nelson stud operation

Upload: johnknight000

Post on 03-Jun-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 NELSON stud operation

    1/11

    ABSTRACT. Microstructural evolution atthe fusion boundary in dissimilar weldsbetween ferritic and austenitic alloys cansignificantly influence both the weldabil-ity and service behavior of the dissimilarcombination. A fundamental investiga-tion was undertaken to characterize fu-sion boundary microstructure and to bet-ter understand the nature and characterof boundaries that are associated withcracking in dissimilar welds. In a previ-ous paper, the evolution of the fusionboundary during the onset of solidifica-

    tion was discussed. In this paper, the na-ture and evolution of the fusion bound-ary and surrounding regions in dissimilarmetal welds during subsequent on-cool-ing transformations in the fusion zoneand heat-affected zone (HAZ) will be dis-cussed.

    A model system consisting of a high-purity iron base metal and 70Ni-30Cu(AWS A5.14 ERNiCu-7) filler metal wasused to study this behavior. Using thissimple Fe-Ni-Cu system, fusion bound-ary microstructures were developed thatwere analogous to those observed in

    more complex engineering systems. Transmission electron diffraction analy-sis and orientation imaging microscopy(OIM) revealed the orientation relation-ships between adjacent HAZ and weldmetal grains at the fusion boundary weredifferent than the cube-on-cube relation-ship normally observed in similar metalwelds. The room temperature fusion

    boundary in the system studied exhibitedgrain boundary misorientations consis-tent with common FCC/BCC relation-ships, i.e., Bain, Kurdjumov-Sachs andNishyama-Wassermann. A theory de-scribing the evolution of the fusionboundary is proposed and the nature andcharacter of the Type II grain boundaryis described.

    Introduction

    Cracking phenomena associated with

    welds have been a recurring problem thathas received considerable attention bymany researchers over the last fourdecades. It is well accepted that weld-related cracking normally occurs alonggrain boundaries. Such grain-boundary-related cracking phenomena includeweld solidification cracking, weld metalliquation cracking (microfissuring) inmultipass welds, HAZ liquation crack-ing, reheat (stress relief) and strain-agecracking, and ductility dip cracking inboth the weld metal and HAZ. Unfortu-nately, the materials that exhibit the

    greatest propensity for these crackingphenomena are those that are vital to thenational infrastructure, i.e., aluminum al-loys, nickel-based alloys and stainlesssteels.

    Cladding or dissimilar metal weldshave by no means been immune to suchfailures and exhibit some unique crack-ing phenomena not observed in weldsbetween similar materials. In fact, crack-ing or disbonding along or near the fu-sion boundary in dissimilar ferritic-austenitic welds has been a persistent

    problem for more than 60 years. Despitethe persistence and potential conse-quences, the evolution, nature and roleof weld metal interfaces in promoting ormitigating weld-related cracking are notwell understood. The implications of boundaries and structures with regard tocrack growth rates, fatigue, stress corro-sion cracking, etc., have been researchedextensively in the materials sciencearena. However, in spite of the recurringproblems and economic losses, there ex-ists a lack of understanding regarding therole of boundaries and structures in pro-

    moting or mitigating weld-related crack-ing. Therefore, a fundamental investiga-tion was undertaken to investigate thenature and character of those boundariesand structures near the fusion boundaryin dissimilar ferritic-austenitic welds.

    A previous paper (Ref. 1) addressedthe nature and character of the elevatedtemperature fusion boundary at the onsetof solidification. This paper will presentin detail the effects of on-cooling trans-formations and the nature of the fusionboundary and surrounding microstruc-ture within the austenitic and alpha fer-rite temperature ranges.

    WELDING RESEARCH SUPPLEMENT |267-s

    W E L D I N G R E S E A R C H

    SUPPLEMENT TO THE WELDING JOURNAL , October 2000Sponsored by the American Welding Society and the Welding Research Council

    Nature and Evolution of the Fusion Boundary inFerritic-Austenitic Dissimilar Metal Welds

    Part 2: On-Cooling Transformations

    BY T. W. NELSON, J. C. LIPPOLD AND M. J. MILLS

    On-cooling transformations in the fusion zone and heat-affected zone were studiedfor their relation to crack initiation

    KEY WORDS

    Dissimilar WeldsGrain BoundaryCrack GrowthOn-CoolingPhase TransformationFerritic-AusteniticWeld Cracking

    T. W. NELSON, formerly with The Ohio StateUniversity, is now with Brigham Young Uni-versity in Provo, Utah. J. C. LIPPOLD and M.

    J. MILLS are with The Ohio State University,

    Columbus, Ohio.

  • 8/12/2019 NELSON stud operation

    2/11

    Background

    Despite the widespread application of

    dissimilar metal joining, there is a historyof weld-related failures associated withdissimilar joints. In the power generationindustry, dissimilar welding or claddingis used in nuclear steam generator chan-nel heads, pressurized water reactors,tube sheets and reheat piping. Claddingin these applications is primarily used asan erosion/corrosion inhibitor in an ele-vated-temperature, high-velocity steamenvironment. The petrochemical indus-try also utilizes dissimilar welding/cladding for similar benefits. Claddingwith austenitic stainless steel for corro-

    sion resistance is used extensively inhydro-desulfurization reactors and otherpressure vessels (Refs. 213). For theseparticular applications, austenitic stain-less steels or nickel-based alloys are cladonto a carbon or Cr-Mo steel substrate toinhibit corrosion and high-temperature

    oxidation, save ma-terial costs and pro-long the life of thecomponent. How-ever, in some in-stances, these weldcladdings have failedcatastrophically nearthe weld fusionboundary. Failuressuch as these cost in-dustry millions of dollars each year inrepair or replace-ment (Refs. 14, 15).

    Research activitiesand reported failuresassociated withDMW date back tothe early 1930s(Refs. 14, 15). Of the

    failures reported, whether produced inthe laboratory or in actual fabrications,cracking has been observed in several lo-

    cations near the fusion boundary region. These locations include 1) the weld fu-sion boundary, 2) the weld heat-affectedzone, and 3) grain boundaries in the weldmetal oriented parallel to the fusionboundary. These parallel boundaries,sometimes referred to as Type II bound-aries, are shown schematically in Fig. 1,labeled Type B. Many of the failures re-ported occur along the Type II bound-aries, which tend to be oriented parallelto the fusion boundary and less than 100m away from the fusion boundary. Thismorphology is contrary to that observed

    in similar metal welds, in which solidifi-cation grain boundaries are orientedroughly normal to the fusion boundaryand are the extension of HAZ grainboundaries at the fusion boundary as aconsequence of the epitaxial nature of solidification nucleation.

