Mechanisms of Bonding for Solid-StateWelding ProcessesJerry E. Gould, Edison Welding Institute

SOLID-STATE WELDING PROCESSESencompass all the methods in which metallicbonding occurs without the presence of resolidi-fied liquidmetal. These processes range from coldmethods (cold-pressure welding) to hot upset pro-cesses (forge processes) to diffusion processes(diffusion bonding). Typically, these processestake advantage of applied strain and/or heat tofacilitate joining. Joining is largely the result ofintimate intermetallic contact in the absence oflocal protective films.This article focuses on the underlyingmechan-

isms of bonding for these processes, with particu-lar emphasis on mechanisms for the forge-typeprocesses, while mechanisms for both the coldand the diffusion processes are considered inother articles. Specific mechanisms for differentstages of these processes are identified and quan-tified using best-available theory. Further, thesemechanisms are used to understand the roles oftemperature and strain in facilitating bondingwith these classes of joining technologies.There are three categories of bonding

mechanisms for the forge welding processes tobe considered:

� Contaminant displacement/interatomicbonding

� Dissociation of retained oxides� Decomposition of the interfacial structure

Modeling of contaminant displacement/inter-atomic bonding is largely taken from the cold-pressure literature and adapted to the forgewelding processes. This modeling suggests thatwith increasing surface strain, bond strengthscan asymptotically approach that of the basemetal. The cold-pressure models can be adaptedto forge welding processes by considering therole that the developing temperature field hason the distribution of strain. Generally, how-ever, after the surface strain is applied (upset),there are still residual oxide particles trappedin the bondline, and thermal dissolution of theseparticles can further improve joint performance.Thermal dissolution modeling has been

adapted from the carbide dissolution (in steels)literature, with stability data collected from

the diffusion bonding literature. Resulting mod-els show the importance of the size and distri-bution of the residual oxide particles, as wellas the role of the thermal cycle.Finally, the third underlying mechanism of

forge welding processes is the decomposition ofthe interfacial structure. Following the forgingrequired to displace contaminants, the bondlinecan be characterized as a highly dislocated,high-energy structure. Improvements in weldperformance can be made by decomposing thisstructure and reducing the residual bondlinestrain energy. Decomposition can occur eitherby recovery or recrystallization, depending onthe thermal cycle employed. There is evidenceto suggest that decomposition by recrystalliza-tion gives better bond performance. Decomposi-tion by recrystallization can be promoted byappropriate thermal cycles as well as appropriatedistributions of strain following upsetting.

Solid-State Welding Processes

Solid-state welding processes are the oldest ofwelding processes, with the official AmericanWelding Society definition of forge weldingrequiring an anvil and a hammer (Ref 1). Solid-state welding processes have proliferated, particu-larly over the last several decades, as new powersystems have developed. General classificationsof these processes include cold-pressure welding,externally heated hot-pressure processes, resis-tance processes, friction processes, arc-heatedprocesses, and diffusion processes. In this article,mechanisms of bonding are described for thoseprocesses using both mechanically applied strain-ing andheating.Detailed examinations of bondingmechanisms of the other processes are available inthe literature. These include the cold-pressurewelding processes (Ref 2–8) and the diffusionbonding processes (Ref 9–12). Specific variantsof the other bonding processes are as follows:

� Externally heated hot-pressure processesa. Forge weldingb. Gas pressure weldingc. Induction hot-pressure welding

� Resistance processesa. Flash butt weldingb. Resistance butt weldingc. Projection weldingd. Mash seam welding

� Friction processesa. Inertia weldingb. Continuous-drive friction weldingc. Linear friction weldingd. Friction stir welding

� Arc processesa. Percussive weldingb. Magnetically impelled arc butt welding

These processes can be thought of as havingtwo generally separable stages. These include aheating stage and an upsetting stage. As such,these welding methods can be generally classi-fied as heat and forge processes. Distinctionsbetween these processes then are largely inhow heat and forging are applied. Inevitably,however, heat is first applied. Mechanistically,this heat is used for two purposes. First, heatingthe workpieces reduces the yield strengths ofthe base materials and permits forging to occurwith high degrees of strain at reduced upsettingforces. Second, if heating is properly applied,upsetting creates high degrees of strain over avery localized area (at the bonding surface).Once the appropriate heating has been accom-plished, forging (or upsetting) commences.Upsetting also has two major functions. Theseinclude collapsing asperities to create intimatecontact, and displacing/dispersing protectiveoxides and films to facilitate metal-to-metalbonding. Residual heat content/heating is alsoconsidered advantageous, to further consoli-date/homogenize the joint.

Mechanisms of Solid-State Bonding

To attempt to define the specific mechanismsof bonding for solid-state welding processes, itis first necessary to have an understandingabout the microstructural and surface condi-tions of the workpieces planned for joining.On a microscopic scale, the surfaces for

ASM Handbook, Volume 6A, Welding Fundamentals and ProcessesT. Lienert, T. Siewert, S. Babu, and V. Acoff, editors

Copyright # 2011 ASM InternationalW

All rights reserved

www.asminternational.org

bonding have been well categorized as irregularand covered with various oxide and contami-nant films (Ref 2–12). In addition, there maybe microstructural/compositional irregularities,which further complicate the joining process.A typical representation of the prebond surfacecondition is presented in Fig. 1. This surface isgenerally characterized as three layers: the basematerial, a layer of mechanically and/or chemi-cally affected metal, and surface oxides/con-taminant films.There are a number of mechanisms that can

proceed to form a bond between such surfaces.The most important of these is that asperities onthe surfaces must be collapsed to form intimatecontact between materials. In forge weldingprocesses, creation of this intimate contact isdone mechanically; that is, local yield stressesare exceeded on the contacting surfaces, andsurface deformation is used to create the con-tact. For diffusion bonding processes (not cov-ered extensively in this article), such surfacecollapse is done under relatively low forcesand relies on creep and surface diffusionalmechanisms to consolidate the surfaces.Once the surfaces have come under intimate

contact, bonding still cannot initiate until a num-ber of other criteria are met. The most importantof these is how oxide and surface contaminantfilms can be affected to allow intimate contactof the underlying virgin materials. Generally,there are two mechanisms for this. For forgewelding processes, contaminant films can be bro-ken up as a result of mechanical action. In addi-tion, it is also possible to break down metaloxides by dissolution into the matrix. This is amechanism particularly important in diffusionbonding (Ref 13) but also plays a role in otherthermally assisted forge processes.Even when base materials are in intimate

contact, there are additional changes that mustoccur to facilitate an adequate joint. First, crys-tallographic matching across the boundary mustoccur (Ref 14). Obviously, most forge weldingapplications are between parent materials withrandomly oriented grain structures, and so, thisbond surface must take on the characteristic ofa series of high-angle grain boundaries. Gener-ation of this dislocation structure can occurmechanically (Ref 6) thermally (Ref 9, 10), orby a combination of the two. At this stage ofthe process, intimate solid-state bonding hasundoubtedly occurred; however, the localizedhigh-angle grain-boundary structure is

relatively unstable and is unlikely to yield anadequate joint. As a result, the last stage ofthe process is to relieve these local bondlinestresses, typically with some sort of thermalassist. Depending on the treatment, this localconcentration of strain energy can result in afinal bond structure ranging from local recoveryto recrystallization (Ref 15).As mentioned, these mechanisms collectively

permit solid-state bonding between metallicmaterials, although not all mechanisms are usedby all solid-state welding processes. Generally,these mechanisms, particularly as they operatewithin the group of forge welding processes,can be classified in three general areas. Theseinclude surface deformation mechanisms, con-taminant dissolution mechanisms, and interfa-cial structure homogenization mechanisms.These are described in detail, including theirdirect relationship to the forge welding pro-cesses, in subsequent sections.Contaminant Displacement/Interatomic

Bonding. As briefly described previously, sur-face deformation mechanisms have two func-tions: to collapse surface asperities and todisplace surface contaminants. It is of interestthat the best information on the role of surfacestrain and its effect on the extent of solid-statebonding is available in the cold-pressure weld-ing literature (Ref 2–8). Several authors haveexamined the roles of surface condition,mechanisms of interfacial breakdown, anddegrees of subsequent bonding for cold-pres-sure welding applications. Collectively, initialbonding, related to surface straining, appearsto progress through the following stages.

Extension of the Contacting Surfaces

For any bonding to occur, it is essential thatcontaminating oxides/films be disrupted. Thisis, of course, accomplished by application ofcontact surface strains. It is equally important,however, that surface oxides/films be in a con-dition in which they can be readily broken upwhen the surface strain is applied. Table 1 listshardnesses for some common oxides. Of inter-est here is the difference between the hard-nesses of the aluminum or copper oxides.Figure 2 shows some fractographs of bond sur-faces for cold-pressure welds below criticalbonding deformation for aluminum, copper,and silver. It is clear from these results thatthe aluminum oxide fails in a brittle manner,while the copper oxide fails in a shear manner.

A more general plot showing the relationshipbetween oxide/metal hardness ratio andrequired deformation for bonding is presentedin Fig. 3. One method of improving the charac-teristics of surface film fracture is to locallycold work the base metal. It has been demon-strated that cold working the surface usingscratch brushing both minimizes the extent ofcontaminant films and creates a local layer ofheavily cold-worked material, which, on strain-ing, can fracture and carry more ductile oxidefilms (Ref 6).For most conventional forge welding pro-

cesses, extension of the contacting surfaces isdone with a combination of heat and force.For this stage of the process, local strain is themost important factor. However, how that strainis distributed is a strong function of how thethermal field is applied. Figure 4 shows someresults from numerical simulations of theflash-butt welding process (Ref 16). This plotshows how contact surface strain is affectedboth by the amount of upset used and the levelof flashing acceleration employed. Flashingacceleration directly controls the heat distribu-tion in the workpiece, with higher flashingaccelerations indicating steeper, higher thermalgradients (Ref 17, 18). For conductivity materi-als, particularly aluminum and copper, strainlocation provided simply by the thermal gradi-ent is difficult. In such cases, pinch-off diesare recommended (Ref 19). Pinch-off dies sim-ply use the constraint of the die (rather than thethermal profile) to create localization offorging. The function of pinch-off dies is shownschematically in Fig. 5.

Separation of the ContaminatedAreas

It is established that the onset of bondingoccurs with applied surface strain as surfacecontaminants are separated and virgin basematerials are allowed to contact. Considerablework has been done, again largely in the cold-pressure welding area, attempting to quantifythe separation of these contaminants and theresulting bond quality. In examining the roleof contaminated surfaces, all workers agree thata critical strain must first be achieved that thissurface ruptures (Ref 2–8). There are discrepan-cies, however, on how this rupture occurs.Mohamed and Washburn (Ref 6) suggest thatseparations of the two contaminated surfacesare unrelated, while Wright et al. (Ref 7) andlater Bay (Ref 8) suggest that surface contami-nants impinge on either side of the joining andtherefore separate as pairs. Local bonding isthen accomplished by extrusion of virgin mate-rial into the spaces between the separated con-taminated surfaces. Each set of authors hasdeveloped models based on their assumptionof interfacial breakdown. In each case, theunderlying assumption is that the strength ofthe joint is a direct function of the fraction of

Fig. 1 Schematic representation of workpiece surfaceconditions in the prebonded state

Table 1 Representative hardnesses ofsome metal oxides at room temperature

Metal Hardness, HV Oxide Hardness, HV

Al 15 Al2O3 1800Cu 40 Cu2O 160Ag 26 Ag2O 135Au 20 . . . . . .

