Metallurgy, Corrosion & Surfacing Technology Group, TWI Ltd, Granta Park, Great Abington, Cambridge, CB21 6AL, UK. T: +44 (0) 1223 899 000, F: +44 (0) 1223 894 717.
Paper presented at Euro PM 2008 Congress and Exhibition, Mannheim, Germany, 29 Sept - 1 Oct 2008.
Powder metallurgy (PM) processes have high productivity and are ideal for making near net-shape parts of especially complex geometries from a range of materials, which maximises material utilisation, and hence minimises or eliminates secondary operations such as machining. Secondary operations are common for components made via liquid metal processing, and result in an additional step in manufacturing with substantial cost and waste implications. Despite this obvious advantage of PM processes, however, the joining of materials synthesized from powders has been associated with difficulties related to their inherent characteristics, such as porosity, contamination and inclusions, at levels, which tend to influence the properties of a welded joint.
This paper presents a critical review of the current state-of-art of welding PM components. It also seeks to identify preferred joining processes and identify apparent technology gaps in joining of PM parts, in terms of initial processing and attendant materials issues, with an emphasis on offering solutions to welding problems.
In PM materials processing techniques all or some constituents of a part are employed in particulate form with certain characteristics of composition, morphology and size, and compacted into a high precision product. The ability of PM to produce high quality, complex parts with close tolerances and high productivity presents significant advantages, such as energy efficiency, with potentially low capital costs. PM is widely used for a range of applications, such as dental restorations, implants, bearings and automotive transmission parts, from biomedical to automotive industry sectors.
PM components are becoming increasingly attractive as substitutes for wrought and cast materials in various applications. However, it is possible to increase further the use of PM by exploiting the ability to manufacture complex geometrical configuration by joining PM parts to one another or to other cast/wrought products. The main issues restricting the welding of PM parts have been porosity, impurities and the fact that some PM parts have high carbon content. Of these, the most important characteristic of a PM part for welding has been porosity, created either deliberately to make a porous part or incidentally due to insufficient densification. Powder particle characteristics (such as particle shape, size and surface area) determine the porosity or relative density of a powder compact, which in turn influences several important physical properties of the preform, such as thermal conductivity and hence hardenability, as well as thermal expansion. It has been noted that porosity can also act as a trap for impurities/inclusions that could potentially have an impact on secondary operations such as welding when entrapped impurities in the pores could be deleterious to the weldability of the PM part, e.g. by encouraging solidification cracking. PM parts have been reported to be susceptible to cracking in the heat affected zone when welded, due to the porosity of the preform and limited area of inter-particle bonding giving low ductility adjacent to the joint. Hence these locations may be unable to resist the thermal stresses generated as a result of contraction in a fusion weld. In addition,welding is commonly associated with resultant distortion, whereas PM parts are known to provide good dimensional and geometrical accuracy and thus, if machining operations are to be avoided, distortion must be minimised via selection of appropriate joining technology.
Therefore, widespread success with welding of PM parts requires understanding of the influence of porosity, chemical composition, impurity level and overall cleanliness, upon weldment properties such as weld metal and HAZ cracking,ductility and toughness, residual stresses and distortion. These issues are addressed in this review.
Summary of Different Joining Processes for PM Parts
Joining processes applicable for PM parts can be categorised as solid-state and liquid state. The solid state processes such as diffusion bonding and brazing have been predominantly used for lower density porous parts. In comparison, parts with higher densities or minimal porosity are typically treated as fully dense wrought materials, and these are typically welded using fusion-based joining processes, including arc welding, ie gas tungsten arc (GTA),gas metal arc (GMA), electron beam (EB) and laser welding. Further joining techniques such as adhesive bonding and shrink fitting may be used for some applications but are not considered here.
Arc welding of PM parts may give porosity with an associated detrimental influence on the weld integrity. It has been reported that gas metal arc (GMA) welding of powder compacts, which are not fully dense, can result in porous welds, and weld toe cracking, the latter presumably resulting from low ductility of the original PM part. As would be expected, the density of a PM part and its composition are reported toaffect the tendency for porosity during welding. [2,3]
In addition, it has been reported that welding of ferrous PM parts can form a soft pearlitic microstructure as a result of slow cooling, due to porosity reducing thermal conductivity, which can allow strains to be accommodated that would otherwise give rise to cracking. In addition, hydrogen can diffuse out of the metal into and through the inter-connected pores, hence reducing susceptibility to fabrication hydrogen cracking, to the extent that sintered steels are considered to be more resistant than wrought steels.