    The dissimilar test sample shown inFig. 2 was made on ASTM A508 basemetal, clad with Type 309L stainless steelon the first pass, followed by Type 308Lstainless steel on the second pass, andsubsequent passes with AWS ENiCrFe-3.Shielded metal arc welding (SMAW) wasused as the welding process in this testsample. In this figure, the Type II bound-aries can be seen running roughly paral-lel to the fusion boundary. This morphol-ogy is unconventional relative to similarmaterial combinations and has shown tohave important implications with respectto both fabricability and structural in-tegrity. From Fig. 3, it can be observedthat the fracture path follows along the

    Type II boundary. This is the fracture pathobserved by many investigators (Refs. 24)when investigating clad disbonding.

    Much of the research to date has in-vestigated the effect of hydrogen on TypeII boundary failure or implicated hydro-gen-induced cracking as part of the

    mechanism (Refs. 215). Most of theseinvestigations have been successful atpromoting cracking in dissimilar metalwelds by cathodically charging samplesto very high hydrogen concentrations.However, some have proposed that it isthe very high pressures that promotecracking in these investigations, and thathydrogen, although a contributor, is notthe underlying mechanism (Ref. 8). Otherinstances of Type II boundary crackingcannot be related to hydrogen-inducedcracking because of the absence of hy-drogen in the fabrication or service envi-

    ronment. Although considerable workhas attempted to determine the underly-ing mechanisms responsible for this formof weld cracking, little research has fo-cused on understanding the nature andevolution of those weld metal interfacesalong which cracking occurs.

    268-s | OCTOBER 2000

    RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT

    Fig. 1 Schematic representation of fusion boundary morphologiesobserved in DMWs (Ref. 1).

    Fig. 2 Weld metal microstructure in clad component. Fig. 3 Fracture profile showing failure along Type II boundary in an actual clad test component.

  • 8/12/2019 NELSON stud operation

    3/11

    Although not well understood, the re-gion of the weld near the fusion bound-ary has proven to have important impli-cations on the failures associated withDMWs. It has been hypothesized (Refs.216) that microstructural and chemicaltransitions exist from the fusion boundaryacross these Type II boundary regions. Anumber of factors support the existenceof this transitional region: 1) differentcrystal structures between the body-centered cubic (BCC) ferritic base metaland the face-centered-cubic (FCC) weldmetal; 2) diffusional mixing of alloying

    and impurity elements from the weldmetal into a stagnant boundary layer ad- jacent the fusion boundary; 3) changingbase metal dilution (BMD), which affectsthe composition gradient in the weldmetal at the fusion boundary; and 4) thediffusion and growth kinetics during mul-tipass welds and long postweld heattreatments (PWHT).

    This transition region, its crystal struc-ture and composition, can ultimatelygovern the ability of a welded compo-nent to be successfully fabricated andperform as engineered in its intended ser-vice environment.

    Few previous investigators have ad-dressed or proposed any mechanisms re-garding the nature and evolution of the

    Type II boundary. Recently, one hypoth-esis was developed by Wu, et al. (Ref.16), with regard to the formation of the

    Type II boundary. They proposed the Type II boundary was a result of a changein the primary mode of solidification. Ac-cording to their hypothesis, primary so-lidification from the body-centered-cubic (BCC) substrate occurs as ferrite(BCC), and, at some distance from the fu-sion boundary, the mode changes to a

    primary austenite as a result of the in-creasing amount of austenite stabilizingelements toward the center of the weld.

    This solidification mode change wouldcreate a Type II boundary. The distance atwhich the change in solidification modewould occur would be governed by theslope of the composition gradient withinthe transition region adjacent the fusionboundary. Other than this theory, there isli ttle information in the literature regard-ing the evolution of the fusion boundaryin DMWs and Type II boundaries. There-fore, a fundamental investigation was un-

    dertaken to investigate the nature andcharacter of those boundaries and struc-tures near the fusion boundary in dissim-ilar metal welds.

    Experimental Approach

    Material Selection

    A simple ternary system that wouldrepresent the solidification and phasetransformation behavior of more com-plex dissimilar alloy combinations wasselected for this investigation. High-purity iron was used as a base metal as itexhibits both the delta-gamma andgamma-alpha transformations that occurin C-Mn and low-alloy structural steels.A 70Ni-30Cu, single-phase (FCC) binaryalloy with low impurity content was se-lected as the filler metal. The chemicalcompositions of both the base and fillermetals are listed in Table 1.

    Welding Conditions

    Gas tungsten arc welding (GTAW)using a cold wire and argon shielding gaswas used for producing single-pass bead-

    on-plate welds. Current, voltage andtravel speeds were held constant for allwelds at 250 A, 11 V and 6 in./min (2.5mm/s), respectively. Wire feed speed wasused as the primary variable for control-ling and changing the base metal dilution(BMD), where BMD defines the percentbase metal composing the weld metal.Wire feed rates ranged from 10 to 100in./min (4.242.3 mm/s), producingBMDs ranging from 1080%.

    Microstructural Characterization

    Numerous transverse and plan viewsamples were removed from welds of various BMDs for metallographic analy-sis. Standard metallographic techniqueswere used to prepare optical and scan-ning electron microscopy (SEM) samples.Because of the different characteristics of the base and weld metal, two etchants

    WELDING RESEARCH SUPPLEMENT |269-s

    Fig. 4 BSE photomicrograph showing the Type II boundaries (indi-cated by arrows) extending beyond the martensitic transition region at the fusion boundary into the fully austenitic weld metal (65% BMDweld in Fe/70Ni-30Cu).

    Fig. 5 SE photomicrograph showing Type II boundaries (indicatedby arrows) running parallel to the entire fusion boundary (48% BMDweld in Fe/70Ni-30Cu).

    Table 1 Chemical Composition of Baseand Filler Metal

    Element Iron ERNiCu-7 ERNiCu-7actual AWS A5.14

    Fe Bal. 0.25 2.5 MaxNi 65.57 62.069.0Cu 27.84 RemainderCr 0.015 Not Specif.C 0.020 0.048 0.15 MaxMn 0.320 3.58 4.0 MaxSi 0.010 0.85 1.25 Max

    Ti 1.99 1.53.0Al 0.02 MaxS 0.013 0.0003 0.015 MaxP 0.010 0.02 MaxNb+Ta 0.005 Not Specif.Creq 0.0 1.3 N/ANi eq 0.8 68.8 N/A

  • 8/12/2019 NELSON stud operation

    4/11

    were used for metallographic prepara-tion of samples. These included 1) 4%nital, which etches the base metal; and 2)an electrolytic solution of 5 g Fe 3Cl, 2 mL

    of HCL and 99 mL of methanol for theweld metal. Optical microscopy andscanning electron microscopy (SEM)were used for microstructure characteri-zation of structures and boundaries alongthe fusion boundary. Optical metallogra-phy was performed at magnifications upto 400X and SEM analysis up to 1600X.

