si, em e\? MVR tn aces_-PREFACE Interest in the phenomenon of superplasticity and its commercial potential as the basis of an important forming process has significantly increased in xecent years and this book is written as an introduction to this rapidly expanding subject. It is aimed at final year undergraduate and graduate students in metallurgy, materials science and materials engineering, but should also prove useful to those in industry concerned with superplastic forming/diffusion bonding technology. The book has been developed from a series of lectures offered at the Materials Science Centre, University of Manchester. The’ topics covered include: the main characteristics of superplastic materials; processing to develop superplastic microstructures in a range of materials; experimental techniques to assess the extent of superplasticity; mechanisms of superplastic deformation; cavitation during superplastic flow; superplastic forming and diffusion bonding. Where appropriate the concepts presented are accompanied by worked examples. It has not been possible to cover the topic of superplastic forming as completely as would have been desirable this field is now based in engineering since much ©: e wb mechanics and,in particular, finite element modelling, the details of which are beyond the scope of this text. John Pilling Norman Ridley July 19881 Lt Le 13 2 21 22 2.3 2.4 3 - Superplastici CONTENTS AN INTRODUCTION TO SUPERPLASTICITY Background Mechanical aspects of superplasticity Types of superplastic materials SUPERPLASTIC MATERIALS: Introduction Grain refinement by mechanical working Grain refir rai Aluminium alloys Tron alloys Titanium alloys ment by phase transformation efinement by phase separation in ceramics Superplasticity in other materials CHARACTERISATION OF SUPERPLASTIC DEFORMATION BEHAVIOUR Introduction Step strain rate tests Determination of strain rate sensitivity Determination of the true stress as a function of strain for deformation at a constant strain rate Determination of an activation energy for superplastic flow Summary awe 8 13. 14 7 25 35, 38 39 49 55 644 PHENOMENOLOGY OF SUPERPLASTIC DEFORMATION 4.1 Introduction | 4.2. Non-superplastic flow — region III 4.3. Superplastic flow — region II 4.4 Summary 5 CAVITATION AND FRACTURE 5.1 Introduction 5.2 Fracture by unstable plastic flow 5.3. General characteristics of cavitation 5.4 Cavity growth 5.5. Analysis of cavity growth 5.6 Void growth by coalescence 5.7 Role of hydrostatic pressure on cavity growth 5.8 Cavity nucleation 5.9 Summary 6 SUPERPLASTIC FORMING AND DIFFUSION BONDING . 6.1 Introduction 6.2 Forming processes 6.3 Component design criteria 6.4 Determination of the pressure—time cycle for superplastic forming 6.5 Diffusion bonding 6.6 Summary BIBLIOGRAPHY INDEX1 AN INTRODUCTION TO SUPERPLASTICITY 1.1 Background Superplastic materials are polycrystalline solids which have the ability to undergo large uniform strains prior to failure. For deformation in uniaxial tension, elongations to failure in excess of 200% are usually indicative of superplasticity, although several materials can attain extensions greater than 1000%. The highest elongations reported are 4850% in a Pb-Sn eutectic alloy, Fig- 1.1 [1] and greater than 5500% for an aluminium bronze [2]. Observations of what appeared to be superplastic behaviour were made in the late 1920s by Hargreaves [3,4] and Jenkins [5]. However, the most spectacular of the earlier observations was that by Pearson in 1934 [6]. While working on eutectics, he reported a tensile elongation of 1950% without failure for a Bi-Sn alloy. Pearson also examined the bulging characteristics of his materials using internally pressurised tubular specimens. Following these observations there was little further interest within the Western world in what was clearly regarded as a laboratory curiosity. Nevertheless, studies were carried out in the USSR and the term superplasticity was coined by Bochvar and Sviderskaya in 1945 to describe the extended ductility observed in Zn-Al alloys [7]. (The term superplasticity is derived from the Latin prefix ‘super’ meaning excess and the Greek word 4tional superplasticity in Pb-62%n Visas in tension at 413K to an elongation Figure 1.1 eutectic pu: of 4850% [1]. "plastikos’ which means to give form to.) In 1962 Underwood [8] reviewed previous work on superplasticity and it was this paper, along with the subsequent work of Backofen and his colleagues [9] reported in 1964, which was the forerunner of the present expanding scientific and technological interest in superplasticity. Padmanabhan and Davies [10]. These author: methods of asseccie apthe experimental deformation Chand the superplastic also sumne characteristics of materials. They superplasticite the observed features of hypotheses provernd compared them with the many 1980. ‘the presen oa i2 the literature up to 0. sopenpisseiuelvely”on micesd2 concentrate Sticity since this has been the most 2bonding and superplastic forming (DB-SPF) of titanium alloys is used to produce fan and compressor blades for acroengines. Aluminium alloys can be used in the fabrication of airframe controL surfaces and small-scale structural elements where low weight and high stiffness are required. Non aerospace applications of Al alloys include containers with complex surface profiles and decorative panels for internal and external cladding of buildings. 1.2 Mechanical Aspects of Superplasticity The most important mechanical characteristic of a superplastic material is its high strain rate sensitivity of flow stress, referred to as 'm’ and defined by oaks (1a) is the flow stress, € the strain rate and k anaterial constant. If superplasticity is regarded as a type of creep behaviour then equation (1.1) may be rewritten as € = kom (1.2) where n, the stress exponent for deformation is 3ow ic behaviour, m wo lastic lg jm. FOE super? 0.3 and for the x than OF. egic materials m es in the Vf superP 17 wresence of a nec ina : 9 t Poile straining leads to O fate and, for a high value oe 1 e in the flow stress within locally “arp increas ce the neck undergoes Strain epahibits its further ; ma hn strain rate sensitivit, Thus a hig! neck development ang ment « e to develop! igh resistance Tongations fers a hi ile e. £05 in the high renplastic materials. i 1.1) that if the It can be seen from equate & ieasured and then relationship betwerliiy the slope of the plot ig plotted logexstrain rate sensitivity of the flow stress m, In practice most superplastic materials show a sigmoidal variation of the flow stress with strain rate. In Fig. 1.2 it is evident that the strain rate sensitivity passes through a maximum. a value of m>0.3 delineates the superplastic regime (Region II). Both the high and low strain rate ranges exhibit values of m in the range 0.1 to 0.3. The region at high strain rates (Region III) is generally believed to correspond to conventional xecovery controlled dislocation creep (power law the shcene eemation within this region leads to development of high die 1peees, and _to the . the graine Grystett islocation densities within material ig inereasea ees texture within the elongation occure a and significant grain S during deformation. an the superplasti i uniform strains are Gogims Region II, where high have so far faileg 4o°ootVed, experi dies Fi ed : ‘perimental stu controlling mechanic’ identity a unique rate roteeiet that Grain bowegeesormation. It is clear, ation mak ndary slidi : Sig ch" fabemeie cutie Ji emain equia, ° 7 ‘ins mecar af hsgied ehroyenst, © fegion LiT'the $2 banding develon iPitially gnosozmation and 1} "icrostructure) @ Ore unigge” Microstructura ®: Crystallogre sm, equiaxed aphic texture may be 41 AN INTRODUCTION TO SUPERPLASTICITY a STRESS, MNm? 10 0 10 10 STRAIN RATE, s! Figure 1.2 Measured variation of the flow stress with strain rate for a superplastic Al-Mg eutectic at different temperatures and grain sizes [16]. reduced during deformation in this region. transmission electron microscopy studies have shown only limited evidence for dislocation activity within the grains of superplastically deformed materials. The flow stress, o, decreases and the strain rate sensitivity, m, increases with increasing temperature,T, and decreasing grain size, d, (Fig. 1.2). The elongation-to-failure in this region tends to increase with increasing m [11]. The origin of the low strain rate regime (Region I) is at present unknown. The experimental evidence available at these low strain rates is limited and often contradictory [12]. It has been suggested that the decrease in the strain rate 5i train rate sivity with decreasnd threshold stregc™y apps ae the effects of micro. for : oF i i rowth ha: deformations Stability (grain ge wth hardening), structura ly, the similarity Til has pee sPoRent alternative Me Regions I and H en Usea n=1/m) Lata Region I also ae recover, to imply a dislocation creep. OF er experiments, controlled ere “nowever, have shown that at lo, investi gee Ss the strain rate sensitivity can i “taking values close to unity and inereby implying diffusion creep. z e evooriin ntal evidence relating to Region I wil) oxpespsidered in more detail in Chapter 4, ah Pahari te, Of cgercture and gee Taek my is very sensitive to potting ene temperature Gt size. In genera’ Jin size of the material has a Seer effect on the variation of flow stress wit, strain rate (Fig. 1.2). Increasing the .temperature decreases the flow stress, particularly at the lower strain rates corresponding to the transition from Region II to Region I. The maximum strain rate sensitivity has been found to increase with increasing temperature and the strain rate of maximum ‘’m’ moves to higher strain rates. The increases in 'm’ values are much greater in Region II than in Region III. The strain rates at which sSuperplasticity is normally observed lie in the range 10-5 to 10-3/s, although this is more usually between 2 x 10-4 and 2 x 10-3/s. These strain rates are less than those used in conventional hot deformation processes. The mechanic 1.3 Types of Superplastic Materials For superplastic behaviour a ma i t terial must be capable of being processed into a fine equiaxed grain structure whi i : : deformation. ich will remain stable during still observed for even higher. In th microstructure, peasence of a suitable Superplasticity occurs over 4 Py1 AN INTRODUCTION To SUPERPLASTICITY narrow range : above 0.5 of temperatures which generally lies There : pseudo single pain types of superplastic alloys: class of material and microduplex. In the former working and heat ra combination of hot and cold fine scale dist. iboceer oe is employed to develop a recrystallisacns ution of dispersoids so that on of the order SES the alloy wily haye aarare size a £ x less. Ideally, the dispersion of particles will also prevent any further grain growth during superplastic deformation. The ee the bees strengthened aluminium alloys based on the xxx series (Al-Cu), 7xxx series (Al-Zn-Mg) a hie on) series Al-Li alloys can be: Jacluded up. er materials include version strengthened copper alloys where silicide or aluminide particles are used to limit grain growth. Some steels and ultra-fine grain ceramics such as UOz, can also be classed as essentially single phase. The aluminium alloys, which froma commercial viewpoint are the most important of the pseudo single phase materials, can be further subdivided into those which are recrystallised prior to superplastic forming and those which acquire their fine grain structure only after a limited amount of deformation at the forming temperature. The microduplex materials are thermomechanically processed to give a fine grain or phase size. Grain growth is limited by having a microstructure that consists of roughly equal proportions of two or more chemically and structurally different phases. This latter group of materials includes a/f titanium alloys, a/t stainless steels, a/B copper alloys, eutectics and some ceramics. In the case of the ceramic materials, tensile elongations greater than 200% have been reported. Although these elongations are small by comparison with those commonly attained in metallic materials, they are much larger than the 1 to 2% normally observed in structural ceramics [13-15].2 SUPERPLASTIC MATERIALS 2.1 Introduction It is well established that a fine grain size is an essential prerequisite for superplasticity. An understanding of the basic metallurgical principles underlying grain refinement and grain growth is therefore important to the development of superplasticity in materials which would not normally be superplastic. Several methods are available for grain refinement including phase Separation, phase transformation, and mechani n i : cal working with recrystallisation. It should be possible in principle to develop fine grai: microstructures using thermal ereatmaiit 1 However, the imposition of mechanical working. dus ng any stage of the heat treatment ma: a required eed Process which produces ‘thee oe uires rain si steps. g: size in a fewer number of processing Unfortunately mers n el: ievi i is not in itself Y achieving a fine grain sizePoth Decade of superplastic deformation in materials. ngle phase and microduplex in this Chapter the metallurgical principles erancnenee the thermomechanical and thermal tyeatme ts that are capable of refining the grain followed De terial are described. This is then Slows by a more detailed account of the proces lures that have been applied to develop superplasticity in aluminium, iron and titanium eee TOS: The Chapter concludes with a summary of | observations of superplastic behaviour in se and other materials, including ceramics. 2.2 Grain Refinement by Mechanical Working 2.2.1 Duplex alloys Grain refinement in equilibrium duplex alloys is accomplished by hot working the material close to the temperature range where superplastic deformation is to be carried out. In most cases the temperature is as high as possible concomitant with a microstructure consisting of approximately equal volume fractions of the two phases. If the phases have different deformation characteristics, such as one being harder and more brittle than the other, then working would fragment the harder phase. The softer phase would then be forced to infiltrate and separate the harder phase. The phases can recrystallise or spheroidise during the hot working process (Fig. 2.1). Alternatively, if the two phases have very similar mechanical responses, working first elongates the grain structure and then fragments it by the development of intense shear bands. Again the original structure can be reformed, but on a much oe scale, by lisation. In some cases, the fine recrysta structure is developed from a cold rolled material by recrystallisation during heating immediately prior to superplastic deformation. h during both heat treatment and . Gran geoeic flow is restricted as the individual phases have different chemical compositions. In 9elongated grains - Be warm work hot work ae ga Id work equiaxed grains / v fragmentation of _— Serr microstructure by fracture of shear band formation re-xtal harder phase ie i tion of the i hematic illustra t Figure 2odtural changes that occur during grain refinement by mechanical working of duplex alloys. rain growth can occur solutes must Sittuse teen the snatien grains of each phase through or around the other phase to the larger grains. As alloy additions partition to the phase in which they are most soluble, the rate of mass transfer of the alloy element through the other phase is restricted by its limited solubility. the grain size is then said to be ‘segregation stabilised’. A duplex superplastic material of very high structural Stability would be one in which the component metals (or ceramics) showed little or no solid solubility. Examples of duplex materials which can be hot worked to develop fine grain microstructures include a/f titanium alloys [29,30] a/B copper alloys, ultra-high carbon steels [31-33], a/t duplex stainless steels and eutectics such as Al-Ca (34), Al-ca-zn [35], Pb-sn and Bi-Sn [6]. 2.2.2 Pseudo single phase alloys Grain refinement ag a recrystallisation extensively int ae fesult of warm working and hie a tments has been used ‘evelopment of superplastic 102° SUPERPLASTIC MATERIALS oeunnton based alloys. The alloys are termed entirely ope Dhase Since they consist almost with ini , €@ solid solution strengthened matrix present to py ol ane of a precipitate phase which is grain groote al . ise the microstructure against two types on The alloys are constituted such that working but gparticles are present during warm two, remains ae ueually only one, the finer of the temperature. @ superplastic forming The high density of dislocations which form during warm working would normally tend to rearrange themselves by climb to form dislocation walls and possibly subgrain boundaries. In the absence of any fine particles the dislocation arrays would migrate by climb allowing the microstructure to undergo continuous recovery and ultimately recrystallisation. However, the presence of fine particles, which are usually less than 0.2 um in diameter, prevents recovery by exerting a drag on the migrating dislocations, dislocation walls, and subgrain boundaries. Mechanical working in the presence of the fine particles therefore generates and maintains a large amount of stored energy and introduces into the microstructure a large number of potential nucleation sites for subsequent recrystallisation. Alloys which contain predominantly fine particles develop a fine grain equiaxed microstructure during the initial stages of superplastic deformation by in situ recrystallisation. The particles and solutes which prevent recrystallisation during warm working at lower temperatures cease to be effective in restraining dislocation migration at the higher deformation temperatures + The majority of the grains which form on heating to the superplastic deformation temperature have very similar crystallographic . orientations. because of the texture introduced into the material during warm working. As high temperature deformation and hence grain boundary 11reheat fine particles warm work inhibit re-xtat On == — ee ke elongated grains ae “equiaxed subgrains high density of with high cislocation submicron particles density spf _- =e —_ @ large particles ; tricron and submicron _—disrupf plastic flow rapid heating particles small particles large particies prevent recovery dissolve Figure 2.2 Schematic illustration of the . microstructural changes that occur during grain refinement by mechanical working of pseudo single phase alloys. sliding ‘*> proceeds, the misorientation between the grains increases and leads to the formation of high angle grain boundaries and a true superplastic microstructure which can undergo grain boundary sliding (Fig. 2.2). A second group of pseudo single phase alloys contains both fine and coarse particles during processing. The presence of larger particles, which are usually greater than 1 um in diameter, provide nucleation sites for recrystallisation and cause -localised distortions in the orientation of plastic flow during warm working. The resulting (+) For the moment it will be assumed that grain boundary Sliding is the mechanism by which superplastic flow occurs. The validity of this assumption will be examined in detail in Chapter 4. a2 SUPERPLASTIC MATERIALS eetercts in crystallographic orientation of the OF een eet Wery fine scale, lead to the nucleation rystallised grains with widely varying orientations and the formation of high angle boundaries as the embryonic grains impinge- Materials which contain predominantly the larger particles are statically recrystallised to produce the fine equiaxed microstructure by annealing prior to superplastic deformation (Fig. 2.2)- Alloys which are refined by recrystallisation include Al-Cu-zr (2xxx series alloys including Supral), Al-Mg-Zr (5xxx series), Al-Zn-Mg-Cr (7xxx series) and Cu-Al-Si-Co (Coronze). 2.3 Grain Refinement by Phase Transformation Several studies have shown that repeated thermal cycling of a material through a phase transformation can result in a very fine grain size. The mechanism of grain refinement is the nucleation of the reaction product at several sites on the grain boundaries of the parent phase. The product phase then grows as the transformation proceeds, replacing the single parent grains by a fultitude of smaller grains. Repeated cycling through the phase transformation further refines the structure until a saturation grain size 5 ved Fig. 2.3). A similar e J rescbed fie; Talis stew ta fanetotie avi temperature such thet it phase is produced. The high unrecrystallisee Py within the parent phase results dislocation density “nucleation sites for the in a large nil’rhe parent phase transforms product ppese ei in product on cooling. i a fine grain produ directly te lastic deformation would take place Ideally Smperature range where the two phases are : t the grain size is _— present in order = if Sugerplastic deformation is stabilised. Howe’ect a temperature where the i t to be carried gle phase then additional steps need to be take prevent grain wth - the ke growth - @-d-_ mn to prevent gr duction of fine dispersoids to pin the grain introdu' i ies. 13high temperature heat phase heat ( ture _f nucleation of high low temperati Paces ce on on F gran boundaries ~& — & nucleation of low temp. saturation phase on grain boundaries i i ion of the Figure 2.3 Schematic illustrat. h . fierostructural changes that occur during grain refinement by phase transformation on thermal cycling. Controlled rolling coupled to phase transformation | has been successfully exploited in high strength | low alloy steels (HSLA), ultra-high carbon steels [26] and could equally be applied in the Cu-Al based alloys where the eutectoid transformation (B->a@+t2 or B->a+Kxxx) Occurs on cooling [2]. 2.4 Grain Refinement by Phase Separation It is often possible to anneal duplex materials at a temperature where only one phase is stable. Quenching the resulting single phase structure will then either effect a martensitic transformation or produce a supersaturated solid solution. A subsequent annealing treatment will result in the seperetica ene two equilibrium phases from the sites are available then a fine eticient nucleation microstructure is produced, A Mastencitie titanium alloys from thi 1 e B phase £i i the a geenetty of potential Tueleatien’ ohioe toe S and hence aids the change of the p'-T 14,2 SUPERPLASTIC MATERIALS high temperature metastable solid phase solid solution or martensite GBs quench anneal equilibrium microstructure pase: separation Figure 2.4 Schematic ill} i : ustration of the microstructural changes that occur during grain refinement by phase separation from a non-equilibrium microstructure. to the equilibrium a+B phases, (Fig 2.4). Another example of grain refinement by phase separation is the spinodal decomposition of gn. 22%Al. Slow cooling the Zn-Al alloy results in a lamellar eutectoidal decomposition product with very poor superplastic properties [36]. If however, the alloy is quenched to room temperature a supersaturated solid solution of Al in gn is formed which decomposes in a spinodal fashion producing a very. fine grain, highly superplastic, microstructure [37,38]. variety of methods available for microstructures, only a very r of upwards of 100 distinct alloys ore number Gensive superplasticity are, or have the potential to be, exploited on a commercial scale [10]. These include medium to high strength aluminium alloys, some duplex titanium alloys and ultra-high carbon and stainless steels. Despite the wide - developing fine grain 15008 - 00g o00T - 005 000T - 00S Ost< oozT OSsT 00% - o0z 588 004 oosT< o0zT - 008 009 oozt ~ 009 000T ~ oor (shorTe aseyd efburs opnesd zuozz0e5 z0moy :skorte xetTdnp :uotqoes zeddq) Aq unoys st rnoravyoq oFseTdzedns yoTym Tepun suoFaTpUOD oY. S5°0 ~ 0 S970 80 - S*0 6°0 SE°O - £70 90 £°0 - Sh°O0 59°0 ost Slo-Ste 0bS-005 ozs STS oss 09€-02€ oes 0€S-08% 096 086-074 StS-0S oLe 0z5~08% Qo 225° 0-TTE-ROp-TW 225° 0-TTE-TV| (0608) 221° 0-6N9*0-NDZ*T-F1S* Z-Te (oto) ng¢*t-6WS*e-uze" 9-TY (s2e2) 29@°0-n2g*1~OHS*Z-uzg"S-TY, 212€° 0-ONT -Uz6-TY| SHE-UZI-TY, 22° 0-On9-TE (Tezdoon) AST *0-UWL* 0-N29° 0-BHS-TE| (0zz Teadns) ept-o- 78° 0-BNE* 0-226 ° 0-n29-TW (oot Teadng) Sue * 0-226 0-099-TY, ugs-P0G¢-I¥ NOEE-TH ssXoTTe eseq-untutum Tunre 3 Jo Axemmg [+z eTqes2 SUPERPLASTIC MATERIALS 2.5 Aluminium alloys een el tam alloys which have either been plasticity, developed or processed for super- ptmicten 7 only two are used extensively in (includin spplications; Al-7475 and Supral [39] although oa: ‘upral 100, Supral 150 and Supral 220), arthough there is substantial interest in the use Alloys auch Ge oe alloys such as |A1-8090. 548 ve A ae as Supral 5000 (Al-2Mg-0.4zr), Formal decorati leopral (Al-5Mg-0.15Cr) are used for The ive panels in architectural applications. compositions of a number of pseudo single phase and microduplex aluminium alloys which have been shown to be superplastic are given in Table 2.1. 2.5.1 Dynamically recrystallised alloys In the Al-Cu-zr alloys such as Supral 100 and Supral 220, a fine grain structure is achieved by in situ recrystallisation during the early stages of superplastic deformation rather than immediately following warm working [40,41]. The processing route for Supral 220 is shown schematically in figure 2.5. During the initial homogenisation of the cast alloy, metastable cubic ZrAls particles are precipitated and most of the copper and magnesium is taken into solution. The ZrAls particles have a volume fraction of ~5% and are distributed homogeneously with a particle size of between 5 and 10 nm. During warm working at 300°C, recovery and recrystallisation of the structure is prevented by the ZrAls particles, and to a lesser extent by the Mg2Si particles and the copper and magnesium in solid solution, Fig. 2.5. During the early stages of superplastic deformation at 460°C the structure dynamically recrystallises. Recrystallisation occurs ‘in situ’, the low angle subgrain boundaries which form on heating to the super- plastic deformation temperature become progressively more high angle and thus able to contribute to grain boundary sliding (Figs. 2.5 and 2.6). ess of 1500% are i ations in exc Tensile elong: 1 220 with a strain rate. commonly attained in Supra 17sensitivity of 0.5-0.6 at 460°C. Moreover, the high elongations are obtained at strain rates between 10-2 and 10-2/s and thus superplastic forming can be carried out quickly, minimising both forming times and the deleterious effects of Strain hardening due to grain growth. Indeed there can be advantages to forming at high strain rates during the early stages of deformation to ensure that a very fine grain size is produced, since the recrystallised grain size decreases with increasing initial strain rate [42]. The forming can then be completed at a lower, more optimum, strain rate for superplasticity. However, the presence of coarse particles of CuAl2 and primary ZrAls, which tend to be clustered in bands within the processed sheet can act as sites for cavity nucleation during ' superplastic flow (see Chapter 5). The extent to 500 solutionise ° uj a > & warm work « Ww a = au BE 0 (cold work ) by ‘a He na ; y ‘dynamic’ or in-situ’ recrystallisation. 