Source: Ref 6, Table 1

172 / Fundamentals of Solid-State Welding

the bond area converted by actual base mate-rial/base material bonds.The simplest of these models is that devel-

oped by Mohamed and Washburn (Ref 6). Thismodel assumes a completely brittle contamina-tion layer and no coordination of contaminantson either side of the bondline. The physical rep-resentation of this model is presented in Fig. 6.The resulting equation for strength is:

f ¼ CR

Rþ 1

� �2

(Eq 1)

where f is the ratio of the joint strength tothe parent material strength, R is the surfacestrain, and C is a constant to incorporate con-taminant mismatches and contaminant hardness.

Fig. 3 Relationship between oxide/metal hardnessratio and the critical deformation for bonding

during cold-pressure welding. Source: Ref 3

Fig. 4 Thermomechanical modeling results showingthe relationship between flashing acceleration,

upset distance, and contact surface strain for flash-buttwelding mild steel

Fig. 2 Fractographs showing the faying surfaces of cold-pressure-welded aluminum, copper, and silver at subbondingstrains. (a) Aluminum, 3% deformation. (b) Aluminum, 6% deformation. (c) Copper, 5% deformation.

(d) Copper, 23% deformation. (e) Silver, 25% deformation. Source: Ref 6, Fig. 1

Mechanisms of Bonding for Solid-State Welding Processes / 173

The model proposed by Wright et al. is slightlymore complex. This model was generated forroll-bonding applications, so largely plain-strainconditions exist. The physical representation ofthis model is presented in Fig. 7. This modelassumes matchup of contaminants across thebondline and attempts to account for a degree ofprebonding deformation. The resulting equationfor joint strength is:

f ¼ C 1� 1� Rf

� �21� Rtð Þ2

" #(Eq 2)

In this case, C is considered an empiricalhardening factor, Rt is the threshold deforma-tion for bonding, and Rf is the total deformationof the process. Equations 1 and 2 are similar,asymptotically approaching a maximum bondstrength as the total deformation (R or Rf)approaches 1. The most complex analysis isprovided by Bay. This model includes theeffects of contaminant films and subsurfacehardened layers and is diagramed in Fig. 8.The resulting equation for joint strength is:

f ¼ ð1� bÞY p� peso

þ bY � Y0

1� Y0p

so(Eq 3)

where f is now the ratio of weld tensile strengthto base material tensile strength, b is the frac-tion area covered by contaminant films, p isthe applied pressure, pe is the threshold pressure

for bonding (extrusion pressure), and so is theyield strength of the base material. Y and Y0are the surface exposure and threshold surfaceexposure, where the surface exposure is de-fined by:

Y ¼ 1� 1

1þ X(Eq 4)

where X is the degree of expansion of the con-tact area.These models, of course, show a greater

degree of complexity, because a greater numberof bonding factors are included. It is importantto recall, however, that these models have beendeveloped for cold-pressure processes, andthese complexities may be more or less relevantfor conventional forge welding processes. Onefactor of note is the extrusion pressure(described as pe in Eq 3). This factor is includedeither directly or indirectly in each of thesemodels. However, for conventional forge weld-ing processes, extrusion pressures will fall dra-matically with temperature and may be lessof a factor. Also to be questioned is the roleof subsurface cold-worked layers, which, inconventional forge welding processes, willprobably anneal substantially before any mac-roscopic deformation occurs.

Realignment of the Grain Structuresfor Bonding

There is considerable evidence that crystallo-graphic matchup across the bondline is alsoimportant at this stage of bonding (Ref 3, 6,14). Detailed work (Ref 3) suggests that contactbetween similarly oriented close-packed ornear-close-packed planes most readily bonds.For aluminum, (111) to (111) and (110) to(110) were found to bond readily, while (111)to (100) were found difficult to bond. However,most structural materials are polycrystalline,so such ideal crystallographic matchups arerelatively uncommon. To accomplish bondingrequires some localized crystallographic re-orientation. The model here is the one of aseries of grain boundaries. Grain boundariescan be thought of as a complex dislocationpileup, accommodating the misorientationbetween grains over a very small distance.The types of macroscopic surface strains andlocal intercontaminant extrusion described hereare ideal sources for dislocation generation andundoubtedly contribute to the generation ofthis bondline structure. An example of this dis-located structure is presented in Fig. 9. Thisparticular example is a resistance butt weldon steel, showing evidence of a residual bond-line. This region is characterized by relativelyhigh internal strain energy and may be aquality concern. Reactions of this region tothe applied thermal fields typical of the forgewelding processes are described in a subsequentsection.

Metal

Metal

MetalOxide

Oxide

Metal

(b)

(c)

(d)

(a)

Fig. 6 Schematic illustration of interfacial breakup asproposed in the Mohamed and Washburn

model. (a) Original interface. (b) Fracture of brittle oxidefilm. (c) First requirement for welding: formation ofoverlapped oxide-free metallic areas. (d) Secondrequirement for welding: extrusion of metal through thegaps created in the oxide and some relative sheardisplacement at the points of contact of oxide-freemetal. Source: Ref 6

Fig. 7 Schematic illustration of interfacial breakup asproposed by Wright et al. Source: Ref 7

Fig. 8 Schematic illustration of interfacial breakupas proposed by Bay. (a) Interfacial surfaces.

(b) Onset of extrusion and thinning of contaminant film.(c) Welds. Source: Ref 8

Fig. 5 Use of pinch-off dies in upset welding processesto localize strain

174 / Fundamentals of Solid-State Welding

Thermal Dissolution of Oxides/Contaminants

The preceding discussion indicates thedegrees to which bondline strain can be usedto create a solid-state bond. However, implicitin that discussion and related modeling weretwo related facts: some level of contaminationwas always present in the joint area, and jointstrengths could only asymptotically approachparent material strengths. The relationship hereis straightforward. As long as contaminantsexist in the joint, they reduce effective bondarea and act as initiation sites for subsequentmechanical failures (Ref 20).To achieve improved joint properties, partic-

ularly in industrial applications, some furtherreduction in the residual bondline contaminantcontent is advantageous. Fortunately, for manymetallic systems, oxides are soluble in thematrix at elevated temperatures. The degree ofsolubility of a specific oxide in MxOy in amatrix of metal “A” at equilibrium can bedefined by the solubility product:

Keq ¼ ZðCMÞxðCoÞy (Eq 5)

where Keq is the equilibrium solubility product,CM and CO are the compositions of the metal“M” and oxygen in the matrix metal “A,” andZ is a proportionality constant relevant to theactivity coefficient. If the oxide is of the matrixmetal, this expression reduces to:

Keq ¼ ZðCoÞy (Eq 6)

with Z a different proportionality constant. Thissuggests that the solubility product is similarly apower function of the maximum soluble oxygencontent in the base material. For oxides of thebase materials, Eq 6 suggests that the solubilityproduct can be estimated from the maximum sol-ubility of oxygen in the matrix as taken from theappropriate phase diagram. In addition, the shapeof the oxygen solvus provides some indication ofthe temperature dependence for the solubilityproduct. In a similar manner, Eq 5 suggests that

the solubility product for non-base-materialmetal oxides relative to the base metal (as wellas the temperature dependence) can also be esti-mated from the phase diagram. Here, the solubil-ity product is estimated from the maximumsolubilities of the secondary metal and oxygenin the base material.If Raoultian behavior of oxygen in the base

material is assumed, the proportionality con-stants in Eq 5 and 6 become equal to 1. Then,knowing the stoichiometry of the oxide presentand using the appropriate phase diagram,approximate solubility products for some dif-ferent metals can be done directly. Table 2 listsapproximate solubility products for some stan-dard engineering materials with their mostcommon oxide. These solubility products arecalculated for approximately bonding tempera-tures estimated for forge welding processes(0.9 Tm). Materials shown include aluminum,iron, and titanium. These solubility productscover approximately 30 orders of magnitude,indicating, on one extreme, the difficulty of dis-solving aluminum oxide into an aluminummatrix as well as the relative ease with whichtitanium can dissolve its own oxide.Similar calculations can be done for nonmatrix

metal oxides. The approximately solubility prod-uct for Al2O3 on iron is calculated and comparedto the similar calculation for the oxide of iron(Fe2O3) in Table 3. These calculations were doneat the approximate bonding temperature for forgewelding iron. In this case, the stoichiometry forthe two oxides is similar (x = 2, y = 3), so the dif-ference between the two solubility products isdirectly related to the solubility of the aluminumin the steel. This fact appears to account for therelatively low solubility of Al2O3 in iron.Such solubility products and diffusivities of

oxygen in the matrix have been used as a basis

for modeling oxide dissolution during diffusionbonding (Ref 21). However, this analysis waslargely based on continuous oxide films andfocused on the maximum thickness of thesefilms for relatively long (diffusion bonding)heating cycles. Such an analysis does not takeinto account the breakup of the oxide film intodiscrete particles caused by the applied surfacestrain during forge welding processes, or therelatively short heating times.A better analysis can parallel that done by

Ashby and Easterling for the dissolution of car-bide particles during welding (Ref 22). Thatanalysis examines the dissolution of discreteparticles. The approach used attempts to esti-mate the roles of both the solubility of the car-bide constituents into the matrix and diffusionof these constituents away from the decompos-ing carbide. The approach is based on theassumption that distinct spherical particles canbe dissociated completely into a volume ofmatrix with radius l. Further, the particle willdissociate into this volume at a locus of timesand temperatures defined by:

l ¼ ðD�t�Þ1=2 (Eq 7)

where D* and t* are the combinations of thediffusivities (D, a function of temperature,defined at T*) and times (t*) over which theparticle can be completely dissolved into thevolume matrix defined by l. Combining thisapproach for examining the role of diffusioncan be combined with an expression for thetemperature dependence of the solubility prod-uct of the particle, to examine particle dissolu-tion behavior. The discussions on solubilityproducts for oxides detailed previously can beused to adapt these equations for oxide particledissolution. The resulting governing equationsinclude for base-metal oxide particles:

Ts ¼ B

A� lnOð Þyf

h i (Eq 8)

and for non-base-metal oxides:

Ts ¼ B

A� lnMð Þx Oð Þy

f

h i (Eq 9)

where Ts is the dissolution temperature, A and Bare the temperature coefficients for the appro-priate solubility product, and f is the matrixvolume fraction affected by the decaying oxide,defined by:

f ¼ 1

1þ t�t exp� Q2

R1T� � 1

Ts

� �h i3=2 (Eq 10)

In this expression, Q2 is the activationenergy associated with the appropriate diffusiv-ity (oxygen or metal + oxygen), and R is theideal gas constant. With some estimate of t*and T*, Eq 10 combined with either Eq 8 or 9

Fig. 9 Resistance butt weld on mild steel indicating ahighly deformed zone down the bondline

Table 2 Estimated solubility products foroxides present on some commonengineering materials assuming Raoultianbehavior of oxygen in solution in the parentmaterial

Metal Oxide Keq

Al Al2O3 3 � 10�29

Fe Fe2O3 1 � 10�15

Ti TiO2 1 � 10�1

Calculations are done for temperatures approximately representingbonding temperatures for forge welding processes (0.9 Tm) (internallygenerated).