For ferrous PM parts, it is also worth mentioning that, when subjected to a steam treatment (heating the parts to a temperature in the region of 550°C and exposing them to a water vapour), a thin layer of Fe3O4 is formed both on the outer surface and on the surfaces of the interconnected porosity, giving improved corrosion resistance, increased surface hardness, compressive strength and wear resistance. This or a similar post sintering heat treatment of ferrous sintered parts will, however, prevent satisfactory welding. The problem is due to a resulting oxide film, which is slightly porous, probably contains moisture and has an insulating effect and hence is potentially a source of weld metal porosity and a potential cause of cracking in weldments.
Arc welding (GTAW and GMAW) has also been attempted for non-ferrous metal matrix composite (MMC) systems, for example, SiC particle-reinforced Al. Gross porosity and delaminations in both weld metal and HAZ were reported to be frequent. It was claimed that despite large volumes of particulates in the Al alloy matrix, a wide range of MMCs can be fusion welded, with weldability similar to Al. 
There are advantages of laser welding for PM parts, as it is a highly automated process, offering precision and control. Welding speeds of several m/min are possible, with a low heat input, resulting in a small HAZ, and limited thermal distortion and residual stress. 
However, various defects such as blowholes probably due to entrapment of gases that cannot leave the melt during rapid solidification were observed in laser welding of sintered steel parts in general, together with occurrence of hydrogen cracking of medium carbon parts cold and hot cracking have been observed in laser welding of sintered steel parts. It was recommended by one author that a filler wire could be used to prevent all these kinds of defect, although filler wire addition is an additional complication in laser welding. Low C-steels (typically up to 0.3%C) can be satisfactorily laser welded and it has been reported that a sintered steelpart can be acceptably laser welded to a wrought counterpart, as long as both components have low carbon content. However, laser welding of medium C steels (typically 0.3-0.6%C) appears to be difficult, due to the formation of a hard, brittle and hydrogen crack sensitive martensitic structure in the joint, due to rapid cooling. Pre-heating may help by inducing a softer bainitic microstructure with some fine pearlite,which will increase toughness and hence improve the defect tolerance of the joint and reduce sensitivity to hydrogen cracking.
Gas carburised and oil quenched steels are considered to be difficult to weld, because of significant blowhole formation, probably due to entrapped oil and gas in the pores. Elsewhere, it was noted that smooth, discontinuity-free welds can be produced at slower travel speeds and lower beam powers. It was also reported that beam weaving laser welding technique would suppress porosity and that increased width to depth ratio of the molten metal was beneficial for the escape of bubbles in the weld zone. 
Laser welding of sintered austenitic stainless steel (grade 316L) was reported to be very easy, resulting in good joints. In trials with sintered Al alloys, it proved difficult to obtain a sound joint because of porosity formation, resulting in spongy welds. It is thought that oxidation was the probable cause, possibly creating locally overheated spots when laser energy is absorbed by the oxides more than the metal and another possibility is that the oxide absorbs moisture and this is released during welding.
EB welding is normally carried out in high vacuum (eg 10-6 mbar). Therefore it is a batch process and hence may be expensive and thus restricted to high value parts only. It has a tendency to give high cooling rates and high hardness in C-steels, similar to laser welding, but is likely to have a greater tendency towards pore formation due to the vacuum encouraging trapped gas to try to escape during welding. It is reported that porosity increased with reduced travel speed. However, it has been demonstrated that weld metal porosity content in sintered ferrous compacts with a range of porosities can be controlled by beam parameters and a non-vacuum EB welding process has been used for sintered parts. 
Any residual films, such as heat treatment quench oil trapped in the pores of a PM part, were found to have a detrimental effect in EB welding. A fine-grain size (< ASTM 8-9) in asintered PM part was reported to be essential for good EB weldability, presumably due to improved ductility, in high temperature PM superalloys. This is related to increased ductility and toughness of a fine grained material for absorbing strains upon solidification. Cracking has been observed in EB welding of PM superalloys designed for aerospace applications (engine components such as turbine discs) which required further investigation. 
However, despite the problems described above, EB welding has the ability to give low distortion, again similar to laser welding, or at least uniform distortion effects, which is important for preserving the dimensional stability of near net shape PM components.
Resistance Projection Welding
Projection welding is one of the most widely applied welding processes for sintered PM parts. An advantage of projection welding is the limited distortion associated with welding. It has been emphasised that one potential difficulty for the success of projection welding is the likely presence of an oxide layer on parts, for example ones subjected to steam treatment, typically for ferrous sintered parts. Such a layer could prevent satisfactory welding through an insulation effect at the interface and as a potential source of moisture and hence porosity.[15,16] It is therefore recommended that any steam treatment on PM parts, as described earlier, should be done after welding. It has been possible to weld high carbon PM steels and case hardened parts with resistance projection welding. 