18rostructure of Supral in (a) b) after dynamic — Figure 2.6 Mic! as-rolled condition and ( recrystallisation [40]. aere-xtal solutionise s00r hot work quench warm work TEMPERATURE, °C TIME Figure 2.7 Schematic processing route for Al-7475 used to generate a fine grain microstructure by ‘static’ recrystallisation. which cavitation damage may develop during superplastic flow can, if it is not controlled by other means, be quite high and may often approach 10% by volume at high strains. However, at commercial levels of strain, typically €<1.2, cavitation is more likely to be around 1% by volume. 2.5.2 Statically recrystallised alloys High elongations to failure are also exhibited by Al-Zn-Mg-Cr alloys such as 7475 [53]. However, unlikd Al-Cu-Zr alloys, the 7xxx series alloys require a more complex thermomechanical processing treatment to develop a fine grain size in what is not normally a superplastic material, (A typical processing schedule for 7475 is illustrated in figure 2.7.) A solution heat treatment consisting of 3h at 480°C is used to dissolve all but the Cr-rich dispersoids. The latter particles, probablY 202 SUPERPLASTIC MATERIALS MgsCr2Alie, CrAlz, or Cr2Als, are insoluble and typically 0.1-0.2 um in diameter. The alloy is then annealed at 400°C for 8 h to produce an overaged dispersion of M phase (a mixture of MgZn2 and CuMgAl) and T phase (a mixture of MgsZnsAl2 and CuMgsAle) precipitates with diameters between 1 and 2 um. The latter particles have two functions. Firstly, to effect inhomogeneous deformation during warm rolling and secondly to act as nucleation sites for the recrystallising grains during subsequent heat treatment. The material is warm rolled at around 200°C, following overageing, creating areas of intense deformation and substantial lattice reorientation around the undeformable M phase precipitates (Fig. 2.8a). After warm working, a high temperature annealing treatment is applied to produce a fully recrystallised microstructure. The fine scale dispersion of large precipitates and a rapid heating rate result in a large number of recrystallisation nuclei, and thus a small grain size is developed. Grain growth during annealing and subsequent superplastic deformation is restricted by the drag effect imposed on the grain boundaries by the fine insoluble Cr-rich dispersoid phase [54,59] (Fig. 2.8b). Similar processes have been developed for other Al-Zn-Mg alloys including 7010, 7050 and 7075 and Al-Li alloys with Zr, Mg and Cu additions. In the Al-Li alloys grain boundary pinning is provided by a fine dispersion of submicron sized particles of ZrAls. The alloys are solution treated at around 550°C for 3 h, water quenched, then overaged at 310°C to develop large particles of T phase (Al2LiCu (Ta) and AlsLisCu (T2)) and § phase (CuMgAl2). After warm working, a rapid recrystallisation heat treatment is carried out at 550°C to produce the fine grain size necessary for superplasticity. 2.5.3 Powder metallurgy alloys Al-Li-zr based alloys containing between 2 and 4 wt-% lithium have been converted into powders by 21Figure 2.8 Microstructure of Al-7475 in (a) the warm rolled condition showing the non-uniform pia flow around the coarse intermetallic (b) the pinning of the recrystallised grain boundaries by the fine intermetallic particles {40}. o>2° SUPERPLASTIC MATERIALS sphere can pecation/atomisation techniques. The Broduction of siepiacs 2,Precursor material for the Ene powdate se ab and ultimately sheet alloys. and then hot sees Tally cold compacted, degassed outruded ‘ee pressed in vacuum. The compact is then material. Althoond 400°C to give a fully dense seein cine Though the extruded slabs have a fine fhe grain boungY 222 Ret normally superplastic as angie type pees are predominantly of a low angle t¥pe. Additional thermomechanical processing, sesuiced pat described in Section 2.5.2, is usually zoauired im order to develop a fully superplastic For the peoduce oe 60); The powder metallurgy route ox the production of Al-Li based alloys may be Pore unifors the nowmal ingot metallurgy route as a more uniform product with a higher Lithium content The direct powder route may however prove impractical for producing large volumes of material since fine aluminium powders are highly explosive and need special handling facilities. An alternative method of producing larger preforms is that involving metal spraying where a jet of molten alloy is atomised in an inert environment, and then condensed onto a semisolid surface. The preform is built up rapidly as additional semimolten droplets of metal are forced into contact with the surface of the preform itself. The spray cast billet can then be upset forged or éxtruded to both consolidate and remove any porosity Which developed as the billet was being built up. hen be followed by warm rolling This treatment may t 0 to around 90% reduction to give an alloy of the ly recrystallising type (e.g. dynanicany Yere thermomechanical processing to Al-4Cu-3Li-0.52r) or e p give a fine grain material by the static | recrystallisation route described in the previous section. 2.5.4 Eutectic alloys uminium alloy systems solidify ina peyton ogy including Al-CuAl2 [19,44] z hol erate CaAla [34] and Al-Mg2Si [62], but Al-Si [61], Al- 2Figure 2.9 Microst: ructure of Al-5Ca-5zn in as-cast condition and (b) after hot worki2° SUPERPLASTIC MATERIALS only the fo have commnary Al-Ca-Zn eutectic has been considered can be chill —. potential [35]. The Al-5Ca-5zn alloy cee eT) Sree Stet tO Produce @ lamellar structure with jamellar spaci: um, in which the Al-CaZnAl, inter- 2.9a). The: 1a, ng is between 0.2 and 0.4 um, (Fig. hot and old Soria structure can be broken up by perticles ag oe ling to give partially spheroidised Pere grain ct ‘aZnAls. On heating to about 450°C a Fine grain structure forms by in situ recrystallis— ation of the aluminium matrix (Fig. 2.9b). A grain Semele vit tan ote on ee ee erate rate deformati es greater than 0.5 to be recorded at Blongatione temperatures between 500 and 550°C. been“ obt ns to failure of the order of 600% have n Obtained at strain rates of 2 x 10-2/s. Grain we luring superplastic flow is minimal because of the high volume fraction of intermetallic particles present in the microstructure. 2.6 Iron alloys Iron based alloys are probably the most versatile and, in many contexts, the most important of all structural materials. It is not surprising, therefore, that superplasticity has been developed in a number of these alloys. The compositions and properties of several superplastic iron based alloys are given in Table 2.2, and also in the veviews by Ridley [63] and by Walser [64]. The Yarious alloys can be divided into two main typest particle stabilised and segregation stabilised. 2.6.1 Particle stabilised alloys This group consists of primarily plain carbon and low alloy steels. The fine grain microstructure required for superplasticity can be developed by thermal and thermomechanical processing and is stabilised mainly by carbide particles. ntaining up to eutectoid carbon rsed cementite partictes ine i small ferrite grains can be developed by nates i eings quenching to produce martensite, followed by tempering at 550 to 700°C. However, during deformation at 700°C, the cementite For steels co n content, finely dispe: 25(s0zz avs) UHE-OWE-INS * §-TIZZ-AI| (ogaue easeay) OWE-ING-2DG* BT -2d, NPI *O-OWE-INS * 9-295E-0F (po £-nI) Tibb*O-ING* 9-2092-0a 26° 0-uNG *T-2z} (sg@ wnzTezzez) OWE -INS-TI9T-2T | 000T - 006 o000T - 00s 000T OS0T - 006 00L< 00sz - 006 o00T< 09% 096 £eL 009 056 9001 - 00% |} TOG *T-29°T-8r, oost ~ oorr 008 058 ~ 0S2 oost - o0z os9 2(6*T-E*T)-ag, = 008 TSE-IDS*T-IE"T-az] oozt - ost 0s9 9TZ eet SOL uyw/3 eqez uTez3s ssd0rre pastrrqeas eporazed ~ uorzoes aeddn) (sdotre pesttrqeas uorzebexbes - uotz2as zomoy Wa zopun suorazpucs eyz Jo 1000°C (Fig. 2.12). The duplex microstructure is developed by hot deformation from 1200 to 1250°C in the ferrite phase field followed by annealing at 900°C to precipitate austenite, or by hot deformation, cold work and recrystallisation at 900°C to give a fine grain equiaxed microstructure (Fig. 2,12). commercially available materials include IN744, avesta 3RE60, SAF 220 , whi 5, all of whic! are capable of elongations to i Ss 1. ig failure in exces Maehara and Ohmari [69] e i xa i in a duplex Stainless stool ot eeuperplasticity Fe-25Cr-6, 5Ni-3Mo-0 rear ween , +14N, The 1d by forging and hot : material was processe solution treating at 2eng to thick plate then 1250°c ; > 50% cold rolling reductions Gn gebeneng, sven a2 SUPERPLASTIC MATERIALS = TEMPERATURE n o Ni % CrMoTi hot work TEMPERATURE TIME Figure 2.12 (a) Schematic (Fe,Ni)-(Cx-Mo-Ti) phase diagram showing the region where superplasticity is observed. (b) Thermomechanical processing schedules which can be used to develop a microduplex Microstructure in Fe-Ni-Cr-Mo alloys. 312° SUPERPLASTIC MATERIALS deform i on areee at an optimum initial strain rate of elongation ‘wo peaks of high superplastic 900°C ang asetee and ~2000%) were observed at apord and C, respectively. Detailed studies Selection tthe high elongations were due to the eveiuetol 2 fine equiaxed microstructure of 900°C and of a+ at 1050°C. Cavitation is observed to devel i saporplastic defenantion of dapibx sealaloee ar coctated some materials cavities are mainly — oe gage aaeteed Ti(C,N) inclusions and are believed te Ihave nucleated during the original thermo- _ processing applied to the material in order to develop the fine grain microstructure [66] (Fig. 2.13). The small additions of titanium to materials such as IN744 combine with the residual carbon to avoid the formation of CrasCe precipitates on interphase boundaries which would normally lead to cracking during hot working. Cavitation in SAF2205 has been reported to be mainly associated with triple junctions involving alpha, gamma and sigma phases, while in 3RE60 the Gavitation behaviour is similar both above and below the temperatures at which sigma phase forms. esses of the particle stabilised iron based alloys under optimum superplastic conditions tend to be much higher than those of the aluminium alloys described in Section 2.5. The higher stresses make it impractical to bulge form parts from such alloys since the gas pressures required would be too high to handle on an industrial scale. such alloys are much more suitable for superplastic isothermal forging which be formed in materials lex shapes to e m allows empoth strong and tough in their as-formed condition. The flow stresses of the microduplex i eels under optimum deformation, stainleee re comparable with hess of the res aluminium alloys: However, the hig! tempe tures required for superplastic forming coupled wi inherently good cold formability of these materials may limit the application fo} superplasticity in stainless steels. The flow str 3300zT - 00S O0%< 0s9 0L9 OzL o0s< o0s< 00L< OOTT ~ 006 sT8 056 - STB 056 - STB 0L8 - 0F8 ONST-TL ung-T TWE-ATT~29ET-FL TWE-USE-ADE-AST~F 0E6 - 088 TSS*0-USZ-OWP-TVF-FL uss 2-I¥S-Th ed@-Ab-TW9-T | ING~AP-T¥9-TL| ODe-Ab-TV9-TL ATQT-OWS-TWS* b-Th ONE-T2b-USE-TVI-TL AS. (sdorre § ATT eseq-untuegy3 uF peatesqo st A: aT OT 3SeT ny - uofz00s remy :sdotre 9/0 ~ wor dzodns yoTya repun suorzzpuos ay JO goes zeddq) sXorre Axeouungs €°Z eTTed 34; 2 SUPERPLASTIC MATERIALS 2.7 Titanium alloys It is somewhat ¢, 2 proceteed) oat fortuitous that (conventionally x such as Ti-6al—4 Ti-6Al-2Sn- Vv and pega ae when hot rolled to sheet can deformation pa tp SupexPlasticity during Since the fine eth phase field [18,29,74-78]. superplasticn se Toport by Lee and Backofen [29] mieren icity has been observed in a range of € uplex titanium alloys and to a lesser extent in fully B titanium alloys (see Table 2.3). 2.7.1 Microduplex alloys Superplasticity in titanium based alloys is dependent on a number of factors, including the grain size, its distribution and aspect ratio, the relative proportions of the a and ® phases, the rate of grain growth and to a lesser extent the texture and inhomogeneity of the starting material. Such a large body of data reflects the extent of commercial interest in superplasticity in titanium alloys. Increasing the grain size results ina shift in the position of Region II (the superplastic regime) to lower strain rates and an increase in the flow stress [18,29]. Strain enhanced grain growth is commonly observed in ~i-6Al-4vV and leads to an increase in the flow stress and a reduction in the strain rate __ sensitivity with increasing strain [18] (Fig. 2.14). id hases have very different deformation the and eels. The cph a phase has fewer slip systems and a self-diffusivity which is approximately two orders of magnitude slower than pee oin the bec B phase. It might therefore be expected that the greater the volume fraction of B phase, the easier it would be to relax the phases Senerated during grain boundary sliding, end thus the greater the extent of superplasticity. However, the rate of grain growth in alloys ts sow ey ining a large volume fraction of B phase high and a substantial volume fraction of ¢ ato harder less accommodating @ phase is requires: e inimise grain growth. The effect of variat e the volume fraction ® phase on the elongation 3540 24 %/min 12%/nin | 8m sosnin | 0:6 %/min d=10ym TRUE STRESS, MNm? 7 7 70 15 20 TRUE STRAIN Figure 2.14 Effect of strain enhanced grain growth on the variation of flow stress with strain for Ti-6Al-4V at 920°C (note: grain growth is greatest at the lowest strain rate) [79] failure and strain rate sensitivity is shown in figure 2.15 [76]. One modification to superplastic ti i alloys has been the addition of low levels (up t9 a wt-%) of transition metals, such as Fe Ni and Co that have high tracer diffusion coefficients : [53,54]. This has enabled the temperature at which superplastic deformation is carried out to be reduced by ~100°C from around 915°C to 815°C. The xeduction in forming temperature has been possible for two reasons. First]. i 1 F Y, the additions ilise the 8 phase allowing the optimum volume fraction of 40% to be attained at a l. e Ower temperati id secondly, the additions increase the self-diffusio® 362 SUPERPLASTI¢ MATERIALS ‘000 © Ti-GAl-2Sn-4Zr-2Mo © Ti-6Al-4V 800] x 3 S 600 & g SB 400 dt tu 200) 01 02 03 04 05067 VOLUME FRACTION B-PHASE Figure 2.15 Rffect of the volume fraction of B phase o titanium ai rate enabling the normal drop in self-diffusion rates with falling temperature to be countered. The lower forming temperatures also enable reductions in energy costs and tool wear to be made. Furthermore, grain growth is less extensive at the lower temperatures and the extent of strain hardening is significantly reduced. 2.7.2 Fully 6 alloys 's, after conventional processing, ghe fully 8 eirain size and the tensile elongations observed, although high by normal standards (>1008) are low when compared with those shown by tl . a+sp alloys (75,77,84]- Examination of the stress strain relationship of alloys such as 37Ti-15v-3Cr-3Sn-3Al over the norma ee) superplastic strain rates (1 doe he after initially shows a very high Sa eased att ee ee aemuion ae stant. The microstructural emains con: r cniiges that occur during the mortensng Drocess include the development of a subgrain Ss mus ure, In effect, the ® titanium alloys achieve © high i ‘tion and elongations as a result of disloca diffusional mass transfer on the scale of the subgrain size rather than the grain size. 2.8 Superplasticity in ceramics Superplasticity is not normally associated with fully crystalline ceramic materials although it should be recalled that hot glass (an amorphous structure) is the ideal superplastic material, with a strain rate sensitivity of unity and a potential] for virtually infinite deformation in either tension or compression. However, with the advent of ultra-fine ceramic powders, the mechanical and microstructural characteristics of superplasticity which are readily observed in a wide range of fine grain metallic materials have also been observed in ceramics. Like their metallic counterparts, two classes of superplastic ceramics exist: single phase and microduplex. The single phase ceramics are normal] repared from high purity powders which have been milled, However, tensile Studies of ili i . polycrystalli ia that clongations ef (tot7=)g( 85) have. donons¥rated 7 n are easi i eramercially viable strain rates (2 aes on Ow stresses less than 1 x 1074/s) wit that the principle Limitatiny On Bug UES appeat2 SUPERPLASTIC MATERIALS resent wit! 7 | Poundarieses? the microstructure to pin the grain Duplex ceramics firstly, those wh: residual glass 5. partial crystall from the additio: can be divided into two subgroups: ich contain a low volume fraction of uch as that resulting from the ‘sation of a ‘glass ceramic’, or n of sintering aids, and secondly, the fully crystalline materials comprising equal” volume fractions of two chemically distinct phases (e.g. 2r0, with 20 wt-3% Al,0,). Systematic studies of superplasticity in microduplex ceramic materials are even more limited than for single phase ceramics, despite the potential attractiveness of such materials with regard to segregation stabilisation of the microstructure. Investigations of high strain deformation in duplex ceramics have been carried out using crystallised 8 spodumene (lithium aluminosilicate glass) [13] and Zr02-Al20s composites [14,86]. In both instances strain rate sensitivities of 0.5 and greater were reported at strain rates between 10-* and 10-3/s and for the case of ‘B-spodumene tensile elongations up to 400% were observed. The strain attainable in compression was, in both materials, virtually infinite. Strain enhanced grain growth during superplastic flow was still found to occur in the duplex materials but was much less than in the single phase ceramics with comparable initial grain sizes. astic properties of a number of single 1 4 : 2 eee aaa | ex ceramics are summarised in phase and microdupl Table 2.4. 2.9 Superplasticity in Other Materials 2.9.1 Nickel alloys 1 based alloys have been shown to A number of nicke” “atnermomechanically processed ic wh be superplasty microstructure [92]. The to a fine Oee oe ickel alloys are similar to i ctures of the nickel alloy: The tone microduplex alloys, with a volume fraction of the second phase between 30 and 50%. 39(xetdnpozoru - votz008 zeMoq ur peazesqo st znoraeyeq of aserdzedns 00ST-00%T OSST-O0bT OSTT-OOIT OT9T~00ST OST~SLET OorrT OsTT~oorr OsTI-OorrT 008T-00TT toseyd erburs ~ uo yoTYA zepun suorz ©0°Tw-©084-t022 ©0°TW-F02Z, F300s zeddn) sterzeqew opuezes Fpuoco out 30 Arewums 52 orqer 402 SUPERPLASTIC MATERIALS Two types of alloys exist, by conventional casting and those which are f firstly those produced working and secondly, abricated via the powder route. ilekelsbose ase structure of the conventional in the single hee 8 produced by first hot working initial ace oe ase region to break down the the tempest ructure. Hot working is continued as eae rae ure is allowed to fall into the the _Phase field. Nucleation of the second phase during cooling would normally occur virtually homogeneously throughout the microstructure. Since hot working can continue below the recrystallisation temperature a short anneal high in the two phase field prior to cooling gives rise to an equiaxed microduplex structure (Fig. 2.16). If the kinetics of precipitation of the second phase during hot working or annealing are slow then the material is given an additional cold working treatment prior to recrystallisation within the two phase field to generate a duplex structure. [93,94]. Depending on the alloy composition, the microstructure would consist of equiaxed grains of nickel solid solution,which is strengthened by Co, W or Mo, and a second phase which could be either a-Cr (95} or, if the alloy contains Al, f§isAl [96,97]- The average grain size o e Nea abed material would typically be in the range 2 to 7 um. The range of temperatures and strain rates over which the Ni-base alloys are superplastic are summarised in Table 2.5. roup of nickel based alloys are those wre aocca by the powder route. These include a materials such as IN-100 [98-100), IN-713 i } an MAR M-6000 [102]. The alloys are produces rom compacted and hot extruded powders. A: cor a ena extrusion, the grains are generally eu axed a 3 grain size stabilisation is provide by high volume fraction of NisAl ae mene conventional 211078, :Yaispersion of Ox or particles, Seualiy yttria (¥20a). The particle size is generally betw > 2 a1 Gena,TEMPERATURE TIME hot work2 SUPERPLASTIC mMaTERTALS (ae tN) T= (35°38 ozt oss orn TT ore - ost 000-006 jowe-Pac~Taz-Mb~T¥S~20ST-FN (4b2-W BER) FHT ~P.LE~TWS-M6-O06-128-TN FUT ~GNE~OWS-TY9-ZIET-TN (00z-W ueK) TS-TWS-4901-O9TT-MET-FN TWe-T 2 -048-296E-TN osor sor 08rt-0sor 086 o08< (ec-nr) 006 TLE-TWE ME -OD8-TI9T ~ TN (oot-Nx) OS0T-026 OWE-TLS-TWS-ODST-2I0T -TN| , — 1 — # urw/% Do szoy_| uoravéuora oqez urerag | oxnqezeduoz | ory sdor[Te paeseq-Taxotu euos zo sofzredozd ofasetdzedns oy jo Axeumng g*z eTqes 43lastic Ni- _an flow stresses of the super Peothermal Base the high £ oe favour their us inherent ductilit, alloys teMdoye with their high s7oyy fabricated,’ forging hapes to lloys are f, i complex §| urce a Y Or, Boe cample, IN-100 powder fe integral blades, on the outer the blades neat treated to under service ‘Jaq nce are circumfere! 2 eebesistanc improve their cr conditions [97]- din MAR cuperplagticity bee 2220, tefeat peoceaging, of I= 4] b: erm wrenches which had been extruded from bundles of rapidly solidified ribbons. 2.9.2 Copper alloys Superplastic behaviour has been reported for a conga of a/® copper alloys including brasses, aluminium bronzes and nickel-silvers, and also for the essentially single phase copper dispersion alloy, Coronze (CDA-638). For the binary brasses and aluminium bronzes the duplex superplastic microstructure is obtained by hot working the materials in the a + 8 phase field or during cooling from the 8 to the a + B® fields. The optimum deformation temperature for the brasses is “600°C with maximum tensile elongations of “500% being obtained for an alloy with approximately equal volume fractions of the two phases. However, the alloys cavitate and since soon een en ot puperplastic flow is believed to vily withi cavitation increases as the velane’ gene, LeveL OF phase increases. For the aluminine t strongly inhibited. The a/8 nickel i See Table 2.6, are has IN-836 and | See Ya ; : processed by nd IN-629, | * Hti6, lati Ere teat aang a tie | juenc! 44 , Silvers suc2 SUPERPLASTIC MATERIALS 5 %/min 2%/min oN Js ELONGATION, % a ee eT it 460 500 540 580 620 TEMPERATURE, °C Figure 2.17 Variation in the elongation to failure with temperature of IN-836 [70] to room temperature to retain the single phase structure. After a heavy cold rolling reduction, ~85%, the material is finally annealed in the two phase field to recrystallise the a phase with a grain size of 1-2 um and precipitate a dispersion of fine, d ~1 ym, B phase particles. The optimum deformation temperatures are 600°C and 570°C (Fig. 2.17) respectively, but the alloys are prone to cavitation due to the low volume fraction of B phase in the alloy and the relatively low deformation temperatures [105-108]. The dispersion strengthened copper alloy CDA-638 is a cobalt modified Si-Al bronze. It is a pseudo single phase alloy in which a fine @ grain size of ~1 um is produced by recrystallisation following heavy cold deformation. The grain size is stabilised by a fine dispersion of cobalt | silicide particles and the alloy is superplastic at 45avitat | rly prone to c ion sg particula 500°C, but is PI coe: has been observed j, rplasticity "Composition ceptional superP f nominal ¢ icrostructure ca, complex al-bronze OF ne stable anence which n Cu-LOAL5Fe oF 4 processing sed eee og oc loped by a P ec, anneali ’ pe Gives rolling at 900°C, ANPTts cand finally involvelling at 640°C, 00°C. The room temperature er stallisation at 8 oe gw phase with a grain fiersstructure consists a fine dispersion of ree of 2-3 um containing #5). The material can be aluninide particles (X phases); o'sso°c which spant superplastically sets + K phase fields. Tensile the . . clongations of 20008 can pe readily obtalned at strat rates as high as 10-7 to. bei Siongation of 5,500% without ee to-e/s at soon initi train rate of x ¢ : ior ote tanpaverure, the microstructure contains about 40% a, 30% 8 and 30% K by volume and its stability is due to a combination of segregation and particle stabilisation. Na 2.9.3 Eutectics and eutectoids Superplasticity has been reported in over 100 different alloys. It is beyond the scope of this book to detail all the recorded instances of the reader is referred to the review of Padmanabhan and Davies [10} for further details. However, the properties of four eutectic alloys (Mg-33Al, Pb-62Sn, Bi-d4sn and Ag-28Cu) together with those of a commercially used eutectoid| alloy (Zn-22A1) are given in Table 2.7, @n-22A1 all fine grain microstructure ; ° alloy, a ecomposition of a mari cored by the spinodal from above the sol Lh2 SUPERPLASTIC MATERIALS BIT arr‘ orr z EIT-60r 801° 90T 400’SOT uy peazosqo st Agzorasetdzedns yorum xepun suorztpuoo ey7 Jo AreuNg *L-Z STERIL 00ss - 000T Ooze ~ 002 one 089 - 00e OfL ~ 059 osz ~ 002 of o0r- 02 _00% - OSE uyw/3 eqer ure73g +so[T@ proz093na/oT3003Ne 0s@ - 0s¢ 009 ~ OLe ous 5°0 - ££°0 029 - 0% eqez uTer3g «sXotre aseq-teddoo yo serzzedozd orzsetdredns ey3 zo Axeuung 9 nogz-by Teee-usz! uspe-Ta usz9-qa\ 47 TNS-eaS-T¥OT-n9) (8¢9-waa) 00 0-FSz-T¥P-nD| (6Z9-NI) une * 0 TNST-UZgE-nD| °e erqen3) SATION OF CRC DEFORMATION, 3.1 Introduction i tal procedures which, | i ter, the experimen e In he adpoted to measure the extent to which a nat fal may be deformed superplastically. are | de cribed -K-number of worked examples are included | {n the text to guide the reader through the various| interpretative procedures. As was stated in the introductory chapter, the most important mechanical characteristic of Superplasticity is the high strain rate sensitivity of the flow stress,/m) and its variation with yl strain rate, temperattre and grain size. In an q ideal-material, where the microstructure remains | can be measured by | te tensile tests at | Suring the steady oe the experimental ning low strain rates in log (o) =m log (€) + oe The Slope of a 1 inst Ogarithmi against true mic plot o. Sensitivity m— Tte gives Of ths true stress (3.1) 42locity testing, the true thus the flow gtrces, the specimen elongates and unaltered, would dentertetnt, ther effects remain This situats d crease with increasing strain. constant ican, is the reverse of that found in cross—section Sresp testing where the specimen Seen oneomal azea decreases with increasing Sontinacu: f est piece thereby experiences a g Sly increasing stress during testing. lowever, the problem in superplasticity is much more pronounced than in creep as uniform elongations of hundreds of percent are common and it is therefore possible for the strain rate to decrease by more than an order of magnitude during constant velocity testing. Most engineering materials are microstructurally unstable at elevated temperatures. Hence, it is impossible to determine a meaningful flow stress from a constant strain rate tensile test, since the flow stress often increases with increasing strain du@ to the effects of grain growth. It is therefore important, where comparisons are to be made, to determine the flow stress at a constant structure. To do this, step strain rate or strain rate jump tests have been developed. 