Table 3 Comparison of solubility productsfor Al2O3 and Fe2O3 in an iron matrix at theappropriate bonding temperature for forgewelding iron (0.9 Tm)

Base material Oxide Keq

Fe Al2O3 1.8 � 10�19

Fe Fe2O3 1.0 � 10�15

Internally generated

Mechanisms of Bonding for Solid-State Welding Processes / 175

(as appropriate) defines an implicit relationshipbetween the time/temperature profile for theprocess and the degree of oxide dissolution.In these expressions, t* and T* are direct

functions of the oxide particle size and distribu-tion. Values for these can presumably be esti-mated from the original distribution of oxideson the bonding surfaces and estimations of sur-face strain, as described previously.From these equations, some qualitative esti-

mate can be made of the role of both the degreeof forging and time-temperature profile on bondquality. With increasing strain applied to thecontacting surface, both particle size and den-sity will inevitably fall. These factors reduceamounts of oxygen (and potentially secondmetal) that must be diffused and increase thekinetics of oxide dissolution. Extended heating(welding) times are important in that, again,diffusion is promoted. Increasing welding tem-peratures are not only advantageous for increas-ing rates of diffusion but also for increasing thesolubility product for the dissolution reaction.

Breakdown of the InterfacialStructure

A third mechanism of bonding results fromthe decomposition of any interfacial structure.As described previously, straining of the bondsurface, extrusion of material around residualoxide particles, and matching crystallographicstructures across the bondline result in a highlydislocated bondline structure. This highly dislo-cated structure is of relatively high energy aswell as planar. An example of such a highlydislocated bond is shown in Fig. 10. This struc-ture obviously develops during straining thecontact surface. However, the presence of vari-ous particulates from the contaminated bondsurfaces may also stabilize this structure.Decomposition of this structure is largely a

thermally assisted process. To develop thishighly dislocated structure, considerable energyfor deformation is required. Much of thisenergy is stored in the interfacial structureitself. With varying degrees of activation

energy, this structure can quickly decomposeto a lower energy variant. Parks (Ref 15) hasdone considerable work to understand thebreakdown of contacting interfaces. In hiswork, he suggests two regimes for breakdownof this interfacial structure. These parallel theconcepts of recovery and recrystallization.Recovery of the interface implies a realignmentof the dislocated structure to reduce the overallstrain energy of the system. This is typicallydone at relatively low temperatures, permittingonly local movement of the dislocations thatmake up the boundary structure (Ref 23–25).During recovery, these dislocations realignthemselves into dislocation cells. An exampleof such cells is shown in Fig. 11 (Ref 26). Parksfound that very little effective bonding occurredif interfacial decompositionwas limited to recov-ery. Rather, substantial bond strengths werefound if higher annealing temperatures wereused, resulting in bondline recrystallization.This is shown in Fig. 12. Recrystallization isessentially the nucleation and growth of newgrains. Provided activation energies are highenough, this mechanism of interfacial decompo-sition shows the greatest reduction in bondlineenergy and is suggested by Parks as essential forforming high-integrity bonds.During welding, residual stored energy (as

local deformation) can play a role in the kinet-ics of recovery and/or recrystallization. Parkshas demonstrated that actual bonding tempera-tures can be reduced depending on the degreeof deformation in the material. Required recrys-tallization temperatures as a function of thedegree of deformation for a range of materialsare shown in Fig. 13.Obviously, the extent to which this interfa-

cial structure can decompose is a function ofboth the amount of strain applied and the tem-perature cycle experienced. Increasing amountsof strain (upset) obviously increase the amountof work in the material and promote subsequentbreakdown of the interfacial structure. Time attemperature, however, provides the activationenergy to allow this aspect of the bonding pro-cess to proceed. It is interesting from this dis-cussion that greater levels of upset may permitbonding at shorter times and lower tempera-tures. However, in practice, extended times

and temperatures are almost always advanta-geous, permitting maximum homogenizationof the joint microstructure.

Comparison of Solid-State BondingProcesses

The above discussion suggests that, in sum-mary, solid state processes employ one or moreof three mechanisms to accomplish bonding.These include disruption of contacting interfacesfor nacient metal contact, diffusion related disso-ciation of residual contaminants, and breakdownof the remaining interfacial structure. These fun-damental mechanisms, are driven by processmechanisms, specifically temperature, time, anddeformation. Obviously, temperature and timeboth drive diffusion reactions (promoting bothcontaminant dissolution and breakdown of theinterfacial structure), while deformation pro-motes interfacial disruption. This approach wasused by Fenn (Ref 27). In the developed con-struct, Fenn created a ternary diagram with axesof temperature, time, and deformation. For thisdiagram, the axes are represented as conceptualfractional values. Here, temperature can be con-sidered as the fraction of the absolute meltingtemperature, time as a dimensionless fraction,and deformation as the relative collapse of thetwo components. On this diagram, Fenn thenplaced hypothetical ranges for a number of solidstate processes. The resulting diagram is shownin Fig. 14. Limits of the diagram include diffu-sion bonding (all time and temperature) and, coldwelding (all deformation). Of note, most conven-tional solid state welding processes (flash weld-ing, friction welding, upset (or forge) welding)fall toward the middle of the diagram, utilizingcomponents of all the mechanisms describedabove. Of note, the diagram is not material spe-cific. As a result, how individual processes fallon this diagram will be strongly affected by thesubstrate welded, shifting to match the specificcombination of mechanisms most advantageousto individual material systems.

Summary

This article provides a systematic look atmechanisms of bond formation during solid-state(forge) welding processes. Discussions have beenlimited to those processes that can be character-ized as having two stages: heating and forging.Explicitly excluded were those processes that donot use heating (cold-pressure welding processes)or forging (diffusion bonding processes). For theforge welding processes, three distinct mechan-isms of bonding have been discussed. Theseinclude contaminant displacement/interatomicbonding, dissociation of retained oxides, anddecomposition of the interfacial structure.Contaminant Displacement/Interatomic

Bonding. This mechanism of bonding relatesto displacement of contaminants by local strainat the contacting surface. Displacement of these

Fig. 10 Interfacial structure on resistance-projection-welded mild steel

Fig. 11 Dislocation cells in a dynamically recoverediron microstructure. Source: Ref 26

176 / Fundamentals of Solid-State Welding

contaminants allows exposure of clean surfacesfor direct interatomic bonding. The basics formodeling this mechanism were largely takenfrom the cold-pressure literature. Although sev-eral models are available with increasing levelsof complexity, all predict that bond strengthsasymptotically approach that of the base metalwith increasing surface strain. For the forgewelding processes, the developed temperaturedistribution also plays a role, increasing metalplasticity, assisting in localizing strain at thebondline, and reducing required upset loads.Thermal Dissolution of Oxides/Contami-

nants. The applied surface strains describedpreviously permit considerable bonding butleave a residue of oxide/contaminant particlesdispersed over the bond surface. As a mecha-nism for further facilitating bonding, many of

these particles can be thermally dissolved inthe matrix. The relative solubility of specifictypes of particles can be assessed directly byexamining solubility products between the con-stituent elements of the particle compared withsolution in the base material. For base-materialoxides, this solubility product is only a functionof the solubility limit of oxygen in the matrix.This analysis was used to indicate relative stabil-ity of a range of base-metal oxides. This exami-nation was extended, using previous work donefor dissolution of carbide particles in steel, toexamine the kinetics of dissolution. This analysisincorporates both solubility product and diffusiv-ity factors. The results indicate the effects ofresidual oxide particle size as well as the roleof the severity of the thermal cycle for dissol-ving these oxide particles.

Breakdown of the Interfacial Structure.The side result of the first two mechanisms is ahighly dislocated interfacial bond structure. Thisstructure results largely from the application ofbondline strain but can be stabilized by the pres-ence of discrete oxide particles. This structure isof relatively high energy and can be a detrimentto weld quality. Decomposition of this structuredoes improve bond quality. The mechanismof decomposition, however, depends on the ther-mal cycle employed. For relatively short or low-temperature cycles, the structure may onlyrecover, resulting in a series of dislocation cells.At higher temperatures and longer times, recrys-tallization of the metal at the bondline can alsooccur. Some results suggest that recrystallizationof the bondline structure results in better weldproperties. Increasing stored energy at the bond-line (causedbyhigher levels of strain) also appearsto aid the kinetics of recrystallization and improveweld quality.

REFERENCES

1. “Standard Welding Terms and Defini-tions,” ANSI/AWS A3.0-94, AmericanWelding Society, Miami, FL, 1994

2. D.R. Miller and G.W. Rowe, Fundamentalsof Solid Phase Welding, Metall. Rev., Vol28 (No. 7), 1962, p 433–480

3. R.F. Tylecote, Investigations on PressureWelding, Br. Weld. J., Vol 1 (No. 3),1954, p 117–135

4. R.F. Tylecote, D. Howd, and J.E. Fur-midge, The Influence of Surface Films onthe Pressure Welding of Metals, Br. Weld.J., Vol 5 (No. 1), 1958, p 21–38

5. L.R. Vaidyanath, M.G. Nicholas, and D.R.Milner, Pressure Welding by Rolling, Br.Weld. J., Vol 6, 1959, p 13–38

6. H.A. Mohamed and J. Washburn, Mechan-isms of Solid State Pressure Welding,Weld. J. Res. Suppl., Vol 54 (No. 9),1975, p 302s–310s

7. P.K. Wright, D.A. Snow, and C.K. Tay,Interfacial Conditions and Bond Strengthin Cold Pressure Welding by Rolling,Metals Technol., Vol 1, 1978, p 24–31

8. N. Bay, Mechanisms Producing MetallicBonds in Cold Welding, Weld. J. Res.Suppl., Vol 62 (No. 5), 1983, p 137s–142s

9. K. Inoue and Y. Takashi, Recent VoidShrinkage Models and Their Applicabilityto Diffusion Bonding, Mater. Sci. Technol.,Vol 8 (No. 11), 1992, p 953–964

10. Y. Takashi, K. Inoue, and K. Nishiguchi,Identifications of Void Shrinkage Mechan-isms, Acta Metall. Mater., Vol 41 (No.11), 1993, p 3077–3084

