EDITORIAL

Papers on aluminium alloys in the MST archive

There have been some excellent papers on aluminium published in our journals in the past 40 years or so and my main objective has been to identify ones that, in some cases, represent landmarks in our understanding of aluminium metallurgy and, in other cases, are just ‘good reads’ – by which I mean ones that tell convincing stories elegantly and that I enjoyed reading. This guest editorial was commissioned not only to celebrate the silver anniversary of MST but also to react to the full digitisation of MST and one of its predecessors, Metal Science (Journal), MS(J); I have, therefore, included MS(J) in my survey and also explored papers in the other predecessor of MST, Metals Technology, MT. I hope that Maney will eventually send MT to the digitisers!

In the following sections, I deal with both traditional metallurgical topics and with new areas of research that have emerged in the period under consideration.

The early years of MS(J) and MT coincided with the expanding use of transmission electron microscopy in metals research and I was particularly impressed by a few papers that used in situ techniques to measure physical parameters.1^–3 Two intriguing papers on intergranular embrittlement of aluminium by liquid gallium make interesting reading, especially if one has a need to cause a sample to separate at its grain boundaries.4,5

Rapid solidification

Rapid solidification techniques emerged in the 1960s, whereby melts could be cooled at ∼106 K s⁻¹, yielding extremely fine microstructures and abnormally high solute supersaturations. Thin ‘splats’ were formed from droplets sprayed onto conducting substrates and were then comminuted, compacted and (usually) extruded to bulk forms. Al–Cr–(Fe or Zr) alloys can be produced having abnormally high supersaturations of solute that allow extremely fine dispersions of very stable precipitates to be produced. One thrust of the research was to produce alloys that were mechanically strong and stable at temperatures above ∼200°C (the limit of off-the-shelf alloys) and Gardiner et al.6 reported some quite spectacular high temperature mechanical properties in an Al–7·5Cr–1·2Fe alloy that had been vapour deposited (the ultimate splatting technique)! Forming these materials into useful products (even a tensile specimen) is not easy because they are very strong and have low ductility at ambient temperature, so they must be warm worked; a balance needs to be made between ease of forming and loss of properties due to microstructural coarsening. Saunders and Rivlin7 survey phase diagrams of such alloys and decomposition mechanisms are treated thoroughly by Zhang and Loretto.8 Such high solidification rates are not easy to obtain under commercially viable conditions and much research has been done using lower solidification rates (about 10³–10⁴ K s⁻¹) obtained by gas or water atomisation or by the Osprey/Cospray route to forming billets. In several cases, authors added transition metals (e.g. Zr, Ti, Mn, V, Sc) to alloys having compositions close to those available commercially and demonstrated much-improved mechanical properties.9^–15 Sadly, these products seem not to have made any commercial impact.

Superplasticity

Superplasticity sprang into the world in the early 1960s and the first plausible model of the deformation process was published in MSJ by Ball and Hutchinson16 in 1969; they proposed that grain boundary sliding led to dislocation pile-ups, the back-stresses from which inhibited sliding, and the rate of sliding was controlled by climb of dislocations from the pile-ups to sinks such as grain boundaries. That paper is still quoted 40+ years later and it led to a multitude of other models that are critically reviewed by Todd.17 Almost all known superplastic alloys at the time that Ball and Hutchinson published their paper were eutectics or eutectoids containing large volume fractions of second phase particles that inhibited grain growth at high temperatures but, in the early 1970s, two groups on opposite sides of the world published papers using almost identical methods for making commercially available Al based alloys highly superplastic. These are the TI/BA group in the UK and Matsuki and co-workers in Japan; both used Zr to refine the castings and to inhibit recovery and they also applied high levels of work to the ingots. Matsuki et al.18,19 gave their materials a recrystallisation anneal before superplastically deforming them whereas the TI/BA group20,21 raised the as-rolled sheet quickly to the deformation temperature; they found that recrystallisation occurred ‘dynamically’ and could yield a finer, stable grain size. TI established a completely new company (Superform Metals) in 1973 to form components from their material. A sobering thought for academics is that it took 30 years for a detailed understanding of the ‘dynamic recrystallisation’ process to emerge (Ridley et al.22) but that did not stop Superform Metals making and selling about 2·5 million components in the interim.