3.2 Step strain rate tests ermine the strain rate range over which a | notevial might_be expected to exhibit superplastic ‘shaviour, a series of step strain rate tests are various temperatures. Strain rate or; ests were originally developed by uP [9] for the d ination of straii ially a tensile specimen is Initi t_velocity until a steady load carried out at veloci um Backofen et al. rate sensitivity. i @éd at a cons i (eon cared. As many materials do not have 24 uniform starting structures tate before tne first i ly < wed to accumulat is usually is nade. After the velocity jumpy which can be either upwards or downwards, 4 49prt Peed Figure 3.1 A schematic load - time plot for a velocity jump test on Supral 220 at 460°C (Vi=0,02 mm/min; Va=0.05 mm/min) ie - Pe, i e adopted for tive loads for a velocity 503 CHARACTERISATION OF SUPERPLASTIC DEFORMATION BEHAVIOUR decrease. Provided that the velocity jump is small eniain SOHStaRE, Cece sensitivity can be assumed to : over the range experienced by the test piece oe strain rates _ log(Pc’/Pc) log(Pd/Pd’ ) = 3.2 log(V2/V:) loa(Vi/V2) ee) for the velocity increase and decrease, respectively. Using figure 3.1, which is a diagrammatic representation of the load-time plot and is not to scale, _an example of the procedure used to determine ‘m’ would be as follows: The velocity of the crosshead of the tensile test machine is increased from_0.02 mm/min to_0.05 mm/min and the load allowed to stabilise before returning the crosshead to its original speed. The load-time record shown in figure 3.1 has been recorded with a chart speed of 20 mm/min, The load-time plot at the initial velocity is first extrapolated past the point at which the velocity jump was made, to rejoin the trace where a second Steady state load has redeveloped at the original testing velocity. The point of maximum load, c’, at the increased velocity is located, and the load Po’; noted. (In this example the maximum load was kg). The time, tz, taken to attain the maximum load from the point at which the velocity was changed, a;-is calculated from the distance between the point c’ and a. = = 14 min (3.3) chart speed 20 mm/min Next, the time, ta, that would have been required at the initial testing velocity to reach the strain corresponding to point c’, is calculated. distance (a-c’) 28 mm t= Va . ta = te x — = 1-4 mi n x 0.05/0.02 = ae (3.4) Ue 51 :istance along tof the eelateg then be the prowvn ee, art line a-d Cheats see. 20 mm ip 2 33 min Soy speed an e ehia point, Pc, eS be oe eee 2 tne ease the load was found thi the corresponding di Using eqn (3.2) +3/3.8 log(Po'/Pe) _ log(5.3/3.8) Pose ts mn log(V2/Vi) log(0.05/0.02) i ined for the velocit, me Gace te m= "0.36, Altornatively the load-tink pict ae the increased testing veer nee be Extrapolated back to the point at whic! e velocity change was made and the corres ponding loads measured [11]. Then from figure 3.1, "7/2. log(5.5/3.9) ne log(Pa‘/Pa) . = 0-30 cd log(V2/Va) log (0.05/0.02) A similar procedure can be adopted if the tests are carried out at a constant true strain rate rather than a constant velocity (21). To measure ™ over a range of strain rates, a ormed at a velocity which ce. i : uate in the middle of the erest until a stead: state . The crosshead speed reduced te se ar to be measured at i 3 2 typical loag- responding tmeftgure 3.2, HPL true stresses Calculated fron fhe We Strain rates are tee® sanpiseieS and instamexiUM loads. crosshecn Sample. The q, EBgths of the best Plo t est fit curve toa Fete. then gives the st ain ress against frue_strais S variation with Strain rate interest. aa Sensitivity m, and rate in the range of. tray h¥E $5 {418 S tyes! Wo Vad 52 are——_—_S—Ss (CHARACTER. ; ISATION OF SUPERPLASTIC DEFORMATION BEHAVIOUR LOAD ——~ of load vs- time for a est carried out on Al-8090 Figure 3.2 A record fe the crosshead speed in repeated velocity jump & at 500°C (figures indica mm/min )~yapra ueuoeds wu L6°% = <5 uu £0°~ = sseuyoTyy uewToedgs Swi oT = Yzbuey abneb ueuyoeds Sere peat Or = paeds 31eY2 9L°T- BET zee €T 87's o0set T'6r 0 oz vO'Z- —-TE'T —-9S*0Z LTT eBs's 0zz6 40°8T "0 oT ze'e- Te" b°9T 89°6 ZPLTS OvLb LS°LT z ors OL°%- OT*T = "ZT 00°86 80°9 ot0z LS°9T € 07% 86°7- 8670 Lb°6 OT'9 6TE'9 vot L6°ST s OT L2'€- #80 669 GOH EZS"9 6ES Le°ST zr s°0 S9"E- 2470 = BZ*S_-GOTE 9BL"9 zz £98 "FT Iz z°0 ¥E"E- 09'0 LOE EBTZ F869 STT Loe PT 0z TO €@*p- BHO, = SOTE-OZ"ZzZBO"L sas Lbe*HT of s0°0 €9'b- SETO = 9Z"Z_ GOTT LSTTL 9°€Z $960" HT 8z zo"0 €6°P- «9770 ZBTT «EET «98TTL 6°IT SOPO"#T LT T0°0 @2*S-TZ70,— OTT «OZ*T «FETE ¥6°S — SEZO“PT Lb s00~0 €Z°€- S80 ZL ES 902% s6s O°FT 08 s°0 680° OT ot en nn nn ee aE aEEEEEIEEnEyaINTIEIEEENEP SNOT SREP SUEEDENErd (g-umn) (6%) (eum) — (8/90) (uur) (umr) —— (wyur/unn) To Ta Te 79 Zeger TT ‘yqbuet Tp Ta ‘poeds (ro)901 ssez3g peoT very = urer3s uewroeds eouezsTg peoyssoz7 (43) 501 a "eyez pue ssexqs onz3 8y3 eqeTnoTes 09 pesn oxnpecord 9y3 GuTzexasnTTT Z°E eANBTF ut umoys S03 dumf AzyooTea eyz Fo pr0de7 parzetnqeL, T°E STqeL 543. CHARACTERI: ‘SATION OF SUPERPLASTIC DEFORMATION BEHAVIOUR The procedure used to a pro jetermi i sensitivity from a series of velocity sone vane exp S is explained in the follewang’ se ti ction. 3.3 Determination of strain rate sensitivity he chart shown in fi : : n figure 3.2 is against t. a record against tine (equivalent. to displacement) ce superplastic aluminium alloy tested using the re ereieit ned above. To determine the strain rate goneitiv ty and its variation with strain rate, the true, flow stress and instantaneous strain rate Phe polite & et xcmined at each crosshead velocity. Bo s corresponding to the velocity changes are first marked on the chart and the instantaneous length of the sample, 1s, is calculated using la = la-a + vads/chart speeed (3.7) where vi is the crosshead velocity immediately prior to the velocity change, di is the distance on the chart covered at that velocity and ls-1 is the length of the test piece at the previous speed change. the initial gauge length of the test piece used to obtain figure 3.2 was 10 mm and the sample was Strained to an clongation of 40% prior to the first velocity change. Using figure as an example then the length of the sample at the initial locity change would be \14.mm and at the second velocity 7 235 mm (= and ae ee oeyi0y, see velocity change 14.0 - 7 vepie 3/1. Thus for the third velocity change 1s = 14.0235 + 17x 0.01/10 = 14.0405 mm (3.7a) tee . The instantaneous strain rate, €4, is then given by aur é 8 yes = va/ls (3-8) on tie ab which for the above example would be (3.8a) = 1.19x 10-5/s Ey = (0-02/60)/14-0405 = te the true flow stress, 1 ee cad, ps, at that velocity +8 ‘meas the steady ured from 55Ss STRESS Nmm* e wo 10? 0 1! STRAIN RATE s" Figure 3.3 The variation of true stress with true | strain rate (plotted logarithmically) as derived | from the 'm’ test record given in figure 3.2, the chart and the cross-sec calculated from the instant. sample assuming both consta: cross-sectional area, i.e, tional area, as, aneous length of the nt volume and a uniform as = leas/le B62 9% eH (3.3) & Ue, ~ wheré 1. and a. are the original gauge length and cross-sectional area, respectively. Continuing the example, — ; 10:0 x (4.97 x 2.03) 9 =7. a 14.0405 nasa as = The true stress, es, is then calculated from 56 anCHARACTERI, ‘SATION OF SUPERPLASTIC DEFORMATI( (ON BEHAVIOUR \ [7 — Superplastic regime —____ I | I 0° 10% 1 162 Fe STRAIN RATE, sé Figure 3.4 The variation of strain rate sensitivity with strain rate as calculated from the polynomial expression defining the best-fit curve in figure 3.3. os = Ps/as (3.10) hence 9.81ms-? os = 2.35 = 1.816 Nmm-? = (3-104) 7.186 mm? d from the load versus time record The data obtaine ° ad versus ble 3.1 shown in figure 3.2 is summarised i ana jated values of stress ant the cet ehmi of true stress together with strain eo Logarithm e ee tast two columns of mp s.a) sero in figure 3.3 together with Table 3.1) is sho 57ih the data points, The + rve throug! cate poate the best fit Gnsitivity at a givens rain rate te pevalculated either fy othe eed ane Es to Ye Ci in | valua g t 7 tot Car ye oe Pression defining the curve. The Peleulated variation of plotted in figure 3.4. | +m! with strain rate te i tress as a funct. ination of the true 5 on 3.4 oe eteaia for deformation at a constant Strain rate Conventional tensile tests are carried out at constant crosshead velocity. The large tensile strains which can be accumulated in superplastig materials result in a considerable reduction in the true strain rate during testing. A decrease in the true strain rate would, because of the high Strain rate sensitivity of superplastic materials, normally produce a marked decrease in the true floy| stress. To allow a steady state deformation stress a (3.11) raiser? 3.5 (chart in ain is given by 130 mm —_ (10/69 mm/s ) = 0.6084 (3.112)CHARACTERISar: TON OF SUPERPLASTIC DEFORMATION BEHAVIOUR LOAD TIME ——~ Figure 3.5 A load-time record for a constant strain rate tensile test carried out on A1-8090 (€=7.8x10-4/s; 520°C) calculated by assuming that the strain is uniform i.e. a = ao exp(-€) (3.12) Hence, a = (width * thickness) exp (~€) a = (5.02 mmx 1-98 mun) exp(-0.6084) = 5.409 mm? (3.12a) The true stress, ° is then given by eqn (3.10) and so 4.1 kg x 9-81m s-? = 7.44 Nm? (3.33) ° 5.409 mm? 59<2 TRUE STRESS Nmni 10 b 05 +0 a 20 TRUE STRAIN Figure 3.6 The variation of true stress with trug Strain derived from figure 3.5. The var derived it is ap independ material dation of true stre from figure 3.5 is figure 3.6 and| parent that Ww stress . is almost ent of strain, Im an ideal s the flow Stress j SS with tyr ue Strain shown in grain growth Walle 4 SS, parti t hil infer tp Pi icularly a g e development of pter 5),3 CHARACTEI RISATION OF SUPERPLASTIC DEFORMATION BEHAVIOUR 3.5 Determination of . . supexrplastic eof im activation energy for in eretastic flow, Gas, can be detecnen Ow, Qap, can be d i let ee cerene dependence of the colatiananif bee flow str s and strain rate must be mown a aad a n eupizical level. (The detailed fora of such 2m quation will be discussed in Chapter 4 Following the, Conventional ghaceshologionl a ona tn rate aig temperature deformation, the crepe seee it eevee fo OTT eet function of temperature nthe Stren? in oat ie oe eee etion at a given strain rate. To é aA exp(-Qsr/RT) 3 ouem oe (3.14) which (if the temperature range, dT, is narrow) can be rewritten as in(é) = B+ 1/m In(o) - Qee/RT (3.15) where A and hence B are material constants. Therefore, providing that the strain rate sensitivity remains constant over a range of temperatures then the netivation energy is simply obtained from the slope of an ‘Arrhenius plot of ln (strain rate) against the reciprocal of the absolute temperature at constant stress. Unfortunately, 4 number of errors can be readily introduced into the determination of activation energy for superplastic flow. These can be summarised as follows? re interval, 5T, is not °c or more. the temperatu: varies with 1. In practice insignificant, usually 30 2. The strain rate sensitivity, ™ temperature at a given stress. . A, and hence B, contains 4 number of 3. The constant nd ee erns temperature dependen’ e: e 4. Eqn (3-14) ‘assumes a constant structure. In many gs the phase proportions change duplex material 61ences in dramatically with small diffe: temperature. 5. It is often argued that the measurements be made for @ constant value of the modu} hou, ther than the actual © ‘tra, iy compensated stress ra The modulus is temperature dependent. (3.14) is not known 6. The exact form of eqn The factors noted above often make it diffic interpret or assign the measured activation ult ty to a particular physical process and in man Snerg, | instances the measured activation energy is! strongly temperature dependent. itser; The variation of the flow i ° stress with | measured using a step strain rate tent fee Late | fine Sys material is shown schematically | +7 for test: i sts conducted over a ange er” T p Lose (4.4) where Arrz is a constant, Dees is the effective lattice diffusion coefficient, Dv the volume diffusion coefficient, Do the dislocation core diffusion rate, ac the dislocation core cross-sectional area. The presence of a deviatoric applied stress field will result tween the normal stresses on some grain boundary surfaces, introducing a gradient in Chemical potential which in turn induces diffusional mass transfer to reduce those differences. Diffusional transport both through and round the grains results in a shape change and hence creep [124,125]. The strain rate for true diffusion creep, Region 0, is given by . DasexGb [ b I [=] (4.5) = —— Eo = Ro a G and 18 (4.6) Dasexz = Dv + —= Dor i he grain where Ag is a constant, 1 t ms ha grain boundary diffusion coefficients 2s re, boundary width, and 4 the ¥ 69at sufficiently low stresses, ew creep win, i islocation . Soe oes, ne activation energy for diffusion temper be that for grain boundary at fusion Sith a qrain size exponent p= 3, wnt ate vated t vares lattice diffusion wee minats with Cee te the vacancy creation or a sorption processes rather than va eae aoe elutes rate ing, as might occur 1 . controllingy are present in the grain boundaries, Phen tl may become interface deformation rate eeetrolied. The strain rate would then vary as the square of the applied stress [123]. 4.3 Superplastic flow - Region II jate strain rates (Region II) the flow process is less well understood, although there is agreement on the microstructural features associated with it. Strain is accumulated by the motion of individual grains or clusters of grains relative to each other by sliding and rolling. Grains are observed to change their neighbours and to emerge at the free surface from the interior [126]. During deformation the grains remain equiaxed, or, if they were not equiaxed prior to deformation, become so during superplastic flow. Textures become less intense as a result of deformation in Region II, while the converse is normally observed in Region III. The motion of individual grains is dependent on both the normal and shear stresses acting on their grain boundarie and is therefore dependent on the shape and oracehne ou of the grains. Translation and rotatio! stochastic in nature, occurring in different directions i different locations.’ © “+*etent extents, at At intermed Many attempts have been m: ave lade to de ies capable of predicting both the mcehenT ay and, es of superplastic deformation However, none have yet be a Unlike recovery controlled qo one ee tely successful, dislocation climb i. beee od ae ap ten, gzeeDs BP i S accept d Pted mechan een recess several eaters iminin ave been propose for superplastic 70qe 4. PHENOMENOLOGY OF SUPERPLASTIC DEFORMATION grain boundary sliding \ e055 One oo — Ce) grain emergence to fill vos swapping of nearest neighbours ain boundary sliding. © 4.2 Unaccommodated gx in the grains will be Figur’ ghe size of the holes betwee! Teauced as a consequence of grains moving the plane of motion shown perpendicular to [131,132]- 127,128], mukherjee [ 130} and ee reviews by Langdon [12, deformation (S$ Kherjee [129]+ Kashyap and Mul Gifkins [131])- x in a If grain poundary sliding was to occu: | completely rigid system of grains then voids would develop in the microstructure [132] (Fig- 4.2). The holes or cavities would expand or contract as grains, moving in three dimensions: approached oF . However, many superplastic - ein boundary sliding ities are receded from them materials do not cavitate- mmodated « cavi = therefore acco! reer ee observed, their distributio! homogeneous and while they would accomodate sliding, cavitation is not as Likely 27 son ox accommodation mechanism as either diffne dislonetion activity. If the accommodath 71ici i t the defo. sufficiently rapid a _th eformat, Pomperature, then grain boundary sliding itse),% could be the rate controlling nea [133], diternatively, if grain peel then iaing was intrinsically rapid [134,1 . accommodating processes would be rate limiting, i ly limiteg icrostructural studies have found onl evidence of dislocation nctivity within Ene GEaing of materials deformed superplastica ate Ough a, the strain rate approaches that of the ransition to Region III this becomes increasingly apparent [136-141]. By way of contrast, large grained non-superplastic microstructures deforming under the same conditions of strain rate and temperature as those associated with Region II show extensive dislocation activity, often with a well defined Substructure present. The dislocation density in the subgrains themselves is, however, generally very low. .The fine grain size of superplastic materials, coupled with low flow stresses at a given strain rate and temperature, ensures that the equilibriun subgrain size is greater than the grain size (142]. It might then be arg hat the material is ‘deforming by conventional dislocation creep and that the dislocations are observed as most of them are trapped within the grain boundaries, However, the absence of dislocations within the grains may also be cited as evidence in support of superplastic flow accommodated solely by diffusion. In examining the models which to explain superplastic flow, distinguish between cause and have been proposed it is important to effect. Often, it is neePI 4 PHENOMENOLOGY OF SUPERPLASTIC DEFORMATION 4.3-1 Dislocation Models when grain boundaries sli wnetiop wherever that guider stress concentrations relaxation of the stress concent etre pon emission of dislocations from ntrations! by’ the onitneix absorption by another can be boundary and vate at which the dislocations. pei lttea (source control), the rate at which th ome the grains (glide or lattice climb conttol) or the rate at which they are al i i (grain boundary climb canteen the ed pea has been assumed to occur relativel Pa Sdty in superplasticity since there isa Tack oF either ‘strong’ obstacles or significant solute ar. er effects within the grains at the ‘deformation’ temperature [38,127]. Pile-ups of dislocations adjacent to the grain boundaries are thought to develop and provide a back stress against which the sliding grain(s) would have to work to emit further dislocations along a particular slip plane. Climb of the leading dislocation from the pile-up into the boundary wo allow another dislocation to be emitted and enable a small increment of grain boundary sliding (= strain) to be accumulated (Fig. 4.3). Stress expo! ents and grain size dependencies of 2 would be predicted together with ah activation energy commensurate with that of grain boundary diffusion. If only one source per grain were activated then flow would be boundary climb controlled. By allowing grain boundary ledges to the absolute magnitude act as dislocation sources, of the predicted strain vate can vary by upwards of d is therefore source two orders of magnitude an' h f dependent. The deformation rate is given by : peebGh pby2 777? és =Asxx —— [-] [=] (4.7) kT a G accommodated = alternative view of dislocation - ow arises if grain poundary s1i ing ii ose accomodated by dislocation cli and glide within the grain boundaries themselves (243/248) - comm Pile-ups of grain poundary dislocations cou. 01 in the grain poundaries at triple points. 73climb of lead dislocation dislocation pile-ups in sliding and blocking gris glide and climb of grain boundary dislocations Figure 4.3 Dislocation pile-up models of superplastic flow. The rate at which the grains slide past each other can be controlled by (i) the pores of a pile-up of lattice dislocations (Ball and lutchison [38]), (ii) The emission of lattice dislocations from grain boundary ledges (Mukherje? boundary aislocarions (oLeneya: See gree Ee [144}). tons (Gifkins [131], Falk et al. 74. PHENOMENO! OGY OF SUPERPLASTIC pero RMATION jation of the i i880Ce tion oe tlons ing dislocati yattil® Nid glide i 1 0r boundary di On into either jon could go le in the other by, islocations gecting the triple poin oundaries ater ling proces i tt, would contro g P: s (Fig. 4,3), then be the rate jveral arguments have been rai . sslocation pased models of supe geinet ow, namely? The dislocation pile-u threshold stress for ce parniseric not predict a Region I. ic deformation, i.e. 2, There is no implicit mechanis: i crystal lattice of either the sliding or accommodating grains could rotate in the lattice pile-up model of Ball and Hutchison [38]. 3. Grain elongation is implicit in any model favolving dislocation glide/climb on a limited number of slip systems. 4, Dislocation pile-ups are not observed experimentally. Furthermore, at the high temperatures at which deformation takes place pile-ups would not be expected since the average stress is low. i If grain rotation was accepted as resulting from a non-balanced system of unrelaxed grain boundary shear stresses (See Section 4.3.3) tne? Syariations elon i ot be observed: andom fe aa would 3 of rotation would in the direction an i to cause slip to S slip systen (>30°) nO witch from one i be large another. Since the rot can ; net change in grain shape would then be apparent an Grain rotation would, on the other hand, Pe eoeion implicit feature of the grain poundary diss nee vain todel, again, the sh boundaries would be able, £° mo ausing a shear within e bounemid Tesultant torque on the grains WOOT an o the crystal Tettice of the grain and 8°)" shows the @rain motion would be observe’ oa the micro! Z ®ffect of superplastic strain OM cyearly OPP sli 2 perp eric alloy. ; Yiiy slide a Pb-62Sn eutectic alloy, ‘rains ‘ot roy PDz625n emraphs that the g 75seep Eee en - *Lopt] upezqgs uate ‘pue uoTjejueTIo uteIH Jo uoTeTITeA pernseou Suy (9) [¢pr] upexys ofaseTdzedns Surseezouy yaTM UTPTIS uTerb go uoTssexbord peazesqo (q) pue (e) oT OEANe usz9-qd ©U} UT uTeIQS otqsetdzedns butTseezouT eA OUL pp OANSTI uate uoTyequeTzo uTeIb Jo uoTIeTT NUS 3MYL z b 0 0z- =X 9 upib 7 = NY Nolvioy NIve9 3 5 i 5 g ve s a s é 5 3 § g 8 z 8 & a ° 77mantle model of superplasticity nd Figure 4.5 Core ai (Gifkins [131]) and rotate, but also move perpendicular fo fhe free surface [145]. Measurements of the orientation of specific grains has shown that the rotations can vary by as much as 40°, Figure 4.4c [146]. The objection raised against dislocation ‘accommodated flow based on the absence of either dislocation activity or dislocation pile-ups can be countered by the following arguments. Firstly, the stress generated at the head of the pile-up could Superplastic flow is of. and mantle! model, as een referred to as the 'core boundary Sliding is assumed to oc 78¢o 70 nm wide! the mantle would b only 30 4.3.2 Diffusion Models tt has also been envis, istribut nvisaged th redistributed by diffusional flow [iar ita) e stre: 1148), i Pleantial on adjacent grain bowbamagte Chace! teat occur Sliding ot compression to" sen C : Ss is ensi change in grain shape ag marten by a gradual hange in n matter is aif usio Grain boundary migration rect ms orig. equiaxed shape but ina ietatee *he ed’ orientation (Fig. 4.6). ; equiaxed grain shape Ms trerecentiog of ae Gadel of Ashby and Verrall. Furthcrnoze, because, transient but finite increase in grain boundary ° area results from the shape change, the model predicts a threshold stress for superplastic flow The strain rate in the superplastic region is that due to diffusion but operating under an apparent stress, o*. This stress is equal to the applied stress less the stress necessary to create that additional grain boundary energy. 2 DaseeGb ¢ b 42 7 €x-zx = Ao [-] [=] (4.8) | xT a G If the sliding results in grain switching then no Tyidual three-dimensional matter how small the individud p shape change steps arer a finite increase in the grain boundary area would still be required. 2 work done in creating the new grain, boundary surface would be some fraction 0: © ould have £0 : i area an instantaneous grain boundary tress. The increased be supplied by the extern the grain’ boundary energy is used toe pat the energy Subsequent grain boundary migrate york done per Subsequent gr@vform of heat. The vot grains in grain would vary a8 47+ ihe numbers the threshold unit volume varies 5 1/a? and nit volume, stress, which is the ress will vary as 1/d. The threshold re diffusion niporaaeede flow predicted PYInitial state 20 on ~s () 4 7 diffusion stage (3) (°) [ w & grain boundary \.B3 migration stage “\ | ye Cle) : a €=0-41 ) Final state Figure 4.6 Grain switchi i f ing thr i ional (aaa n) en eetulated by Ashby?and verral! (#1 (bi-ba-bs), *f4ed by Spingarn and Nix [148] 80Ee" _ 4. PHENOMENOLOGY OF SUPERPLASTIC DEFORMATION gcomnodation model varies as the inverse of th of the a : grain size. AS with the case of the disl i several objections have Seen wetecd based models, accommodation of superplastic flow s ° . e giffusion, namely: jolely by 1, the diffusion paths origi and verzall [147] eee aaa che proposed by Ashby Jace in different Seat ene opposite ence peethe same grain boundary. As Tee bari py the stress acting perpendicular Me ain PY ndary this is physically impossible trig. 4.6) 9, peformation in the Ashby-Verrall Model is not symmetrical. 3. If grain boun solely by diffusion then the lat! sore yidual grains cannot rotate. os ingen in figure 4.6 are only apparen from grain boundary migration. 4. the strain rates predicted bY meat two orders of magnitude 5. glongated grains should be apparen microstructure: 6. The threshold stress decreases with increasing grain size, experimental evidence. Moreover, +! significan 7. The grain c ik once giving a ™ imum strain of 0.55. (Emaxe=0- for the two-dimension@ illustrated in figure 4.6)- dary sliding is accommodated tices of the rotations and result equation (4.8) are too fast. in the ter modified by aths were 1 . if that each grain within maintaining jistic shape The diffusion Spingarn and Nix [148] 8° the cluster underwent the same change, d a more rea. symmetry of deformation an transient (Fig. 4-6)- i iven The grain size dependence of strain ate, 2 give by eqn (4.8) falls betwee 2 and 2, HOROTEES oniy observation of a stress exponent 0! i? ri Set ol as a result of the transit ch a theeneional | Stress (Region te 5 bperating under an apparent flow (Region II, 81824 PHENOMBNOLOGY OF SUPERPAGn re PEFORMATIO} 8 to conventional di 851 islocati, nes), rather than a distj On creep (Regi ie?th view of the indepenact superpiast eo esion and dislocation creap mee,of both 7 er, before examining multimechanism models of superpl, astic tl z mption that grai: fl WN the presul rain bound Tae. iow, ae Bea rate controlling process 1s coeeng tect re i 4.3.3 Grain Boundary Sliding examine the process of grai: — sonsider a two-dimensional array of someting Ht covagons orientated at random with cemmect t Hi fenotely applied stress field (Fig. 4.7) whe a h ind shear stresses acting on each grain ong, nonal { ant pe balanced by stresses acting throughout the 4 continuum body. The magnitude of the shear stresses i or normal stresses can only be evaluated if an if gssumption is made with regard to one of the types q of stress [149]. If it is assumed that the grain | boundary shear stresses are fully relaxed i (ra"te=te=0) then the grain boundary sliding | tate will be controlled by the rate at which the in boundaries are 4) axial tension, the 4 en grain boundary is normal stresses on the gra relaxed. For the,case of un. i average stress, 0, on a giv i i & = om (3/2 - 2sin0) (4.8) | ; i Je where o is the applied stress, and 9 is the ang. tetween the applied stress and the normal te ees grain boundary. If, for example, diffusion veimply the normal stresses then the stzain rate i that due to diffusion creep, 4-° : DaseeGb pb 726 2 (4.10) Gre 215 ——— [ = i kK? a2 'G it Thi . 