11. T. Enjo, K. Ikeuchi, and N. Akikawa,Effect of Oxide Film on the Early Processof Diffusion Welding, Trans. JWRI, Vol10 (No. 2), 1981, p 45–53

12. T. Enjo, K. Ikeuchi, and N. Akikawa,Effect of the Roughness of the Faying Sur-face on the Early Process of Diffusion

Fig. 12 Weld strengths as a function of annealing temperature for a range of materials. Source: Ref 15

Mechanisms of Bonding for Solid-State Welding Processes / 177

Welding, Trans. JWRI, Vol 11 (No. 2),1981, p 49–56

13. A. Nied, General Electric Company,Research and Development Center, Sche-nectady, NY, private communication, 1991

14. V.M. Zalkin, Theoretical Problems of ColdPressure Bonding of Metals, Svar. Proiz.,Vol 11, 1982, p 41–42

15. J.M. Parks, Recrystallization Welding,Weld. J. Res. Suppl., Vol 32 (No. 5),1953, p 209s–222s

16. J.E. Gould and T.V. Stotler, “An Examina-tion of Morphological Development duringFlash Butt Welding,” EWI CooperativeResearch Report MR9602, 1996

17. E.F. Nippes, W.F. Savage, J.J. McCarthy,and S.S. Smith, Temperature Distributionduring the Flash Welding of Steel, Weld.J. Res. Suppl., Vol 30 (No. 12), 1951,p 585s–601s

18. E.F. Nippes, W.F. Savage, S.S. Smith, J.J.McCarthy, and G. Grotke, TemperatureDistribution during the Flash Welding ofSteel—Part II, Weld. J. Res. Suppl., Vol32 (No. 3), 1953, p 113s–122s

19. Resistance Welding Manual, 4th ed., Resis-tance Welding Manufacturers Association,Philadelphia, PA

20. W.F. Savage, Flash Welding—ProcessVariables and Weld Properties, Weld. J.Res. Suppl., Vol 41 (No. 3), 1962,p 109s–119s

21. Z.A. Munir, A Theoretical Analysis ofthe Stability of Surface Oxides duringDiffusion Welding of Metals, Weld. J.Res. Suppl., Vol 62 (No. 12), 1983,p 333s–336s

22. M.F. Ashby and K.E. Easterling, A FirstReport on Diagrams for Grain Growth inWelds, Acta Metall., Vol 30, 1982,p 1969–1978

23. J.D. Embury, A.S. Keh, and R.M. Fisher,Substructural Strengthening of MaterialsSubject to Large Plastic Strains, Trans.Metall. Soc. AIME, Vol 236 (No. 9),1966, p 1252–1260

24. J.H. Cairns, J. Clough, M.A.P. Dewey, andJ. Nutting, The Structure and MechanicalProperties of Heavily Deformed Copper,J. Inst. Met., Vol 99, 1971, p 93–97

25. A.L. Wingrove, Some Aspects of RelatingStructure to Properties of HeavilyDeformed Copper, J. Inst. Met., Vol 100,1972, p 313–314

26. J.E. Pratt, Dislocation Substructure inStrain Cycled Copper as Influenced byTemperature, Acta Metall., Vol 15 (No.2), 1967, p 319–327

27. R. Fenn, “Solid phase welding-an oldanswer to new problems?” Metallurgistand Materials Technologist, Vol. 16 (No.7): 1984, pp 341–342.

Fig. 14 Construct of time-temperature-pressure regimes of solid state welding processes proposed by Fenn (Ref 27).This diagram includes dimensionless values for temperature, time, and deformation, and contains suggested

ranges for specific solid state processes.

Fig. 13 Recrystallization temperatures as a function of degree of deformation for a range of materials. Source: Ref 15

178 / Fundamentals of Solid-State Welding

AN OVERVIEW OF WELDING IN SOLID STATE

Mihaela Iordachescu , Elena Scutelnicu

Caminos, Canales y Puertos, Dep. Ciencia de Materiales, Universidad Politecnica de Madrid, Espana Dunarea de Jos University of Galati, Romania

miordac@vahoo.com

ABSTRACT The importance of the Solid State Processes (SSP) has increased in the last

decade due to the industry demands of improved properties of joined/surfaced materials, combined with cost reduction and energy saving. New and/or micro-scale solid state processed materials are used by aerospace, automotive and electrotechnics industry. Nowadays, classic SSP are mainly applied to light materials, but progresses were also reported in steels. In this field, the tools design, the technology and practical techniques surpassed the fundamental approach of the materials solid state processing. The SSP parameters evaluation is based on different experiments, approaching the material flow in the large plastic deformation domain. The paper approaches the solid state welding/joining and surface processing. The envisaged SSP are solid state joining processes as Cold Welding (butt and spot welding), Friction Stir Welding - FSW, and surface processing, Friction Stir Processing - FSP. Therefore, the investigation targeted the deformation and flow pattern of the parent metal in case of cold welding and FSW/FSP, processes parameters evaluation and correlation, local analysis of the material structural transformations, and material hardening.

KEYWORDS: solid state processes, microstructure, hardening, material flow

1. INTRODUCTION

In the beginning of the XXth century, together with the development of resistance welding process, it was noticed the decisive influence of the pressure in joints achievement, leading up oriented researches in the field of cold pressure welding . Nowadays, the process is used to achieve the joints of the high voltage networks' wires, as well as for joining several parts of the cryogenic equipment.

Cold welding (CW) process can be easily and comfortably achieved, being practically the result of the pressing force applied between two metal sheets appropriately and carefully cleaned. This process requires important materials deformation, obtained by using high pressing forces, able to generate upsetting pressures 10 times greater than the maximum material's yield strength. As fig. 1 presents, CW can be achieved mainly by two methods: spot, respectively butt CW

The complexity of the cold weld formation reported by W. Zhang and N. Bay was studied, in case of aluminium bars butt colds welding, by M. Iordachescu

In 1891, Bevington realised the opportunity to use friction to generate heat for both forming and

welding. Of significant importance was the ability to produce a product in solid state. The use of friction for welding came in the 1950's, when Bishop [9] reported many applications of Russian origin. On a more worldwide scale, the process gained acceptability for high volume production and its ability to join a wide range of materials from the 1960's. The automotive industry adopted the process to weld bimetallic valves, rear axle and front wheel drive shafts, while the electrical industry was welding copper/aluminium connectors in large scale

Another major milestone was reached in 1991 when Thomas Wayne from The Welding Institute (TWI) in UK patented FSW, extending the opportunities to use friction heating and material flow to join sheets and plates in solid state . The process principle is illustrated in fig. 2a.

FSP is a new solid state processing technique (fig. 2b), which can locally eliminate casting defects and refine microstructures, thereby improving strength and ductility, increasing resistance to corrosion and fatigue, enhancing formability and improving other properties . FSP can also produce fine-grained microstructures through the thickness to impart superplasticity. FSP provides the ability to thermomechanically process selective locations on the

structure's surface and to some considerable depth to enhance specific properties. This was mainly

2. COLD WELDING PROCESS

2.1. Cold Welding process parameters

Cold welding process can be obtained as result of applying a pressing force on two metal sheets, appropriately cleaned. The process requires important material deformations. Easy deformable metals as Aluminium or Copper (or their alloys) can be cold-welded, but the process can be also achieved between dissimilar metals (Aluminium-Stainless Steel, etc.). Butt cold pressure welding rises very interesting theoretical and practical issues relate to joint achievement, material deformation, material flow and cold hardening during deformation, and also to material thermal response during the up-setting process.

The butt cold pressure welding procedure of the aluminium bars depends on the following determinant

accomplished by adapting the technology developed for FSW

technological factors: clamping dies selection, preparation of bar contact surfaces, the initial standoff value, up-setting force, clamping force, bars deformation, and welding equipment adequate selection.

The selection of clamping dies have to be adapted to the bar cross-section shape. Furthermore, a large contact surface between bars and clamping dies must be ensured, to avoid the bars sliding at up­setting.

Figure 3a presents a cross-section through the clamping die. It can be noticed that this is made of three distinct pieces, machined inside for allowing contact with the bar exterior surface. The space between these pieces makes possible the initial clamping of the bar. During up-setting, some of the bar material fills this space, creating longitudinal burrs. The cogged active surface of the clamps prevents the bar sliding in the clamps during up­setting (Fig.3b). The clamping length used during the

SPOT COLD WELDING

BUTT

COLD :JZf̂ r^^F^*" ' WELDING ' ' • • • \

2 _ - - J

a) b) Fig. 1. Cold welding process variants: a) spot cold welding; b) butt cold welding; 8 - material displacement; F -

upsetting force; 1- pressing/clamping devices; 2 - samples to be welded

1/ FRICTION STIR

PROCESSING

a) b) Fig.2. FSW/FSP variants: a) FSW; b) FSP; F - pressing force; sr - rotating speed; Sj- advancing speed

experiments was LB = 40 mm (an empirical technological prescription indicates as minimum value LB = 4 • d, where 'd' is the bar initial diameter [1,2]).

The geometry of the clamp active side (d! =1.4 • d, a = 5°, (3 = 60°) was designed to ensure appropriate material flow and joint strength. Figure 3c presents the initial position of the clamps during up­setting, and final one, respectively.

The bars contact surfaces preparation, their smoothness, alignment and perpendicularity are necessary for preventing their eventual relative sliding and compressive buckling. Furthermore, the cleaning and degreasing of the bars extremities were necessary before welding. The cleaning of the bars extremities with a rotating wire brush followed the mechanical cutting of the bars samples.

Fig.3. Clamping device: a) cross-section; b) longitudinal section of the clamping area;

c) longitudinal section of the joint area, before (left) and after (right) applying the pressing force, 'F'; Fs -

squeezing force

The initial standoff represents the initial length of the non-clamped end of the bar to be welded. An optimum positioning of the samples in the clamps is described by an initial standoff capable to ensure accurate up-setting that produces welded joint of good quality. The bars standoff is experimentally determined according to the base materials qualities. The correct standoff value experimentally determined is of 10mm, equal with the diameter of the aluminium bars. An excessive standoff doesn't lead to a correct pressing, causing the bars buckling occurs.

The up-setting force, 'F ' , is the actuation force produced by the hydraulic motor of the toggle-lever press (F = 63,000 N). All the others technological parameters of butt cold welding process of aluminium bars were determined at this up-setting (pressing) force value.

A special design clamping device ensured the bars self-blocking for low values of

clamping/squeezing force (Fs = 8,650 N), before upsetting. During upsetting, the bars material actuates towards the clamping dies, developing forces of similar magnitude as the upsetting force. Thus, reaction forces of important values are generated in the clamping device, without increasing the squeezing force. In conclusion, whilst the necessary squeezing force is about 7.30 times smaller than the upsetting one, the reaction forces have the same order of magnitude as the pressing force. Consequently, the actuation of the squeezing devices must be designed to provide the 'Fs' value, whilst the clamping device itself should be able to carry out the bigger loads generated by the reaction forces.