The revelation that alloys close in composition to those available ‘off the shelf’ could be rendered superplastic, and the realisation that many superplastically formed components do not need an alloy capable of 1000% elongation, led others to induce superplasticity in several standard alloys by suitable processing; the most successful were AA 7475, used in aerospace applications, and AA 5083, a general purpose alloy with good corrosion performance; several papers characterising these alloys have appeared in MST.23^–25 Grimes et al.26 pointed out that the next major challenge for Al alloy development is to produce commercially viable alloys capable of being deformed at much higher strain rates, making feasible the forming of components for high volume products. So far, this has not been achieved commercially although the latter paper gives a strong indication of how this could be achieved. One problem is that there is a lack of funding for such research and, perhaps more importantly as Grimes27 later pointed out, ‘it is, sadly, the case that the majority of university-based work has been concerned with the minutiae of superplastic forming mechanisms. Such work has had very little influence on the development of commercial superplastic alloys or the processes by which such materials are turned into engineering components.’

A common characteristic of superplastic alloys is that they cavitate as a result of deformation; this can be detrimental to service properties. For all practical purposes, the growth of cavities in superplastic Al based alloys is understood,28,29 but the issue of the nucleation of these voids is still unclear. In most alloys, some voids are formed at inclusions during processing to sheet but the possibility that voids are also nucleated during the component forming process cannot be ruled out. An excellent paper by Martin et al.,30 who used X-ray microtomography to examine cavity growth and coalescences in AA 5083, warns that, because of the complex geometries of coalesced voids, it may not be valid to conclude from measurements of void volume fractions on planar metallographic sections that voids are actually nucleated during forming. More detailed research along those lines is needed.

Grain refinement

Grain refinement, not necessarily to the level required for superplasticity, has been reported in numerous papers in MST and its predecessors. I deal first with grain refinement of castings. It has been known for a long time that this may be achieved by adding intermetallic aluminides, and Clyne and Robert31 examined the effects of adding, via master alloys, Al₃Ti, Al₃Nb and Al₃Zr particles to commercial purity aluminium melts. Later, an extensive study of master alloys containing Al–Ti–B and their ability to grain refine aluminium alloys was reported by Arnberg et al.32 Since those early years, a large literature on this subject has built up, which is expertly reviewed by Quested.33 TiB₂ particles are not very effective at nucleating aluminium; solute is required to provide constitutional undercooling and growth restriction.34,35 A major advance in understanding of this process has been made by Greer’s group in Cambridge. In an inspired experiment,36 TiB₂ particles were added to an Al₈₅Y₈Ni₅Co₂ alloy melt and then rapidly quenched to give an amorphous matrix. The (0001) faces of the TiB₂ particles were found to be covered with thin epitaxial layers of an Al₃Ti-like phase on which α-Al was formed, again in epitaxial orientation. (The advantage of rapid solidification is that diffusion-controlled changes due to slow cooling and recalescence are virtually eliminated). The observation that aluminium nucleates readily on the basal planes of the (coated) TiB₂ particles led to the ‘free growth’ model. This posits that, once the radius of the nucleated solid has grown beyond half the diameter of the TiB₂ facet on which it sits, it can grow freely into the melt; the undercooling to achieve this state is inversely proportional to the facet diameter, so the largest TiB₂ particles are the most effective. The consequences are neatly summarised in Quested’s review33 and are commercially very significant.