4g) except the Ts may be compared with eg (ast g aifeusion®} StrecgS BO threshold stress: PY. of the norma SS relaxation, the local ¥' 83at the triple points to anywhe orn very te and -0.72 times the aPPlicg steel Sat the centre of the grain sees sere For Miiaxial compression, diffusional ma is ran Port would be predicted to occur from imately grain boundaries orientated a saan ia the perpendicular to the applied s rate hariee triple point, to the centre of pouneelied stres orientated mainly parallel to te hee Ss. If the applied stress was tensi cue opposite diffusional flow would occur in. be aoe direction. The predicted diffusion 2 ser an excellent agreement with the modifi: oe nas x diffusion accommodated superplastic J (Fig. 4.6). tractions on the If on the other hand the normal n | gain boundaries are relaxed by vapid diffusional flow (Gaz0n=0<=0/2) then the average shear stress, t, acting parallel to the boundary would be given by % = -ow (2sin6.cosé) (4.11 and the corresponding strain rate by . v3 qn-l fo yn é2-al—] [=] (4.12 2 G where A is a rate constant, and n the stress exponent characterising the stress relaxation process. If all the boundaries are equally resistant to sliding then the strain rate is independent of the orientation of the grains. If the boundary viscosity varies with, for example, misorientation or precipitate density, then . 1 €xr = yaa [(3a-3») (2080-1) + (SatSn-28e) (2sino.cose)] (8) where the grain centre shear rate, Sa is ' 84UPERPLASTIC DEFORMA‘ TION equal grain centre offset no sliding (x=0) sliding (X<%q)> sliding only (X= Xe) rotation only (@ = 5) rotation (<0) no rotation (w=0) figure 4.8 Sliding and rotational dis required to generate a given grain eee erst petween two adjacent hexagonal grains. (a) slidin perwe(xa), (b) rotation only (ws) and (c) a 3 orlure of sliding and rotation (Xds is given by [151] : DeviG pal po a5 & a —— [=] [=] (4.4 kT a 7 Dislocation | n boundaries into th? daries begj the stress ig increazen to become Progress toward the’ cent!? c flow gra i dually tends | region wh me is trend Controlled’ creer, mre | Strain rate yo a2 tO one wRefOX® a transition fron! © in Region § o,2?>ds. If the Given by: | (4.39) ional regime, |4 PHENOMENOLOGY OF SUPE) RPLASTIC D| EFORMATION 1 | 1 100+ deds d>ds ”_ fet | t>f>0 4 | iw : m05 | 0 o 10 | fe | dowdy n fo | 1 iL | L : \ 10 0 10° 10° STRAIN RATE Figure 4.12 Schematic illustration of the ot€ curve y the sub-grain transition model of predicted (de, sub-grain diameter; Spingarn and Nix (148,151). d, grain diameter) Region II, where d>ds is é : se +20 ex = f€x + (1-£)Exz2 (4.20) n size to the whe. 5 to of the subgrain § : re f is the ratio rnef stress with strain is shown grain size. The variatio fs tate predicted by the transitional mode) that the in figure 4.12, from which it can be S00" 7) regime Strain rate sensitivit he transitionr Te also is greater than in either Region) i ox Ollows from eqns (4-18) t? (4.20) “Region of m,p and Qe» will vary throughout REI)? edicts Unfortunately, the slip transition m0 93an activation energy for superplastic flow that i equal to that for grain boundary eer ete an lattice diffusion. Similarly, a gra n r e exponen of 0 to 1 would be expected in Region rather than a value of 2 to 3 which is observed experimentally. | | * | | An alternative explanation for the apparent similarities between deformation in Regions I anq III relies on the observation that the quoted graj, size of a material is in reality the mean value of a distribution of grain sizes, or that the Physica) | properties of the material are the mean of those of | two distinctly different phases. The variation of stress with strain rate would depend on the relative proportions of the two phases, or the form of the distribution of grain sizes and on whether the response of the ‘composite’ was that | obtained by an isostress or isostrain rate rule of mixtures, i.e. o = E £104 (€) (4.21) i | for an isostrain rate model, or €= 5 fies : (0) (4.22) for an isostress model, (Fig. 4.13) (152) ane peomeltional ‘noile) for superplastic flow based v microstructure i i etendpnieee precures is attractive tres see nts. Firstly, only the t t WO CO} i high ae deformation processes one eeguira 14 on the relativ i two microstructural extremes (ine os oe fe grains) or the divergence of erties eee the two phases, any value for th (n=1/m) from 1 to 5 can b %4 PHENOMENOLOGY OF sup ERPLASTY © DEFORMArT ‘ON faz fy 05 ISO-STRAIN RATE 9 faarfyOp m=02 ISO-STRESS = foarte STRAIN RATE Figure 4.13 Schematic illustration of the o:€ curves predicted by the rule of mixtures model | isostrain rate, --- isostress) with only conventional diffusion creep (m 1) and dislocation creep (m=0.2) operating in each of the two phases of the microstructure (curves 4 and b show the variation of stress with strain rate for phase a and b, respectively) .- i i i endent. Stress would be highly grain size Oeesis nrovides However ; stiona. , neither transitio & mpertyenation, £00 aod gia eenance of an ae teristic Superplasti ic flow, suc ®quiaxed grain shape, that are 5° chara’ Superplasticity- osi in two phase 4.3.5 Superplastic deformation © Pnase alloy, ” materials wh, | ers of duplex mat ich There are Largs ree of superplasticity than the show a highe: hase alloys. The mechanical ang peede foal fentures of deformation are sinilar , i alloys those of the single phase Tice ‘at a lower deformation generally takes » eample “0.6%, for s temperature, er tantan aed a/p copper ree as compared with “0. .. £ aluminium alloys. e . phonomaneLogical equations describing fuperplastic deformation which have been given 2 the previous sections have assumed a single DI microstructure, yet such equations were often derived from measurements made using duplex ; materials. The material parameters represented in these equations were either those of the softer | phase or those given by the rule of mixtures. A | much more elegant realisation of the ‘average’ | properties approach is embodied in the principles | | | 0 | of the transitional model set out in section 4.3.4, provided an isostrain rate rule of mixtures is used (Fig. 4.13). However, such an assumption would appear contrary to the observation that superplastic flow seems to be accommodated almost exclusively within the softer of the two phases (an isostress combination). Differences in the materi. phase of a duplex microst-: the two phases deform at al properties of each | ructure would require that | quite different strain x han unity) o ining hard pndeformable particles (for exampis theratning | i | both O/B titanium and a/ copper alloys Grain aents sliding would then occur by trode within 4 grater Beaadon the ® phase in the vicinity of the 8 phase would be more Gliding dislocations in the eek interphase boundaries than ie the pee? ay A B ions mi, ctescted to pile tp in focus MONE therefore ba Phase boundary (IPB) until glide ‘qithin % |4. PHENOMENOLOGY OF SUPERpra, ISTIC Dj nase could contin; EPORMATTO| ne P i : : ue. 7 N ensing distocation intersecting’: ability of faoxe#s° ncreased Ure erection of @ grain would o poe on the B® phase sles” Glide Gis barder depepe given by [152] 0 The strain rate weutd ould 1 - fa 42D te ee [ ; —=[- I gin kr ba, [=] (4.23) exe fa is the volume fracti i the grain size of the B enact @ phase and (4.23) is neo dissimilar to that bredienas garlier analys?s of Gittus [153] whe licted by the esftp within the interphase boundar re dislocation the rate controlling process. Agein a assumed to mt of 2 is predicted since the 7 a kre Pap from a dislocation pile-up as ‘the lel invokes jimiting step in the deformation puicess support for a model of superplasti ; seperials based on accommaation solele within the softer of the two phases has been provided b ch < analysis of Chen [154]. He showed that a better © agreement between the experimentally determined age pehaviour of Ti-6Al-4V and that predicted trom theory could be obtained if diffusion in the two phase alloy was governed by interdiffusion (D = CpDa + CaDv) rather than ambipolar diffusion (D = DaDe/(CpDa + CaDp), i-e- the fastest rather than the slowest diffusing species governs the overall rate of accommodation. 4,3.6 Continuum Model chen has also proposed a continuum mechanics model of superplastic flow based on the transformation plasticity theories of Eshelby [155,156]. It was assumed that a grain switching event similar to that envisaged by Ashby and verrall and others: : {147-149] was the strain producing step, but that these events occurred stochastically i al simultaneously at all joints with aicrostricture The grain switching events were oe only assumed to occur at random pace 97creel 0 p strain Figure 4.14 Schema predicted by the t: Chen [156]. The ‘1 that postulated by °: are relaxed by mac of the applie extrapolated to lower d stress. behaviour (Region IIL) rates 2 1 10° STRAIN RATE tic illustratio: ransformation p Ocal’ strai Ashby rientated Toscopic £ 98 12 n of the o:€ curve lasticity model of in generating event isalso with respect to the put *"since the individ directy, flow Mal grain 9,30, Of plasti were totally random, after a switching evenee ex of events there wou], Sufficient] Binee they would cancel the gfsg8° Ret strain out. However, Ollowing each grain Of each other the reozientated grains would seiqy witching event within their original locations, qo Rave to fit’ switched grains in their new congigy coomodate the jocalised concentration of straig guration, a large surrounding matrix. The local strain Lops in the relaxed by plastic flow throughout thee be naterial and the direction of that ites ody of the be biased by the applied stress in relaxation would flow direction. macroscopic qghe local strain, generated by the itchi event, then becomes redistributed and vantetee ne causes a very small displacement of the outer surfaces of the material in the flow direction. The relaxation of the local stresses produced by subsequent grain switching events, regardless of their orientation, each contribute a little to flow in the direction of the applied stress (Fig. 4.14). The resultant constitutive equation for superplastic flow, assuming that the long range stress relaxation process is recovery controlled dislocation creep, is given by (4.24) n/(n+1) Daas2Gb [ bo eo €xr = A Cex kT aG where €.. is the unconstrained transformation strain associated with the grain switching event, and n is the stress exponent of the) ea rang relaxation process (in this case n = ). ider the The continuum interpretation does nor ning process detailed mechanism of the grain switching Te Since that transformation does not i). “Ter in is Control the rate at which a macroscop) ia eccumulated. As the macroscopic & autelaxation of accumulated only from the long ze py that @ strains (or stresses) generar” "or that event ®witching event then the orientatt 99Grain switching can ehin ther eated, back and forth at the ae ee enarates a onal} i ‘ture and eac! - Strain in the macroscopic flow ae ane The modg, does not, apeeee Soe iby end Verrall and of | trictions as that 0 | Spingarn and Nix [147,148]. It can, be geen (irom | eqn (4.24) that the stress and grain a Ponenty | are non-integer yet close to 2,and t. thee th activation energy is somewhat greater at of grain boundary diffusion. | is irrelevant y | | 4.4 Summary In the preceding sections it has become apparent. that no single process can explain completely the mechanical aspects of superplastic flow, as | manifested in the measured stress-strain rate characteristics. Nor could it account for the topological aspects of superplastic flow, namely the maintenance of an equiaxed grain shape, grain rotation and translation. Intuitively this would be « @xpected since numerous Processes are » simultaneously operative within a microstructure which is far from uniform. Nonetheless, it has been ’ demonstrated that grain boundary sliding becomes increasingly significant i i Superplastic flow ma: Microscopic aspect In the absence of an signifi i | n L nifica dislocation structure, : aE ary girs | short Path lengths. the @irectie oree ener | to the accommodation Processes will be sensitive chemistry, and could foerah eeundany Structure are | degree of certainty, especially wicted var aany the various mechanisms woulg ¥oas the way in which4 PHENOMENOLOGY OF SUPERPragnre DEI FORMATION yoscopic aspect ac’ sponse to h 'gcain switching’ events. the transformation suc! sci othesis does not re, uire a detailed plasticity nypor the grain switching process except unders does occur and results in a localised that “tration of strain, te kinetics of the long concenvelaxation Of that strain throughout the range Tum, biased by the applied stress, define the cols of dieplacement. of the'sirerety surfaces of sate of Ofsphacen hence the perceived macroscopic We wsacetraln rate beliavicur ace given Eaperature’ for a given teal, grain cis’, temp 4015 CAVITATION AND FRACTURE 5.1 Introduction When a superplastic material fails during tensile deformation it is either the result of unstable plastic flow or a consequence of the growth and ‘iinterlinkage of internally nucleated voids. In the former process, inhomogeneities in the cross~ sectional area of a test piece lead to a localised increase in the strain rate and the difference in the cross-sectional area increases~ The rate at which the discontinuity in the cross-sectional area increases depends on both the rate of strain hardening and the strain rate sensitivity [157-159]. In superplastic materials, where true strain hardening is minimal, any neck which is present will always grow, although the rate of growth decreases with increasing m. Unstable plastic flow normally results in the material pulling out to a fine point prior to failure (Fig- 5.1). Where failure occurs as the result of the nucleation, growth and interlinkage of internal voids, the fracture surface is much more abrupt. The value of the strain rate sensitivity is important in determining the rate at which the vols aeen and hus to some extent controls the elongation to failu i i cavitation. ne in ‘systems which: undergo 102oO ——~<4 >—__—q a SF Figure 5.1 Shadowgraphs of the fracture i superplastic alloys after 900% tensile lengetiod wiunstable plastic. flow in Ti-sal-av and (b) , pseudo~brittle fracture by cavitation in Supral 220. 5.2 Fracture by Unstable Plastic Flow The value of the strain rate sensitivity index, ‘nm’, has been shown to have a strong effect on the ductility of superplastic materials (Fig. 5.2). In general, the higher the ‘m’ value, the greater the elongation to failure f16011 although this rule is by no means universal‘ Rai and Grant [46], for example, observed that the Strain rate at which the maximum elongation occurred in a superplastic Lara eutectic alloy did not coincide with that at whic the maximam strain rate sensitivity was mea G1) (optimum strain rate). It has been suggester Ue that the elongation at failure depends on the Value at high rather than low st Ricrostructural evolution during. ave a marked effect on the strasn rate Sensitivity. 4030 Ti-6al-2Sn-4Zr-2M0 © Ti-6AL-AV ‘© Supral 220 © AL-T475 @ Coronze 638 © Al-6090 OIN Toe % ELONGATION SPF‘d parts Figure 5.2 The predicted uniform Strain to failure as a function of strain rate Sensitivity together with corresponding data f, s or several commercial superplastic alloys. The Shaded are, range of strai superplastic forming those observed in co; 1045 CAVITATION anp FRACTURE ca) relationships definin, fra re during superplastic flow haoereit n= fied [161- ]- ane total tensile elong i aerjure of a superp. astic material will poet fe po rised of both the uniform strain be s cumulated within any necks which devel Pde en velop during ; tensile i consider now a ten Specimen in whi i comil neck for which the initial crenach there is a area varies from ac to aac, where a ost Z ost 2 goa % Bost 5 & aL supRat 220 z €2117x103s") Fort t:733K ft tt dence 000204060810 1214 16 STRAIN 1 18 2-0 Figure 5.4 (a) The variati, re 5. ) on of strai sensitivity with strain rate after deformation to various strains at ¢ Strain rate of 1.17% 10-3/s T riation of strain itivi i eteain during deformation at a stegusttivity a +17 x 10-3/s (Supral 220). waem wate if 1105 CAVITATION ayp FRACTURI cE 53 general Characteristics of Cavitati, ; on ite the large plastic s i 3 veh ned in superplastic maton which can be obteniished that cavitation may cece 45, 20¥ well esrrpiastic flow. The alloys Y occur during oe been observed include those benea cavitation 49,170-174], COpper [105,175-181}, ron “ruminium #3166,70,71)182], lead [183], silver tie titanium [185-187] and zinc [188-191]. th 1, si ¢ cavitation in superplasticity has bean » eviond extensively both from an experimental and Feviewed theoretical viewpoint [192-196]. In general cavities nucleate at the grain boundaries and thei subsequent growth and coalescence invariably leads to premature failure. However, and more important] from a practical viewpoint, the presence of ¥ cavities in superplastically formed components could adversely affect their mechanical properties. there is also evidence that cavities may develop from defects which pre-exist, and which are usually produced during the thermomechanical processing required to develop a superplastic microstructure (195). An important requirement for cavitation during superplastic flow is the presence of a local tensile stress. Under the conditions of homogeneous compression cavitation is not observed, and cavities which are produced during superplastic tensile flow are removed during subsequent compressive flow [197]. Superplastic closed die isothermal forging of Ni-base superalloys such as In100, starting from hot pressed powders which are heavily worked by extrusion, is used in the manufacture of turbine discs to give a sound cavity free product of uniform microstructure. It has also bees denonstrated that the superimposition © 1 and vdrostatic pressure during both uniaxial 2 or biaxial superplastic tensile flow 42 43>, damage fliminate cavitation [198]. The cavitation Coralso ich develops during superplastic ratie pressing ® removed by a post-forming hot Teor ‘the voids (HIPpin hich wi {199} ig) treatment whl itation it is . rol cavi therg In order to cont isveta poth the fore necessary to un 15 CAVITATION AND FRACTURE microstructural and deformati influence its occurrence and sage watch In the majority of studies fees variation of the total a cavitation only the a e* . me f. i i with strain at different strain nates acd cavities temperatures has been studied. It should be noted that the volume fracti Boe written as tion of cavities, Cy, can be Cy = EB Nya Va (5.6) where Nva is the number of cavities per unit volume having a volume Vi. The Tbe of voids of any particular size is related to the number of pre-existing voids and to the nucleation rate, while the volume of a particular void is controlled essentially by the growth rate. However, during the latter stages of deformation when the cavitation level is high (approaching 10%), the growth process is affected by the spatial distribution of the voids through void coalescence. Hence growth becomes, in part, dependent on the nucleation process. It is therefore evident that the variation of the volume fraction of voids with strain, strain rate and temperature, will not be a simple one. The effect of changing the strain rate and temperature at which deformation occurs may be further complicated by other factors such as grain growth and/or a change in phase proportions [105,175]. The morphology of cavities formed during superplnetic flow varies from one material to the next and even in the same materials deformed at a different strain rates. In general, three types cavities have been observed in superplasti materials (Fig. 5-5)- These are: i ii ~100 um. is with radii up to Elliptical voids elongated parallel ee the aspect tensile axis with Lengths up to ratios from 2:1 to 103}. 3. Groups of angular or “i up to 10 pm in length, # clusters of grains- : Spherical void ck-like cavities, each terlinked around 113cavity morphologies have been cited as. evidence for the operation of different void gow mechanisms. The circular section voids have often been taken to infer diffusional growth, while the elongated elliptical section voids are thought ty be indicative of strain controlled growth. However there have been few systematic metallographic 1 studies (or quantitative analyses) of the evolution of void size and shape during superplastic flow, Unfortunately, observations of cavities at the nucleation stage are usually difficult to make, Cavities do not form uniformly throughout the microstructure and information concerning | nucleation has to be deduced from specimens in which the voids have grown to a size where they are resolvable, either by direct observation or by the measurement of a physical property of the base material. Since cavities do not always nucleate during superplastic flow and, with only very specific exceptions, will always grow during tensile superplastic deformation, the mechanisms of cavity growth will be examined prior to a consideration of the nucleation process. 5.4 Cavity Growth 5.4.1 Diffusion controlled growth A cavity located on a grain boundary, whether nucleated during su perplastic flow or pre-existing, directed vacancy diffusion, by C f mn of the surrounding matrix or by a combination of both these mechanisms. The former process has been analysed by several workers [200-209] for the case of Spherical voids under pniaxtal tension and leads to a relationship of dr Dgw6Q 2r 1 rere Ps am | (5-7) dé ker? rv" In(l/2r) ~ 3/4 where o1 is the maximum princi, s pal stress local to the grain boundary, Q is the atomic volume, Tr the surface energy and r the Cavity radius. The 114S CAVITATION anp PRACTURE 3 () Figure 5.6 Schematic illustrati u ration of th i (void) growth process for (a) diffusion controlled growth, (b) strain controlled growth. < cavity growth is provided by the ential between an atom dary and that on the free the maximum principal driving force fo: difference in chemical pot on the stressed grain boun surface of the void. When stress is tensile, atoms move from the void surface to the grain boundary and the void will grow (Fig. 5.6). If the stress is compressive then atoms move in the opposite direction and the voids will sinter. derived in the literature differ ast term containing solely i voids. In rand 1, the size and § : i i £ equation deriv. tionships of the form of (5.7) ie pelateen a ebumed that the voids are widely Spaced on fixed poundaries orientated anaes perpendicular to the applied stress axis Tee tne thenselves small relative Fo the grain 5i5) chet the rad of vol poet jocreases as the inverse the rate of void 9=° us slows square of the cavity xadin® and: jority of ‘ the ma Considerably as the voids in "5ids have dimension, | i ials, the voi he. | superplastic mater the grain ote tne pound’ the eeraps Os iebally located onith respect Hs Ties | Shick oe eorientated randomly Te Miller ana 8 Spelzed stress, It was Pronot emetox was the sane a | when the vo. would be | She gen izes shen the goovel rate you enhiancad by additional mass tri face of the voig boundaries intersecting the our © skewn to ke [210]. Such enhancement was Jeter size to bee iti the ratio i . geain size [2i1)- The rate of void growth is then given by [212], dr DerSQ Fon 5 [ : é€ (5.8) | de kT d2 | , i i he previous where d is the grain size. Unlike t pre . equation (5.7), the rate of change of void radius with strain is independent of the void radius, r, Equations (5.7) and (5.8), which were initialy derived for the case of uniaxial tension,can be written in more general terms allowing the effects of multiaxial states of Stress and alternative geometries of deformation to be considered, (Commercial superplastic forming usually involves balanced biaxial or plane Strain rather than uniaxial deformation. ) The maximum Principal Stress, o1, can be redefined in terms of the von Mises equivalent Stress, on, (where on, = 1/V2((ox-02)24(G2-05)2 *(Ga-01)2)a72) and the superimposed Pressure, Py, dS. &nd the extent of grain values for }ding. The d termination of appropri — Scussed in Section 5.4.3. 1165 CAVITATION AND FRACTURE lasticity or i pede? B y strain controll ed growth it: cowth by : ror cavity g h by deformat: surrounding matrix in uniaxial tension th , the gelationship dr n [ 3r Se ef[r-= ] : de 3 2ow (5-10) has been proposed, where i parameter, is dependent on beer the a ionen sate state and the geometry of deformation 1192 193). gnlike diffusion controlled cavity growth, th of void growth increases linearly with void si a and is independent of the strain rate. Plastiate, dominated void growth is controlled by the mean ¥ ‘ stress (Om = (Gatoatos)/3) rather than the t maximum principal stress. The parameter M in equation (5.10) has been determined theoretically by several workers for the case of uniaxial tension,with values ranging from 1.22 [213] through 1.25 to 3 [214]. Following the analysis of Rice and Tracey [215] it can be shown that n Bom = = 0.5585 einn[ << ] 3 2on ~3€2 30% + 0.008 —] cosh [= | te-12) €.- Es 20 where om is the mean stress, C¢ the von Mises equivalent stress, 1, é2 and €s are the Rice and Tracey showed that principal strain rates. nd T h N would be equal to 0.9 for uniaxial tension, 1.65 for plane strain and 1.92 for balanced biaxial tension in a perfectly plastic solid (m=0). Alternatively, Budianski et al. predicted that 9 Would take different values depending on the magnitude of the strain rate sensitivity [216]. For example, in balanced biaxial tension, which would arise in the case of blowing 4 hemisphere from a circular 117i . calculated t terial, Budianski et asa when m= O.3nt feo 93 Squal 1.75 when moo! (317) have also 3 ang 2 a when m=0- growed? e t on the strain Shon i in that the value of Oe will vary vite aa sensitive y jationship given by Stownded to rate rial tension ean be readily Tiowing im riya ae SREY a be shown tha MS and Ashby [202] it can t [218] 2-m Om 1 ) an? [=] sinh l= [=] (5.19) 2 m 2+m on or [193,216] | 3 L/m F Om M = = (140.932m - 0.432m?) [ =] (5.13) 2 On The term containing the ratio of the mean stress, Om, to the von Mises equivalent stress, os, in equations (5.12) and (5.13) defines the triaxiality of stress local to the grain boundary. The form of this term is dependent on the | geometry of deformation and may be rewritten in terms of a geometric constant, Ks, and the superimposed hydrostatic pressure, P. Thus i One ks P | = (5-E] oy To establish the values of k, deformation need be petstie case of rigid grains s and ky _flow, two modes of | whneidered. Firstly, the os id e@ there i. Eogndy sliding and secondly, the cage of piealy ate g ins, where the state of ean boundaries ig dependent e extent of they cap As the voids are located erience the loc, they al rath to a large none and thus their rapes “ge weh is total strain ent, dependent on th fe ee £ the Carried by the grain b ude 3 oundaries. o&5 CAVITATION AND FRactURE Figure 5.7 Geometry ado, e pted by Cock: [202] to determine the Diese etste locates 2 freely sliding grain boundary. a 5.4.3 Determination of stress state local to the grain boundary consider first the two-dimensional arrangement of hexagonal grains shown in figure 5.7 (see also figure 4.7 [149}) [202]- he imposed stresses 0% and o2 are applied at some arbitrary angle @ with respect to the orthogonal axes defined by oa2 and gee Resolving the forces due fo the remotely imposed stresses and those acting on the continuum body, 1a and o22, then Ora = OM2 cos?6 + OM2 sin? (5.15) ges = ow, sin?e + 02 cos?6 (5.16) ‘ iding then the If the boundaries are freely sliding th shear stresses will be fully pelanets i.e. Ta=th=te=0 and the stress, Ser. Norn] co the grain boundary Wall be [149,202] 17 oe = 1/2 (0% + goa) + (0% ~ gm2)cos20 (5-17) 19The stress on the grain boundary will vary in a i 60°. The iodic pattern with a repeat every average stress on the boundary is obtained by integrating equation (5.17) over pnt lite ot to n/g (30°), since there is an equal Prope hoe the he hexagonal grains being present in e. the Uoetoh alternative ‘flat’ orientation or in ipright orientation. 