The bar deformation, '8 ' , is the ratio between the one-bar standoff variation (during up-setting) and its initial standoff. Previous research consider that cold pressure welding process by single up-setting can be obtained only if a minimum deformation (Smin — 0,7 for aluminium) is exceeded during

pressing. The adequate selection of the welding

equipment depends mainly on the necessary up­setting force value, capable to ensure the achievement of cold welded joint.

2.2. Cold Welding material flow

Different tests on butt pressing of aluminium bars of 10mm diameter were performed. The up­setting process was stopped at different values of bars deformation, for better understanding the material deformation process and the cold welded joint formation.

Experiments have confirmed that bars cold welding occurs at the deformation S = 0.75. Continuous pressing at higher deformation values led to corresponding decrease of standoff, with an increased certitude of obtaining good quality joint.

Fig.4. Cold weld in case of 8 = 0.68

The study intended also to determine the deformation critical value when the joint tensile strength surpasses the base material ultimate strength. Thus, in case of 8 = 0.68, the formation of cold welding was noticed, but the joint had a poor strength due to the small area of the weld (fig. 4). At tensile tests, the samples failure occurred, without

elongation, due to material loos of elasticity (cold hardening caused by pressing), in the weld area. Moreover, a weld critical area was defined as the weld area when the joint strength is equal with the base metal strength. Once surpassed the weld critical area, the joint failure initiates in base metal.

Fig.5. Ultimate strength of butt cold welded aluminium bars vs. bar deformation

Figure 5 presents the experimentally determined diagram of ultimate strength of butt cold welded aluminium bars as a function of deformation. Three domains corresponding to different deformation ranges provide information on the progress of cold welding process. It can be noticed that butt cold welding cannot be achieved at deformations inferior to �min = 0.68. Figure 5 also illustrates the moment when the product of the weld area multiplied to the correspondent stress is equal with the value of the ultimate strength of base metal, for the deformation � = 0.73. Good quality joints are possible for bigger deformations, with respect to other technological parameters, such as the preparation of the contact surfaces or the standoff value. Figure 6 shows the macro and microscopic images corresponding to the cold weld formation; the material flowing lines are visible on the joint macrostructure. Macro and microscopic images featured the aluminium behaviour during CW, showing: - The increase of the material flow in the up-setting

force direction, on the longitudinal axis of the bars, in accordance with the deformation value. The grains are compressed on the direction of the predominant stress developed in the longitudinal axis of the bar, a typical forging structure being thus obtained. The initial grains form (typical for the drawing manufacturing process) modifies by pressing. The increased values of the normal stresses couple at high deformation values lead to grains refinement and furthermore, to their reorientation on radial direction.

- The material flowing outside the clamps and weld seam and burr formation, are presented in fig. 6a. Due to internal stress values, grains slide mainly in

the radial direction. At higher deformation values, as the microscopic images present, the flow lines orientations on radial direction are observed. Initiation of the typical forging subgrain structures (fig. 6b), with dimensions lower than 0.3 �m, allows for the fusion of the two lattices, thus achieving the cold welding.

-

a)

b)

Fig.6. Macro and microscopic images of CW joint (99.5%Al), (� = 0.75); a) CW joint macrostructure; b)

CW joint microstructure

At microscopic level, the butt cold pressure welding process of the aluminium bars is produced due to the up-setting force when, in the contact area, the value of the normal stresses couple allows the initiation of the subgrain structures, with dimensions less than 0.3 �m, capable to fusion and create a common lattice.

3. FSW/FSP PROCESSES

3.1. FSW/FSP material flow and temperature

Nowadays, new techniques as solid-state Friction Stir Welding – FSW are currently used for obtaining different aluminium alloys qualitative joints. Although the welding may produce high tensile stresses (up to the yield stress) balanced by lower compressive residual stresses elsewhere in the component, FSW results in a much lower distortion and residual stresses owing to the low heat input characteristic of the process . Recently, a derivative from FSW, Friction Stir Processing – FSP namely, was proved as being useful for inducing directed, localized, and controlled materials properties in any arbitrary location of components.

Basically, the FSP/FSW process has three stages: the penetration of the tool, when the local plasticity properties of the material quickly changes with temperature and the tool travel speed is characterised by acceleration from zero to the working value (fig. 2b-a,b,c); the working stage, when the travel speed and the pressing force are constant, as well as the rotating speed and the tool angles (fig. 2b-d); the tool retracting phase, when the travel speed is decreased by zero value and the tool is removed from the workpiece.

FSP/FSW process is characterised by some main technological parameters, namely: tool geometry, tool tilt and concordation angles (angles of the tool axis with the vertical direction in the longitudinal and transverse plane, respectively), rotating and travel speeds and plunging/working force.

The material flow during FSP/FSW is quite a complex deformation process of practical importance for tools design and materials microstructure transformations. Therefore, an overall pattern of the material flow hasn't been reported yet. As example, the paper approaches the processing of as-cast AA 6061. The processed layer macrostructure is presented in figure 7.

Fig.7. FSP macrostructures of AA6061 as cast aluminium alloy; 1 -6 microstructures positions

The base metal (BM) microstructure of as-cast AA 6061 consists of Al solid solution dendrites along with coarse silicon and intermetallic phases. Shrinkage porosity is also prevalent. FSW/FSP closed the shrinkage porosity and homogenized the as-cast microstructures by breaking up and evenly dispersing initial phases. Moreover, the resulting microstructures do not have a uniform grain size distribution for any one set of process parameters. Grain size varies from the top to the bottom as well as from the advancing to the retreating side. The differences in grain size likely are associated with differences in both peak temperature and time of application of temperature.

Figure 8 shows the typical features of all different zones in a single processed layer cross-section of as cast AA 6061 under processing condition of 1,120 rpm for the rotational speed and 320 mm/min for the welding speed. The positions 1-6 from fig. 8 are located in different micro structural zones. The micrographs show that the microstructure of the processed layer is complex and highly dependent on the position within the processed zone.

This result arises because of the large local variations in the plastic flow and from the thermal history resulted from the material interaction with the tool.

The microstructure in the stir zone is characterized by refined grains in a discrete series of bands and some precipitate mainly distributed at the grain boundaries. There is also still some debate concerning the origin of the annular rings observed within the nugget zone attributed to an abrupt variation in the grain size and precipitate density The nugget zone grains suggest effective strains together with a microstructural evolution that occurs by a combination of hot working and a dynamic recovery or recrystallization. The temperature reached in the nugget zone is known as being situated in the range of 450-500 °C for the 6061- Al alloy Distinct precipitates and coarsened grains are observed at the deformed regions of TMAZ. HAZ grains are severely coarsened by FSP (Fig. 8: 1, 4).

The characteristic annular-banded structure is distinctly observed to be asymmetric and more obvious on the advancing side (A) of processed zone as shown in fig. 8, positions 1, 2, 4. A severe deformation has also occurred along the top surface of the processed layer where the shoulder of the tool is in contact with the material. The flow lines from fig. 8 - positions 4, 2, 5 seem to represent plastic deformation increments that develop as the rotating tool moves through the processing line. Although, it is well known that the material is transported from A to the retreating side (R), Colligan showed that with a threaded pin tool, the material from the upper part of the processed zone is pressed down, whereas the material from the lower part processed zone is moved toward the top surface. The material may travel many cycles around the tool before being redeposited. A little flow of material was observed near the bottom of the processed zone.

The effect of processing parameters on temperature was investigated by Arbegast and Hartley

. They reported that for a given tool geometry and depth of penetration, the maximum temperature depends on the rotation rate, while the rate of heating depends on the traverse speed. A higher temperature on the advancing side was noticed.

From different experimental investigations and process modelling, several conclusions can be underlined about the FSP/FSW thermal profile: - the maximum temperature developed within the stir

zone is below the melting point of the materials; - tool shoulder dominates the heat generation during

FSP; - the maximum temperature increases with increasing

tool rotation rate at a constant tool traverse speed and decreases with increasing traverse speed at a constant tool rotation rate. Furthermore, maximum temperature increases with increasing the ratio of tool rotation rate/traverse speed.

- the maximum temperature occurs at the top surface of the stir zone.

Fig.8. Typical features of all different zones in a friction stir processed single layer cross-section of as cast AA6061: 1- flow patterns in the appendage zone; 2 -nugget zone; 3 - the retreating side of TMAZ; 4 - the

advancing side of TMAZ; 5 - nugget bottom side; 6 - processed layer bottom side; (200x)

FSW Classification

(Sample:

It Q | tHAZ [iv [4 TMAZ

_ .. \lQ 1 lHAZ • Coldo . =} . tv tTMAZ

AA6065-T4, 1=3.9mm) Q[rpmJ

v[mrn/minj

Q [rpm] v [mm/minj

2000 ~ 500 ~

1600 800

Fig.9. FSW/FSP typical material flow patterns

The characteristic annular-banded structure is distinctly observed to be asymmetric and more obvious on the advancing side (A) of processed zone as shown in fig. 8, positions 1, 2, 4. A severe deformation has also occurred along the top surface of the processed layer where the shoulder of the tool is in contact with the material. The flow lines from fig. 8 - positions 4, 2, 5 seem to represent plastic deformation increments that develop as the rotating tool moves through the processing line. Although, it is well known that the material is transported from A to the retreating side (R), Colligan [15] showed that with a threaded pin tool, the material from the upper part of the processed zone is pressed down, whereas the material from the lower part processed zone is moved toward the top surface. The material may travel many cycles around the tool before being redeposited. A little flow of material was observed near the bottom of the processed zone.

The effect of processing parameters on temperature was investigated by Arbegast and Hartley

. They reported that for a given tool geometry and depth of penetration, the maximum temperature depends on the rotation rate, while the rate of heating depends on the traverse speed. A higher temperature on the advancing side was noticed.

From different experimental investigations and process modelling, several conclusions can be underlined about the FSP/FSW thermal profile: - the maximum temperature developed within the stir

zone is below the melting point of the materials; - tool shoulder dominates the heat generation during

FSP; - the maximum temperature increases with increasing

tool rotation rate at a constant tool traverse speed and decreases with increasing traverse speed at a constant tool rotation rate. Furthermore, maximum temperature increases with increasing the ratio of tool rotation rate/traverse speed.

- the maximum temperature occurs at the top surface of the stir zone.

3.2. FSW/FSPparameters and their influence on the processed material

The main result of the research regards the influence of the friction stir main parameters (the tool rotational and advancing speed) on the material flow pattern around the tool (fig. 9) . In the case of hot conditions, the visco-plastic material flow is more concentrated around the pin and the heat affected zone is wider resulting in a basin shape nugget. In the opposite, under cold conditions, the thermo-mechanically heat affected zone is wider and the heat affected zone is smaller.

B4

i 82

V, BO

r1-. . • —

-'^>"*T

w S, .^•J

y

.\i. . "V>", »--* •v-

16 -14-13 -10-8 -8 -4 -1 0 2 4 6 B 10 12 14 IB

Distance from the nugget centre [mm)

a)

100 •

95

90 .