    Transmission Electron Microscopy Analysis

    Discs were removed along the fusionboundary from transverse sections usinga 3-mm-diameter disc punch. Standardgrinding and dimpling techniques were

    utilized for reducingthe thickness of thedisks. Following dim-pling, samples wereion milled from bothsides until perforationof the foil wasachieved. Transmissionelectron microanalysis(TEM) was performedusing standard bright-field, two-beam andmicrodiffraction condi-tions (Ref. 17).

    Electron BackscatterAnalyses

    Electron backscatterdiffraction analyseswas performed on both

    TEM thin foils and bulkmetallographicallyprepared samples.

    Samples were loaded in the SEM at anangle of 70 deg from the incident beamtoward a phosphor detector. The electronbeam was then rastered across the sam-

    ple in a hexagonal grid at specified in-crements. Kikuchi signals were automat-ically analyzed using the OrientationImaging Microscopy (OIM) software(Refs. 1820). This software calculatesthe Euler angles of the electron backscat-ter diffraction (EBSD) patterns with refer-ence to the sample normal, then storesthe position and angles of each pattern.

    This data was then used for various grainboundary and texture analysis.

    Results

    The ability of the base and filler metal

    combination used in this investigation toreproduce those microstructures com-monly observed in engineering materialshas been demonstrated previously (Ref.21). The microstructures observed alongthe fusion boundary exhibited a fullymartensitic microstructure at high BMDs(>80%), a mostly austenitic weld metalmicrostructure with a band of martensitealong the fusion boundary at mediumBMDs (80%>BMD

  • 8/12/2019 NELSON stud operation

    5/11

    ary migration or transformation product,as observed in Fig. 7. This is usually a re-sult of the composition differences be-tween the cell cores and cell boundariesthat change the nature of etching or trans-formation between these two regions.

    As mentioned previously, it seemed asif the Type II boundaries traverse severalSGBs even though the solidification sub-structure was not evident within the TypeII grains. In Figs. 7 and 8, it is evident theoriginal SGBs extend to the fusion bound-ary as indicated by the arrows. Althoughthe crystallographic portion of the SGBs,referred to as migrated grain boundaries,

    was eliminated by the Type II boundary,compositional traces are still evident bydifferences in etching characteristics.Each SGB represents the boundary be-tween two weld metal grains, therefore,as observed in these figures, the Type IItraverses several weld metal grains.

    Similar results are shown in Fig. 8,where the transformation product alongthe cores of the original solidification cellstructure delineates two different solidifi-cation grains. This is characterized by thedifference in angles subtended betweeneach, and normal to the fusion boundary

    interface, as illustrated in this figure. These results indicate the Type II bound-ary migrated from the fusion boundaryinto the weld metal within the austenitetemperature regime. Although the crys-tallographic portion of the SGB has beeneliminated by the Type II boundary, therestill exists a compositional difference be-tween the matrix and a boundary,whether it is a SGB or a solidification sub-grain boundary. This produces the differ-ential etching between matrix and grainboundary or prior grain boundary andmartensite along cell cores as has beenshown in Fig. 8.

    Although it is difficult to locate a spe-cific boundary in a TEM foil, with the as-sistance of OIM analysis and careful ionmilling, several Type II boundaries were

    located and analyzed. Intersection of a Type II and solidification grain boundary,shown in Fig. 9, was observed in a 48%BMD weld (same sample used for OIManalysis). The Type II boundary exhibitsno particular details that would distin-guish it from any other grain boundary inthe weld metal, other than it is orientedroughly perpendicular to SGBs in theweld metal. However, it is interesting tonote the discontinuity of the Type IIboundaries when they intersect SGBs.

    This feature was observed in many of theoptical and TEM photomicrographs.

    Despite the analysis above, this doesnot present any statistical basis for gener-alizing all orientation relationships alongthe fusion boundary. The possible grainboundary pairs/misorientations alongany given fusion boundary are numer-ous. To determine any statistically pre-ferred misorientations at the fusionboundary using TEM poses a daunting, if not impossible, task. Therefore, OIManalyses were employed as a more effi-cient means to obtain and evaluate grainboundary information along the fusionboundary.

    Several samples were analyzed by

    OIM to evaluate larger regions of the fu-sion boundary. Although very useful,OIM could not be performed on all sam-ples by virtue of the ability to resolve in-

    dividual backscatter patterns from veryfine structures, i.e., martensite. There-fore, it was not possible to perform thisanalysis on those samples where marten-site was prominent along the fusionboundary.

    The OIM results obtained from weldsproduced at 48% BMD, which exhibits afully austenitic weld metal microstruc-ture, are shown in Figs. 1012. A repre-sentative grain map reproduced fromdata collected over a region approxi-mately 1440 x 1525 m with a step sizeincrement of 5 m is shown in Fig. 10.

    Grain maps are similar to a photomicro-graph of a polished and etched sample;however, large and small angle grainboundaries are delineated in the OIMmap by the thickness of the lines. Lighterlines represent misorientations between5 and 15 deg, and heavier lines representmisorientation greater than 15 deg. Oncedrawn, these maps provide an interactivetool for extracting various lattice andgrain boundary information from loca-tions of interest. For this investigation, re-gions of interest include the fusionboundary and Type II boundaries.

    Intensity plots of misorientations gen-

    WELDING RESEARCH SUPPLEMENT |271-s

    Fig. 9 TEM photomicrograph showing intersection of SGB(white arrow) and Type II boundary (indicated by black ar-rows) (48% BMD weld in Fe/70Ni-30Cu).

    Fig. 10 Image quality grain map of fusion boundary region in 48% BMD weldin Fe/70Ni-30Cu. Note the Type II boundaries decorating the entire fusionboundary.

  • 8/12/2019 NELSON stud operation

    6/11

    erated from data sets are presented in Ro-drigues-Frank space in Figs. 11 and 12.Misorientations of the entire data set ex-hibit a relatively random distribution of grain boundary misorientations withsome higher densities observed toward45 deg about (lower right cornerof r 3 = 0.011 plot), 27 deg @ , 45deg about (upper right corner of

    r3 = 0.011 plot) and 60 deg @ (upper right hand corner of r 3 = 0.300plot), as shown in the intensity plots inFig. 11.