1/6 oo de _ 0 Ge = ee 1/6 do 0 _ 1 3v3 Fes om [2428 2 2n Ge * 4001/3 - om2/3 However, in a real materi: occur in three dimens that there is an equa. boundary will lie in Oa = 4/3 om, ~ 1/6 ( Normalising with respect. equivalent stress we obt. O21 1 [ 36%, ga, Cz 2 On On In order to obtain the 1, is now assumed that the the same as the remote qd. 0%2 + ows) to the von Mises ain Ocal mean Stress, local deformation leformation rate, 120 al grain motion will ions not two. We now assume 1 probability that the sliding i.e. (5.20)5c, AVITATION aND prac. ‘TURE geviatoxic component (gs stress fields is the >) °f the Io, me. Ag cal and remo: te oa 1 se=— - = (Ga + 3 3 °°? F 93) = (a, - Sx) (5.23 then *23) O21 - Ox = 0%, ~ coy (5.24) gubstituting equation for Om (5.21) for 61 and solving 30m3 ood om ~ = OH + Ome 2 2 OO, + OMe Om = (5.25) 2 Now divide through by om to obtain th degree of triaxiality im the local FOM. Om One 1 a rr 5.26 Cx 2 L Oz Oz { ) he remote and local extent of £ the three remotely applied nd ow3) it is for any Having defined both t! triaxiality in terms © principal stresses (0% 2 ai now possible to obtain kp and ks deformation geometry- 5.4.4 Uniaxial tension with a superimposed gas pressure P. i the 4] plied stress system Ba beer ena fen in the presence of a for uniaxial deformatio 121iven by confining pressure, Pp, is giv o-P . ° 0 -P 0 (5.27) Osa = | 0 9 ” For the case of zero grain boundary sliding, the ratio of the maximum principal s equivalent stress is ee 8 5/3 - Boe (5.28) Ox Cz and the ratio of the mean stress to the equivalent stress is Me On -P—-P- Pp me PE 1/3 = P/ow (5.29) Ox 3on ronpeans (5.22) and (5.28) it is possible to obtain the ratio of the i maximum principal stress to the equivalent stress for A the case of freely sliding grains; Oi/oe = 1/2 (3(62-P)/op — (1/3 - P/oz)] = 1/2 (8/3 ~ ap/ony Oa/oe = 4/3 Poe (5.30) and from eqns 5.26 mean stress look to ene? 12429) he slidi to the equivalent Stress; Baling x the ratio of the rain boundaries Om/oz = 1/2 Uon-P)/on 4 (1/3 ~ pg ' = = 1/2 (4/3 - 2P/on) Om/Oz = 2/3 . P/or (5.32)5 CAVITATION anp FRACTUI IRE comparing equations (5,2 yations (5.9) and (5, ) to (5, esa lysis, values of septa). then znd atieh bande, eee rmatzon are cbten iskss2 for prentding super Pet ulge forming in. ned. However Seformation of sheet matesiay owes, the snapes so +t is also al into t ad ional geometrie. nei strain deformation. 8, balanced Biaxial and veane 5.4.5 Balanced biaxial deformation consider the forming of a hemi . emispheri ; under oaceure P ‘opting pressure PHSP against a i _P. If the thickness of being formed is uniform and the Seatae ee eee h, is @ then the generalised state of stress eurvauee defined by 8Pe/2h -P 0 0 oat 0 8Fp/2h- PO (5.32) 0 0 -(P + 8P/2) practical forming operations Mises equivalent stress is Tt then follows that for zero For the majority of >>h; hence, the von approximately 6Pe/2h. grain boundary sliding 903 /ou = 3/3 - P/o= (5.33) and o%,/on = 2/3 - P/on (5-34) For the case of freely sliding grain poundaries or/onm = 3-5/3 - P/oz (5.35) and ou/on © 28/3 - P/o® (5-36) 123skoS3.5, for balanced biaxial forming, 3: Bad, Hence 2skaS2.5- 5.4.6 Plane strain forming ingular pan der the bulge forming of a rectangular Pen thore the length of the pan is a Geformation Tes idth. Under such conditions ‘is negligible a, peaventie aoe length of the pan is negligibl ae 9h therofere bo approsinatod to plane Strat back Pressure ip Fr 'thes eke sucte of stress within '@ Sheet of thickness hand curvature @ is defined by 8Pg/h ~ P ° 9. ony = 0 8P/2(p/h-1/2) - P 0 (3.37) o ° ~(P + 8P/2) The von Mises equivalent stress is a y2bPe/ah since g>>h. It then follows thae foe zero Sui Poundary Sliding, the ratio of the resete sami Principal stress to the equivalone stress equivesensti© Of the remote mean strees to the equivalent stress, is given by Om /on = 23/3 ~ P/on (5.38) and Pproximately OK/o2 = V3/3 ~ Pion (5.39) Tespectively, whi le for th. Svein boundaries we fing “8 Of freely sliding on/on = 2.8V3/3 ~ p/o, (5.40) and On/om = 1.5V3/3 ~ po. Hence, for p} (5.41) V3SkeS1.5v3" lane strain forming, 2V3kes2.5V3 and 1245 cavrtarroy AND ne relationships for 4, and the mean stress touts eaximum are summarised in Table 5.;° %¢fo FRACTURE Principal stres, p a mation geometry jn general, the value o¢ bot! particular geometry, h ke and ke, values derived above will lie between th for a first approximation, pe obtained by linea: extremes cited above. The limits e t : (for diffusive growth) and ks (for strai controlled growth) are summarised in Table 5.2 5.5 Analysis of Cavity Growth When the rate of change of void radius i strain predicted by each of the three possible void growth mechanisms (eqns (5.7), (5.8) and (5.10) is calculated, then for the majority of superplastic materials which are deformed under conditions where optimum superplasticity is observed, the growth rate of an individual void is dominated in the first instance by diffusion (egn (5.7)). However, once the void attains a radius between 0.5 and 1.5 um then strain controlled growth (eqn (5.10)) _ dominates. Only in materials with a very fine grain size in which the optimum strain rate for _ superplasticity is <10-*/s does superplastic diffusional growth (egn (5.8)) become significant. Bach of the three equations deserthing oe rate of void growth per unit strain will predic J different variation of void volume wath stra: Providing that there is little void au eat 1 ere coalescence during deformation then tne Ne ae fraction, Cv, will be an integer multip: Volume, V, of an isolated void and the two ar 4 therefore equivalent. B Piet ould then enable the volume tion with strain s identified doninent void growth mechanism to pe ieeasion 1196]. For the case of conventional Gttsrter aia Controlled growth the void volum increase linearly with strain‘ € dx/d€ a 1/x? hence Gv & 425 (5-42)EEE antl = = =o/*%o SuTpyTs ATSeTT sold - 9/5 Sofa Oe no BuTPTTS ON sO/d — EAC/E Bo/d - €A/T FO/d - €/@ F0/d - E/T yqoxrb Aqtaeo peTTorquoo uTez4S - =o/to butTptts ATees wore od eoytuo BHuTPTTS ON so/d - Z/G *O/d - 9/L = wav Bo/d - T sO/d - T ™o/d - EA/Z yqmorbh AAgTAeO peTTOTWUCD UOoTSNFFTG — yeTxeta TeTxetTun ssezqs Tepow a R ufeazyg ouetTd ‘[6pL] ez00q zo pue [z0z] Aqusw ‘6 BSutptrTs askteue ey} wory paatiep se Axepunoqg uTez! epost? @ ES pue ‘Axepuncd uteab pTbTz e OF TROOT sessezys ey. Fo (wx03 eqeutxoidde) seTouepuedep eanssezd ey} Jo Azeuums T°¢ eTqes 126‘TURE 5 CAVITATION AND pRac. a RR £AS*z Ene ia woputeen [eTxetTq psouerted eee €A a z ste € i : E yeTxetua = oe 7 0=* pou = ax uot zeEULojJed 5 nu (t=x) Burprts Arepunog utes 3001 pue (g=x) Sutptis Kzepunog utex6 ou 20x pue y3MoIH paTTorzUeD UTeIYS pue sATsNZZTp Toy yQMoIH Aqtaeo TOF yULYSUCD OTZJeUIOeH ey} JO sonTea Suy3qwTT eyR jo Axeuwmg 7g oTqeL10 ° a © IN 629 @ IN 836 @ IN 744 VOLUME OF CAVITIES, % om Fe-Ni-Mo-Ti ot m SUPRAL 220 © Ai-8090 0 05 10 15 20 25 STRAIN Figure 5.8 Experimentally determined i i relationship between void volume fraction and strain for sever superplastic alloys deformed within Region II. 1285 CAVITATION AND FRACTURE I the voids are larger t giffusional growth continues yoid volume fraction will in of strain: e the grain si ze and to dominate, then the ‘rease as the cube root dr/d€ = constant hence Cy a €273 (5.43) yf void growth is strain cont: yolume fraction would be onpectad’to tea the: vold exponentially with strain: DEreAse: dr/d€ a x hence Cy a@ ef (5.44) or Cy = Coexp(Ne) (5.45) ghe variation of the volume fraction i i strain has been measured for several oF rete etice alloys including o/B copper alloys [105,106,217, 919-221}, low alloy steels [63], a/t stainless steels [66,222] and several aluminium alloys [198,218,223,224]- In all cases the variation of the void volume fraction with strain is best described by 4n exponential relationship indicating that for the most part void growth is plasticity controlled (Fig. 5.8). During superplastic flow, void growth volving plasticity control would not secur in the manner assumed by Hancock [214] or Tait and Taplin [225] where the individual voids, i much greater than the grain size, grew in an essentially isotropic matrix. re and practice, wl i sions 0 e grail cavities are com) is tooo sliding. This res ‘in equiaxed. The while the grains t 1 uiaxed: Pived ignifi eviations rain most signif icae 3 id volume fractions and low the voids are small strains. Under these ©2) and 28s ated a iffusional growth would be expected to dominate. Us ts tune noneurements of the variation £ void vols © tor fraction with strain exist for iow strat Sraction wie onditions outsice superplastic regime: 129SUPRAL 220 Composite 7 Plasti¢ growth] (m=055) GROWTH RATE, m Superplastic q diffusion growth | Diffusion growth a CAVITY RADIUS, m Figure 5.9 (a) Calc cat; growth rate with atrain t e araeton of cavity and combined mechanisns, °" ‘#fusive, plastic 1305 CAVITATION AND ppp CTURE SUPRAL 220 Experimental CAVITY RADIUS, m 0 04 08 i620 b STRAIN of cavity radius with ove mechanisms together rements. (Supral 220, = 10MNm-?)« (b) Calculated variation Strain for each of the ab with the experimental measu T=460°C, €=1.2x10-3/8, Ox 131i dius with strain and the the, change 37. ord growth rate (S455 Sie voiq eae eqns +T)a +8) ang | zadieee calcu ene effect of surface tension hag | Oe ) cuded, have been plotted in figure 5.9 ea data for Supral 220 as an example (T = 460°, ée 117% 10-3/35 o=10Nmm-?). In calculating the variation of the radius as a funtion of jotrain it was assumed that the void. prer sxe ed an that nk the start of deformation it had the minimum siz, | consistent with thermodynamic stability (r. = | 2r/o). It can be seen from figure 5.9a that for voids with radii up to “0.3 um, the rate of void growth by " diffusion is the greater and thus dominates. However, for void radii >0.3 um and <1 um both the plasticity and diffusion controlled processes proceed at a similar rate. For voids with radii in | excess of 1 um, growth occurs almost exclusively p, | the plasticity mechanism. It is evident from figure | "Sa x the voids to reach the sizes (Fig. formed ere In order either the ir void coalesa, ger than assumed or must occur during ic flow is, for the m f the voluse ct+ Therefore, the sate lepend sole). an isolated void with | “t¥ on its volume. When evo 132 |+10 Voids upseey J 220, T = 460°C 2) ms voidg coalesce the voi ther of those from -s§ volume would ine: d that of either o: ver, the effe rate of amce during superplastic i ex a ee econ Sion on che Bie mand ore olde per popu arid ithin 2 material. The i i he voids wi ; cortale distribution of t within a sy i se coalesc' systen effect of Patributed voids, on me growth rate can be estimated from [ id coalescen unit density ( az aCya(SE) N(0-13r - 0.37(ax/d€)25E + (dr/de), ir ov ae 1 = 4Cvb(SE) NSE (5.46) where Cy is the instantaneous volume fraction of i i i 11 ids, N is defined in egn (5.12), 6€ is a small Tnorenent of strain, (dr/d€)s is the rate of void growth per unit strain of an isolated void and __8(8E) is given by (SE) = (1 + NEE/3 + (N&E)?/27) (5.47) At low volume fractions of voids (Cv<0.01) void coalescence has a negligible effect on the mean void growth rate, but becomes much more significant at volume fractions greater than ~0.05. Unfortunately, it is not possible to calculate the variation of the volume fraction of voids with strain without a quantitative understanding of the functions describing the distribution of void sizes, the overall population density o. ids and their variation (if any) with strain’ te a quantification of void nucleation, However by al observations of the si distributions and populatio, Teies op eeiae ‘ n i superplastically deformed allogen ate? of voids © Semiquantitative analysis of the evolution of superinpoced wager which can include the effects of posed hydrostatic pressure can be made , 5 ronda per u ase alloys, is between 1024 and 5 x 1015/m3 m wate5.7 Role of Hydrostatic Pressure on Cavity Growth Experimental results have shown that: L, eauoved by subverts superplastic flow can be compressiv 2. Annealing of cavitated natecial aftes tune reduces the volume fraction of voids [226] $ 3. Post-forming hot isostatic pressing has the potential to remove completely any voids formed during superplastic forming [199,266]. unfortunately, the annealing of superplastically formed parts is only capable of removing the very small voids, where surface tension is capable of driving the diffusion process necessary to sinter the voids within a relatively short period of time. Similazly, the hipping process may be limited in application in view of its cost, its restriction on component size and the possiblity that the voids will reappear during any subsequent heat treatment [227]. From an examination cavity growth prese! of the equations describing inted in sections 5.4.1 and ona 5.4.2, it is apparent that the superimposition °: conéifing ges, peegrUreycisr the rate or, writ i res : = . deformation she Sauces poth the maximum principal stress gad the mean stress « The erorts ae ew i i i ction 5-° ® 9 1 that ste cited £P than diffusion is the dominant 28 ekSaen mechanism duree? superP'Setexmining the void growth me imple criterion for Sea easily deformation. 2 o) o£ pydrostatic preseuy i iate i nam decived Erom equation (5-14) (5.48) Om/on < 0 135only uncertainties that remain are in the ectablishment of the appropriate form of Aa equation to represent ox/oz (see Table 32) and of the magnitude of the flow stress of the material under investigation. Equation (5.10) then corresponds to the condition that f takes a zero or negative value. Quantitative investigations of the role of superimposed pressure on the accumulation of cavitation damage during superplastic flow are limited. The majority of the available data relates to commercial aluminium alloys such as Supral 220, Al-7475 and A1l-8090 (198,218, 223,228), Bampton and co-workers [198,228] reported on the effects of hydrostatic pressure on cavitation in Al-7475 deformed in uniaxial, balanced biaxial and plane strain conditions. Similar results have been obtained in subsequent studies on Al-7475, Al-8090 and Supral 220 in both uniaxial and balanced biaxial deformation [218] (Fig. 5.11). In general, increasing the Superimposed pressure was found to: 1. Decrease the rate at which th P of voids increased with strai + Decrease the level of cavitation at a given strai e 1 itat rain Bis Displace to higher strains the strain required 4, Prior to the detection of «, ation, + Increase, to a limiting val i fallen” J va , the strain to © volume fraction ne aN C n a The latter r n 1s consistent with ¢ i heer dominated model of void growppeeticity parts ie use of equations (5.12) or (5.13) "to define n 413A5 CAVITATION AND FRACTURE CAVITY VOLUME FRACTION E ee 0 os CSSCS*« STRAIN Figure 5.11 The variation of void volume fraction with strain in Al-7475 at different superimposed pressures for deformation in balanced biaxial tension [218]. since the cavity growth rate factor is zero when Ou/on is zero (Fig. 5-12)- lastic It has been found for 4 number of SuperP: aluminium alloys that the cavity growth rate factor, N, is strongly dependent on the applic State of stress with cavitation being vir ually Sliminated for superimposed pressures 0: jaxi tress a i of the uniaxial flow s' Ngo S12) C218] ne the variation of the (Fig, . again 2 Velde Helen oF voids’ with strain under 137o—0 Experimental 08 1 06 O4 02 0 02 Oh 06 oe 10 PRESSURE / FLOW STRESS Figure 5.12 The variation of the normalised cavity growth rate parameter (A/No) with pressure normalised with respect to the flow stress, as calculated for different values of th i factor Ks. Values determined experimentally for several aluminium alloys deformed in uniaxial tension are also shown (0). 1385 cavirary ‘ON AND aifferent confining ote FRACTURE experimentally to 5b SSu: controlled model of i Stent with a iezal rain In order to calcul ate the pressure on the rat effect of superi damage it is necessary to comsation of cavitath voids in unit volume v2 estimate the number of “5x 1024 voids/m> pre- _it is assumed that are the minimum size Sense et, and that the voids stability at the onset of deformeryen ne motynamic 3P7(o-P) 3, then it se poe deformation (i.e. ro = volume fraction of voids ata toed mine eee” level of superimposed pressure. In figure. agg oe the variation of the volume fraction of orden, strain is plotted for Supral 320 acer oleae strain rate and temperature where the strain 2 te sensitivity, m, is at a maximu ase of e . . For the case of P=0, the volume fraction of i aly ; v void i with strain, reaching 18 at an elongation of 2508 and “8% at an elongation of 500%. In contrast, the application of a hydrostatic pressure equal to half the flow stress, P=5Nmm-?, limits void growth to such an extent that the volume fraction of voids would be less that 0.1% at an elongation of 650%. Metallographic studies have been made of the void size distributions and void morphologies which develop during the forming of aluminium alloys at various levels of superimposed pressure. The work has shown that the number of voids per unit volume is little affected by the pressure although the size to which they grow decreases dramatically with increasing pressure (Fig. 5.14). The restriction on void growth imposed by the confining pressure is coupled with a change in void morphology. At zero ures the voids are normally or low confining press! d observed to have grown in a cracklike manner. Voids which were initially spherical spread along the grain boundary pecoming elongated, often th lengths far greater than theix height. The change 139SUPRAL 220 S So 0 = rs} z oc L 42 Ww = 5 oa S > E = S 0 55 a 04 08 12 1 20 STRAIN Figure 5.13 Calculated variation of void volume fraction with strain for different superimposed pressures. (Supral 220, T = 460°C, € = 1.17x10-3/s, oz = 10 MNm-?) in void morphology from spherical to cracklike is a consequence of the inability of surface diffusion to redistribute matter from the poles of the void (low curvature) to the void neck (high curvature), at a rate commensurate with that at which atoms leave the neck along the grain boundary [229-230]- fhe spreading of a void parallel to t i boundary allows the rate of increase Hd ts major axis to be large, while its rate of change of volum 1405 CAVITATION anp PRACTUI RE -3 mm SUPRAL 220 NUMBER OF CAVITIES 0 4 8 2 6 D0 % 28 32 & VOID DIAMETER, um Figure 5.14 Measured void size distributions of Supral 220 at 450% elongation with varying superimposed pressures. i i i .7) [230]. Thus at is ll that predicted by eqn (5 low confining pressures, void coalescence occurs frequently and large voids quickly develop. nce formed, the larger voids grow as @ consequence © the plastic deformation of the Sur ee ix, iffusion, and the grow . rather than PY @ycally: However, at higher, confining pressures, ane, vole saneycifeuston a, cemnee i x boun i ‘pata griving force £0 orefore limited, and the large 141(Axqeuo0zTsusp Aq pojyoeqep uoTzeyTAeo ou - p/u) ae SCT°0 = POTTEF = peTTeZ = petTey = peT Tes 009 %20°0 RETO POTTeF poelTyes perres 00s %20°0 SIT" %SS°0 BI'T peTTes 00% p/u %t°0 %SE°0 %S"0 80'S oo€ psu %80°0 820 8570 39°T 00z sore ve eT £70 0 (s) (z-wun) eznsseza urez3s *S/e0T Pue 2,025 3e PeUTOFep 0608-TW jo yo eq CUO UT UOT}eRTARS Jo JUe}xe eYR US eAnssead Jo Joesze ey €°S STqeL 1425 CAVITATION anp FRACTURE networks which are observeg at li ow pressures are prevented from dovelen(er ning oping. flow stress to eliminat i qgherefore the use of hydrostatne ny sued superplastic forming has been to init th _ development of cavitation to such an axcent those voids which do form (Cv<0.18%) would b Phat relatively innocuous, being small, spherical and phesstipsrerne If it is assumed that ~508 of the tota Strain is attributable di grain boundary sliding, then the nisin fe hydrostatic pressure that would be required to eliminate cavitation would be P = 0.75 o(€) (5.49) for balanced bi-axial tension, and P 2 0.72 o(€) (5.50) for plane strain deformation, where o(€) is the measured uni-axial flow stress at the forming strain rate at the point of greatest strain. (The values quoted are midway between those for no sliding and those predicted by the models of Cocks and Ashby, and Beere (Tables 5.1 and 5.2). In practice, would not normally exceed and thus the use of a confining pressure is limited to the slow strain rate forming of alloys such as Al1-7475 and 41-8090 whose flow stresses at strain rates of 2x10-4 to 5x10-*/s are of the order of 3-7Nmm-*. i ducing The effectiveness of back pressure in re the level of cavitation in superplastically formed components, and of increasing the Toxatng potential of marginally eaeteearas ten high 5.3 st, is readily appar | 3. Feeag na eeee chat the elongation 10 fot se incr n= 300% to 76008 while canlention i ‘reduced from 5% py volume at zero 143coalescence ic i i £ void Figure 5.15 Schematic illustration o . nucleation as a result of unaccommodated grain rotation. ~ superimposed pressure to nil at an elongation of 300% with a superimposed pressure of 3.05 N mm-2 (450 psi). 5.8 Cavity Nucleation 5.8.1 Introduction Strain is accumulated during superplastic deformation primarily as a result o£ grain boundary grains themselves, The deformat. topologically and the formation voids is therefore intimately 14. mechanisms of deformation, move relative to each other th v ; en the i voids becomes geometrically necessary [loa ign 35 1132, (Fig. 5.15). Fortunatel- and translational displacesee ver? the rotational ements of be accommodated by the redi ©, Of the grains can - . ist: i hears a narrow zone (mantle) areeeg", of matter magaries, with the interior of the phe, grain proponeny essentially undeformed [131y aes. been = that the width of Such a nantie nenda only of intergranular inked to the Tf rigid grains are to 1445 CAVITATION any FRACTURE of the order of 0.07 ¢. 0 grain diameter, 1 ine grain superplastic materials @rei", __” a 7 1s snall atl indeed (30-70 nm). The acconmodatia® ain boundary sliding, as hag been shown be achieved in several ways (128): "can be imes the 1. Grain boundary and volume di . 2, Glide and climb of grain bose eut: 3. Glide and climb of 1, attice di i across and/or around pegocetions both the grains. the driving force for the ab provided by the variation, on a local scale, in the magnitude of the unrelaxed grain boundary normal tractions and shear stresses [134,232]. The presence of microstructural defects such as grain poundary ledges, triple points and intergranular particles, can contribute substantially to raising the stress level local to the sliding grain boundaries. When the kinetics of the accommodating processes fail to match the requirements imposed by the deformation rate then the stresses do not relax sufficiently quickly and cavities may nucleate. Ove processes is The factors which influence cavity nucleation are those which relate to the microstructure of the _ alloy such as grain size, the type, volume fraction and distribution of hard particles, the proportions and physical properties of the major phases, and those which are associated with the deformation conditions, for example, strain, strain rate, temperature and stress state, It will ibecons evident ever, that there is a interrelntronehip between the microstructural and deformation factors. An expression has begn below Proposed [195] in which the strain rate, or ticle which void nucleation at a grain bone ES °f diameter D, is likely to be inhibite teed diffusional stress relaxation can be pred? : is is given by : 11.502 Dgnd (5-51) Eo = pea xdD? kT 145where d is the grain size, x the spaetson of the total strain carried by the grain boundaries, p., the grain boundary diffusion coefficient, 6 the grain boundary width, and o _the ae at the imposed strain rate. Hence, if the critica strain rate was greater than the imposed strain rate, cavity nucleation would be unlikely. It is clear from the above relationship, which provides a rationalisation of the interaction between the many microstructural and mechanical aspects of superplastic flow, that deformation at higher temperatures and/or lower strain rates would minimise cavitation. However, the latter conditions are not commercially attractive since forming times would be long and energy costs high. Furthermore, structural instability at elevated temperatures can result in substantial grain growth and particle coarsening and a loss in superplasticity and/or enhanced cavitation. The minimum cavity radius which will be stable under an applied tensile stress o, in the presence of a superimposed pressure, P, is given by rv 2 2P/(o-P) (5.52) where T is the surface energy. If the maximum principal stress is similar in magnitude to the flow stress, then for the majority of superplastic materials the critical cavity radius would be approximately 100 nm, or less if nucleation occurred at a stress concentration. It is unlikely that voids would nucleate spontaneously by vacancy condensation, as the number of vacancies required to form a stable void would be too large. Approximately 2 x 108 vacancies w. to form a spherical void of rad grain size of the material is typical of that found in superplastic materials, i.e. 5 to 10 um, then a vacancy concentration of “10-5 would be necessary if all the vacancies in the grain were tO | condense simultaneously to nucleate a single void: | (By way of contrast, the vacancy concentration at | 0.87, is of the order of 10-7 to 10-5 even | when the effect of the applied stress is taken | 146 | jould be required ius 100 nm, If the5 CAVITATION AND FRacrurE count). However, it nto ac wever, it has been 5 grain boundary sliding does not occur ah onlin: Fee ee arto pereee and on a very localised le. O Slidin scarvated could be sufficient me; remsient stresses Bi hi small void froma highly Youagtst? nucleate) -A [134]. Once the gener ex: : verersaturation of vacancies stresses were relaxed by sliding the void would be unstable, but would require a finite time to decay: It is possible that the nucleated void could pe stabilised by further sliding [229]. However, in ractice it is likely that the majority of the cavities pre-exist, and are nucleated at second hase particles during the intense deformation associated with thermomechanical processing. the majority of studies on cavitation in superplastic alloys have shown that the number of voids observed by optical microscopy increases with increasing strain. However, it should be recognised that optical microscopy has only a limited resolution, and that during the initial stages of deformation the smaller voids will not be resolved, even though they are present. Thus