A popular technique for microstructural refinement is equi-channel angular extrusion in which a bar is forced through a die containing a bend (∼120°) that imparts redundant strain to the sample without change of dimensions. Multiple passes allow very large effective strains to be achieved. Bowen et al.,37 working with an Al–0·13Mg alloy, described in detail the microstructural changes obtained by imposing effective strains up to ∼10. Hayes et al.38 used this technique to obtain true grain sizes in an Al–3Mg alloy down to ∼0·2 μm diameter and these specimens obeyed the Hall–Petch formula. However, below a grain size of ∼0·6 μm, the ductility and uniformity of deformation deteriorated. Equi-channel angular extrusion is an excellent research technique but is not readily adapted to production of commercially relevant quantities of material. A pertinent question is, ‘what processing conditions, using a conventional method such as rolling or extrusion, are needed to produce very fine grains of a given size?’ This was answered by Gholinia et al.,39 using an Al–3Mg–0·2Cr–0·2Fe alloy; they produced a ‘processing map’, plotting log(strain rate) versus temperature, for obtaining grains 1–2 μm and 2–3 μm diameter. Unfortunately, the processing windows were rather restricted and not at all practical. Further work along these lines needs to be done with different alloys.

Recrystallisation

Recrystallisation of Al and its alloys has figured frequently in our journals. A series of papers in the March/April 1979 issue of MS gives a good indication of the state of knowledge at that time and of the range of techniques being used. Humphreys40 gave a comprehensive review of recrystallisation mechanisms in two-phase alloys. Furrer and Hausch41 described the microstructural changes during rolling and annealing of a commercial Al–1·0Mn alloy, showing that new grains are nucleated by coarse second phase particles and that fine precipitates inhibit recovery and lead to an increase in the recrystallisation temperature. Gardner and Grimes42 demonstrated that Al–Mg and Al–Zn alloys could recrystallise ‘dynamically’ during rolling; they also demonstrated that these alloys were mildly superplastic and that the addition of Zr can increase the effect dramatically. Later, Humphreys and Chan43 elegantly illustrated the effects of particle size on recovery and recrystallisation in an Al–6Ni alloy in which the particle size was altered by thermomechanical treatments. With fine (∼0·3 μm) particles, the microstructure coarsened on annealing but with large particles (∼1·7 μm) particle-stimulated, discontinuous recrystallisation occurred. In a seminal paper following up that work, Humphreys44 included recovery, recrystallisation and grain growth in an analytical theory of cellular microstructures. It is worth emphasising that experimental studies of recrystallisation (and other transformations) has been revolutionised by the electron backscattered diffraction technique developed by Humphreys. A comparison of two recent papers on recrystallisation45 with that of Furrer and Hausch41 reveals the large amount of extra information that it is now possible to obtain.

Precipitation

Papers on precipitation in aluminium alloys have appeared steadily since the first volume of MSJ. In the past 25 years, the range of techniques used to study this effect has increased considerably and multi-stage heat treatments have also been developed, where the conventional artificial aging process is interrupted and further heat treatment takes place at either higher or lower temperatures. Retrogression and re-aging involves a brief period of heating (after the first artificial aging step) at a high temperature close to the solvus; this leads to re-solution of the smaller clusters/zones so that, after fast cooling to the regular aging temperature, the remaining zones/precipitates grow larger in the enhanced supersaturation. The retrogression and re-aging process has been applied to aerospace 7000 series alloys46 and to the Al–Li AA 8090 alloy.47 In a different approach, the normal aging treatment is interrupted and the alloy is quenched to and held at a low temperature (about 25–65°C) for an extended period before returning to the normal aging temperature. This modified heat treatment can lead to superior mechanical properties, including fracture toughness, for a range of wrought alloys48 and has been improved further by omitting re-aging after the lower temperature treatment49 (whether this is commercially attractive remains to be seen). A good example of the benefits of a multi-technique approach is seen in Gao et al.,50 where various methods are used to study the precipitation sequence in Al–Li–Cu–Mg alloys, with either Mn or Zr additions, subjected to multi-stage aging.

Mathematical modelling

There has been a major growth in mathematical modelling of materials and their processing during the past 40 years. An authoritative paper by Ashby51 details the basic principles of the subject; two fairly recent papers summarise the current state of this topic52,53 and are well worth reading, even though they are not specific to Al. Predicting the microstructure and mechanical properties of Al sheet rolled from cast blocks has been pursued in depth over a long time and with considerable success by Sellars and co-workers54^–57 using a combination of theoretical argument and calibrating experimentation. Construction of such a mathematical model requires understanding of several inter-connected processes such as work hardening, recovery, precipitate growth and coarsening, recrystallisation and grain growth (static and dynamic) and, inevitably, numerous approximations have had to be made; this leaves the way open for others to make further refinements.58^–61 A more recent paper by Ahmed et al.62 shows that significant strides have been made on this topic of late.