    Using the grain map in Fig. 10, mis-orientations between HAZ and weldmetal grains at the fusion boundary wereanalyzed. The highlighted points (indi-cated in yellow) in Fig. 10 represent thoselocations at which individual misorienta-tion data were taken. Misorientation dis-tributions were calculated from thesedata and plotted in Rodrigues-Frankspace as intensity plots. This plot shows

    tendencies toward 35 deg @ and

    45 deg @ indi-cated by the increase inintensity at the upperright-hand corner alongthe diagonal (),and at the lower right-hand corner of plots r 3= 0.0110.056, respec-tively, in Fig. 12.

    As various orienta-tion relationships maybe represented in Ro-drigues-Frank space, itis important to pointout those of particularinterest to the presentinvestigation. As theangle/axis pairs for theBain, K-S and N-W ori-entation relationships

    are approximately 45 deg @, 35deg @ and 45 deg @, re-spectively, these correspond directly tolocations within the Rodrigues-Frankspace plots. As the diagonal in r 3=0.011represents the rotation axis, theK-S and N-W orientation relationshipswould lie at roughly 35 and 45 deg to-ward the upper right-hand corner alongthis axis. Likewise, the horizontal line inthis plot corresponds to the rota-tion axis, and the Bain orientation rela-tionship lies at the lower right-hand cor-ner, i.e., 45 deg @. Therefore, thestronger tendency exhibited in Fig. 12would be toward the Bain orientation re-lationship, indicated by the higher inten-sity. This same trend was observed in sev-eral other samples of different BMDs.

    Discussion

    In similar metal welds, growth of the

    solid in the molten weld pool is initiated

    by arranging atoms in the liquid phase onthe existing crystalline substrate, therebyextending it without altering the crystal-lographic orientation (Refs. 2224). As aresult, misorientations between adjacentHAZ grains are continuous across the fu-sion boundary into the weld metal, cre-ating solidification grain boundaries(SGBs) as solidification proceeds. Thus,there is no crystallographic misorienta-tion between a weld metal grain and theHAZ grain from which it grew, producinga cube-on-cube relationship, i.e.,{100}//{100}, // for cubicmaterials. In an autogenous weld onhigh-purity iron, the original solid to formwould exhibit this orientation relation-ship with the HAZ grains from which itgrew. However, it becomes increasinglydifficult to maintain this relationship as adissimilar filler metal is introduced, cre-ating differences in composition, latticeparameter, structure, etc., as would bethe case in most DMWs

    Nature of the Fusion Boundary within theAustenitic Temperature Range

    In a previous paper (Ref. 1), the natureand character of the fusion boundarywithin the -ferrite temperature rangewas discussed. This paper mainly ad-dressed the nucleation and growth phe-nomenon during the initial stages of solid-ification illustrated in region A of Fig. 13.It was proposed that the initial solid formsat the fusion boundary by a heteroge-neous nucleation event, a result of differ-ences in composition, lattice parameterand crystal structure between the sub-strate (BCC) and weld metal (FCC). Asthese solid nuclei form, they adopt a clos-est match orientation relationship with

    the partially melted substrate grains at the

    272-s | OCTOBER 2000

    RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT

    Fig. 11 Misorientation results of entire region plotted in RodriguesSpace (intensity). Note the higher intensities at 27 deg @ [110], 50 deg@ [110], 45 deg @ [100] and 60 deg @ [111] ( 3).

    Fig. 12 Results of misorientation analysis from Fig. 10 between weldmetal and HAZ grain at the fusion boundary. Results indicate tenden-cies toward 35 deg @ [110] (K-S) and 45 deg @ [100] (Bain).

    Fig. 13 Schematic illustration of fusion boundary microstructuralevolution in dissimilar metal welds. The curved lines represent thelocus of the transformation temperatures.

  • 8/12/2019 NELSON stud operation

    7/11

    fusion boundary from which they nucle-ate. As a result, the fusion boundary is nolonger a simple transition betweenwrought and cast material, but is madeup of random grain boundary types of large misorientations between HAZ andweld metal grains. The nature of this ele-vated temperature boundary affects theevolution of the fusion boundary and sur-rounding microstructure during the on-cooling phase transformation. (Note: forconsistency and brevity, in the followingdiscussion the authors will use terminol-ogy and symbols consistent with the Fe-C phase diagram when describing thevarious phases, phase transformation andinterphase boundaries in the HAZ; forthe elevated temperature BCC iron fer-rite; for FCC iron austenite; and for thelower temperature BCC iron ferrite.However, when referring to the weldmetal, which is a nickel-copper-richphase, FCC will be used.)

    As the weld and surrounding HAZ

    cool, the - interphase boundary in theHAZ approaches the fusion boundary il-lustrated in Region B of Fig. 13. In an au-togenous weld on iron or steel, the fusionboundary offers little or no resistance tothis interphase boundary, as it is simply achange from wrought to cast material. Asthere is no difference in composition,structure and orientation at the fusionboundary, the - interphase boundarykeeps advancing toward the centerline of the weld without hindrance.

    However, in a DMW the -ferrite inthe HAZ does not advance beyond the fu-

    sion boundary, as the weld metal is sta-ble FCC. When the - interphase bound-ary intersects the fusion boundary, thefusion boundary now becomes ametastable -FCC boundary between twophases (HAZ iron-rich austenite andweld metal nickel-copper-rich FCC) of vastly different composition. When thisoccurs, several questions arise regardingthe nature and mobility of this -FCCboundary. These will be discussedbelow.

    In homogenous welds, the weld metalgrain size is roughly equivalent to theHAZ grain size adjacent to the fusionboundary. Unlike homogeneous welds,the weld metal grain size in DMWs maybe smaller or larger than the HAZ grainsize (Ref. 20). Likewise, the austenite inthe near HAZ may exhibit a much largergrain size than the -ferrite. As a result,when the - boundary reaches the fu-sion boundary, the larger austenite HAZgrains may traverse several weld metalgrains and SGBs.