the number of yoids will appear to increase with increasing strain. For example, a void will grow at a rate in determined by eqn (5.10). Hence, for a material wi a strain rate sensitivity equal to 0.5, @ strain equal to “350% elongation would be required before a void with an initial radius of 100nm Pee ton has optically resolvable (the limit of resolution ha been taken as 0.5 um). It 1s therefore ce , when making experimental measurements a distinguish between true nuc Ca eoay Teesent in the detection of voids which are a eattesolvable. material but which are initially 5.8.2 Hard Particles . wm a clear Experimental investigations have SPOW! & cocona correlation between the presence 0. oni croduplex Phase particles and cavitation, vee not readily materials which are particle free © clearly Cavitate. Such a correlation has, Fe Pb-Sn denonstrated for the non=cavi ty ciement leads to eutectic, The addition of a thi 147“seToTqzed Azepunog uter6 eyy ye sdn-eTtd uoTzeD0Tst woz: pue ‘seTzepunoq utezb 6UTPTTS uo seTopaa0d 03 queselpe suoyaeaqueoue> ssezqs wory {Arepunog utezb ey YITA SUTM TO spueq dtts FO suoTJOeszZejUT |Y3 Wor} HuTsTIe sebpetT ‘sjutod eTdtz3 utex6 ye sap wozz SutT3Tnsex suotqerquUesUOD ssons je azeeTonu Aeu seTqtTae) 9T°g¢ emnbTy *0 a dSNd (P) ASVONNOB NIML (2) 4889 ‘ , VAC AavaNnog-8ns *, A \_ ee ‘0 ‘0 So IN3YBHOONI 39037 (4) NOILONAP BIdIaL (2) ans Ae aaviad -\ isao sao, t Le ans AuvaNOD3S | Qe 1485 CAVITATION aND FRACTURE the formation of intermeta] nardnesses, €.g. SbSn, Agssn on subsequent deformation 2 opserved to have nucleated a nase/matrix interfaces and lic phases with i varyin and CugSns [183]. he cavities were i . boundar: idi level of cavitation at a given strain wos ay toe increase, and the correspondin i failure decrease, as the hardness, seiune feocti and size of the intermetallic particles was een increay in are presence of large particles of i(C,N) in a/t steels [63], or of pri and CuAl2z in Al-Cu-zr aie [40],,of MgiCenl ce and Fe and Si rich inclusions in Al-7475 [174], and of cobalt silicide particles, CoSi and CoSi2 in ' CDA-638, appear to be largely responsible for cavitation in these alloys (Fig. 5.16). It is interesting to note that a/f titanium alloys do not readily cavitate. The impurities which are responsible for the formation of hard particles in other materials, such as carbides, nitrides and oxides in steels, and sulphides in copper base alloys [105,220,233] are not as important in titanium, as the impurity elements readily pass into solution at elevated temperatures. It has been shown that the particle/matrix interfaces on sliding grain boundaries are the most likely sites for voids to form [208], since stress concentrations, poor bonding and a large radius of curvature of the interface, combine to reduce substantially both the critical void volume and the ener: barrier to nucleation (Fig. 5.16). The steady state cavity nucleation rate, Jr has been predicted by Raj and Ashby [208,234] to be: -4EuP ARE Dex6 oxo x ] (5.53) j- =F oa2kT oid 227 i inci ting imum principal stress ac T the mean surface energy, Fy is which relates the volume of the eristic dimension such as radius, 149 where o, is the m on the particle, the shape factor void to a charact5 CAVITATION anp FRACTURE SUOT}ORZF SumToa eseyd w GuTAren WITH uz—no g/p Joy suTeIS UAIM SPTOA Jo uoTAORIJ auMTOA ey} JO UCTIeTTeA pemseeW (9) *SeseezouT eseyd g Jo UOTIOeIZ auUMTOA ey} Se buTsverOUT vere UT uoTZONper ayy ‘pebueyo ere suoTzrodozd eseyd ayy se seTTeA OsTe soRzns emMyoerIj ay} JO eouereedde ayL (p) *e-pLL"¢ emn6TT NOLLWONOTS % a oe o0z oo SAILIAVD 40 AWNI0A %i ucleation sites on the and be the density oF hontial term is the most grain etn een (5.53). This infers that nucleation contd ba more likely at higher stresses, lower surface energies, and when the void seni was small with respect to its effective In the absence of particles it has pean araied that the stresses, (including transient a ex stress concentrators such as triple points)s and the necessary vacancy supersaturations are J 3 attainable during deformation at elevate . temperatures, and that conventional nucleation by vacancy condensation is impossible [235]. However, since cavities cannot be observed directly at the nucleation stage, the association of cavities with intermetallic or non-metallic particles can also be interpreted in terms of pre-existing defects such as regions of decohesion (embryonic cavities) which have developed during processing but which have a minimal volume at the commencement of deformation, 5.8.3 Phase Proportions and Characteristics Studies of a/8 brasses (17,221j have shown that the extent of cavitation during superplastic flow is strongly dependent on the relative proportions of each phase present in the microstructure, with cavitation decreasing as the volume fraction of ® phase increases (Fig. 5.17). It can be seen in figure 5.174 that the cavitation behaviour also influences the fracture mode. The alloy with 72% @ phase undergoes substantial Cavitation and the won leecence gad interlinkage of cavities leads to a pseudo-brittle fracture with a 1 on ional area, whereas the alloy wi Fa phe ee that superplastic flow is a hat Ccommodat, imaril: within the softer B phase [17,236]) the ne phase acts in the same wa: discussed in section s.g/3, shtetmetallic phases C n Shan satisfactorily Predicted cavity evolutnee aye 15>5 CAVITATION anp FRACTURE Gefen soft’ phase the evol 00! ase (or an equi: ution of a the yolume fraction of hard a shanen reduction in megner accommodate deformation eee which would Eupsequent cavitation rate, + thus reducing the e a/B titanium alloys suc! . meacturally analogous to Che aye pent are surkably resistant to cavitation rasses, yet are deformation temperature the alloys’ At the optimum doforty volume of B phase which acconmodaves arate poundary sliding and rotation, as the aiaieei er se jn the ff phase is substantially greater chan thane jn the a phase. At lower temperatures, the volume fraction of the B phase is reduced and saviiskion js observed. Even at the optimum deformation temperature, rafting of the a grains into bands fan result in cavitation since the local volume fraction of § is <40% [187]. However, in a/B titanium alloys the volume of cavites is usually ninimal, At higher temperatures, the reduced volume fraction of a phase is less effective in stabilising the 8 grain size and the material ceases to be superplastic. 5.8.4 Grain Size Experimental evidence is available that supports a correlation between increased cavity nucleation and larger grain sizes. In the aluminiye pase alloy Supral 220, where the ‘volumetric growth rate of the cavities is almost independent of strain rate within the superplastic reg i h an fracti i as found to increase wit! : ion of voids w ize (Fig- 5-18): Moreover, increase i ain si se tn the a erplastic flow has been grain wt i sul growth during Pox ‘the ‘appa invoked as the reason f' ral Melestion cf cavities in both al~7475 and opal 220 [238,239]. In the highly superP Sfmal when the loy, cavitation was found to be m™ yeased quite Wain'size was less than 5 Wi. fe th excess of 5 um ferkedly with initial grain sizes © 1). rent’ continuous 453CAVITATION (%x10) GRAIN SIZE (um) Figure 5.18 Measure fraction of voids and the grain size after deformation to 300% elongation at various temperatures within the superplastic regime (Supral 220) 20 GRAIN SIZE CAVITATION 0 atopy db 400 450 500 TEMPERATURE °C d variation of the volume 1545 CAVITATION AND FRACTURE ne importance of microstruc: z s, elation to cavitation is readiie sppoeeaey eft geudies on Fe-C alloys. Plain carbon eutect: id reels with a fine grain ferrite/cementi: gtoia gructure show extensive cavitation at the /FesC interfaces [33]. During superplastic geformation, there is considerable coarsenii f the cementite particles and this is accom anit @ by an increase in the ferrite grain size. B . ¥ increasing the carbon content of the steel to 1.6 att, and hence markedly increasing the volume fraction of cementite, a very fine grain size can be stabilised. The structure can be further stabilised by the addition of “1.5 wt% Cr which artitions to the cementite and increases its resistance to coarsening. The microstructure of the high carbon alloy is extremely resistant to cavitation, despite the increased volume fraction of cementite particles. 5.8.5 Strain Rate and Temperature It might reasonably be expected that increasing the deformation temperature and/or reducing the strain rate would reduce cavitation. The flow stresses would generally be lower at the slower Strain rates and higher temperatures, while there would be either a greater period of time to relax boundary sliding at the stresses generated by grain or the diffusion processes the lower strain rates, d at the higher temperatures. would become more rapii However, the effect on cavitation of changing the strain rate or temperature during superplastic deformation varies widely depending on the material. The differences which arise are often the microstructure of the result of changes in the struc material, particularly the grain size. In the aluminium alloy 7475, increaeong the tts temperature and decreasing the s rain rate resu ina decrease in the overall level of cavitation at a given strain. In Supral 220, decreasing the strain rate -2 © 10-3/s results ina a from 100 (8S 11 level of cavitation at a decrease in the overa . t ; Given strain, but further reductions 5% the strain rate toward 10-5/8 result ina steadily 155forza #4 ys 488% 106mm [op+0 50% Ez sax10te! )ny #3 30x 108mm" Ap+0.49% NUMBER OF CAVITIES mm Ea ssxi0%e1 SAL 2g 77108 mm sy Lapro sem > SS pay | I Ly i EEE SE Eee ee " 2 3 * 7 CAVITY DIAMETER jem Figure 5.19 Measured void size distributions in a Ni-silver as a function of strain rate at 743K fora strain of 0.92. increasing level of cavitation. that the increased levels of cay. lower strain rates result from i growth since the value of the strain rate sensitivity and hence f, the cavity growth rate factor, is virtually constant over the range of strain rates from 10-5 to It is unlikely itation at the nereased cavity 1565 CAVITATION ayp FRACTUR| EE effect is obse gal . rved on . segormation temperature, increasing the ydies of cavitation i : $05, 220, 2331 and in ai” %/ nickel silvers é own that both the volume gistributions of the cavities are i strain trig. and temperature, and gitdependent of strain foria 5.19). As was shown soe only on growt ing superplasticity is ter, void gid, as such, the observed strain controlled ; : inde, of cavitation on both strain rate and conserneure night infer a constant number of cavity ppereyuee these observations could be int e e: whergne view that oither the majority of voids” pre-exist, or inclusion particles dominate void nucleation in these systems. ” 5.9 Summary It is clear that there is a strong interrelationship between the occurrence of cavitation, the microstructure of the superplastic material and the conditions under which it is deformed. There is also a large body of evidence to support the opinion that many of the cavities which become apparent during superplastic flow pre-exist, and are a natural consequence of the thermomechanical processing which has to be applied to the material in order to generate the initial fine grain microstructure. rous potential cavity i imental since @ nucleii need not be cons dered detrimen clear understanding of the cavity growth process has enabled the deformation conditions, ¥ necessary to prevent cavity SEO ey S cowth identifizd. The extent to whie cavity grain occurs is controlled primarily Ys Feation of a imparted to the material. 7h? appliceree the hydrostatic pressure Of 0° Po uuired aeformas igh Uniaxial £low stress at the requir’ icted and strata rate and temperature © found to be suffici Cavitation from those © However, the presence of nume:rtantly, since the ¢), stress of super jols ie generally ve one low, the use of hydrostatic pressure in commercial forming operations becomes an extremely viable i method of improving post-forming mechanical properties. normally apparent. more impo. plastic mater 1586 sUPERPLASTIC FORMING AND DIFFUSION BONDING g.1 Introduction ghere is considerable interest in the application of superplasticity, particularly in the aerospace industry. Unlike conventional materials, superplastics are extremely resistant to neck formation when deformed in tension, and this enables large uniform plastic strains to be attained without failure. This behaviour, as has been shown in the previous chapters, is derived from the high strain rate sensitivity of the flow stress. The high strain rate sensitivity is developed at commercially viable strain rates (5x10 to 1072/s), while the stresses required to attain such strain rates are generally very low. The exploitation of superplasticity is, however, limited by several factors. Firstly, the reluctance of designers to adopt more novel forming procedures and new materials; secondly, the need for careful process control to confine forming to the rather narrow range of temperatures and strain rates over which a given material exhibits a high degree of superplasticity, and thirdly, the relatively small number of alloys presently available that can be both processed to give a fine stable grain size and that will also be capable of attaining the required strength, fatigue resistance and toughness, in the as-formed condition. superplastic forming has uminium alloys such as 159 The bulk of commercial een carried out on al0, 220 and 5000, A1l-7475 and Lita) sopra) 590), and on titanium alloys a as Ti-6Al-4V, (see Tables 2.1 and 2-34 at aluminium alloys are usually reer decendi temperatures between 430 and 54 ii epee A ‘J on the composition, while the ee mJ Perth mh require temperatures between 86 1922 ce e strain rates employed lie between 2x os n¢ 10-3/s, though in the case of the Suprat group of alloys strain rates approaching S$ May be seg The flow stresses of superplastic materials, which generally lie between 2 and 10 MNm~?, enable gas pressure to be used to drive deformation rather than liquid hydraulic or mechanically applied loads. Superplastic sheet forming processes are therefore fundamentally different from the processes that are used in conventional metal forming, and have more in common with the techniques used for shaping thermoplastics. 6.2 Forming processes ~ 6.2.1 Simple female forming The most simple form of superplastic forming involves the free bulging of a sheet, which is clamped around its periphery, into a female mould (Fig 6.1 (1)-(3)). During the initial stages o.: . ag £ : ; ; ‘ deformation, when the sheet is not in contact with to def i -section more uniform, Since faking the formed are usually the last regi fe corners of the die i gions to hb i Yintting 2ncnt t8 developed at these point ane a n simple 4 : gepect ratio of the die (the wate forming are the | pth) and the corner radii +0 of width to Simpl. i i ple female forming is generally preferred over 160 t6 SUPERPLASTIC FORMING AND DIFFUSION BONDING (oFZeUSYoS) Hutwr0z eTeusz oTdutgs 1°9 eanbta (y) (e) (z) (T) rir 161Figure 6.2 Female drape forming (schematic) other forming techniques when the aspect ratio of the component is low or when the convex surface of the finished component has to be a Specific shape, regardless of thickness, Stiffening features such as deep pockets or grooves can easily be in- corporated into the design of Parts (Fig. 6.1 (4))- Simple female forming is Probably the most common method of forming. 6.2.2 Female drape forming This process consists of bulge forming a sheet into a female mould in which one or more male details are placed (Fig. 6.2). As the sheet is bulged, the polar region will be the first to make 16>remale drape forming is prefe. forming when dimensional tolezances ote oncave | surface of the finished component are im ortant. > prape forming is also limited to parts with relatively low aspect ratios. A number of small dies can be placed within the same female forming tool allowing several parts of a component to be formed at the same time. The parts are then cut from the formed sheet and assembled into the finished component. Despite the higher wastage of material, this route can give cost savings compared with the forming of each part separately. A number of techniques have been developed to reduce the thickness variations inherent in components formed by either of the processes described above and to enable parts with substantially higher aspect ratios than 0.3-0.4 to be fabricated. 6.2.3 Reverse bulging the sheet to be formed is first blown away from the female mould into which it will eventually be formed. After producing the initial dome, the forming pressure is reversed forcing the bubble to fold in on itself fea a fonale moulds During the initial bulging, deformatto! ; concentrated primarily in the pole of the forming | bubble. However, on reversing the pressure, i additional deformation occurs within fee ae ene Instead, plastic flow is concentrated within zone adjacent to the clamping atee is eee a he in almuch more uniformythcknes ieee es te formed shape than could” be obtaine ¥ ae ce emale forming (Fig- §-3)- The height to 163 In reverse bulging,(2) (4) Figure 6.3 Reverse bulging (schematic) about 10% greater than_the depth, of the final formed component. If the height of the bubble is too great then folds or wrinkles are likely to develop in the finished part, 6.2.4 Plug-assisted forming mn be used to pre-stretch +4). As the tool6 SUPERPLASTIC FORMING AND DIFFUSION BONDING (oT3PWeYOS) Guywr0y peystsse-Hnta p'9 emnbra (T) (e)(a) (3) sure locks thin regions cf side wall against the tale tool. Tool movement preferentially stretches (a)6 SUPERP; LASTIC FORMING ayp DIFFUSION BONDING central regions of that s. Pri event any significant seret ing oe et forces of the sheet in contact with deformation is trans. 6.2.5 Snap-back forming snap-back forming is the male formi i plug-assisted forming, the plug defining the fivas form of the component (Fig. 6.5). The sheet to b formed is first blown into a bubble away from the male tool. Once the bubble has formed, the tool is moved up into the bubble. Deformation is theeeb transferred from the pole to the sidewalls of the bubble. As the male plug continues to move into the bubble, the forming pressure is reversed forcing the bubble to collapse onto the plug. A combination of friction and forming pressure lock the sheet in contact with the tool and effectively prevent any further deformation. As the Plug continues to move, deformation switches to the relatively undeformed material adjacent to the flange. Parts with aspect ratios of the order of 0.7, and greater, can be formed by this method. 6.3 Component Design Criteria 6.3.1 Selection of starting gauge Component stiffness, rigidity and strength are often the foremost mechanical design criteria. Superplastic forming offers considerable scope for the incorporation of stiffening features such as ridges, grooves and bosses into components, as only one tool rather than a closely matching pair of tools are required for bulge forming. Moreover, Superplastic forming can be combined with diffusion bonding to produce monolithic cellular structures with higher torsional rigidity and strength to weight ratios than could be formed by conventional fabrication methods. Considerable savings in Materials, tooling and production costs are 167ible over the traditionally accepted routes of ossil t fetal fabrication. ifi fatigue endurance ang Be statement of Ptecein is becoming athe a8ingly Teeceant fn materials which are to eho oeeenna teed bearing members of a Se ererore Sten ngth and fatigue requirements are ee iniaun thickness responsible for establishing nent. of material, hman, within a comp* eet a The starting gauge, he, required ge mes on the specified minimum thickness will eps che ones y component size, tool aspect ratio an YY which it will be formed. In genera min X surface area of component he = Fa (6.1) plan area of component where Fi is the thinning factor associated with a particular forming process (asually determined by experiment). 6.3.2 Determination of hy yarostatic pressure requirements Once the starting gauge has been * maximum true strain that will b, forming can be calculated. It is then necessary to find out whether or not the chosen material can be puperplastically formed to the required strain, and to determine the appropriate minimum corner radii such that this strain is not formed component is established, the © attained during For example: It is desired to £ i e I orm a part in plane strain from Lital-x (Al-8090) in which the maxim ue strain would be -1,3 Th Cavitate at t) @ mney loy is known to ormation rate Cavitation data Sile tests at that he imposed a (8x 1074/5 at 520°¢), Using obtained from uniaxial ‘ten, 1626 SUPERPLASTIC FORMING aND DIFFUSION BONDING train rate, the volume fractio i st¥an elongation of 267% (a tre equigenese astic strain of 1.3) is found 0 strain rate jump test showed that th ef. was equal to 7.5 Nmm-? and the strain cate” aoe sensitivity, m, was equal to 0.45 at this strain rate. ghe values Of om/os for uniaxial an strain deformation are obtained lon pare 5.2, by setting P, the superimposed pressure, equal to’ gero. (In this case ou/on = 0.5 for uniaxial tension and 0.72 for plane strain, values half way petween the limits for rigid and freely sliding grain boundaries as calculated from the models of Beere and Cocks and Ashby). The value of the cavity owth rate parameter is then calculated using equation (5.12). Hence for uniaxial deformation; 3p 140.45 2-0.45 fa = -[ ] eann[[ Jo.s0 ] = 3.27 (6.2) 2k 0.45 2+0.45 — od and for plane strain deformation 3p 140.45 2-0.45 = [——] sinn[2[—— }o.72 | = 5.04 (6.3) 0.45 240.45 Now 2 The volume fraction of voids which would be expected to form during plane strain deformation, is calculated using eqn (5.45), i.e. Crepes Cope = Cru exp(E(Npa-Ma)) (6.4a) Continuing the example and substituting € = 1.3 and Cea = 0.004, the maximum volume fraction of voids that would be expected to develop during forming would be: Cope = 0.004 exp(1.3(5-04-3.27)) = 0.040 (6.4b) 169f this level of cavitation is ape hi th (>1%) then the required bac! poring teat caiculated from eqn (5.26), ete a étoais cee be the highest measured stress up coring auras 3 for uniaxial deformation at Sie ‘Gee titer rature. L Taies090) eebibits virtually Boon che paxdening during superplastic flow at 520 ig cavitation pressure required to eliminate the damage would be (Table 5.1): = pt] w ° (6.5) Oz 4v3 On which can be simplified to P 2 5/4V3 os (6.6a) i.e P2 0.72 x 7.5 Nmm-2 2 5.40 Nmm-2 (6.6b) If the pressure required is hig} safely applied during forming, as is likely to be the case here, then a certain amount of cavitation in the finished component must be “accepted. The volume fraction of voids which would be expected to form at the maximum operating pressure can easily be calculated using equation (6.5). If for example the maximum hydrostatic pressure is 3.5 MNm- than that which = (500psi) then SMS RDS hw a: 3.5 3 3.5 - =ff--22], . 2 Ub ys =] [s, =] = 0.2556 SUPERPLAST. ‘TC FORMING anp DrrFystoy BONDING pstituting the result obt; ained abo ve for gu : Sl OF into equation (5.12) anq soins vin Ape = 1.59 one fracti (6:8) gne volume ion of voids sroocted, 26,8, true etrain of 13 after ton a si i 2 i ee? would be uperimposed pressure of 3.5" Cure = 0-004 exp(1.3(1.59-3.27)) = 0.0005 (6.9) which is almost negligi’ jevel of cavitation which sioake be ee ed she the part can be set (for example Cy saat ane an the minimum hydrostatic pressure required ae Ea this level is evaluated for the point of a furain. From equation (6.4a) the required value of fee is first calculated: 1 0.005 Npa = Mut 1s — | = 3.2740.17 = 3.44 (6-10) 1.3 0.004 the next step is to calculate ox/O=2 from eqn (5.12). a 140.45 2-0.45 One sae) [2] oom PLES LET 2b 0.45 240.45 ox Thus om/On = 0-523 (6x41) Next, eqn (6-5) is solved for P. Thus Oe 2 [ 5 =] = 0.523 (6.124) oz 2 & 2v3 on i.e. . (6-12b) (213 psi) P = 1.49N mm? 1716.3.3 Establishing the minimum corner radii Superplastic forming, while capable of reprog, fine Betail, cannot produce very sharp corner, and careful consideration needs be given to determining the minimum radii of curvature that can be sustained without excessive thinning or wrinkling of the sheet. Several empirical relationships have been established to enable Buch radii of curvature to be determined and are given elsewhere (241-243). Ucin, 3 6.4 Determination of the Pressure - Time Cycle for Superplastic Forming A number of theoretical analyses have been developed to predict the pressure-time cycle required to form simple shapes and/or describe the thickness variations in the walls of those shapes (244-250). Unfortunately, production parts frequently have complex shapes which do not readily lend themselves to even an empirical analysis of the pressure- time cycle that would be required for forming at strain rates close to the optimum for superplasticity let alone Predict the resulting variations in wall thickness. However, the parts to mated to one of two cular section ‘top lar section ‘trough’. geometries represent or plane strain idealised shapes, either a cir hat’ ora cylindrical/rectangu The two simplified deformation either balanced biaxial tension forming of thin walled shells. deformation geometries 6.4.1 Superplastic forming of a hemispherical diaphragm To calculate the pressure-time form a hemispherical shell with Xr, at an approximately constant will be assumed that the diaphra, 172 cycle required to @ final radius of strain rate, €, it gm maintains a6 SUPERPLASTIC FORMING AND DIFFUSION BONDING | | i Figure 6.6 Geometry used for hemispherical bulge forming. orm cross-sectional thickness. The stress state in a shell of uniform thickness, h, has already mn in eqn (5.32) from which it follows that unifi with. been give: &6P on == [£+2] (6.13) 2 h ror the majority of forming operations @>>h; hence, oe = 6PE/2h (6.14) From the geometry shown in figure 6.6 it is found that ep? = x2 + (gH)? (6.15) @ = (x?+H?)/2H (6.16) The surface area of the spherical segment, A, Which defines the surface of the dome is given by A = 2ng2(1-cos®) (6.17) 173i ibtended by the is the solid angle sul h cement ‘at the centre of the sphere. Assuming that deformation occurs at constant volum wr?he = 2me?(1-cosé)h (6.18) Recalling that cos@ = (g-H)/g then H = x V(ho/h - 1) (6.19) and . x(ho/h) (6.20) 2 V(ho/h = 1) If the dome is further assumed to maintain a uniform thickness then that thickness, h, at any time, t, is given by h = he exp(-€t) (6.21) From eqns (5.1) and (6.14) we obtain 8P, € am =f. [£] aan 2h k’ Substituting egn (6.20) for @ and eqn (6.21) for h, the differential Pressure, 6P, required to form a hemispherical dome at a constant strain rate, €, -is obtained, € am . . 4 [-] ho/r exp(-2€t) V(exp(€t)-1) o yg u (6.23) material constant k" cou e appropriate Strain de; mmeeieee ae Riessscet Pressure-time cyel 77475 at 515°C and a Strain 74 are shown in figure Ciel Fate of 5 x an 1746 SUPERPLASTIC FORMING AND DIFFUSION BONDING CONSTANT FLOW STRESS CONSTANT 'm’ 3 25) PRESSURE (psi) GRAIN GROWTH { CONTACT WITH HARDENING TOOLING vemt-spiene ———-! ro Hat ht 0 i 10 15 20 25 30 35 TIME. (mins) Figure 6.7 Calculated pressure-time cycle for amm thick Al-7475 at 515°C at a nominal strain rate of 5x 10-*/s, with and without the inclusion of strain hardening due to grain growth. 