Friction welding is a solid phase process mainly used for wrought products in a range of geometries. Particular advantages are the absence of flux, filler or need for a protective atmosphere, which makes the process extremely attractive. The friction welding process is highly suitable for welding PM parts, due to the promotion of pore closure, which can potentially lead to a pore-free weld interface, and are fined microstructure, particularly for Al-based PM parts and MMCs. Particularly for Al alloy PM components, friction welding is also useful in breaking any oxide layer on the particles via the deformation within the bond area which in turn contributes to improved bond strength. One potential disadvantage of friction welding may be associated with the change in microstructure due to re-orientation and deformation of sintered metal grains, which may create a potential weak region in the joint, thereby reducing fatigue performance. 
In brazing, the PM part will typically act like a sponge, which can potentially draw away the brazing alloy from the joint into the porosity in the part, leaving insufficient material for bonding. Techniques have been developed for avoiding the absorption of the brazing alloy by the sintered steel and for the incorporation of braze alloy into the sintering process.[19,20] One of the most common brazing problems is the presence of secondary products, eg due to flux reactions that would stop further infiltration, by blocking the pores. Significant factors affecting the brazed joint strength are reported to be surface condition of the particles (as is the case for other techniques) and the part's surface roughness. New brazing techniques such as laser brazing and new brazing alloys, have been developed for joining sintered components ina mass production environment. Assemblies can consist of PM to PM or PM to wrought or cast structures.
Ferrous parts can be readily diffusion bonded provided they are of a fairly small size. Sintering and diffusion bonding can potentially be carried out in the same furnace. A eutectic reaction can be employed to provide a transient liquid phase for bonding at the interface. Secondary reaction products such as oxides, which may form as a result of the reactions, may reduce the bond strength. However, it has been observed that high bond strength can be obtained by activating diffusion via the addition of suitable elements, such as Cu in ferrous PM parts. Nevertheless, rather low strength joints can be expected from diffusion bonding, which may be limited to certain geometries and alloy compositions. Several applications to date have involved diffusion bonding of light materials such as Ti alloys and MMCs. More conventional C-Mn steels have not been commonly diffusion bonded.
A broad range of powder metallurgy parts are available, in a wide range of alloys, and there is no single best way to join them. However, there are a number of welding characteristics of PM parts that are somewhat different to those associated with wrought or cast equivalents, either as a consequence of the PM production route or the typical applications of PM parts. For example, as PM parts are used in a variety of high precision applications, it is desirable to weld with a process that gives minimal distortion. This favours low heat input processes such as laser and EB welding but any low heat input process will also inevitably give rapid cooling and hence high hardness in steel parts, particularly for higher C contents. It is not clear whether sintered parts can be EB welded in a vacuum however, considering inherent porosity where gases and impurities can be retained and entrapped in the weld. Reduced pressure EB welding, which only requires a vacuum of the order of 10-3 mbar, may be more suitable for welding sintered PM parts, to overcome difficulties in achieving adequate vacuum but is still under development.
As with any welding, the main requirements for welding of PM parts is that the process should not introduce defects. Powder metallurgy parts contain porosity, either deliberately, and hence with a fairly high volume fraction or as a consequence of the inability to obtain complete densification, in which case it is typically at a low level. Any porosity in the PM part will tend to trap contaminants and gas, which can cause pores in the weld metal and introduce species that increase sensitivity to both hot and cold cracking mechanisms, eg sulphur and phosphorus contamination will encourage solidification cracking, whilst moisture and carbon contamination will encourage hydrogen cracking. In order to minimise these problems, cleanliness of parts is vital. In this respect, avoiding steam treatment will be beneficial, and degreasing important prior to welding. Where contamination exists, use of a filler metal that is more tolerant to contamination than the parent material, eg a nickel alloy, may be beneficial, in which case arc welding processes are preferred. One possible advantage of an interconnected porosity may be that hydrogen can diffuse out via the open porous structure, in welding PM steel parts, which may make them more resistant to hydrogen fabrication cracking.
Where present at significant levels, porosity of the parent material may lead to tearing of the material adjacent to the weld, simply due to the development of plastic strain beyond the capacity of the PM part, perhaps exacerbated by geometric effects at the joint. In such cases, use of low heat input is preferred, to reduce the amount of material strained and friction welding may be advantageous, as the compression involved tends to close pores. Indeed friction welding may be generally useful for PM parts due to the compressive force involved and the fact that friction welding squeezes the original surface layer, which may be contaminated, out of the joint.
It is apparent that powder particle characteristics, which influence densification of a PM part and therefore its final porosity, have not received much emphasis in relation to welding studies. In order to achieve a better control of porosity, and minimise its detrimental effects in welding, consideration should be given to the influence of powder particle characteristics such as particle shape, size and surface area on the density and porosity of a powder compact for improved weldability.
The following conclusions are drawn:
- Welding is widely used for a range of PM components for diverse applications across several industry sectors but limitations exist due to the inherent porosity, contamination within the pores and the effect of porosity on the ductility of the material and hence its ability to withstand strain in a weld heat affected zone.