    Not only are the grain sizes different,but the fusion boundary is now made upof numerous grain boundaries of variousmisorientations between the iron-rich

    austenitic HAZ and the nickel-richaustenitic weld metal grains. Because so-lidification of the austenitic weld metalmay have initiated heterogeneously, andthe austenite grain size in the near HAZ

    may be quite large, it is likely that thelarger austenite grains in the HAZ willtraverse more than one weld metal grainand SGBs at the fusion boundary whenthe - interphase boundary intersectsthe fusion boundary as illustrated in Re-gion B of Fig. 13. As a result, the fusionboundary becomes an austenite-FCC ( -FCC) grain boundary made up of variousrandom misorientations between theaustenitic HAZ and weld metal grains. Atthis point, the fusion boundary is nolonger a simple transition betweenwrought and cast materials, but is a -

    FCC boundary between the iron-richHAZ (metastable austenite) and thenickel-rich weld metal grains (stableFCC). When this occurs, growth of theaustenite in the HAZ can no longer pro-ceed by transformation, as there is no -ferrite present in the fully austenitic weldmetal. However, the newly formed -FCCfusion boundary can migrate as a grainboundary under normal grain growthmechanisms.

    Mobility of the Austenite-AusteniteFusion Boundary

    The mobility of the -FCC dissimilarboundary, now superimposed on theoriginal fusion boundary, is greatest inthe austenite temperature range. This isdue to the fact that only short-range dif-fusion is required for this boundary to mi-grate. In DMWs, the weld metal has beenstabilized by the additions of austeniticstabilizing elements (Ni and Cu) from thefiller metal. For the initial -ferrite, or thelower temperature -ferrite, to migrateinto the weld metal, long-range diffusionwould be required to reduce the stabilityof the FCC weld metal phase for it to be

    transformed to a BCC ferrite phase. How-ever, long-range diffusion is highly un-likely to occur during the rapid thermalcycles associated with welding. There-fore, the newly formed -FCC dissimilar

    boundary located at the fusion boundaryis most mobile when the fusion boundaryregion is within the austenitic tempera-ture range because only short-range dif-fusion is required.

    Several factors could affect the mobil-ity of the -FCC dissimilar boundary.

    These include 1) steep thermal gradient,2) composition gradient, 3) interfacialstrain across the fusion boundary, and 4)elimination of grain boundary area, i.e.,solidification grain boundaries (SGBs) ormigrated grain boundaries. From a ther-modynamic standpoint, the driving force

    behind the continued migration of the -FCC dissimilar boundary is the reductionin Gibbs free energy. Those elementscontributing to an increase in free energywhich could be reduced or eliminated bythe migration of the -FCC dissimilarboundary include 1) the compositiongradient at the fusion boundary, and 2)interfacial strain energy.

    Several of the above-mentioned ele-ments have been investigated in variousgrain growth studies. Yoo, et al., investi-gated the effects of thermally inducedstress/strain on grain growth in Ag (Ref. 25).

    They observed grain boundary migrationdistances greater than 40 m inquenched and annealed Ag samples.

    They also observed higher heating-ratesand annealing temperatures producedgreater distances of grain boundary mi-gration. They concluded the thermalstrain produced as a result of quenchingand rapid heating was sufficient to pro-duce grain boundary migration, similarto strain-induced boundary migration, inmechanically deformed materials. Otherinvestigators have observed similartrends in their studies of strain-inducedboundary migration (Ref. 26) and abnor-

    WELDING RESEARCH SUPPLEMENT |273-s

    Fig. 14 Free energy and chemical potential changes during downhill diffusion. A weldedblock of different compositions; B free energy diagram for welded blocks.

  • 8/12/2019 NELSON stud operation

    8/11

    mal grain growth (Refs. 2729). Randleconcluded anomalous grain growth was

    induced in Ni by the application of 2%pre-strain prior to annealing (Ref. 27).Likewise, Gastaldi, et al. (Ref. 28), andSrolovitz, et al. (Ref. 29), concluded thedriving force for anomalous grain growthis mainly strain related.

    The presence of high peak residualstresses in weldments is an accepted andwell-known fact associated with weld-ing. This may be compounded with thedifferences in coefficient of thermal ex-pansion between base and weld metal inDMWs. Schimmoller observed tensileresidual stresses in as-welded roll bondclad plate on the order of 20 ksi (Ref. 30).Likewise, Nho, et al. , observed peak ten-sile residual stresses in clad componentson the order of 48 ksi in Type 309L cladmaterials (Ref. 31). Residual stresses of this magnitude are in excess of the yieldstrength of the Type 309L filler metal,commonly used filler for cladding orDMW. Residual stresses of these magni-tudes are typically associated with differ-ential strains, sufficient to augment nor-mal or even anomalous grain growth atelevated temperatures.

    Similar arguments can be made re-garding the presence of composition gra-dients and grain boundary area in theweld metal adjacent the fusion boundaryin DMWs. The electron microprobeanalyses presented in Figs. 15 and 16demonstrate the composition gradientspresent at the fusion boundary in DMWs.From a free energy standpoint, it is ener-getically favorable to homogenize thecomposition in any system, and mini-mize the grain boundary area, as bothcontribute to the total free energy of asystem. The reason diffusion occurs is toproduce a decrease in Gibbs free energy(Ref. 32). A common example used by

    many in describing diffusion kinetics is toweld together two blocks of the same A-

    B solid solution (Fig. 14A), but of differ-ent compositions. The welded block isheld at an elevated temperature in orderfor long-range diffusion to occur. Thismakes a nice illustration, as this is exactlywhat is attained in DMWs. As BMD de-crease, the composition gradient at thefusion boundary becomes very steep asobserved in Fig. 16. Considering themolar free energy diagram in Fig. 14B, if the molar free energy of each alloy is rep-resented by G 1 and G 2, then initially thetotal free energy of the welded block willbe G 3. If diffusion occurs as shown in Fig.14A so as to eliminate the concentrationdifferences, then the total free energy willdecrease toward G 4, the free energy of ahomogeneous alloy. Although there is in-sufficient time for this to occur in a nor-mal weld thermal cycle, there is a strongdriving force for the -FCC dissimilarboundary to migrate into the weld metal,as the rate of solute diffusion (especiallysubstitutional solute) across a grainboundary is greater than matrix diffusion(Refs. 3233). Likewise, the temperaturerange over which this occurs would beroughly 0.50.95T m, where boundarieshave significant mobility (Ref. 32) and va-cancy concentrations are typically veryhigh. Similarly, a reduction in free energycomes from the decrease in total grainboundary area. Therefore, as the -FCCdissimilar boundary migrates into theweld metal, those SGBs that initially ex-tended to the fusion boundary would beeliminated, and the total free energy of the system reduced (Refs. 32, 33).