6.5 Diffusion Bonding 6.5.1 The DB-SPF process a solid state joining process faces to be joined are prought into contact at an elevated temperature. Application of a moderate pressure to each component brings the surfaces to be bonded into intimate contact creating a planar array, of interfacial voids. Diffusion and creep flow 4 Processes can then transport atoms to the voi tn Surfaces from the adjacent areas 5° reducing t! e Volume of the interfacial voids. Given sufficiem 175Figure 6.8 Example of cellular structures which can be formed by diffusion bonding and superplastic forming (Ti-6Al-4v). time, the voids will disappear completely and hence an atom to atom bond is formed across the original interface. Since the bond zone is not melted during the bonding Processes, the microstructure of the bond is identical to that of areas remote from In the context of superplasticity, diffusion bonding is used for selective bonding of sheet material into sandwichlike constructions. Expanding the 1766 SUPERPLASTIC FORMING AND DIFFUSION BONDING ond areas. Those shell stru ; Mttenally atiffenod minic the seniehee ecg present in nature [251], and since the: oe virtually all empty space they have ‘low are Jensities (Fig. 6.8). The stiffened shell we cellular structures have extremely high on i 1 rigidity or resistance to flexure. Finall. e pine the structures are formed from high eteantne alloys, the components formed by diffusion bondi and superplastic forming have extremely high ing figidity and strength-to-weight ratios, piffusion bonding and superplastic i pifnnolegy (DB-SPF) is readily applicable to titanium-base superplastics, since the titanium jattice has a high capacity to take into solution the surface oxide and other contaminants which would normally prevent the formation of a metal to metal bond, when the two surfaces come into intimate icontact. There is considerable interest in extending DB-SPF to other superplastic alloys. However, in the case of aluminium alloys the problems associated with the tenacious surface oxide, which effectively prevents the formation of the all impor t metal to metal bond, has so far limited the extension of DB-SPF. Diffusion bonded and superplastically formed structures are produced from two or more sheets of superplastic material which are selectively bonded [252-254]. The areas in which bonding is to be een printed with 'stop-off’ prevented are silk scr boron nitride in a (a mixture of yttria and polymeric binder). This enables the two sheets to separate at these points during the superplastic forming part of the DB-SPF process. Since the Pattern is printed onto the sheets it is relatively simple to alter it, and hence effect substantial changes in the geometry of the internal stiffening elements (Fig. 6-8)- The printed sheets are assembled in register and either clenped or welded around their periphery. Tf the sheets are welded then prior to finally sealing the sandwich of sheets, the internal regions are evacuated. the sandwich is then heated to the 177°C, for 3 i erature (around 920°C, diffusion porte auch 86 mi-Gal-4¥) 1 rd ihe oxternal tireaces subjected to a ive prestuelerated bonding)” TETSEESEIDy dgpiteghydeantico23y vai0d, compressed gas. After a predetermined periods fe ‘chen pressure is released. The aeeraen stage may be internally pressurised. This tate dion bonding o¢ carried out immediately after diffu' 1, pressure eau as a separate operation. The interna aa me es the outer skins of the sandwich to eee aining opposite directions and to fill eee cagtey lice cavity. By varying the shape of tha peg ee making it a mould) any external shape a wartation xoduced. (The derivation of the require: thee of the internal pressure with time in order 5 a the sheets deform superplastically, was describe : n section 6.4). As the internal sheets are physically bonded to the external ones they are forced to deform as the structure expands and the internal stiffening structure is thus developed. 6.5.2 Bonding mechanisms Since no surface is atomically smocth the placing of two surfaces together will result ina finite number of contact points and the creation of a planar array of highly irregular voids. If the points of contact are atomically clean then a metal bond will form. The bulk of the time required for the diffusion bonding process is spent in bringing the remaining surfaces, which are not within an atomic diameter of each other, into close proximity. In the same manner as the application of pressure speeds up conventional sintering, the application of pressure enhances the rate at which the unbonded surfaces approach, In this context the bonding Process is that which brings the two surfaces into atomic contact, rather than the Process by which the interatomic bond itself forms. 7 ha. 7 attain 100% contact between the syrewuired to n © two surfaces, the actual bond being assumed t. i nee on duck eeeee © form instantaneously 1786 SUPERPLASTIC FORMING AND DIFFUSION BoNDIM IG (A J 4 ( i, Ib , oN, 4 Vv \ Figure 6.9 Schematic illustration of the mass transfer paths which can operate in order to fill the interfacial voids during diffusion bonding. (a) tame dependent plastic collapse of the interface, (b) diffusion from interfacial sources to the void surface and (c) surface diffusion. The mechanisms by which the interfacial voids are removed are as follows (Fig. 6.9): .Instantaneous plastic collapse. When the bonding pressure, P, is applied the points of contact between the two surfaces will be unable to support the applied stress. The contact area will therefore expand almost instantaneously until the stress acting perpendicular to the contact area is less than the yield stress, Oy: This is taken as the starting point for ponding. At this instant, the area fraction bonded, fa, is given by fa = P/oy (5-24) 2time dependent plastic collapse. The stress ‘ime dependent plastic collal plastic flow by within the contact zone will cause 179ither conventional creep or superplasticity, Sinueie flow will force material into the Voids ang thereby expand the contact area (Fig. 6.9a), The stress in the bonded region will decrease as tho area fraction bonded increases, and hence the Yate of void closure by time dependent plastic Collapse will lessen with increasing time. 3. Diffusion. During application of the bonding pressure, the contact interface between the two sheets experiences a normal stress. Since the free surface of the void cannot support a normal Stress there exists a difference in chemical potential] between the interface and the void surface. The gradient in chemical potential drives diffusion such that atoms move from the contact interface to the void surface and the void volume decreases accordingly. The mass transfer path can be both via the interface itself (boundary diffusion) or through the lattice (volume diffusion) (Fig.6.9b)or via the surface or vapour phase (Fig. 6.9c). 6.5.3 Kinetics of diffusion bonding The first attempt to predict the time required to form a sound bond be tween two rough surfaces was made by Hamilton [255]. It was assumed that the surfaces to be bonded consi 6.10). Hamilton calculat the average net section stress through eqn (4.1). The bonding time, t, was calculated by dividing the strain required to bring the asperities into an overlapping position (Ex = -0.693) by tne average strain rate. Since the stress is proportional to th 1e then applied pressure, P, 0.8kT G an tr. [— (6.25) Azz Dgs»fG v3P6 SUPERPLA: ‘STIC FORMING AND DIFFUSION BONDING 6.10 Geometry adopted by Hamilton [255] to Figure diffusion bonding in plane strain. model stage! + AS, | celts used in model | poem stagell F944 > erby and Wallach Figure 6.11 Geometry adopted by De a jon bonding in plane strain. [257] to model diffus 181consist of short wavelength asperitne bonding pose on a longer wavelength roughness. 5. Firstly: a process was divided into two stage! c1ength , reduction in height of the long Sel of Hamilton roughness, as in the original mode. helant of th was assumed to take place until a ene shore ie remaining voids was equal to that o. coveed 5 wavelength asperity amplitude. This occu the by time dependent plastic collapse only. 1 a ate’ ‘cond stage of the bonding process, interfacia ‘usion was believed to be the dominant void closure mechanism and the time to sinter the remaining planar array of microvoids was calculated. The total bonding time was taken to be the sum of the two sequential processes. A further refinement of the plane strain model of diffusion bonding was introduced by Derby and Wallach [257,258], who considered diffusion bonding to be analogous to pressure sintering [259]. Again the bonding process was divided into two stages, I, the collapse of a ridge and, II, the sintering of cylindrically symmetric voids (Fig. 6.11). Unlike the previous model, all the bonding mechanisms were considered to operate simultaneously. During the initial stages of bonding, the curvature of the neck of the void was much greater than that of the surfaces remote from the neck. Again a chemical potential gradient resulted in surface diffusion of mass to the neck from regions with lower surface curvature, Hence the void radius in the plane of the bond interface would decrease due to surface diffusion. However, since surface diffusion only redistributes matter within the void, no change in void volume would occur, If surface diffusi of very rapid then the voids wor son is dr Dan6 Set WV 3nro? (1-£a) Ree S 2P y ——— ff . kr b 2f. (1-(1-£,)27m) 30P Dené fa2rekT [ Q * Dv¥e(1~fa) ] (6.26) 1826 SUPERPLASTIC FORMING AnD DIFFUSION BONDING i which f. is the area f. interface already in cont. lied pressure and ro t youghness. traction of the act (bonded), P the © original surface yor superplastic materials, substantiatly smaller than t] which results from cleanin ro: prushing or grit blasting. th the model of Dezb: and Wallach it was assumed that only one ony interface, the bond interface, intersected the surface of the voids and hence only one plane of diffusive mass transfer was available. However when the grain size is smaller than the void size numerous grain boundary interfaces will intersect’ the void surface and a number of additional mass transfer paths for diffusion become available [211]. In effect this is the reverse of superplastic diffusional void growth. For cylindrically symmetrical voids the rate of change of the void radius with time is given by a modification of eqn (6.26) in which the second term, representing the diffusive flux along the bond interface, is replaced by a summation of diffusive fluxes arising from all the interfaces intersecting the void. Naturally, as the void becomes smaller then the number of grain boundary interfaces which intersect the grain size can be he surface roughness the void surface will be reduced. The number of Sica additional interfaces, N, which must be an integer, is given by N = 2 int(xro(1-f.)/d) where d is the grain size. In making the summation, the prea fraction bonded, fa, must be xeplaced by the apparent area fraction bonded — the plane parallel to the bond interface, w ere | the grain boundary interface intersects tl ie void surface (Fig. 6.12). The time required to a full interfacial contact, t, is then _ 1 df. (6.28) t= “Fe Voyoy & dr/dt 183Diffusion zone gb. bond ——gh. -_—___——_ 2% — Figure 6.12 Geometry adopted by Pilling et al. to model diffusion bonding in plane strain [211], AN INDEPENDENT elnTeRING E UMIT Ft) | 1 Il PRESSURE (arbitrary units) TIME (arbitrary units) Figure 6.13 The calculated variation of bonding time with pressure at a fixed temperature. At low pressure, the solution is independent of pressure, but strongly dependent on the surface finish (ro). At high pressure, depends exponentially on6 SUPERPLASTIC FORMING AND DIFFUSION BONDING ere dr/dt is expressed as f i weerial constants (as per equation (6-26) * gnitial surface roughness, ro. ' . some DB manufacturins roc i ponded are welded arcund their peripheries and placed in a sealed box which can be pressurised. she bonding process is no longer one of plane . strain, but is isostatic. While the same ph: sical mechanisms of void closure are operative the metics of each process will differ from that under plane strain conditions. On average, the rate wryoid closure during isostatic bonding isa factor of 2 to 3 times greater than that for plane strain bonding [260] tn deriving the rate equations for each mechanism and model it is important to define the geometry of the interfacial voids and the applied stress state in such a way that there is no discontinuity in the rate equations when the voids become isolated from each other. Moreover, it is important to recognise that the surfaces being joined have a complex topography. Often a short wavelength, low amplitude roughness is superimposed on a much longer wavelength, low aspect ratio, waviness and it is normally the latter aspect of the surface geometry that determines the overall bonding time,(Fig. 6.13). Given the uncertainties in the material property data (a factor of 2 to 5 for most values), the fata (2c ponding that are predicted py the models ea as absolute, but rather as @ should not be regard guide to the ‘process window’ in which . experimentation should be carried out. By isolating the time frame in which ponding would be expected to occur under a given set of bonding conditions cial) (pressure, temperature, surface finish ani mater 1) actual experimentation and destructive evalua the bond properties can be minimised+6.5.4. Testing of piffusion Bonds i ly be reli The quality of diffusion bonds ce Chavacteriatio; assessed by comparing the fractu. nt metal afters of the bond with those of the pare ne. alerasons. simulated bonding heat treatment cyc Font l. evaluation of the bond quality is eee oide (aes incapable of resolving bond line micr 200 uni a) since the wavelength of the sound bn (* nin much greater than the defect size an H L nent” No sheets, the transit times are extremely s. have re recently, improvements in signal processing enabled some headway to be made in detecting _ disbonds in which the two surfaces are often ine intimate contact [264]. Once the conditions required to produce an acceptable bond quality have been determined by destructive testing, strict process control is normally employed to ensure i. reproducibility of the bond properties. This is often supplemented by proof testing of lugs which have been designed on to the component and which are removed after the bonding cycle is complete [265]. A number of mechanical tests have been devised to assess the quality of diffusion bonds. In the case of bonds produced in thick sections, conventional tensile, rotating bend fatigue and impact test pieces can be readily produced and the fracture strengths compared with that of the parent material. In practice, parent metal tensile strengths are often achieved in bonds with more than 85% interfacial contact, while parent metal fatigue endurance normally requires a complete absence of Breet ee ctovoida, However, the most iscriminating measure of bond i i i by impact testing, whion poor bose ky ots gobtsined stranttte oneal contact may exhibit low impact Unfortunately, impact testing ; x g is not wi applicable because the majority of ditteet en bonds heet materials the fracture strength Of the joint is met ee eae ee ing ned lap shear test a i standard exists. Furthermore, tnncagh ne eee piere parent metal fracture strength i¢ ot guarantee that 1866 SUPE! SUPERPLASTIC FORMING AND DIFFUSION BONDING pigh quality bond has been ond will be capable of resisting pect a that the guperplasree orming. It is this latter ring | fnich might prove to be the most discri capability wet for evaluating diffusion bonds metwasn the in gneets + 6-5-5 piffusion Bonds in Titanium Alloys gitanium alloys such as Ti-6Al-4V, qi-6Al-28n-42Zr-2Mo and Ti-4Al-4No-26n-0.5Si (IMT 550) are readily bonded at temperaturos between 880 274'940°C with applied pressures of 0.6 to 2MPa for rocess times up to 3 hours. The temperatures and stresses used for bonding are normally close to the conditions under which optimum superplasticity is opserved, Table 2.3. In general, as the bonding temperatures and times are increased, the integrity t¢ the bonds increases. However, excessive temperatures and long process times can lead to grain growth during the ponding cycle and a eduction in the subsequent superplastic formability of the alloys. The relative ease with which solid state diffusion bonds can be formed in titanium-base alloys is reflected in the growing body of fiterature and its acceptance 3° a mainstream manufacturing technology [ 252,253, 266-2691- The ease with which titanium can be bonded may be attributed to the ability of titanium to take into solution both its oxide and other surface ted to pressure at elevated contaminants 270). It has been found that parent temperatures n at pa metal atures | gerengths can be attainoe i oitaniwt ai i figure 6-14, but that © iffusion bonds [268], £4907" those of the paren’ strengths are often . metal [271] 1 (Fig. 6-15)- The latter can be improved by heating to the 8 transus after ponding 35 complete, allowing grain poundary migrations ce 6 hence a disruption © the planar pond inte: , occur. 187PEAK STRESS (MPa) LIFE (cycles) Figure 6.14 Fatigue properties of Ti-6Al-4V after bonding for 2 hours at 950°C under an applied pressure of 0.69 MPa (100psi) [268]. 6.5.6 Diffusion Bonding of Aluminium Alloys Unlike titanium alloys, the development of a DB/SPF technology for aluminium alloys has proved difficult because of the existence of a tenacious surface oxid However, studies have shown that ‘solid state’ diffusion bonds with fracture strengths in excess of those attained using polymeric adhesives can be realised. Two types of bonding procedure have emerged. Firstly, that involving surface modification, such as the deposition of a silver coating on a nominally oxide-free aluminium surface and secondly, the use of large scale deformation in the bond zone to fracture th b act e tenacious layer of alumina on the aluminium surface, 1886 SUPERPLASTIC FORMING AND DIFFUSION BON. IDING 20 @ THERMAL CYCLE = OB IMPACT ENERGY (J) 0O7 12 20 60 BONDING PRESSURE (MPa) Figure 6.15 Impact strength of IMI-550 (see Table 2.3) after ponding for 2 hours at 950°C under various applied pressures [271]. Silver coated surfaces have been produced on clad 31-7010 [272,273], Al-2014 and ‘A1-6082 [274], Al-5056, Al-5657 and Al-6351 [275]. The coating prevents the formation of alumina and any surface oxidation will lead to the formation of a silver oxide (Ag20) which is soluble under pressure at ve elevated temperatures- The aluminium sur: aces to coated are polished to <1 um finish then sputte cleaned in vacuum using argon ions. The silver ox by coating is applied either py vapour depo Silver ion plating- ; i at Bonding is generally carried out in vactomperatures Pressures ranging from 1.5 to 5 MPa 189of 450 to 550°C for times up to 1 aoe ae 00 ee Clag Al-7010, bonding has been carried pe 20 te 300°C using pressures of 130 to 140 Bas Tn | cases, the bond shear fracture streng 7 ‘eased with an increase in the bonding temperature and the extent of deformation in the bond zone, Despite the large number of alloys and conditions investigated, bond shear fracture strengths in the range 50 to gq Nmm-? were generally observed. Post-bonding heat treatment, usually to the T6 or equivalent condition, resulted in increased bond fracture strengths of 140 to 180 Nmm-?. Solid state bonding without the use of a protective coating has been restricted to two alloys, Supral 220 (a fine grain derivative of Al-2004) [276] and Al-7475 [277-279]. Bonding temperatures and times were limited by the amount of grain growth which could be tolerated during the bonding cycle before the subsequent SPF behaviour of the material was impaired. Bonds which were fabricated under static compressive loading gave shear fracture strengths in the range 40 to 150 Nmm-?. Again, the bond strength could be improved by the application of a post-bonding heat treatment with shear fracture strengths as high as 250 Nmm-? being recorded. The conditions under which solid state bonds have been produced in aluminium alloys are summarised in Table 6.1 together with the reported fracture strengths. For example, Byun and to 4 bonds per set of roce: i deviations Bers set ot Process variables recorded [279], who examined between 16 m and 20 set of process variables, observed deviations of mi008 to ~60% of the mean fracture Strength. In the latter study, the variability o 46 SUPERPLASTIC FORMING AND DIFFUSION BONDING azz (2) ost Bert ors Trox TSzI-T¥ punoz6 Sere 942 (3) OTT-OF s s STS dep anodea uz pe3setq 3136 SLbL giz (a) SOI~S¢ oF S-T 066-095 euou peaseTq 3725 4007 ece’Zce (2) OST-0S T 9L°% ors auou punozb SLbL 6£z (2) OST-OF eine and oss-sTs ener pe3seTq 3146 SLbL siz (0) — SbT-0s I> O°S-S*z OSS-0S' dep ancdea by — paystjod 9505 bez (a) bL-ES p> ST 0ss-00s peaetd-uor by Peysttod Tw ecz’eiz (a) 08-2. 7 OPI~OET O00€-02%@ pazetd-uot by Peysttod OTOL peTo yeu kaw ’y3Suez3s = eUT, ea eINssera 9, duos, 6uT3209 ystuta eorzins Teyz03UN S(erysuey ‘0 tavoys /2) -peuroz spuoq oy go syybuer3s exnaoez3 oy3 yatA z0y30603 sfoTTe unyuyunte ueonjeq peonpord ueaq eaey Spucd uOTSHIZTP YorYA TopuN suoTaTpU od ey3 Jo Areunins T*9 eTqUs 191DB Ti- 6AL-4V. 927°C, 3hrs, MPa OAL-2475 515°C, Shrs, SMPa PROBABILITY OF FAILURE oF 1 NORMALISED FRACTURE STRESS Figure 6.16 Probability of fracture, as determined by Weibull statistics versus normalised fracture strength (ts/tors) for Al-7475 and Ti-6Al~4v. (Fig. 6.16). (By way of comparison, repeated measurements of the yield strength of copper would give a Weibull modulus of =40, while that of the modulus of rupture of a structural ceramic would be closer to 10), Attempts have been made to bond alumini transient liquid Phase, although this is not’ . errictly. gi stusion bonding. By interposing a melting point depressant between the two i it is possible to form a liqui ayo team stirtaces liquid i zone during the bonding cycle, THeysr im the bond6 SUPERPLASTIC FORMING AND DIFFUSION BONDING yp in the surface layer of the alum: pplication of a moderate pressure otter the di id pnase has formed displaces the bulk of the li fa from the bond zone and disrupts the oxide Durin e remainder of the bonding cycle the concentration of the melting point depressant decreases as it ° qiffuses further away from the bond zone which ysothermally solidifies. only elements which are compatible wi alloy are suitable for forming a frenetont digaid nase. In the case of aluminium the choice is jimited to silver, magnesium, silicon and zinc (copper although compatible, is likely to cause intergranular embrittlement of the aluminium alloys). The melting point depressants are usually interposed between the aluminium surfaces as elemental coatings [275,277] or as aluminium alloy foils [277]. 6.5.7 Diffusion Bonding of Nickel Alloys After titanium and aluminium, nickel-base alloys are, perhaps, the next most important group of superplastic materials, see Table 2.5. However, their use has been concentrated in ‘massive’ rather than thin sheet components, where they have been superplastically forged into complex shapes [280,281]. Diffusion bonding of nickel (super)alloys has recently received considerable attention means of joining superplastic particularly as a 1 alloys of dissimilar compositions, mechanically alloyed materials, and single crystal or direstionally ‘solidified materials. Nickel, like titanium, is capable of taking into Solution its surface oxide when ieee pressure at elevated temperatures, although other surfac! contaminants azé not as readily soluble. Nickel-base Superalloys are normally ponded at temperatures just below the t’ solvus. Depending of the alloy, bonding can be carried out at temperatures between 100te tact 1as0°c with applied pressures 2) the range 1 to 20 MPa. These conditions correspond a to those where superplasticity can pe exhibited. 4193mechanical strength of bonds formed between dissimilar materials can be degraded by a number 9, factors which would not normally be apparent when bonds are formed between like alloys. These are [282] :- 1. Penetration of the single phase t microstructure into the two phase t-t’ microstructure as a result of the diffusion of t’ forming elements into the single phase alloy. 2. Formation of TiC particles/films at the interface as titanium reacts with adsorbed CO on the mating surfaces. 3. Development of Kirkendall porosity on the aluminium-rich side of the joint. The latter factor would not normally be apparent after fabrication as the isostatic pressure applied to form the bond is often more than sufficient to prevent the formation of Kirkendall porosity due to the flux imbalance between Ni and Al [283]. However, during subsequent service at elevated temperatures porosity will develop rapidly in the absence of a confining pressure [283]. The formation of carbides within the bond zone can be avoided by treating the mating surfaces with NHaBFa which decomposes on heating, reacting with TiC to form volatile titanium fluoride and ~ fluorocarbon [284]. It has been demonstrated that nickel-base alloys “such as IN-718 can be diffusion bonded and superplastically formed [285] in a similar manner to superplastic titanium alloys. 6.6. Summary Diffusion bonding and superplastic forming is now an established manufacturing technology. Its application has enabled low weight-high stiffness cellular structures to be fabricated simply and cheaply using titanium-base alloys. Although it is possible to join superplastic aluminium alloys using both solid state and transient liquid phase techniques, the poor reproducibility of each of those processes has so far limited the introduction 1946 SUPER! ‘PLASTIC FORMING AND DIFFUSION BONDING -SPF of alumini of DB luminium allo opa-supexplastic nickel-base alloys superplastic and in be nocessfully diffusion bi sueimilar materials are ded. However, where gempexatures , careful consideratio at elevated given to composition to ensure than ae to be porosity loes not become evident amet e. acknowledgements ghe authors are grateful to Dr. D.J.Lloyd and professor ~.G.Langdon for s i micrographs used in the tee. ee references M.M.I.Ahmed and T.G.Langdon, Metall. Mm ahre 9 etall.Trans. 8A (1977) 2 K.Higashi, T.Ohnishi and y.Nakatami, Scripta Mé 1 19 (1985) 821-824. pee etait 3 F Hargreaves, J.Inst.Metals 39 (1928) 301-327. 4 ¥ Hargreaves and R.J.Hills J.Inst.Metals 41 (1929) 257-283. wn Jenkins, J.Inst.Metals 40 (1928) 41-54. 5 6 C.E.Pearson, J. Inst .Metals 54 (1934) 1211-124 - 1 ‘A.A.Bochvar and %.A.Sviderskaya, Izvest.Akad.Nauk. 9 (1945) 822- 8 E.E.Underwood, J-Metals 14 (1962) 914-919. § lalBackofen, 1.R.Turner and D-H-AVSFYs ‘Trans. ASM 57 (1964) 980-990. vies, Superplasticity 10 K.A. Padmanabhan and G.J.Da pubi. Springer-Verlag, Berlin (1980) i J.Hedworth and yw g.Stowell, J-Materl- 1061-1069. 12 T.G.Langdon, 13A (1982) 689-701. 13° J.G.Wang and 5 Bad, 67 (1984) 385-390 and 399-409. 14 -F.Wakai and H.Katoy adv.Ceram.Materls- 3 (1988) 71-76- 15 .E.Chung and T.J3.Davies, Acta Metall. 27 (1979) 627-635- 1969) 1057-1069. is oenee, Acta Metall 17 (1962) oyag. 41a (1980) au-éa- ci. 6 (1971) Metall.Trans- gamer .ceram.SOc+ 7 M.S) nd B.Baudelet, i 18 Me aghaoh fand C.H.Hamilton, Metall.trans- 104 (1979) \ 699-706. | 19° B.M.Watts and m.g.Stowell, gumater.Sei- 6 (1972) oes genk acta Metall. 34 (1986) 20. 