B5 •

BO -

75 -

70

65 |

60 -

*-,

w </v

-18 -14-15 -10 -8 -6 A -7 0 2 4 6 8 10 12 14 IE

Distance from trie nugget centre [mm]

b)

Fig. 10. FSW/FSP typical hardness profile: a) non heat treatable aluminium alloys

(AA5083-H111; thickness: 4mm); b) heat treatable aluminium alloys (AA6061-T4; thickness: 4.8mm)

The typical hardness fields obtained for the two main different groups of wrought aluminium alloys

are presented in fig.10 a, b. Figure 10a show the hardness profile of a

friction zone of an aluminium non heat treatable alloy. Increased hardness values are found in the processed material area and in the heat affected zone when comparing with the base material. Due to their sensitivity to strain hardening the increase is most significant in the dynamically recovered zone and the thermo-mechanically heat affected zone.

Figure 10b present the hardness profile results in case of processing a heat treatable aluminium alloy. Information about the typical location of the global minimum value of the harness field can be found here, located in the interface between the heat affected zone and the thermo-mechanically heat affected zone. Along the heat affected zone there is, typically, a local minimum hardness value due to processed material over-ageing.

The processed materials hardness profile enables a reliable assessment of its static mechanical resistance.

4. CONCLUSIONS

The main conclusions emerged after experiments related to butt cold welding of aluminium bars are:

- Small clamping force is needed at process beginning, ensuring only the initial bar self-blocking in clamping dies. The actuation of the squeezing devices must be designed to provide this small value force, whilst the clamping device itself should be able to carry out bigger loads generated by the pressed material reaction forces.

- Material flow due up-setting reflects its simultaneous displacement outside and inside the clamps;

- The structural refinement microscopically observed has only mechanical origin.

- At microscopic level, the butt cold pressure welding process of the aluminium bars is produced due to the up-setting force when, in the contact area, the value of the normal stresses couple allows the initiation of the subgrain structures, with dimensions less than 0.3 |j,m, capable to fusion and create a common lattice.

Moreover, after FSW/FSP processing of aluminium alloys it can be conclude that:

- The material flow is a complex deformation process of practical importance for tools design and materials microstructure transformations; an overall pattern of the material flow hasn't been reported yet.

- The material temperature increase within and around the stirred zone; Its distribution directly influences the microstructure of the processed materials, such as solid state transformations, grain size, coarsening and dissolution of precipitates, and resultant mechanical properties of the processed surface.

- Tool geometry is an important factor for producing required materials microstructures, but at the present, tool design information is very limited, protected by patents.

- The processing parameters, including the tool rotation rate, the traverse speed, the spindle tilt

angle, and the target depth, are crucial to produce the bulk material modifications.

- The experiments regarding the influence of the processes main parameters (the tool rotational and advancing speed) on material flow pattern around the tool in visco-plastic conditions indicate a larger heat affected zone than under cold conditions.

- The processed materials hardness profile enables a reliable assessment of its static mechanical resistance

These results are important milestones and constrains for the solid state processes simulation using finite element method modelling.

REFERENCES

Georgescu V., Iordachescu M., Georgescu B., Practica sudarii prin presiune la rece (Coldpressure welding practice), E.T. Publishing House, Bucuresti, 2001.

Iordachescu M., Contributii la sudarea prin presiune la rece cap la cap (Contributions to the butt cold welding), PhD thesis, Dunarea de Jos University of Galati, Romania, 2005.

Zhang W., Bay N., Cold welding - Experimental investigation of the surface preparation methods, Welding Journal 76 (8): S326-S330, 1997.

Zhang W., Bay N., Cold welding - Fractographic investigation of the weld formation, Welding Journal 76 (9): S361-S366, 1997.

Zhang W., Bay N., Cold welding - Theoretical modelling of the weld formation, Welding Journal 76 (10): S417-S420, 1997.

Bay N., Mechanisms producing metallic bonds in cold welding, Welding Journal 62 (5): S137-S142 1983.

Bay N., Cold forming of aluminium - State of the art, Journal Of Materials Processing Technology 71 (1): 76-90 NOV 1 1997.

Constantin E., Iordachescu M., Scutelnicu E., New approaches on Aluminium butt joints design evaluation using FEA simulation, ASR International Conference: Achievements and perspectives in producing welded construction for urban environments, Editura SUDURA, 11 July 2003, Bucharest, Romania, pp. 311-321.

Nicholas, E.D., Friction Processing Technologies, Welding in the World, Vol. 47, n° 11/12, 2003.

Thomas W.M., Staines D.G., Norris I.M., Frias R., Friction Stir Welding- Tools and Developments, Doc IIW-1639-03.

Mishra R.S., Ma Z.Y., Friction stir welding and processing, Materials Science and Engineering R 50 (2005) 1-78, Elsevier, 2005.

Iordachescu M., Iordachescu D., Scutelnicu E., Vilaca P., Ocana J. L., Aluminium friction stir processing - roughness vs. macro/microscopically results, Welding in the World, Vol. 51, Sp. Iss.,pp. 441448,2007.

Vilaca P., Santos J. P., Gois A., Quintino L., Joining Aluminium Alloys Dissimilar in Thickness by Friction Stir Welding and Fusion Processes, Welding in the World, Vol. 49, No. 3/4, pp. 56-62, 2005.

Staron P., Kocak M., Williams S., Wescott A., Residual stress in friction stir-welded Al sheets, Physica B 350, pp. e491-e493, Elsevier, 2004.

Colligan K., Material Flow behaviour during Friction Stir Welding of Aluminium, Welding Journal, 75(7), pp. 229s-237s, 1999.

Mechanism of Solid State Pressure Welding

Study establishes a rational basis for atom-to-atom bonding that supports neither the film theory nor the dif­fusion principle

BY H. A. MOHAMED AND J. WASHBURN

ABSTRACT. The mechanism of pres­sure welding in polycrystalline alumi­num, copper, silver and gold was in­vestigated. The role of the oxide fi lm was studied and it was found that no metal to oxide bonding contributes to the strength of the weld. From scan­ning electron microscope observa­tions a two-stage model has been suggested which could explain the dif ferent behavior of the metals studied.

The first stage of welding involves the formation of overlapped oxide-free metallic areas; this is controlled by: (a) difference on a microscale of the local plastic strain occurring on matching opposite faces of the weld interface, (b) relative hardness of the metal and its oxide fi lm, and (c) mechanical properties of the oxide.

The second stage involves: (a) plastic flow of the metal to the over­lapped areas; the stress at which this can take place is influenced by the stacking fault energy of the metal, and (b) some relative shear displacement at the points where metal cleaned of oxide comes into contact; this is influenced by surface roughness.

The different weldability of differ­ent metals is attributed to differences in stacking fault energy, hardness ratio and plastic properties of the ox­ide. The weld strength calculated theoretically on the basis of measured

The authors are associated with the Inor­ganic Materials Research Division. Lawrence Berkeley Laboratory and the Department of Materials Science and Engineering, College of Engineering, re­spectively, at the University of California, Berkeley, California 94720.

welded area was in agreement with measured fracture strength. It was concluded that the strength attained after a given deformation is deter­mined by the fractional welded area at that deformation.

Introduction Pressure welding is the establish­

ment of an atom-to-atom bond be­tween the two pieces to be joined through intimate contact between oxide- f ree areas achieved under pressure and without the formation of liquid phase.

In order to develop this bond, sur­face films have to be removed or at least reduced in amount. Surface fi lms fall into two categories:

Oxide Film

All metals except gold possess an oxide film at room temperature. In most metals the oxide film reaches a limiting thickness in the range 20-100 angs t roms at room t e m p e r a t u r e (Ref. 1).

Contaminant Film

This film consists of a thin layer of moisture and greases. The best tech­nique which has proved (Refs. 2-4) to be successful in reducing these films is a combination of chemical and mechanical cleaning.

Two theories have been proposed so far to explain the mechanism of pressure welding:

The Film Theory

This theory (Refs. 2, 5) proposes that if two clean metal surfaces are

brought into intimate contact a weld will be created. The theory attributes the different weldability of different metals to the relative hardness of the bulk metal and oxide. Initiation of welding is controlled by the degree of fragmentation of the oxide fi lm.

In the present work, scanning elec­tron microscope photographs show that oxide films on aluminum, copper and silver crack at deformat ions much less than the minimum welding deformation (see Fig. 1); the way this deformation was determined will be discussed.

This suggests that it is not only the fragmentation of the oxide film which controls the initiation of welding.

Energy Barrier Theory

The energy barrier theory (Refs. 6, 7) suggests that even if clean sur­faces are brought into contact no weld will result. The theory states that an energy barrier exists that must be overcome before welding can take place. Parks (Ref. 7) thought that the barrier is recrystallization while Erd-mann-Jesnitzer (Ref. 6) thought that it is diffusion. Semenov (Ref. 6) sug­gested that the energy barrier comes from the misorientation of the crys­tals at the contact surface, since he could weld aluminum, copper and silver at the temperature of liquid nitrogen. It is impossible to assume that welding at this temperature oc­curred due to diffusion or recrystal­lization.

McEwen and Milner (Ref. 8) have shown that immiscible metal pairs can be joined satisfactorily.

In the present work the rate of ap­plying the pressure was found to have

302-s I S E P T E M B E R 1 9 7 5

no effect on either the welding defor­mation or the weld strength. It will be shown that the misorientation factor has an important effect in the ini­tiation of welding.

Experimental Procedure

Materials

Materials for Lap Welding. The materials used for this study in-

-" / %i

i T t ^ X t i r '

eluded high purity aluminum, silver, copper, gold (purity, 99.999%) and commercial purity aluminum (purity, 99%). The dimensions of the strips were 75 mm length, 18.75 mm width and the thickness ranged from 0.8 to 1.2 mm. The overlapping distance was 25 mm.

Materials for Butt Welding. These materials included high purity alumi­num (purity. 99.995%) and 6061T6 aluminum alloy (main alloying ele­ments are silicon and magnesium). The dimensions of the rods were 60 mm length and 9 mm diameter.

Procedure

Welding Dies. Two welding dies were designed, one for lap welding and the other for butt welding. Figure 2 is a schematic illustration of these dies.

Surface Preparation. In order to reduce surface fi lms, all the spec­imens were first degreased in ace­tone and then wire brushed using a motor driven wire brush. Figure 3 shows a scanning electron micro­scope photograph for an aluminum surface. It is seen that the surface consists of a series of hills and valleys.