    Similar mechanisms have been used todescribed the theory of diffusion-induced grain boundary migration(DIGM). Several investigations havedemonstrated the effects of composition

    gradients and coherency strain energiesas possible driving forces for DIGM (Refs.

    3437). Likewise, elevated temperaturesonly act to enhance diffusion kinetic of both solute and solvent, thus, amplifyDIGM (Ref. 37). It has been shown in thepresent investigation that many of thesecharacteristics, i.e., composition andstrain gradients, are inherent along the fu-sion boundary in DMWs. Therefore,DIGM poses a viable mechanism for ex-plaining the driving force for the -FCCdissimilar boundary to migrate into theweld metal, creating the Type II boundary.

    This proposed mechanism of the typeII boundary formation described abovediffers from Wu, et al . (Ref. 15), who pro-posed that the Type II boundary formedduring solicitation. They proposed thatsolidification initiates from the substrateas primary ferrite but after a short dis-tance changes to primary austenite dueto the increase in austenitic stabilizingelements in the weld metal, leaving be-hind a residual high angle grain bound-ary. The original primary ferrite thentransforms to austenite at lower tempera-ture during the on-cooling weld thermalcycle. If this were the case, there shouldexist some FCC/BCC orientation rela-tionship consistent with FCC/BCC inter-facial growth; however, this is not thecase. The mechanism proposed in thepresent paper suggests the Type II bound-ary is a result of solid-state grain bound-ary migration in the austenitic tempera-ture range during the on-cooling weldthermal cycle.

    Summary of the Fusion Boundary within theAustenite Temperature Range

    It is likely all of the characteristics dis-cussed above may be present at the fu-sion boundary in DMWs. Steep thermal

    274-s | OCTOBER 2000

    RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT

    Fig. 15 Electron microprobe analysis across the fusion boundary at 80% BMD weld in Fe/70Ni-30Cu. Note that the bulk weld metal di-lution is reached within approximately 40 m of the FB.

    Fig. 16 Electron microprobe analysis across the fusion boundary ina 46% BMD weld in Fe/70Ni-30Cu. Note that the bulk weld metal di-lution is reached in approximately 10 m.

  • 8/12/2019 NELSON stud operation

    9/11

    gradients are inherent in a weld thermalcycle as a result of efficient heat extrac-tion by the base metal surrounding theweld. Likewise, steep composition gradi-ents near the fusion boundary exist viadiffusional mixing of filler metal alloyingelements within the stagnant boundarylayer. Strain energy at the fusion bound-ary is created by a combination of latticemismatch and differences in coefficientof thermal expansion between base andweld metals. Likewise, the high residualstresses inherent to fusion welding con-tribute to a significant amount of strainnear the fusion boundary. Therefore, it islikely that, within the austenite tempera-ture range, the -FCC dissimilar bound-ary migrates into the weld metal as a re-sult of the strong driving force for graingrowth, producing the Type II boundaryso often observed in DMWs. It is alsopossible that one or any combination of these may also contribute to the crackingrelated problem in DMWs.

    Nature of the Fusion Boundary within the -Ferrite Temperature Range

    As the gamma-alpha interphaseboundary moves through the HAZ to-ward the fusion boundary, illustrated inRegion C and D in Fig. 13, the austenitein the HAZ is transformed to the lowertemperature BCC -ferrite. A detailed un-derstanding of the nucleation and growthof -ferrite within the austenite is criticalin explaining the evolution of the fusionboundary. The orientation relationship

    observed between the HAZ and weldmetal grains at the fusion boundary be-gins to evolve within the austenite tem-perature range, as the austenite in thenear HAZ grows toward the fusionboundary, consuming the BCC delta-ferrite phase, and into the weld metal.

    As the delta-gamma interphaseboundary migrates through the HAZ andinto the weld metal, it leaves behind acoarse-grain austenite region. Within theaustenite temperature range, thesecoarse austenite grains are able to growacross the fusion boundary and into theweld metal. This morphology is illus-trated in Fig. 13 as Region B. At this point,the austenite in the HAZ extends acrossthe fusion boundary, exhibiting a cube-on-cube relationship between the HAZand weld metal, as they are the samegrains. As the weld cools, the lower tem-perature - interphase boundary in theHAZ moves toward the weld fusionboundary, transforming the higher tem-perature FCC austenite in the HAZ to thelower temperature BCC -ferrite.

    It is likely the - transformation pro-ceeds by both nucleation and growth.Initially, the existing -ferrite at the -

    interphase boundary grows behind the - interphase boundary. The morphologyand kinetics of nucleation of -ferrite inaustenite has been well documented(Refs. 32, 33, 38). At higher tempera-tures, -ferrite forms at the austenitegrain boundaries exhibiting an equiaxedor lenticular morphology. These willoften nucleate in an orientation so as tominimize the energy of the interface be-tween ferrite and austenite (Ref. 33). Thisis often accomplished by having a well-defined orientation with one grain whileexhibiting an arbitrary orientation withthe opposing grain. One of the best de-fined orientation relationships observedbetween -ferrite and austenite is that of Kurdjumov-Sachs (K-S). As the tempera-ture of transformation decreases, thesecrystals develop facets on at least oneside, but often on both. Other morpholo-gies include 1) Widmansttten sideplatesor laths, 2) intragranular blocky ferrite,and 3) intragranular plates. The two latter

    morphologies nucleate within theaustenite grains.

    It is likely that several of these mor-phologies may be present in the variousregions of the HAZ. From the micro-graphs evaluated, it appears the predom-inate morphology is intragranular ferrite,giving rise to the equiaxed morphologyobserved in the HAZ microstructure, es-pecially in the near HAZ. However, thepresence of some facets or sideplates wasevident in many of the HAZ microstruc-tures. The intragranular morphology ischaracteristic of a larger austenite grain

    size. In fine-grain samples, essentially allthe nuclei will form at, and grow from,the austenite grain boundaries. In acoarse-grained material, the grainboundaries will be covered by nucleiearly in the transformation, but ulti-mately nuclei will form inside the grainto reduce the supersaturation within thegrain (Ref. 33).

    During a weld thermal cycle, thoseaustenite grains adjacent the - inter-phase boundary will have experiencedthe longest time and the highest peaktemperature within the austenite temper-ature range. As a result, this region willexhibit the largest austenite grain size,Region B in Fig. 13. As the - interphaseboundary moves through the austenite inthe near HAZ, initially the -ferrite willform at the austenite grain boundaries.However, because of the large austenitegrain size, -ferrite will also nucleate in-tragranularly, producing the equiaxed -ferrite morphologies observed. The man-ner in which this nucleation occurs givesrise to the various fusion boundary ori-entation relationships observed in thepresent investigation.