Modelling the effects of thermal treatments on Al alloys, in particular on their tensile properties, has progressed notably in the past two decades. Shercliff and Ashby63 modelled yield strength and microstructural changes in non-heat treatable Al alloys and showed how non-isothermal treatments could be dealt with (for example, welding). Go et al.64 have extended this approach to model the strength of the non-heat treatable alloy AA 5754 as a function of non-isothermal annealing conditions, taking account of recovery and recrystallisation and the degree of reduction of cold rolled sheet; process maps corresponding to commercial heat treatment operations were predicted. A sophisticated mathematical model of strain hardening and steady state deformation of Al–Mg alloys, covering a wide range of processing parameters, is reported by Marthinsen and Nes65 and agreement with experimental data is very encouraging.

Precipitation has also been the subject of mathematical modelling. Poole et al.66 modelled two-step age hardening of AA 7475 and included the effects of deformation on hardening kinetics. Robson67 published a detailed model for predicting precipitate densities and size distributions during isothermal precipitation treatments. Khan and Starink68 developed this model further to predict precipitation kinetics and strengthening in Al–Cu–Mg alloys undergoing non-isothermal heat treatments, which involved heating samples at controlled rates to various temperatures up to 500°C and then cooling linearly to ambient temperature. The precipitate densities and sizes predicted matched closely the experimental results (although the precipitate/matrix interfacial free energy was used as a ‘fudge factor’) and hardnesses predicted from their strength model also showed good correlations with experiment. This theory was then extended to cover the prediction of microstructural factors and yield strengths of fusion welds in Al–Cu–Mg alloys.69 Very recently, Hersent et al.70 applied a similar model to predict hardness and microstructural parameters of material near friction stir welds in AA 2024, again with impressive results. We now have a situation where isothermal and non-isothermal heat treatments can be modelled with fairly good accuracy and can account for microstructural changes that result from welding processes (although there still is a need for calibrating experiments).

In recent years, several interesting papers on modelling of metal slurries have appeared. A quartet of papers by Chen and Fan71 covers the rheological behaviour of liquid–solid mixes under shear flow; here, the emphasis is on the kinetics and mechanics of particle agglomeration and de-agglomeration and good correlations with experimental data are obtained. Solek et al.72 evaluated some complications in modelling die-filling for thixocasting an Al–Si alloy; they aimed to predict accurately the moving front of casts in fairly complicated shaped dies and, although not completely successful, produced interesting predictions.

Aluminium–lithium alloys

A massive international R&D exercise in the 1970s and 1980s was devoted to developing Al–Li based alloys for aerospace that were 10% lighter and 10% stiffer than those in current use. This led to a huge literature and an ongoing international conference programme. Significant commercial applications emerged using, mostly, the AA 8090 alloy – for example, the Westland/Agusta EH101 helicopter. However, target ductility and fracture toughness properties were not fully attained and the R&D activities declined very significantly. Many of the key papers in this area can be found in conference proceedings but only a few in our journals. Some of the early metallurgical studies were published in MS(J), where Noble and Thompson reported basic observations on precipitation in Al–Li73 and Al–Cu–Li74 alloys. The paper by Noble, Harris and Dinsdale75 is an important early study of yielding characteristics in binary Al–Li alloys. In the MST era, several key papers on precipitation in these alloys appeared. Gregson et al.76 discussed the role of vacancies in the precipitation of δ′- and S-phase in Al–Li–Cu–Mg alloys and noted that the very high binding energy between Li atoms and vacancies had a marked effect on the precipitation sequence and kinetics (features that were later expanded upon in two excellent papers by Huang and others on the effects of Li additions to AA 606177,78 and to AA 7075 alloys79). A comprehensive critical assessment of solid state transformations in a range of Al–Li–based alloys by Flower and Gregson80 brought together data from different sources and clarified this complex topic. Of the small number of papers in MST on mechanical behaviour of Li-containing alloys, I draw attention to just three. In a useful study of the effect of surface treatments (pickling and shot peening) on fatigue performance of AA 8090 and AA 7010 (with no Li addition), Gregson et al.81 found that 8090 was less susceptible to chromic/sulphuric pickling than 7010; this was attributed to the more homogeneous microstructure of the Li containing alloy that led to a smoother surface. The other two papers, Srivatsan and Coyne82 and Prasad and Rao,83 both investigate low cycle fatigue behavior of Li containing alloys and show a bi-linear Coffin–Manson power law relationship for Li additions above about 2·5 wt-%, attributed to changes in deformation mode.