Welding. After surface prepara­t ion, the specimens were imme­diately set in the welding die and then the pressure was applied. At the beginning of the experiments, the pressure was applied at very slow rate and then at a much higher rate. It was

lOOmm -* — 5 0 m m

l - T - r H

Specimen

i 20mm I30rr

r—L-J—I _ i _

45 mm

t

- 35mm -4 5 mm - - . -

(o) Butt Welding Die

- 40mm

• ' 3000X

Fig. 1 — Scanning photographs for the surfaces of aluminum, copper and silver of defor­mations below the minimum lap welding deformation. High purity aluminum: (A) 3% defor­mation, (B) 6% deformation. High purity copper: (C) 5% deformation, (D) 23% defor­mation. High purity silver: (E) 25% deformation. All reduced 9%

4 mm >-25mm

(b) Lop Welding Die

Fig. 2 — Schematic illustration of the welding dies

W E L D I N G R E S E A R C H S U P P L E M E N T ! 303-s

11000X Fig. 3 — Scanning photograph for aluminum surface after wire brushing. Reduced 41%

found that the rate of applying the pressure does not have a marked effect on either the welding deforma­tion or the weld strength. The weld­ing time (time of applying and re­leasing the pressure) was then set to be 1 min in all the experiments. The experiments were carried out at room temperature.

Deformation Measurements. For lap welding the deformat ion was measured as a percentage reduction in the total thickness of the two strips. For butt welding it was measured as a percentage increase in the cross-sec­tional area.

Measurement of Minimum Welding Deformation. Figure 4 charts pressure and strength vs strain for aluminum. It is seen that there is a stage of easy plastic flow; the onset of this stage corresponds to a sudden increase in the weld strength.

The deformation at which this oc­curs was taken to be the minimum welding deformation. The minimum welding deformation for the other metals was determined in the same way.

Mechanical Testing. Lap jo ints were tested in tensile shear. Due to the relatively high welding deforma­tion of copper and silver, most failures occurred outside the welding at the weld metal junction. Data for lap welding were obtained for alu­minum and gold only. Butt joints were first machined and then tested in ten­sion.

Scanning Electron Microscope In­vestigations. The specimens were viewed in a direction normal to the weld surface. This investigation was carried out at deformations less than

IO 20 30 40 50 DEFORMATION, %

Fig. 4 — Pressure and strength vs strain diagram for aluminum

Sleeve Specimen

Fig. 5 — Schematic illustration of sleeve used to prevent plastic flow at the Inter­face in butt welding

the minimum welding deformation and at deformations higher than the minimum, i.e., investigating the frac­ture surface of the weld.

Exper imental Results and Discussion

Role of the Oxide Film

To investigate the role of the oxide fi lm the following experiments were carried out:

1. In the case of aluminum butt welding, a hard sleeve was set around the rods, as shown in Fig. 5, so as to prevent the metal from deformation at the interface. If the oxide f i lm does contribute to welding, it would be ex­pected that welding can occur in the presence of the sleeve. It was ob­served that whatever the pressure applied in the presence of this sleeve no welding did occur. Figure 6 is a s c a n n i n g p h o t o g r a p h o f t w o aluminum specimens pressed at the same pressure, one in absence of the sleeve (A) and the other in the presence of it (B). By comparing the two photographs it is seen that in the presence of the sleeve the oxide layer did not break.

In lap welding, gold could be weld­ed at zero macroscopic deformation.

2. By welding silver at 200 C, it was found that the minimum lap welding deformation was reduced from about 75% at room temperature to negligible value at 200 C. This behavior cannot be attributed only to a reduction in the flow stress of the metal but is associated with the dis­sociation of silver oxide (it is well known that silver oxide dissociates completely at 190 C at atmospheric pressure).

3. A thin (about 100 angstroms) layer of aluminum was deposited on a scratch brushed gold surface by evaporation in vacuum. The gold specimens were then exposed to the atmosphere so that the deposited layer consisted mostly of aluminum oxide. It was found that no welding did occur up to about 70% deformation.

These results lead to the following conclusions:

1. The oxide film does not con­tribute to welding; in order to initiate welding this fi lm has to be broken.

2. Plastic flow of the metal is a pre­requisite for oxide breakage.

Mechanism of Oxide Film Fracture

It was concluded in the last section that plastic flow of the metal is a pre­requisite for oxide film breakage.

Consider a spot on the interface where the mating surfaces come into contact. Upon applying a deforming pressure the dislocations most favor­ably oriented with respect to the applied stress start to move on their slip planes. These dislocations con­tinue their motion until the outermost

304-s I S E P T E M B E R 1 9 7 5

loops approach the surface where the oxide film may act as a barrier to their emergence. The stress concentra­tion associated with the pile-ups of dislocations can be relieved by either opening a crack in the oxide or mov­ing pre-existing dislocations, and/or generating new ones in the oxide, de­pending on the relative hardness of the metal and its oxide fi lm and the mechanical properties of the oxide. If the surface is covered with an oxide film harder than the metal, the dislo­cations experience an image force which is a repulsion reflecting the strain energy of the elastically harder material. The stress concentration can be relieved by opening a crack in

the oxide, the oxide ultimately failing in a brittle tensile manner. Evidence for the pi l ing up of d is locat ions against surface oxide fi lms has been provided by Barrett (Ref. 9).

Figures 1A and 1B are scanning photographs for high purity a lumi­num surfaces at two deformations below the minimum welding defor­mation. It is seen that the surface has discontinuities which may corre­spond to cracks in the oxide fi lm. The direction of propagation of the crack is seen to be normal to the applied stress which is a character­istic feature of brittle tensile failures. If, on the other hand, the surface is covered with a deformable oxide fi lm,

Fig. 6 — Scanning photographs of aluminum surface (A) in absence and (B) in presence of the sleeve. Reduced 48%

Table 1 — Hardness of Metals and Their Oxides at Room Temperature

Hardness Hardness Metal HV Oxide HV

Al Cu Ag Au

15 40 26 20

Al203

Cu 20 Ag 20

—

1800 160 135

— (a) Original Interface

Table 2 — Minimum Lap Welding Defor­mation at Room Temperature

Defor-Metal mation, %

^ ^

(b) Fracture of Brittle Oxide Film

High purity Al Commercial Purity Al High purity Cu High purity Ag High purity Au

10 30 64 75

0

Table 3 — Minimum Butt Welding Defor­mation at Room Temperature

(c) First Requirement for Welding I Formotion of Overlopped Ox ide- f ree Metallic Areas

VVVV LVXCVZ

Metal

High purity Al 6061T6AI alloy

Defor­mation, %

130 227

(d) Second Requirement for Welding: Extrusion of the Metol through the Gops Creoted in the Oxide ond some Relative Shear Displacement at the Points of Contact of Oxide-free Metal

Fig. 7 — Schematic illustration of the weld­ing process

dislocations will also multiply in the oxide. In that case the metal and ox­ide undergo plastic deformat ion together. This behavior is possible when the oxide film is relatively duc­tile. In this case, the oxide may fail in a shear manner. It is seen in Figs. 1C and 1D, which are scanning photo­graphs for a copper surface at two deformat ions below the m in imum welding deformation, that the dis­continuities are quite different from those on aluminum surfaces. They are in the form of steps or striae which may suggest that the oxide f i lm on copper fails in a shear manner. Figure 1E shows the surface of silver, the dis­continuities are similar to those on copper surfaces; therefore, silver ox­ide also fails in shear.

Table 1 gives the relative hardness of the metals investigated and their oxides. If the oxide fi lm is completely brittle at the welding temperature, then the proportions of oxide-free metallic area formed at a certain deformation is equal to the surface extension at that deformation. This is not the case if the ox ide is de­formable. It may be concluded that the proport ion of oxide-free metallic area revealed is dependent upon the relative hardness of the metal and its oxide fi lm, and the mechanical prop­erties of the oxide.

The First Requirement for Welding

It was found from scanning electron microscope photographs that the ox­ide films on aluminum, copper and silver crack at deformations much less than the minimum welding defor­mation (see Fig. 1). This shows that initiation of welding is not deter­mined only by the fragmentation of the oxide film as was suggested in the simple film theory. Table 2 shows the minimum lap welding deformations for the metals investigated at room temperature . Table 3 shows this deformation for butt welding.

In the mechanism of oxide fi lm fracture presented in the last section it would not be expected that disloca­tion pile-ups at both sides of the inter­face would have the same distr ibu­tion and would always line up. There­fore, it seems likely that the oxide fi lm on the two surfaces should break independently not as one layer as was suggested by Vaidyanath and Milner (Ref. 10). Although a metallic area is revealed on one surface, yet, the cor­responding area on the other surface may still be covered by an oxide. Welding cannot commence unless freshly revealed areas overlap one above the other as shown sche­matically in Fig. 7.

This may explain why there is a considerable additional deformation required between the first initiation of cracks in the oxide and the initiation of welding. Therefore, the first weld-

W E L D I N G R E S E A R C H S U P P L E M E N T ! 305-s

ing may occur after the cracks begin to overlap.

Fracture Surface of the Weld

Figures 8 to 13 are scanning photo­graphs of the fracture surface of welds. It is seen that the fracture sur­faces of all the metals investigated have the same feature. They are made of concave depressions which is the prominent feature of fracture created by coalescence of voids (dim­ple rupture) , (Refs. 11 , 12). The dimples are seen to be equiaxed which is typical of normal rupture. Each of the voids of the interface

represents an internal crack. During mechanical testing these voids grow until the material between two voids thins down and separates by rup­turing. Figure (14) is a macrophoto-graph for the surface of a butt welded aluminum specimen just before fai l­ure; deformation bands can be seen.

Role of Surface Roughness

The fact that initially rough sur­faces are required for welding sug­gests that bringing oxide free metals into contact does not result in weld­ing unless there is also some shear displacement as the two surfaces

1000X

Fig. 8 — Fracture surface of lap welded high purity aluminum. (A) 29% deformation, (B) 46% deformation. Reduced 50%

come into contact. This was con­f irmed by butt welding gold with dif­ferent methods of surface prepara­t ion; gold was selected due to the complete absence of the oxide fi lm. It was found that, at the pressures where weld ing d id occur in wire brushed specimens, highly polished specimens degreased in acetone did not weld. In lap welding of gold at zero macroscopic deformation there was a critical pressure for welding although the macroscopic deformation is ex­cluded. This behavior should be asso­ciated with the local deformation of high spots since there should be a minimum applied pressure neces­sary to deform these spots. Agers and Singer (Ref. 13) suggested that in lap welding the local deformation at the interface is more important than the macroscopic deformation.

It seems that the importance of the shear d isp lacement , besides in ­creasing the contact area, is that it de­stroys the continuity of any adsorbed oxygen layer which may contaminate the oxide-free area due to trapped air at the interface.

Second Requirement for Welding

The second requirement for weld­ing is to force the metal to flow through the gaps created in the oxide. This flow will result in welding when the metal flowing from one side of the interface comes into contact with the metal flowing from the other side and then some relative shear displace­ment occurs to destroy the continuity of the adsorbed oxygen monolayers

Jam l30"" TB^F^ ,p7ooor Fig. 9 — Fracture surface of lap welded commercial purity alu­minum. (A) 45% deformation, (B) 58% deformation. Reduced53%

F0£&l

Fig. 10 — Fracture surface of butt welded high purity aluminum. (A) 130% deformation, (B) 360% deformation. Reduced 55%

306-s I S E P T E M B E R 1 9 7 5

which probably contaminate the sur­face as a result of the t rapped air at the interface.