- For porous materials, use of low heat input is recommended to reduce strains developed in the heat affected zone and minimise the risk of tearing adjacent to the weld. Where applicable, friction welding and projection welding may be advantageous as they involve compression, which tends to close pores in the joint area. Porosity of the weld metal is likely to result when any fusion based process is used and rapid cooling rates may increase porosity due to the limited opportunity for bubbles to escape from the molten metal.
- Where contamination causes weld metal cracking, the use of a welding process that allows introduction of a consumable filler material, such as the various common arc welding processes, is beneficial. In extreme cases, use of a tolerant nickel based filler metal may be necessary to avoid cracking. Any process that introduces contamination or oxide to porous parts, such as steam treatment is likely to be detrimental to the ability to make sound welded joints and should be avoided. Similarly, cleaning of the joint surfaces prior to welding is beneficial.
- Laser and EB welding have also found application in welding of PM components when the inherent low distortion of these processes is an advantage, ie for high precision parts, but these processes have rapid thermal cycles, which encourage hardening and cracking of the weld metal, with limited opportunity for filler addition, and are not always applicable. EB welding in particular is likely to suffer from porosity and the necessary high vacuum might not be achieved when gas contamination is extensive. Reduced pressure EB welding, which operates with higher gas pressures, is attractive for sintered porous parts.
- J C Thornley, Welding Design and Metal Fabrication, November 1973, Vol. 12, pp 399-402.
- K Couchman, M Kesterholt and R White, Seminar on Secondary Operations (Int. Conf. PM 1988), Orlando, Proceedings, pp33-39.
- M A Greenfield, R F Geisendorfer, D K Haggend and L P Clark, Welding Research Supplement, May 1977, pp43-148.
- J H Devletian, Welding Journal, June 1987, pp. 33-39.
- A Rocca and G Capra, SPEI Vol. 1031 GCL-7th Int. Symposium on Gas Flow and Chemical Lasers, 1988, pp635-645.
- A Joskin, J Wildermuth and D F Stein, Int. J. Powder Metallurgy and Powder Technology, 1975, Vol. 11, no. 2, pp137-142.
- E Mosca, A Marchetti and U Lampugnani, Proc. Int. Conf. PM (Powder Metallurgy),1982, Florence 1982, pp193-200.
- S Chiang and C E Albright, Journal of Laser Applications, Fall 1988, Vol. 1 (1), pp18-24.
- X Zhang, W Chen, G Bao and C Zhao, Science and Technology of Welding and Joining, 2004, Vol. 9, no. 4, pp379-376.
- G M Alexander-Morrison, A G Dobbins, R K Holbert and M W Doughty, J Materials for Energy Systems, June 1986, Vol. 8, no. 2, pp79-79.
- JA Hamill, Jr, Welding Journal, February 1993, pp37-45.
- G W Halldin, S N Patel and G A Duchon, Progress in Powder Metallurgy, Vol. 39, pp267-280.
- J H Davidson and C Aubin, Proceedings: High Temperature Alloys for Gas Turbines, 1982, Liege, Belgium, pp853-886.
- P Adam and H Wilhelm, Proceedings: High Temperature Alloys for Gas Turbines, 4-6 Oct 1982, Liege, Belgium, pp909-930.
- W V Knopp, Society of Automotive Engineers, Automobile Engineering Meeting Toronto, Canada, Oct 21-25, 1974, 740984.
- J E Middle, Chartered Mechanical Engineer, July 1980, Vol. 27 (7), pp55-60.
- L J Johnson, G J Holstand, M J O'Hanlon, 1971, Fall P/M (Powder Metallurgy) Conf. 19-20, Detroit, Michigan, MPIF/APMI, pp193-203.
- W A Baeslack III and K S Hagey, Welding Research Supplement, July 1988, pp1395-1495.
- P Beiss, Powder Metallurgy, 1989, Vol. 32, no.4, pp277-284.
- W V Knopp, Materials Engineering, 1975, 12-75, p34.
- K Okimoto and T Satoh, Int. J. Powder Metallurgy, 1987, Vol. 23, no.3, pp163-169.
- N Janissek, DVS Berichte, no. 243, Proceedings Brazing, High Temperature Brazing and Diffusion Welding Conference, Auchen, 19-21 June 2007, pp1-5.
- H Duan, M Kocak, K H Bohm and V Ventzke, 2004, Science and Technology of Welding and Joining, Vol. 9, no. 6, pp513-517.
- A Akutso and M Iijima, 1985, Modern Developments in Powder Metallurgy, Vol. 16, pp195-208.
- T Tabata, Nasaki, H Susuki and B G Zhu, Int. J. Powder Metallurgy, 1989, Vol. 25, no,1, pp37-41.