Metal matrix composites

The last topic I cover is metal matrix composites (MMCs). Like Al–Li alloys, this has received massive research support worldwide and numerous conferences have been held on the subject; part proceedings of four organised by the Institute are reported in MST (June 1994, October 1998, July 2000 and May 2002). Here, I consider articles concerned with processing methods for generating MMCs and a few applications. Shercliff and Ashby84 point out that MMC development had been more driven by science than by commercial need and they present a methodology that allows one to design a composite to fit projected end use. Shakesheff and Purdue85 go into some detail on the end properties required of composites for applications as missiles (the product needs to work just once – but it must work!); they consider numerous important design factors and discuss a range of possibilities on the basis of then currently available composites. An inventive paper focused on sliding wear and seizure resistance is that by Warner et al.86 Their composite, trademarked GrA–Ni*, consists of an aluminium matrix containing SiC or Al₂O₃ particles and Ni-coated graphite particles that provide excellent friction performance. Impressive results for both dry sliding wear and machining performance are given. Applications for cylinder liners and brake pad rotors are discussed, although it is not clear whether these have been commercialised. In another paper, the use of Al based MMCs in disk brakes for rail application is described in some detail by Zeuner et al.87 Particle clustering in high volume fraction composites was elegantly treated by Murphy et al.,88 who used an Al–9Si–0·5Mg alloy containing 20 vol.-% of SiC particles that had been processed by different routes to give different microstructures exhibiting varying degrees of particle clustering. The microstructures were analysed statistically and a reliable clustering parameter recommended. Interesting papers on processing methods include twin roll casting of Al–Si alloys reinforced with SiC particles89 and a method by which interpenetrating phase structures can be produced using both particulate and fibre reinforcements.90 In another inventive paper, Zhou et al.91 describe methods for producing controlled non-uniform microstructures in an AA 6061 alloy containing submicrometre Al₂O₃ particles that had been made into larger agglomerates; the matrix was squeeze-cast into preforms of the particle clusters, yielding composites that were stronger and more ductile than those produced with the same volume fraction of the original Al₂O₃ particles.

Conclusion

As a postscript, it is worth noting that the global aluminium industry has undergone major changes in the period covered by this editorial. At the outset, there were six major players – Alcoa, Alcan, Alusuisse, Kaiser, Reynolds and Pechiney – plus several smaller companies and no Russian exporters. Now, most of the smaller Western companies have been closed or absorbed by the larger ones, some of which have themselves been broken up and dispersed. Only Alcoa of the ‘big six’ survives (roughly) as it was 40 years ago and it and Russian companies now dominate the world’s commercial field. Huge quantities of Al are now used in beverage cans (∼100 billion per year in the USA alone), Al alloys are being increasingly built into automobiles but Al usage in aircraft is in slow decline. We have witnessed quite exciting scientific advances in aluminium science during the lifetime of MST but, sadly, few of the bright ideas have made their way into the marketplace. Although there must now be concerns about the sourcing of funding for UK university research into aluminium, the increasing popularity of MST with Asian authors will no doubt ensure a continuing supply of good papers on this subject.

M. J. Stowell

Department of Materials Science and Metallurgy

University of Cambridge

[email protected]

This is the third in an occasional series of editorials reviewing the digitised archives of MST and Metal Science. Previous contributions appeared in the January and April 2010 issues.