity starting powder under inert conditions, the mechanical properties of titanium components made by this process will improve significantly. These studies are currently in progress.