High stacking fault energy would be expected to facilitate extrusion of the metal through cracks because of easy cross-slip of glide dislocations.

Except for gold, aluminum, which has high stacking fault energy (~110 ergs/cm2), smallest hardness ratio (Hmetai/Hoxide) and possesses a com­pletely brittle oxide f i lm, was the easiest metal to weld. Although gold has low stacking fault energy, it is easier to weld than aluminum be­cause the first requirement of weld­ing is satisfied everywhere due to the complete absence of the oxide. It is only necessary for deformation to in­crease the area of contact.

Theoretical Calculation of the Weld Strength

Assume that the oxide f i lm is com­pletely brittle, so the metallic area re­vealed, which could be welded, is equal to the amount of surface exten­sion that occurs after the surfaces are in intimate enough contact to ex­clude oxygen.

In butt welding the experimentally measured extension R is given by

R = A - Ao

Ao

where A0 is the original cross-sec­tional area and A is the instanta­neous area at extension R. The true fractional metallic area revealed at a certain extention R is then:

AA

A

R

R + 1

In case of lap welding, if we divide the surface area A0 into small areas A, (i = 1, 2, 3, . . .N) then A0 = NAi and upon applying a deforming stress, the area A; extends to be A, + AAj where

Ai (x, +x,)(y, + y,)-x,y,

A , X I V I

Ax, and Ay, are the extensions in the x and y directions respectively and therefore:

AA; Ax +

^ V i

From the first law of plasticity:

A X i -Wt Az, + =0

Therefore:

AA i Azi

(minus sign means that the strain is compressive)

The total measured reduction

R

and

N = 2-

1

A A ,

At

AA

Ac

the revealed

i

N

1

R

AZ

Z i

true fraction is g liven by:

al

and AA

~ A ~ R + 1 (2)

metallic area

which is in the same form as that for butt welding, Eq. (1). Now, if welding is assumed to occur whenever clean metal surfaces come into contact, then the fractional welded area, f, should be given by:

(D I

Fig. 11 33%

1000X Fracture surface of lap welded high purity copper, 64% deformation. Reduced

Fig. 12 — Fracture surface of lap welded high purity silver, 75% deformation. Reduced 50%

W E L D I N G R E S E A R C H S U P P L E M E N T ! 307-s

Jems

" 1 0 0 0 X * * S * e l ^ * »l

F/g. 13 — Fracture surface of lap welded high purity gold. (A) 3% deformation, (B) 10% deformation. Reduced 50%

Fig. 14 — Macrophotograph tor the surface ot butt welded aluminum specimen just before failure in tension. Reduced 42%

—) + 1 '

(3)

where C is a proportionality constant. If there is a complete matching be­tween the metallic areas over the in­terface, f has its maximum value given by Eq. (1) or (2), i.e.,

R + 1

but if there is no correlation, f should be given by Eq. (3) if the strength of the weld is Sw and the strength of the metal is Sm, then

SW

Sm = fn • f — y

V R + 1 / (4)

where 1 < n < 2. It would be expected that f cannot attain this maximum value because flow of the metal will be restricted at the edges of the oxide gaps.

Experimental values of (f) were ob-ained from tensile tests, where

fexp —

true ultimate of the weld

tensile strength

true ultimate tensile strength of the metal

The average fractional metallic areas revealed were estimated from the fractographs. The true welded area cannot be est imated direct ly be­cause of the deformation of the weld­ed region which occurs during me­chanical testing. The fractional weld­ed area

R R + 1

was calculated from the estimated fractional revealed metallic area

0.6

i r

— Calculated o High Purity Aluminum A 6061 T6 Aluminum Alloy

Deformation (R)

Fig. 15 — Fractional revealed metallic areas estimated from fractographs vs deformation in butt welding of aluminum

0.4

0.2 -

I

—

-

- y

^ / I

I I

A A r

A ^ . CT~

^^o°

Calculated

O High Purity Aluminum

A 606I T6 Alluminum Alloy

I I

I

A

o_ O—-—

-

-

I

Fig. 16

I 2 3

Deformotion (R)

Weld strength vs deformation of butt welded aluminum

308-s j S b H I b M B E H 1 9 7 5

0.4

| 0.3

S 0 . 2 -

•2 0.1 -

-

1

/ o

1

1

/ o

1

1 1 1

— Calcu la ted O High Purity Aluminum

A Commercial Purity Aluminum

1 1 1

-

0.15

0.10

5 0.05 -

0

I I I

— Calculated O High Purity A luminum

A Commercial Purity Aluminum

- \ 1 1

1

1

1

A

1

A

0 6 0.1 0.2 0.3 0.4

Deformation (R) 0.5 0 6

0 0.1 0,2 0.3 0.4 0.5 Oeformation (R)

Fig. 18 — Weld strength vs deformation of lap welded aluminum Fig. 17 — Fractional revealed metallic areas estimated from frac­tographs vs deformation in lap welding of aluminum

R + 1

Figures 15 to 18 show that:

(a) The experimental values of the fractional metallic areas revealed are in agreement with those calculated theoretically f rom R/(R+1).

(b) The measured strength of the weld is in agreement with that calcu­lated theoretically f rom Eq. (3) with n = 2 using the measured welded area.

(c) The proportionality constant C in Eq. (3) is almost constant and inde­pendent of deformation. It has a value of the order of 0.7-0.8.

(d) The experimental values of the weld st rength are comparab le to those calculated on the basis of un-corre lated crack ing of the oxide layers.

It may be concluded from these results that cracks in the oxide layers are uncorrelated and that the weld strength, as a fraction of the strength of the metal, attained after a given deformation is determined by the fractional welded area at that defor­mation according to the relation:

f = C — Y R + 1 /

With reference to gold, the formula which gives the revealed fractional metal l ic area whi le assuming the presence of a brittle oxide fi lm, is not useful. For gold, the area of clean metal in contact will depend only on the initial roughness of the surface and on the deformation of the high spots as the pressure increases. Figure 19 shows the variation of fexp obtained from mechanical tests with deformation.

The Parameter C. It was shown that some relative shear displacement at the points of contact of oxide-free metal is required for welding and that an initial rough surface facilitates this displacement. These points of con-

0.30

0.25 —

0.20

Fig. 19

0.2 0.3

Deformation (R)

Weld strength vs deformation of lap welded gold

0 5

tact at which the stress is highly con­centrated may be cons idered as nuclei of the weld which grow with deformation, and as the number of these nuc le i i nc reases , we ld ing becomes easier. It was also shown that C is almost independent of the deformation.

It is proposed that C depends on the method of preparation of the sur­face, highly rough surface gives high value of C and vice versa. Therefore, one of the principal functions of wire brushing is to increase the param­eter C.

Conclusions

1. Metal to oxide bonding does not con t r i bu te s ign i f i can t l y to weld strength.

2. Initiation of welding of a given metal is determined by cracking of the oxide fi lms which permits contact to gradually develop between clean metal surfaces. Localization of slip into heavy slip bands may be the cause of uncorrelated cracking of the

two oxide layers. 3. Factors that affect the relative

difficulty of pressure welding are: (a) stacking fault energy of the metal, (b) relative hardness of the metal and its oxide film and the mechanical proper­ties of the oxide, and (c) surface roughness prior to welding.

4. The weld strength attained after a given deformation is determined by the fractional area of clean metal sur­face that has been brought into con­tact at that deformation.

5. Some relative shear displace­ment at the points where clean metal surfaces come into contact is neces­sary for welding.

Acknowledgment

This work was done under the auspices of the U.S. Atomic Energy Commission.

References

1. Tylecote, R. F., The Solid Phase Welding of Metals, St. Martin's Press, 1968, pp. 38, 57, 201.

2. Tylecote, R. F., British Welding Jour-

W E L D I N G R E S E A R C H S U P P L E M E N T ! 309-s

nal, Vol. 1, 117, 1954. 3. Vaidyanath, L. R. and Milner, D. R.,

British Welding Journal, Vol. 7, 1, 1960. 4. Donelan, J. A., British Welding Jour­

nal, Vol. 6, 5, 1959. 5. Bowden, F. P. and Tabor, D., Struc­

ture and Properties of Solid Surfaces, Gomer, R. and Smith, C. S., eds.

6. Semenov, A. P., Wear 4, 1, 1961. 7. Parks, J. M., Welding Journal, Vol.

32, 5, May, 1953, Res. Suppl., pp. 209-s to 222-s.

8. McEwan, K. M. B. and Milner, D. R., British Welding Journal, Vol. 9, 406, 1962.

9. Barrett, C. S., Acta Met. 1, 2, 1953. 10. Vaidyanath, L. R., Nicholas, M. G.

and Milner, D. R., British Welding Journal, Vol. 6, 13, 1959.

11. Beachem, C. D. and Pelloux, R. M. N., "Electron Fractography, A Tool for the

Study of Micromechanisms of Fracturing Processes," in Fracture Toughness Testing and Its Applications, STP 381, ASTM, Philadelphia, p. 210.

12. Telelman, A. S. and McEvily, A. J. Jr., "Fracture of Metals," in Fracture, An Advanced Treatise, Liebowitz, H., ed., Academic Press, 1969, p. 156.

13. Agers, B. M. and Singer, A. R., Brit­ish Welding Journal, Vol. 11, 313, 1964.

WRC Bulletin

No. 187 Sept. 1 973

"High-Temperature Brazing"

by H. E. Pattee

This paper, prepared for the Interpretive Reports Committee of the Welding Re­search Council, is a comprehensive state-of-the-art review. Details are presented on protective atmospheres, heating methods and equipment, and brazing proce­dures and filler metals for the high-temperature brazing of stainless steels, nickel base alloys, superalloys, and reactive and refractory metals. Also included are an extensive list of references and a bibliography.

The price of WRC Bulletin 187 is $5.00 per copy. Orders should be sent to the Welding Research Council, 345 East 47th Street, New York, N.Y. 10017.

WRC Bulletin No. 197

August 1974

"A Review of Underclad Cracking in Pressure-Vessel Components"

by A. G. Vinckier and A. W. Pense

This report is a summary of data obtained by the PVRC Task Group on Under­clad Cracking from the open technical literature and privately sponsored re­search programs on the topic of underclad cracking, that is, cracking underneath weld cladding in pressure-vessel components. The purpose of the review was to determine what factors contribute to this condition, and to outline means by which it could be either alleviated or eliminated. In the course of the review, a substantial data bank was created on the manufacture, heat treatment, and clad­ding of heavy-section pressure-vessel steels for nuclear service.

Publication of this report was sponsored by the Pressure-Vessel Research Committee of the Welding Research Council. The price of WRC Bulletin 197 is $5.50. Orders should be sent to the Welding Research Council, 345 E: 47th St., New York, N.Y. 10017.

310-s I S E P T E M B E R 1 9 7 5