    Similar to grain boundary allotri-

    omorphs, intragranular nuclei will nucle-ate in specific orientations so as to mini-mize the interfacial energy betweenaustenite and ferrite. The typical orienta-tion relationships observed between FCCand BCC materials include 1) the Bain, 2)the Kurdjumov-Sachs, and 3) theNishyama-Wassermann relationships.

    The -ferrite that nucleates and growswithin the coarse austenite grains adja-cent the fusion boundary will nucleatewith one of these specific orientation re-lationships. Once nucleated, the ferritenuclei will continue to grow until all theaustenite in the HAZ has been trans-formed. Although the gamma-alpha in-terphase isotherm will continue towardthe weld centerline as the weld cools,growth of the -ferrite in the near HAZwill terminate at the fusion boundary be-cause the weld metal is stable FCC tobelow room temperature as a result of al-loying from filler metal additions.

    The fusion boundary orientation rela-

    tionships observed in DMWs in the pre-sent investigation are a result of abnor-mal austenite grain growth, followed bynucleation and growth of -ferrite grainson specific orientations within thesecoarse austenite grains. Because thecoarse prior austenite grains in the HAZwere able to migrate across the fusionboundary, the crystallographic orienta-tion is not altered at the fusion boundary,even though the composition changesdramatically. Therefore, any -ferrite thatnucleates within these HAZ austenitegrains would exhibit the same orientation

    relationship with that portion of theaustenite grain extending into the weldmetal. Evidence of this was presented ina previous paper by Nelson, et al . (Ref. 21),where TEM analyses demonstrated the re-lationship between base (BCC) and weld(FCC) metal exhibited a Nishyama-Wassermann orientation relationship.Likewise, similar trends were presentedin the OIM misorientation analyses (Fig.12) where the orientation relationships atthe fusion boundary tend toward thosecommon FCC/BCC relationships: theKurdjumov-Sachs, Nishyama-Wasser-mann, and Bain relationships.

    Summary on the Nature of the FusionBoundary in Dissimilar Metal Welds

    As a result of the inherent differencesbetween base and weld metal in DMWs,the evolution of the fusion boundary mi-crostructure and orientation relation-ships is complex. Within the -ferritetemperature range, the fusion boundaryexhibits various grain boundary misori-entations that may or may not includethose orientation relationships character-istic of FCC/BCC interfaces. Within the

    WELDING RESEARCH SUPPLEMENT |275-s

  • 8/12/2019 NELSON stud operation

    10/11

    austenite temperature range, the - in-terphase boundary moves through theHAZ toward the fusion boundary. As thisoccurs, the coarse austenite grains in theHAZ intersect the fusion boundary, cre-ating a -FCC dissimilar boundary be-tween HAZ and weld metal. As a resultof the large austenite grain size, the HAZgrains may traverse several weld metalgrains and SGBs, creating a fusionboundary of various misorientations.However, because both the HAZ andweld metal are austenite when the HAZis in the austenite temperature range, the-FCC dissimilar boundary at the fusionboundary is rendered mobile.

    As the weld cools through the austen-ite temperature range the -FCC dissimi-lar boundary migrates into the weldmetal. The driving force behind the mi-grating -FCC boundary is the reductionin free energy. Inherent differences be-tween base and weld metals in DMWscreate steep composition gradients adja-

    cent the fusion boundary in the weldmetal, and interfacial strain across the fu-sion boundary. As the -FCC dissimilarboundary migrates across the transitionregion in the weld metal, these factorsmay be reduced or eliminated, therebyreducing the free energy of the system.

    The composition becomes more ho-mogenous by diffusion across and alongthe -FCC dissimilar boundary as it mi-grates through the transition region. Thismigrating boundary may sweep or carryiron into the weld metal while at thesame time allowing Ni and Cu to diffuse

    across or along it toward the base metal.All of these factors contribute to the dri-ving force for the -FCC dissimilarboundary to migrate into the weld metal.

    At lower temperatures, the - inter-phase boundary eventually intersects thefusion boundary. In the near HAZ wherethe austenite grain size grew quite large, -ferrite nucleates at austenite grainboundaries and as intragranular id-iomorphs. The -ferrite nuclei continueto grow toward the fusion boundary be-hind the gamma-alpha interphaseboundary, eventually consuming allaustenite in the HAZ. However, growthof the -ferrite is interrupted at the fusionboundary, as the composition of the weldmetal no longer supports the gamma-alpha transformation. For the -ferrite togrow into the weld metal, long-range dif-fusion of substitutional alloying elementswould be required to decrease the stabil-ity of the austenitic weld metal. Over theshort diffusion times created by the rapidthermal cycles associated with welding,this will not occur.

    The orientation relationships ob-served between the HAZ -ferrite and theaustenitic weld metal result from the fact

    the habit planes on which the -ferritenucleate are continuous across the fusionboundary. This occurred when thecoarse-grain austenite grains in the HAZgrow behind the advancing -FCC dis-similar boundary, across the fusionboundary and into the weld metal. As aresult, a cube-on-cube relationship wascreated between base and weld metalwithin the austenite temperature range.

    The lower temperature -ferrite then nu-cleated within these large austenitegrains and grew up to the fusion bound-ary. Although the growth of these grainswas terminated at the fusion boundary,they share the same orientation relation-ship with the weld metal grains as theydid with the prior austenite HAZ grainsadjacent the fusion boundary. This givesrise to those misorientation trends ob-served between HAZ and weld metalgrains, as shown in Figs. 11 and 12. As aresult, the fusion boundary exhibits largemisorientations (>15 deg) rather than the

    small misorientation observed in homo-geneous welds (

  • 8/12/2019 NELSON stud operation

    11/11

    and Hattori, T. 1985. Microscopical crit-ical condition for the initiation of dis-bonding of weld overlaid pressure vesselsteel. Transactions of the Iron and SteelInstitute of Japan 25(6): 505512.

    6. Ohnishi, K., Fuji, A., Chiba, R.,Adachi, T., Naitoh, K., and Okada, H.1984. Study on a stainless steel overlaywelding process for superior resistance todisbonding. report 1: effect of strip over-lay welding conditions on resistance tohydrogen-induced disbonding. Transac-tions of the Japan Welding Society 15(2):129135.

    7. Matsuda, F., et al . 1984. Disbond-ing between 2.25%Cr-1%Mo steel andoverlaid austenitic stainless steel bymeans of electrolytic hydrogen chargingtechnique. Trans. JWRI , 13(2): 263272.

