102 Hot isostatic pressing
Coble and Flemings (1971) confirmed that pores, if sufficiently fine, and in a fine-grained matrix, would gradually disappear given a few tens of hours and a temperature high enough for a 'sintering reaction' to occur. They found that the application of modest pressure, about 20 atmosphere, greatly assisted the process. The development of hot isostatic pressing (HIPping) was the outcome. Much higher pressures were employed, usually nearer 1000 atmospheres, and temperatures as near the melting point as was practical. The conditions for HIP have been defined in an elegant study by Arzt et al. (1983) in which they characterize tool steel, superalloys, alumina and ice.
The significant improvement in mechanical properties, particularly average fatigue life, reported for certain alloys after HIPping is probably at least in part due to the contribution towards the deactivation of entrained double oxide film defects (bifilms) as fatigue crack initiators. We shall present the evidence here for HIPping as a solid-state process for the closing of pores and bifilms.
When a cast Al-7Si-0.3Mg alloy with oxide film defects, like that shown in Figure 10.la, is subjected to HIPping treatment, the applied pressure at temperature close to its melting temperature induces a substantial plastic deformation in the casting causing the defects to collapse and their
of the crystal structure that would be expected to encourage diffusion bonding across the oxide/oxide interface. Analysis of bifilms that have acted as fatigue crack initiation sites have confirmed their conversion to a spinel and confirms that fatigue properties are improved by HIPping (Nyahumwa 1998 and 2000). A further positive finding from this work was that compared to filtered castings, the unfiltered but HIPped castings exhibited higher fatigue performance, despite larger maximum defect sizes, implying some degree of bonding across the crack. The application of HIP to castings shown in Figure 9.12 resulted in fatigue test samples that did not fail. The runouts at the stress 150MPa reached nearly 108 cycles (not shown). At the higher stress of 240 MPa improved fatigue lives were still recorded (Figure 9.13) although it is of interest that in all except the most resistant specimen, the fatigue failures still occurred from oxides, probably healed or partly healed bifilms.
The liquid-state healing mechanism as described in section 2.4, and the HIP solid-state healing mechanism, are suggested to be analogous. However, there are interesting differences. In the liquid state the healing mechanism operates at high temperature (i.e. 700°C for an aluminium alloy) and with only the moderate external applied pressure due to depth (<0.1 MPa). The solid-state oxide film healing process operates at very high-applied pressure but lower temperature (i.e. 100MPa at 500°C for Al-7Si-Mg alloy). In both the liquid and solid conditions the pressure drop will be expected to occur internally within the bifilm as the oxidation reaction proceeds. When all the oxygen is consumed, it is expected that the nitrogen may subsequently be consumed to form nitrides. Pores in this alloy before and after HIPping are shown in Figure 10.1.
Whether bifil ms are actually 'healed' (i.e. effectively welded) may relate to the chemistry of the alloy and its films. Some bifilms appear to heal as we have seen above. However, others appear to be resistant. These important factors are not well researched at this time. Thus HIPping has limitations that do not appear to be widely known or understood. However, evidence for possible mechanisms is discussed below.
In contrast to this beneficial action of HIPping found by Nyahumwa and others as described above, Wakefield and Sharp (1992) observed HIPping to have no beneficial effect on the fatigue properties of Al-lOMg alloy castings, despite the closure of pores and cracks. The inference is that the bifilms in this alloy proved impossible to deactivate, resisting effective bonding. This is attributed to the magnesia (MgO) film formed during oxidation of Al-lOMg and entrained during casting. The magnesia film is (i) thicker, and (ii) has a stable structure that does not transform during HIPping (Nyahumwa 1998 and 2000). This lack of any substantial atomic movements explains the inert nature of the highly stable MgO compound.
Similarly, it would be expected that A1 alloys with very low Mg contents, which would be expected to contain entrained alumina films, would be similarly resistant to HIPping because of the great stability of alumina in the absence of sufficient Mg to convert it to a spinel structure.
During the early development of the Pegasus engine for the Harrier Jump Jet 25 polycrystalline Ni-based alloy turbine blades that had previously been scrapped because of their content of porosity were subjected to HIPping, and were fitted to a test engine alongside sound blades to evaluate whether HIPping might be a satisfactory reclamation technique for blades that otherwise would be scrapped. The HIPped blades failed within a few hours, damaging the engine and forcing a rapid shutdown of the test. The failures had occurred by creep cavitation at the grain boundaries of recrystallized regions in the centre of the castings. Almost certainly the original porosity would be caused by aluminium oxide or aluminium nitride films entrained by the severe turbulence that is usual during the vacuum casting process. (The vacuum is known to contain plenty of residual air to ensure the growth of surface films.) The great stability of the films, formed at the high casting temperature would ensure that they were resistant to any re-bonding action. The recrystallization would have happened because of the large plastic strains that were a necessary feature of the collapse of the porosity. However, the subsequent grain growth would expand grains up to local barriers such as bifilms. Thus the bifilms, effectively unbonded, and so acting as efficient cracks, were automatically located at the grain boundaries from where the failures were seen to occur.