0.N.Senkov and woM.Myshlyaevs 97-106. i 363-369- a ee and B.Baudelet, g.Materl Sci 38) 6. 22 wia.clark and T.H.Alden, : 1195-1206.23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 ripta Metall. 16 C.u.Cacares and D.S.Wilkinson, serip tae ih (3962) ore dv Weiss, Motall.Trans. 3A ( 5-681, -R.Yoder and V.Weiss, snaterl-Sci. Letters C.H.Gaceren ang D-S.Wilkinson, a sss 1984) 395-399. .Trans. 10A ( ) BeWalser and 0.D-Sherby, Metall. Tr 1461-1471. and 0.D.Sherby, T.Oyama, J.Wadsworth, M.Rorchynsky G.Kostoroz, ACSMA 5, Ed. P.Hassen, V.Gerold and ¢Yos ter publ. Pergamon Press, London (3979) Working Technol, J.Wadsworth and 0.D.Sherby, J-M 2 (1978) 53. . 1966 Deke and WIA.Backofen, Trans. THS-AIME 239 ( ) 1034-1040. N.E.Paton and C.H.Hamilton, Metall.Trans. 10A 1979) 241-257. hoyana, 0.D.Sherby, J.Wadsworth and B.Walser, Scripta Metall. 18 (1984) 799-804. 4.43 (1978 J.Wadsworth and 0.D.Sherby, J.Materl.Sci. ) 2645-2649, 0.D.Sherby, B.Walser, C.M.Young and E.M.Cady, Scripta Metall. 9 (1975) 569-574. G.Piatti, G.Pellergrini and R.Trippodo, J.Materl.Sci. 11 (1976) 186-189, D.M.Moore and L.R.Morris, Materl.Sci. and Eng. 43 (1980) 85-92. F.w.Ling and D.Laughlin, Metall.Trans. 10A (1979) 921-928. Bon aiden and H.W.Schadler, Trans. T™S-AIME 242 (1968) 825-832. A-Ball and M.M.Hutchison, Metal Sci.J. 3 (1969) 1-7. B.M.Watts M.J.Stowell, B.L.Bakie D.R.E.Owen, Metal Sci. 10 (1976) 189-206, D.J.Lloyd and D.M.Moore, Supe: Structural Allove, Ed. N.E.Paton and © publ. TMS-AIME, Warrendale (1982) 147-172. T Sakai and J.J.Jonas, Acta Metall. TSaRaL etall. 32 (1984) milton, B:Geary, J.Pilling and N.Ridley, Supersiast n 1 Superplasticity Aerospace ~ Aluminium Ed. R.Pearcs and L-Keliy ae, poet Of Industrial science, Cranfield (2985) ‘127-135 D.L.Holt and W.A. 7esc7¢a, nt WA-Backofen, Trans. asw 59 (1966) K.A.Padmanabhan and G.J. i (1977) 177-184, Pavies, R-H.Bricknell and A.P, (asia) agrees Bentley, J.materl.sci. 14 »Rai and N.T.Gra: it, R-Grimes, CiBaker, MH. Seqegreens OA Aluminium 51 (1975) i], B.Geary, Ph.D. Thesis, Uni K-Ohorls W.watanebe, h.wniversity of Manchester (1985), Alcaro $ (iseaynape! H-Mukaiyaana’y gagch (oe K.Matsuki, Y,Uetani Metal sci. R-H.Bricknell and 6 (1976) egeaseMe IW. Edington, Movers. Metal Sci. 13 Trans 7A52 53 54 55 56 51 58 59 60 61 62 63 64 65 66 67 68 69 70 m1 12 73 74 15 16 1 78 79 80 6 SUPERPLASTIC FORMING AND DIFFUSION BONDING x.Mateuki, H.Morita, M.¥ yaunitses: "Gs M:Yaneda and Y.Murakami, ton Hamilton, C.C.Bampton and Co eer aetic Poiming of structural Allove $4. N.E.Paton and C.H.Hamilton ae ub). THS-ATME, Warrendale (1962) 173-189 Eic-Bampton, J.A.Wert and M.W.Mahoney, Mi tal {3a (1962) 193-198. Y, Metall Trang Pp G.Partridge and A.J.Shakesha: t Je ff, Te Royal Mrerafs Establishnont Fitnbecough (1982) panty a , B.Geary and J. Z é University of Be ea rose Unpublished Research, J.tadeworth, I.Palmer, D.D.Crooks and R.E.Lewis 2 dnt Riuniniumebithiun Conf. ed. B.8.Starke and al ao lers, publ. TMS-AIME, ‘Warrendale (1983) 111-135 Siwedeworth, ‘Merall Trane, 16a (1985) 2312-2332 , ‘Rwert, N-E.Paton, C.H.Hamilt e Metall-Trans. 12A (1981) ere. wewanoney: NeScimes, Superplasticity Ed. B.Baudelet and MSuery, publ. CNRS Paris (1985) Ch.13. D.W.Chung and J.R.Cahoon, Metal Sci. 13 (1979) 635-640. w.otsuka, Y.Miura and R.Horiuchi, g (1974) 1405-1408. N-Ridley, Superplastic Forming of Structural Alloys, Bd. N.E-Paton and C.H.Hamilton, publ. ‘TMS-AIME, Warrendale (1982) 191-207. Warriser end U.Ritter, Supexplasticity 2d; B.Baudelet Bad M.Suery, publ. CNRS, Paris (1985) ch.15. Siwadsworth, J.H-Lin and 0.D.Sherby, Metal Technol. g (1981) 190-193. C.1.Smith, B.Norgate and N.Ridley, Metal Sci. (1976) 182-188- N.Ridley and L.B.Duffy, ICSMA 7, Ed. H.J.McQueen et al. publ. Pergamon Press, Oxford 7195) 853-858. £.U.Engstrom, ‘Duplex Stainless steels, 86", ev engan Liere, publ. Nederlands Jnstituut voor yastechniek,The Hague (1986) 303-307. Lastenara and ¥.ohmori, Metal? 77008 16a (1987) 663-672. A.R.Marder, Trans~ TMS-AIME 245 (1969) 1337-1342. act Brith and N.Ridley, Metal Technol. 1 (1974) publ. 191-198. §.k Srivastava, Duplex stainless steels, Ssu, Metals Park, USA (1982) 1-14. Rechadler, Trans. TMS AINE 242 (1968) 1281-1287- A.Arieli and A.Rosen, Metall.Trans- 8A (1977) 1591-1596. aratkaibyshev, 1-V.kasschkos ‘and R.M.Galeev, SiMaterl.Sci. 16 (1981) 2501-2506. ; Mit Cope, D-R.Evetts and Aonidley, J-Materl.Sci- 22 1986) 4003-4008. . CR hamid eon, ‘Superplastic Forming Ba: g.P.Agrawal, Ubi. ASM, Metals Park (1985) 13-22- Siwert ‘and N.E.Paton, Meta 14a (1983) 2535-2544. Rex’Ghosh and C.H.Hamiltony (1979) 241-250. + anid M..Cope and N.Ridleyr Materl.Sci an (1986) 140-145- Scripta Metall. 10 40781 82 83 85 86 87 8g 89 90 91 92 93 94 95 96 _ 97 98 +99 100 101 102 103 104 105 106 107 J.R.Leader, D.F.Neal and C.Hammond, Metall.Trans, 17a (1986) 93-106. sean R.Kent and C ae i 70 (1985) 123-129, D.S.McDarmaid, Mat.Sci and Eng. 11. 20 (1973 P.Griffiths and C.H.Hammond, Acta Meta. (1972) 35-945. . PiWakal, $-Sakaguchi and Y.Matsuno, Adv.Ceram. Mater, 1 (1986) 259-263. | P.Wakes) H.Kato, S.Sakaguchi and N-Murayama, Yogyo-Kykai-Shi 94 (1986) 1017-1020. Jef Hensler and G.V.Cullen, J.Amer.Ceram.Soc. 51 1967) 584-585. S.e-champon and B.Escaig, J.Aner-Ceram.Soc. 63 (1980) 680-686. K.R.Venkatachari and R.Raj, J.Amer.Ceram.Soc 69 1986) 135-138. M-K-CAnsh” Wea Riodés aiid. A.N\Heuer, J-Anee.Ceran. doe 63 (1980) 46-63. P.C.Panda, R.Raj and P.E.D.Morgan, J.Amer.Ceram.Soc. 68 (1985) 522-529. M.P.Merrick, Superplastic Forming of Structural Alloys, Ed. N.E.Paton and C.H.Hamilton, publ. TMS-AIME, Warrendale (1982) 209-223. R.C.Gibson and J.H.Brophy, Ultrafine Grain Metals, Ed. J.J.Burke and V.Weiss,Publ. Syracuse Univ. Press (1969) 377-394. E.E-Brown, R,C.Boettner and D.L.Ruckie, Superalloys - Processing - 2nd. Int.Conf on Superalloys, Publ. AIME (1972) L1-112, penavden, R.C.Gibson, H.F.Merrick and J.H.Brophy, Trans. ASM. 60 (1967) 3-13. BoH;Kear, J-M.Oblak and W.A.Owcarski, J.Metals 246 (1972) 25-32. Rin Noodford, Metall.trans. 7A (1976) 1244-1245, R-L.Athey and J.B.Moore, Powder Metailar Performance Applications Ed. J.3.Burke and V.Weiss pub) Syracuse Univ. Press (1972) 281-293, ' +H.Reichman and J.W.Smytl oi Betchnan on Smythe, Int.J.Powder Met. R.G.Menzies, J.W.E 15 (1981) 210-216. J.P. Immarigeon and P.H. (1981) T17seaneen? P-H-Floyd, Metall trans. 12a J-K.Gregory, J.C.Gibeling and W.D.wi Superplasti Forming of Structural Alloys: Bay Nee page ngctte KigHamilton, publ THS-aIME, ‘warrendei= (1982). 362 a (Teas aga eae gmmazigeon, Wewallace’ (2902) 362. 14 (1980) 493-499, + Resa Got, X,G:Nekagawa, H. Yoshi zaw ‘ faterl.Sci. and Technol” 6 ‘ D.W.Livesey and N-Ridley, 7 peat) 637-639. 519-526. Hammond, J.Materl.Sci. 21 (1986) ‘dington and G.J.Davies, Metal Sci. 9A (1978) D.W.Livesey and N.Ri 161 1626" idley, Metall. trans 13A (1982) R.D.Schelling and G.H, (1973) 2199-3203" Reynolds, Metall trans 4a 1986 SUPERPLASTIC FORMING AND DIFFUSION BONDING g D.M.Ward, B.J.Helliwell i 100 oon Metal. Foraing (1973) sagt aage eices Metallurgica ~324 §.A.Shei and T.G.Langd : 308 39-646. ngdon, Acta Metall. 26 (1978) 0 $-A-Shei and 1.6.1: ul 3153-1158, angdon, Acta Metall. 26 (1978) §.A.Shei and 1.6.1 ‘i ql 3508-2996, G.Langdon, J.Materl.sci. 16 (1961) 12 -R.G.Fleck, C.J.Beevers j pr eignay daavelmae, and D.M.R.Taplin, J.Materl.Sci. 313. S.W.Zehr and W.Backofen, Tra: ui B.uhudelet ard M-Suory, J Meteri aed, 7 ci90d) Sen aie us R.C.Gifkins, J. Inst Met (1967) Br 7 META) BhaneL bs F.A.Mo! a 6 FsA-Wohamed’ and T-G.Langdon, Acta Metall. 29 (1961) 17 H.Naziri and R.Pearce, Int.. i BNaziri nt.J.Mech.Sei. 12 (1970) 18 -H.E.Cline and D.Lee, Acta Met. 18 (197 - HO Rig Ghosh, Superplaseie Forming of Structural Alloys, Ed. N.E.Paton and C.H.Hamilton, publ. TMS-AIME, Warrendale (1982) 85-103. 120 S.H.Vale, D.J.Eastgate and P.M.Hazzledine, Scripta Metall. 13 (1979) 1157-1162. 121 R.Lagneborg, B.H.Forsen and J.Wiberg, Creep Strength in’steels and High Temperature Alloys, publ. Metal Society, London (1974) 1-7. 122 Eon. Grossman and M.F.Aahby, Acta Metall, 28 (1975) 425-440, 123. H.J.Frost and M.F.Ashby, Deformation Mechanism Maps, publ. Pergamon Press Oxford (1982). 124 RuL.Coble, J.appl.Phys. 34 (1963) 1679-1682. 125 ClHerring, J-Appl-Phys. 21 (1950) 437-445, 126 P.M.Hazzledine and D.E.Newbury, Grain Boundary Structure and Properties, Ed. G.A.Chadwick and B-a-Smithy Publ. Academic Press (1976) 235-266. 127 ALK iMukherjee, Mater].Sci and Eng. 8 (1971) 83°89. 128 AlK.Mukherjee, Ann.Rev.Materl.Sci. 9 (1979) 151-189. 129 -B.P.Kashyap and A.K.Mukher jee, Superplasticity, Ed. B.Baudelet and M.Suery, publ. CNRS, Paris (1985) ch.4. 130 Tle Langdon, Superplastic Forming of Structural ‘Alloys, Ed. N.E.Paton and C.H.Hamilton, publ. ‘TMS-AIME, Warrendale (1982) 27-40. 131 ere Su) Lo lastic Forming of St uctural Alloys, Ed. N.E.Paton and C.H.Hamilton, publ. ‘TMS-AIME, Warrendale (1982) 3-26- 132 wEreeckinit, Metal Sci. 17 (1993) 13-48- 133 WiBeere, Metal Sci. 10 (1976) 5 cages) 3892368 134 Jipenirth and W.D.Nix, Acta Metall. 95 ; 135 Jlnischneibel and P-M-Hazzledine, J.Materl.Sci. 1 1983) 562-570- 136 Pend and A.X.Mukherjee, Materl.Sci and Eng. 45 1980) 61-70. . 137 Ue Slanelson, K.N.Melton and J.W. Edington, Reta Metall. 24 (1976) 1017-1026. s 138 K.N.Melton and J.W.Edington, Metal Sci-J- 7 (1973) 172-175. 139 Betaeiet, R.Pearce, M.H-Brown and K.F.Haley J.Microse. 97 (1973) 229-238. 199140 a4. 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 R.H.Bricknell and J.W.Edington, Acta Metall. 5 (1977) 447-458. Fow'batngcon, K.N.Melton and oo Prog-Materl.Sci 21 (1976) 67-170, J-E.Bird, A.K Mukherjee and J.E.Dorny ‘ ative Relation between Properties and Microstructure, , Ed. D.G.Brandon and Meee ee 51, publ. Israeli Univ. Press, derussian (3969) 255 i tall. Trans. Ue Seiee Peete SL. Dunlop and T.G.Langdon, tall 34 (1986) 1203-1214. ae vaste anit ecbangdon Acta Metall. 27 (1979) 251-257. A.E.Geckinli and C.R.Barrett, J.Materl.Sci 11 1976) 510-521, tee Adhby and R.A.Verrall, Acta Metall. 21 1973) 149-163, SiR Spingern and W.D.Nix, Acta Metall 26 (1978) 1389-1398, W.Beere, Philos.Trans.Roy.Soc.Lon. A288 (1978) 177-196. A.Arieli, T.Kianuma and A.K.Mukherjee, Acta Metall, 30 (1982) 1679-1688. W.DsNix, Rovenslegeie! Pé¢aing, Ed. S.P.Agrawal, publ. ASM, Metals Park (1985) 3-12. M-Suery and B.Baudelet, Res Mechanica 2 (1981) 163-170. J-H.Gittus, Trans. ASME (1977) 244-251. 1.W.Chen, Acta Metall. 30 (1982) 1655-1664. LiprEshelby, Proc-Roy.Soc.Lond. 241A (1957) 376-396. I.W.Chen, Superplasticity, Ed. B.Baudelet and M.Suery, publ. CNRS Paris (1965) ch.5 E.W.Hart, Acta Metall. 15 (1967) 352-355. CIRogemebel, J-Mech.Phys Solids 15 (1967) 359-370, C;Rossard, Revue Metall. 63 (1966) 225-335, prA-Woodford, Trans. ASM 62 (1969) 291.293, A.K.Ghosh and A.Ayres, Metall.Trans. 7A (1976) 1589-1591, Dat Avery and J.M.Stuart, Surfaces and Interfaces JI, fa: J-J-Burke, N.L.Reid and V.Weres : + Publ. Syracuse Univ. Press, New York (1968) 371- D-E.Holt, Ultrafine Grain Metals, Ed. J.3.Burke W.B Morrison, Trans. TMS. 17 (1983) 92-9 1093-1098, M.J.Stowell, Metal sci, 1 milton, ICsMa 5 ¥.Gerold and ¢.kostorz, PUBL, Bessey Ed. P.Hassen, and V.Weiss, publ. Syracuse Univ. Press, New York M.J.Stowell, Metal Sci. Mat.Sci.Technol, 2 (1986) 3.Lbian and makzMukherjee, Grain Boundaties in Enginee Meterials, publ. Claitor paste! roe (1975) 93-105. ishing, London (1979) 905-911, t+ Pergamon pain, 8) 2221-2227, 8. (1970) 355. “AIME 242 (196 J-Lian and M.Suery, B.Baudelet, +Sci. 3etian elet, Mat.Sci. and Eng. 94 (1986) A.K.Ghosh and C.H.K: 200in 472 173 374 175 176 47 178 179 180 181 182 183 164 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 6 SUPERPLASTIC FORMING AND DIFFUSION BONDING g.Pilling and N.Ridle; Foe angakors Poy Genie ot ee ae C.J.Peel b + P.J.Gregson, S.J.Harri ; publ. Institute of M iarets) anc (1985) 184-190. stale, London D.A.Miller and 3232125. T.G.Langdon, Trans JIM. 21 (1980) A.H.Chokshi, J.Materl.Sci ¢.C.Bampton and a acEdington, He tates) ede (ige2) 1721-1727. n, Metall.Trans. 13A t.Chandra, J.J.Jonas : 13 (1978) 2380-2384. and D.M.R.Taplin, J.Materl.Sci. G.L.Dunlop, E.Sha 5 Git -Dunlop, E-Shepizo, D.4-Rotaplin and J-Craner $§.Sagat, P.Blenkinsop and D.M.R Taj Li J-Ingt Metals 100 (1972) 268-274 pene sBelzunce and M. i M JBelgunce an M.Suery, Scripta Metall. 15 (1981) c.W.Humphries i 547702482, es and N.Ridley, J.Materl.Sci. 13 (1978) 8.x; bhet end P.G.Langdon, J-Materl.Sci. 13 (1978) c.H.Caceres and D.S.Wilkins: Ci eaea aise43e. on, Acta Metall. 32 W.B-worrison, Trans, ASM 61 (1968) 423-434. Ds ney ey and N.Ridley, J-Mater1.Sci- 13 (1978) b2o-eaais, Superplastic Forming of Structural lloys, Ed. N.E.Paton and C.H.Hamilton, publ. ‘TMS-AIME, Warrendale (1982) 392. peito and A.Hasegawa, Titanium 1g0, publ. TMS-AIME, Warrendale (1980) 983-992. G.Gurewitz, N.Ridley and ‘A.K.Mukherjee, ICE and Fracture Paper 5a-6, Beijing, China (1983)- M.T.Cope, Ph.D. Thesis, University of Manchester (1984)> H. Ishikawa, D.G.Bhat, F.A Mohamed and T.G.Langdon, Metall.Trans BA (1977) 523-525. M.M.I.Ahmed, F.A.Mohamed and T.G.Langdon, Jimaterl.Sci. 2 D.A.Miller an 1688-1690- D.w.Livesey and N.Ridley, 2257-2266. . N.Ridley and J.Pilling, Superplasticity, Ed. B.Baudelet and M.Suery, publ. CNRS Paris (1985) ch.8. M.Suery, ‘Superplasticity, Ed. B.Baudelet and M.Suery, Ch.9- publ. CNRS Paris (1985) B.P.Kashyap and Mukerjee, Res Mechanic® a7 ni (1986) 293-35 teg.Stowell, Superplast Forming of Structure) Alloys, Ed. N.E-Paton and jamiiton, pub) Bue kTME, Warrendale (1982) 321-336+ gJ.Pilling and N.Ridley, Res Mechanica 23 (2988) 31-63. D.W.Livesey an (2982) 563-559: Cc.Bampton ani 30 (1982) 2043-2053. P.J.Meschetex, Sripta Metall. osium g.Materl.Sci- 17 (1982) g w.Ridley, Met-Scl. 16 d R-Raje acta Metall. R.J-Lederich and P.S.RA0r Te (1984) 833-836~ 201200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 A.C.F.Cocks and M.F.Ashby, Metal Sci. 14 (1980) tee eocks and M.F-Ashby, Metal Sci. 16 (1982) aoe Forehead M.F.Ashby, Progress in Materl.sci. 82) 189-244. | Deel ae D.Rimmer, Philos.Mag. 4 98 ee M.V.Speight and J.E.Harris, Metal Sci. 83-85 . ¥.V.Speight and W.Beere, Metal Sci. 9 (1975) 190-191. : W.Beere and M.V.Speight, Metal Sci. 12 (1978) Lowches 29 (1981) I.W.Chen and A.S.Argon, Acta Metall. 29 ( 1759-1768, R.Raj and M.F.Ashby, Acta Metall.23 (1975) 653-666. G.H.dward and M.F.Ashby, Acta Metall. 27 (1979) 1505-1518. D.A.Miller and T.G.Langdon, Metall. Trans.10A (1979) 1869-1873. : J.Pilling, D.W.Livesey, J.B.Hawkyard and N.Ridley, Metal Sci. 18 (1984) 117-122. A.K.Chokshi and T.G.Langdon, Acta Metall. 35 (1987) 1089-1101 ‘ A,Needleman and J.R.Rice, Acta Metall. 28 (1980) 1315-1332. J.W.Hancock, Metal Sci. 10 (1976) 319-325. J.R.Rice and D.M.Tracey, J.Mech.Phys.Solids 17 (1969) 201-217, B.Budiansky, J.W.Hutchinson and §.Slutsky, Mechanics of Solids, Ed. H.G-Hopkins and i-J-Sewell, publ. Pergamon Press, Oxford (1982) 13. MJ Stowell, D.W.Livesey and N.Ridley, acta Mevall, 32 (1984) 35-42. J,Pilling and N.Ridley, Acta Metall. 34 (1986) 669-679. N-Ridley and D.W.Livesey, Res Mechani : (1981) 73-78, ¥ ss N.Ridley, D.W.Livesey and A.K. Mukherjee, Harbors 19 (1984) 1321-1332, -J.D.Patterson and N.Ridley, J. i Wid; Patterson Y, S.Materl.sci, 16 L.B-Duffy, M.Sc. Thesis, Universit, of Man: J-Filling, Naterl.sci. and Technol’ (1305) aeieaddee? spc etorey, J-1-Petit, D.J.Lege and a4 Hazard , superblasticity in Aerospace = alumirivg Ede R.Pearc id BEET Coe eRinium, . sree seeegtnd E-Kelly, publ. Crantiete AeTait and Do metOsY: Bedford (1986) 67-104. R.A.Tait and D.M.R. i Trega* OPd P-M-R.Taplin, Scripts Merge 13 (1979) A.Varloteaux and M.Suery, Superplasticity j Aerospace - Aluminium, £4, R: Pearce and eae Publ. Cranfield Institute of Technol. sendy (1986) 55-66. es EBdtosy H.Ahmed and R Pearce jupe: i r : + Superplastics Aluminium, Ed. R.Pearce and LeBel Institute of Technology, Bedeens (138 in Aeros: ce - Publ. Cranfield 6) 146-159, 202228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 6 SUPERPLASTIC FORMING AND DIFFUSION BONDING ¢.C.Bampton, A.K.Ghosh and M supexplasticity in neros ASS nie, (Bat R.Pearce and L.Kelly, publ. ¢: i a RePettute of Technology, Bedfccd sae #LB.Evans, Mechanisms of Cree! meee) aeee Elsevier Applied Science, New Yorten PUP ~.J.Chang, K.1.Kagawa, J.R.Ri a i wela metal. 27°(1979) 26sc20¢, SB SiT8, p.Shariat, R.B.Vastava and 7.G. 30 (1982) 285-296. +G.Langdon, Acta Metall. M.H.Yoo and H.Trinkl Saylsel. haus, Metall.Trans. 14A (1983) N.Ridley, D.W.Livesey and isa (584) 1443-2480" A.K.Mukherjee, Metall.Trans, R.Raj, Acta Metall. 26 (1978) 995-1001 =. 6 J.8.Wang, J.J.Stephans and W.D.Ni t 3:5 (Reads Tgoseionn Nix, Acta Metall. M.Suery and B.Baudelet, Superplastic Forming of alors TEST SIRE aca Woe rs publ. TMS-AIME, Warrendale (1982) 105-127. ‘ H.M,Shang and M.Suery, Metal Sci. 18 (1984) 143-152 PENG Peer and a N.Ridley, ICSMA 7, | H.J.MeQueen and J.P.Bailon, publ. P Oxford, (1985) 823-828. P rt A.K.Ghosh, Deformation of Polycrystals - Mechanics and Microstructure, Ed. N.Hansen, Publ. Riso National Laboratories, Denmark (1981) 277-282. N-Ridley, C.W.Humphries and D.W.Livesey, ICSMA 4, publ. ENSMIM, Nancy (1976) 433-437. Pueawle, Superplastic Forming of Structural Alloys, Ed. N.E.Paton and C.H.Hamilton, publ. TMS-AIME, Warrendale (1982) 307-317. R_J.Barnes, Superform Metals, Worcester (1983). Superform Metals Ltd. publication #¥307, Worcester UK. wW.C.Zhang, R.D.Wood and 0.C. Zienkiewicz, Gy, Ba, T.Sheppard, publ. Institute of al-Technolo‘ Metals, London (1986) 635-640. Soysguan and Z.Juan, Materl.Sci and Eng. 64 (1986) 311-125. 3) 340-348. A.RsRagab, Met. Technol. 10 (198 G.c.Cornfield and R.H.Johnson, (1970) 479-490. Mrlnait Int.J.Mech.Sci,12 (1970) 491-497. Wgohnson, T.¥.M-Al-Naib and J.L-Duncany J.Inst. Metals 100 (1972) 45-50- werarene, Int.J-Mech.Sci. 10 (1968) 403-427. \ MIF Ashby, Metall.Trans. 140 (1985) 1755-1769. \ eR Williamson, Superplastic Forme) of Structural Alloys, Ed. N.E.Paton and C.H.Hamilton, ‘publ. ) PMS-AIME, Warrendale (1982) 291-306. i G.W.Stacher, Superplasti Fo. in seated on . N.E.Paton and C.H.Hamilton, of Structural Alloys, Ed publ. THS-AIME, Warrendale (1982) 273-289; 1s a975) BG patent 3927817, Rockwell International ltd. ( 4 CLH.Hamilton, Titanium Science. ‘and Technology, Ed. R.I.Jaffe and R.M-Burte, publ. Plenum Press (1973) 625-647- Int.J.Mech. Sci, 12 203256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 G.Garmong, i BepaeGn and A.S.Argon, Metall.Trans, 6A (1975) 1269-1279. 7 B-Derby and E.R.Wallach, Metal Sci. ag (58a) 356. B.Derby and E.R.Wallach, Metal Sci. 18, (1984 (38 D.S.Wilkinson and M.F.Ashby, Acta Metall. ) 1277-1285. J.Pilling, Materl.Sci. and Eng. 100 (3988) aazenae. S.Reusswig, R.Gleichmann, P.G Zielinski, 0.6. and R-Raj, Acta Metall. 32(1984) 1553-1560 Kimura N.Furishiro and S.Hori, Titanium 80, F agi and Q.Izumi, publ. TMS-AIME (1980) 10 ISEn uae C.I.Smith, B.Norgate and N.Ridley, Scripta Met, 8 (1974) 159-164. . ; G.tober ane S.Elze, Advanced Joining of Metallic Materials, AGARD CP398 (1986) Ch.ll. R.G.Wing and M.Newnhan, in Diffusion Bonding, — Ed. R.Pearce Cranfield School of Industrial Science, (1987) 159-172. . D.Stephen and S.J.Swadling, Advanced Joining of Metallic Materials, AGARD CP398 (1986) Ch.7. D.Stephen, Designing with Titanium, Institute of Metals, London (1986) 108-115. . T.S.Baker and P.G.Partridge, in Titanium Science and Technology, Ed. G.Lutjering, U.Zwicker and W.Bunk, DGM, Munich, 2 (1985) 861-868. V.Alerja, in Titanium Science and Technology, _ Ed. G.iutjering, U.Zwicker and W.Bunk) DGM, Munich, 2 (1985) 933-938. 2.A.Munir, Weld.J.Res.Supp.62 (1983) 3338-3368, T.S.Baker and P.G.Partridge, in Diffusion Bonding, Ed. R.Pearce, Cranfield School of Industrial Science, (1987) 73-89. J.Harvey, P.G.Partridge and A.M.Lurshay, Materl.Sci. and Eng. 79 (1986) 191-199. J.Harvey, P.G.Partridge and C.L.Snooke, J.Materl.Sci. 20 (1985) 1009-1014. R.V.Sharples and I.A.Bucklow, Res.Rep #307/1986, The Welding Institute, Cambridge UK. (1986). A.E.Dray and E.R.Wallach, in Diffusion Bonding, Ed. R.Pearce, publ. Cranfield School of Industrial Science, (1987) 125-128. N.Ridley, J.Pilling, A.Tekin and %.X.Guo, Diffusion Bonding, Ed. R.Pearce, publ. Cranfield School of industrial Science, (1987) 129-142, T.D.Byun and R.B.Vastava, Welding, Bonding and Fastening, Ed.J.D.Buckley and B.A.Stein, NASA Langley Research Centre, Hampton Va. (1984) 231-245. T.D.Byun and P.Yavari, Superplasticity in Aerospace = Aluminium, £d. R.Pearce and L- Kelly Cranfield, (1985) 285-294. ' J.Pilling and N.Ridley, Materl.sci, and Technol. 3 (1987) 353-359. echnod S.H.Reichman and J.W.Smythe, Int J. Powd 6 6 (1970) 65-74. reser Mee R.G.Menzies, J.W.Edington and G.J.Davies, i. 15 (1981) 210216. a ite 204ws ae as 6 SUPEKPLASTIC FORMING AND DIFFUSION BoNLING Y-Blenvenu, T-Heseart, Livan Wouw, w. 5, A-morrloon, Ciltupin Dunding, #4: BR pearee Crantield School of industeial Beience ti Y.Miroks and B.J.Pletha, Private comeunicati T.Maseart, Yo Bienvenu and A-RoEr toon, Pittvelon Punding, td. W Pearce Cranfield Schoul of Induateial Wesence (1980) Sie]0$- WoT Chandler, AK Ghoeh and Wo kahuney, J. bpacece, Rockete, 21 (1964) 61-67, ) dean, on, 205{A) Superplasticity of Structural Alloys [C] Superplasticity in Aerospace - Aluminium [D} Superplasticite | i | [B] Superplastic Forming (E] Superplastic Forming \ (F] Superplasticity [G] Diffusion Bonding (H] superplasticity in Aerospace (I] Superplasticity and Superplastic Forming enn BIBLIOGRAPHY K.A.Padmanabhan and G.J.Davies, Springer-Verlag, Berlin, 1980, Ed. N.E.Paton and C.H.Hamilton TMS-AIME, Warrendale, Pa., 1982. Ed. R.Pearce and L.Kelly, Cranfield Institute of Technology, Cranfield, 1985. Ed. B.Baudelet and M.Suery, Centre National de Recherche Scientifique, Paris, 1985. Ed. §.P.Agrawal, ASM, Metais Park, OH., 1986. AGARD-LS-154, NATO/OTAN, Nuelliy-sur-Seine, 1987. Ed. R.Pearce, Cranfield School of Industrial Science, Cranfield, 1987. Ed. H.C.Heikkenen and T.R.McNelley ‘TMS-AIME, Warrendale, Pa. 1988. Ed. C.H.Hamilton and N.E.Paton TMS-AIME, Warrendale, Pa. 1989 206INDEX activation Energy grain boundary 64 superplastic flo measurement of 61,73,79,92,94 region I anit Aluminium Alloys a alnea 10,23 Al-Ca-Zn «28 ea 16,18 Al-Cu-Zr (Supral) 13,17,89,109,112,130,136 148,153,159,189,191 Al-Li (8090) 16,17,21,59,142,160,170 Ki-ig Bi 23,189,191 Al-Mg-Zr (Formal, Neopral) 16,17 Al-Si 23 Al-Zn-Mg (7475) 16,20,89,136,143,149 153,155,191 | (7010) 16,20,191 16 | (7050) os superplastic forming of 1° aiffusion bonding of Alumina a 39,40. superplasticity ey diffusion bonding ‘, } Aluminides 21,149 Chromium 16, 18,149 ae Copper 21 M-phase a Nickel 21 T-phase 21, zirconium 207 eens25,30 carbides nucleation of cavities 2 149 stabilisation of grain size 321185 iron carbide 30,14 7149 Tic 1149,194 cavitation characteristics 25 rr nucleation ’ effect of pressure 121-124,135 in aluminium alloys 18,137,142 in nickel silvers 45,145 volume fraction 113 experimental observations 111,125 effect of strain rate 155 in diffusion bonds 178 Cavities coalescence 132,140 growth rate 114,117,125,132 growth rate parameter 117,127,137,169 nucleation 144 hard particles 147 phase proportions 152 grain size 153 pre-existing 135 resolution of 147 shape 112 Ceramics 38,40 B spodumene 39 zirconia 38,39 ' Coal. iti escence of cavities 132,140 Cobalt silicide 147 Copper alloys Al-bronze 44,47 Coronze 1,46 IN-836 13,46,89-91 IN-629 45,47 45,47 Creep constitutive laws diffusion 65,69 67,97,99 208in diffusion bonding dislocation eo ' piffusion creep rate of $7469 719188 in diffusion bonding 180 stress relaxation by e 84 cavity growth by 114,125,129 piffusion bonding 175 mechanisms 178 rate of . 180,182,185 testing of bonds 186 of aluminium alloys 188,191 Al-7475 190 A1-7010 191 Supral 189 of nickel alloys 193 of titanium alloys 187 Ti-6Al-4V 187 IMI-550 187 Dislocation s trapping oreen 3 69,89,92,99 climb 66.75 constitutive laws 69,9 ‘oa sources. 74 pile-ups 96 interphase 47 Eutectic alloys 47 Ag-Cu ° 16 Al-cu 23 Al-Ca 23 Al-Ca-zn 47 Al-Mg 47,76 Pb-Sn 1,47 Bi-Sn 13 Eutectoid transformation Fe-alloys (see iron alloys) alloy: 25,149 Fracture 209Glass 43,44 metallic 3 ceramic in size ; “108 Gree particle stabilised 4 9129/35/03 segregation stabilise 9,37, 40. gr exponent 66,97,99 for superplastic flow for diffusion creep 69 cavity growth rate dependence "3s in si 2 sub-grain size . 163 effect on cavity nucleation Grain boundaries _—_ activation energy for sliding 83 inni 19,21 pinning sliding of 11,83 dislocation accommodated 73,98 diffusion accommodated 79,98 cavity nucleation during 144,148 elongation of 68 diffusion in 75,78,83-87,98 migration of 80,187,194 stress on 83,119 Grain refinement mechanical working 9 phase separation 14 phase transformation 13 Hot isostatic pressing 111,135 Iron alloys 96 ultra high carbon steels 10,27 carbon manganese ¥ IN-744 29 Avesta 3RE60 30,149 30,149 Lithium alloys (Al-8090) 16,17,21,59 142,160,170 159, , ’ Lithium aluminosilicate 39 Mechanical working cold rolling 21 10,17 210 “_isothermal warm working 10,17,30 ; 25 Metallic glass 44 Nickel alloys Ni-cr 43 Ni-aAl 42 IN-100 41,194 IN-718 3,44 Mar M-247 194 Mar M-6000 41,43 diffusion b i 41,43 onding of Nos Nickel si silvers 45,47, 156 Nucleation of cavities 144 by vacancy condensation 146 critical strain rate for 145 during grain boundary sliding 144,148 Pb-alloys Pb-Sn 2,47,76,147 Phase proportions in superplasticity 37,94 effect on grain size 9,29,35,39 effect on cavitation 152 Plasticity controlled cavity growth 117,129-132 Pressure hydrostatic 111,116,118,121-124,135,169 effect on cavity growth 135-144 stress state . ans hydrostatic requirements for forming 269 during forming | and aot in diffusion bonding e 21 Powder metallurgy Recrystallisation _ 17,109 dynamic (in-situ) 20,31,41 static 21166-67,91,106 Region I igs oe 67,90-92 strain rate sensitivity 66,89 activation energy . Ir 4,70 Region 83,84,93,97,99 constitutive laws 73,79 B31BR Peete 68,92,94 Region III : silver . 47 superplastic alloys 1e5 ion plating Spinodal decomposition 46 Steels (see iron alloys) Strain enhanced grain growth 36,49,109 controlled cavity growth 117,129-132 Strain rate constitutive laws 3,69,73,79,83 84,92,97,99 constant rate testing 58 iso-strain rate 94 . due to grain boundary sliding 84,92 critical, cavity nucleation 145 effect on cavitation 155 Strain rate sensitivity jump test 49-57 effect on elongation a on 0 105-108 stability during flow io Stress equivalent stress 114,117- 2 flow stress 117-118,121-124 mean stress yeas uniaxial nr balanced biaxial 123 plane strain 124 212isostress model of supe measurement rPlasticity 2s maximum principal stre e ‘uniaxial ~ SS 444-126, 119-124 balanced biaxial 2 plane strain 123 threshold (region I) 66 a5h2t on sliding grain boundaries 82,119 relaxation 83,144 Superplastic flow strain rate sensitivity 3,48,65,110,108,118 activation energy 61,73,79,92,94,99 grain size dependence 66,97,99 constitutive law 3,69,73,79,83 84,92,97,99 transition 92,95 in two phase alloys 12,16,26,34,40,43,45,96 continuum model 99 Superplastic forging 3,33,44,111 Superplastic forming P female forming 160 drape forming 162 reverse bulging 163 plug assisted 164 pressure - time cycle 17a hydrostatic pressure 168 ana diffusion bonding io snap back forming stint 58 Testitonstant strain rate gonna constant velocity 186 non-destructive (bonds) 49,55 strain rate jump Thermomechanical, Progal working/grain refinenent) 34 Titanium alloys Ti-6A1-4V Ti-6A1-2Sn-42Zr-2M0 160,187 34-36 , 62/96/1497 34,37 213| Ni,Co,Fe additions IMI 550 diffusion bonding of fully 8 alloys Zinc alloys zirconia 214 36 34,189 187 34,37 44,45, 67,106~107 38,40