    8. Matsuda, F., Nakagawa, H., and Tsuruta, S. 1986. Proposal of hydrogenblistering mechanism associated withdisbonding between 2.25%Cr-1%Mosteel and Type 309 (stainless steel) over-

    laid metal. Transactions of JWRI , 15(2):207208.

    9. Morishige, N., Kume, R., and Ok-abayashi, H. 1985. Influence of low-tem-perature hydrogen degassing on hydro-gen-induced disbonding of cladding.Transactions of the Japan Welding Soci-ety 16(1): 1218.

    10. Blondeau, R., Pressouyre, G. M.,Cheviet, A., Berthet, J. A., and Du-ranseaud, J. M. 1982. Contribution to asolution to the disbonding problem in2.25%Cr-1%Mo heavy wall reactors.Current Solutions to Hydrogen Problems

    in Steels. pp. 356360. Proceedings, 1stInternational Conference, Washington,D.C., Eds., C. G. Interrante and G. M.Pressouyre. ASM International, MaterialsPark, Ohio, ISBN 0-87170-148-0.

    11. Pressouyre, G. M., Chaillet, J. M.,and Valette, G. 1982. Parameters affect-ing the hydrogen disbonding of austeniticstainless cladded steels. Current Solu-tions to Hydrogen Problems in Steels. pp.349355, Proceedings, 1st InternationalConference, Washington, D.C., Eds. C.G. Interrante and G. M. Pressouyre, ASMInternational, Materials Park, Ohio, ISBN0-87170-148-0.

    12. Frederick, G., and Hernalsteen, P.1981. Underclad cracking in PWR (pres-surized water reactor) reactor vessels.

    Translation CE 7892, London, England,Central Electricity Generating Board,

    Translations Section, pp. 22 (Translationof Paper 20 presented at AIM Meeting,Modern Electric Power Stations, Liege,Belgium, 1981).

    13. Horiya, T., Taeda, T., and Yamato,K. 1985. Study on underclad cracking innuclear reactor vessel steels. Transac-tions of the ASME, Journal of PressureVessel Technology 107(1): 3035.

    14. Thielsch, H. 1952. Stainless steelweld deposits on mild and alloy steels.Welding Journal 31(1): 37-s to 64-s.

    15. Lundin, C. D. 1982. Dissimilarmetal welds-transition joints literature re-view. Welding Journal 61(2): 58-s to 63-s.

    16. Wu, Y., and Patchett, B. M. 1992.Formation of crack-susceptible structuresof weld overlay of corrosion resistant al-loys. pp. 206219, 31st Metallurgist Conference of CIM , Edmonton, Canada.

    17. Edington, J. W. 1976. PracticalElectron Microscopy in Materials Sci-ence, reprinted by TechBooks, Herndon,Va.

    18. Adams, B. L. 1993. Orientationimaging microscopy: application to themeasurement of grain boundary struc-ture. Materials Science and Engineering ,A166, pp. 5966.

    19. Wright, S. I. 1993. A review of au-tomated orientation imaging microscopy(OIM). Journal of Computer-Assisted Mi-croscopy 5(3): 207221.

    20. Adams, B. L., Wright, S. I., andKarsten, K. 1993. Orientation imaging:the emergence of a new microscopy.Met. Trans. A , 24A, pp. 819831.

    21. Nelson, T. W., Lippold, J. C., andMills, M. J. 1998. Investigation of bound-aries and structures in dissimilar metalwelds. Science and Technology of Weld-ing and Joining 3(5): 249255.

    22. Flemings, M. C. 1974. Solidifica-tion Processing. McGraw-Hill PublishingCo.

    23. Savage, W. F., and Aronson, A. H.1966. Preferred orientation in the weld

    fusion zone. Welding Journal 45(2): 85-sto 89-s.24. Samuel, J. 1979. Crystallography

    in fusion-weld-metal solidification me-chanics. Ph.D. Dissertation, RensselaerPolytechnic Institute, Troy, N.Y.

    25. You, C-Y., Han, S-C., and Yoon, D-K. 1995. The grain boundary migration inAg induced by thermal strain. Metallur-gical and Materials Transaction A , Vol.26A, pp. 30483049.

    26. Bailey, J., and Hirsch, P. Proceed-ings of the Royal Society, A267, 11.

    27. Randle, V. 1993. Microtexture in-vestigation of the relationship betweenstrain and anomalous grain growth.Philosophical Magazine A 67(6):13011313.

    28. Gastaldi, J., Jourdan, C., andGrange, G. 1992. Materials ScienceForum , pp. 9495, 17.

    29. Srolovitz, D. J., Grest, G. S., andAnserson, M. P. 1985. Computer simula-tion of grain growth, A, abnormal graingrowth. Acta Metallurgica 33(12):22332247.

    30. Schimmoller, H. 1972. Deter-mintation of tensile stresses in evenlyplated materials, part 2, tension in warm-

    rolled, explosion clad austenitic platedsteels. Materials Testing 14(11):380387.

    31. Nho, S-H., Ann, H-S., and Ong, C-W. 1992. Derivation of theoretical resid-ual stress and stress intensity factors indissimilar weld metal of nuclear vessel.pp. 117123. Proceedings of an Interna-tional Conference on Trends in WeldingResearch, Gatlinburg, Tenn.

    32. Porter, D. A., and Easterling, K. E.1981. Phase Transformations in Metalsand Alloys. Published by Van NostrandReinhold International, Berkshire, Eng-land.

    33. Shewmon, P. G. 1983. Transfor-mations in Metals. J. Williams Book Co.,

    Jenks, Okla.34. Shewmon, P. G., and Meyrick, G.

    1986. The Diversity of DIGM (DiffusionInduced Grain Boundary Migration) , pp.717, ASM International, Materials Park,Ohio.

    35. Guan, Z.M., Liu, G.X., Williams,

    D.B., and Notis, M.R. 1989. Diffusion-in-duced grain boundary migration and as-sociated concentration profiles in a Cu-Zn alloy. Acta Metall. 37(2): 519527.

    36. Rabkin, E. 1994. Gradient and co-herency strain energies as driving forcesfor DIGM. Scr. Metall. Mater . 30(11):14431448.

    37. Rabkin, E., Shvindlerman, L.S.,and Gust, W. 1993. Theory of grainboundary motion during high-tempera-ture DIGM. Interface Sci. (Netherlands),1(2): 133137.

    38. Honeycombe, R. W. K. 1981.

    Steels Microstructure and Properties .Edward Arnold Ltd., London, U.K.