The closure of internal cavities usually causes negligible changes to the overall dimensions of the casting if the pores are small and/or deep seated. For large or near-surface pores, however, the collapse of the surface of the casting in the form of a localized sink may scrap the casting if the depression exceeds the machining allowance. In a severe case, the surface may puncture, opening up the internal cavity to the surface (Zeitler and Scharfenberger 1984).
Naturally, HIPping cannot work if the pores are already connected to the outer surface of the casting. Such pores will never heal. Unfortunately this is all too common. The existence of the various forms of surface-connected porosity is well known to those who HIP castings. In particular, pressure die castings are woefully resistant to the benefits of HIPping because of their many surface connected bifilms (even traversing the so-called 'dense' outer layers of the casting). However, many gravity-filled or even counter-gravity-filled castings exhibit surface-
connected pores as a result of bifilms intersecting the surface.
However, although an excellent start, even good filling of the castings will not guarantee a satisfactory response to HIP treatment. As we have seen in section 7.5.1 poor feeding in a long-freezing-range alloy can easily create surface-connected pores.
For a reliable HIP response, the surface of the casting must be sound.
Finally, therefore, although the mechanical properties of castings usually exhibit an improvement, in the sense that their average properties are raised, the Weibull modulus most often falls. This is a direct result of the healing of many defects, but leaving a few unaffected, for the various reasons we have seen. Thus the scatter of properties is increased. Regrettably, this is one of the greatest disadvantages of HIPping, often overlooked.
It seems curiously perverse that most metallurgical texts continue to foster the erroneous assumption that working eliminates casting defects. Usually, the strains involved in most working operations are too low to effect any significant welding of faults. Bifilms in general are merely pushed around, if anything, growing worse before (if ever) gelling better. The growth of casting defects by plastic working is well known to those who work in the forging industry. There are good reasons for this behaviour.
During the working of cast material, for instance by rolling, it is to be expected that the defects will be elongated in the direction of working. The elongation of the defect necessarily rotates it, to align it along the rolling direction.
In addition, of course, its elongation increases the area of the defect. The newly extended surfaces in the bifilm would necessarily be oxidized by the remnant of air entrained in the defect. The entrapped air would be contained partly amid the microscopic pores between the crystals of the oxide, and partly in macroscopic reservoirs formed by folds and bubbles. During rolling, the continued oxidation of the expanding surfaces would hinder the welding of the interfaces that were being newly created, and by this means assisting the defect to grow as a crack, possibly to several limes its initial size. The great effectiveness of the creation of this newly oxidized extension to the crack would be a consequence of the small oxygen requirement; the oxide would grow to a thickness of only nanometres at the hot working temperature (which is. naturally, much lower than the temperature at which the film was first formed on the melt). In other words, the entrained residual oxygen is highly efficient at these temperatures to continue the oxidation of new area as it is formed.
Miyagi et al. (1985) used ultrasonics to observe this increased area of cavities in 5083 alloy (Al— Mg type) during the early stages of hot rolling. Microcavities near the surface of the rolled plate seemed to close early, but those nearer the centre were long-lived. Micro-examination revealed that they were smooth sided, and appeared to have opened and expanded along grain boundaries. (The association with grain boundaries is a feature to be expected of bifilms.) They were seen to reduce in size only after reductions of over 50 per cent.
Later, if the extension were sufficiently large to consume the remaining air, further extension would result in the welding up of new extensions to the crack. It seems likely that normal forging and rolling do not reach this stage for most alloys, and so do little to heal defects, simply extending and realigning them as described above. Processes such as extrusion and multi-pass rolling may sometimes account for a more complete elimination of the casting defects because of the much greater strains that can be involved. Even so, it seems that many defects remain, as indicated by the evidence from corrosion (section 11.2) among others.
Work by Harper (1966) on the hot rolling of wire-bar copper showed that rounded bubbles of gas are collapsed asymmetrically, forming convex bubbles, from which are pinched off strings of microscopic bubbles only 30 nm diameter. These minute bubbles appear to be completely stable in the solid. They can only be removed by remelting. It seems they do reappear in welds.
Conversely, for continuously cast steel slabs, Leduc etal. (1980) found the wide scatter in porosity was eliminated by working, the steels becoming fully dense at about 75 per cent reduction. Examination of the microstructure indicated that all the pores were eliminated. This is probably understandable in view of the low residual levels of Al in the steels (0.02-0.04 percent) and absence of other strong oxide-forming elements. Any surface oxide would therefore have been expected to be a liquid iron-manganese silicate, and so any bifilms. if present, would have been easily welded shut.
Average user rating: 5 stars out of 2 votes
Post a comment