•nature and evolution of the fusion boundary in ferritic-austenitic dissimilar weld metals, part 1...
DESCRIPTION
•Nature and Evolution of the Fusion Boundary in Ferritic-Austenitic Dissimilar Weld Metals, Part 1 Nucleation and GrowthTRANSCRIPT
ABSTRACT. A fundamental investigationof fusion boundary microstructure evo-lution in dissimilar-metal welds (DMWs)between ferritic base metals and a face-centered-cubic (FCC) filler metal wasconducted. The objective of the workpresented here was to characterize thenature and character of the elevated-temperature fusion boundary to deter-mine the nucleation and growth charac-teristics of DMWs. Type 409 ferriticstainless steel and 1080 pearlitic steelwere utilized as base metal substrates,and Monel (70Ni-30Cu) was used as thefiller metal. The Type 409 base metalprovided a fully ferritic or body-cen-tered-cubic (BCC) substrate at elevatedtemperatures and exhibited no on-cool-ing phase transformations to mask or dis-guise the original character of the fusionboundary. The 1080 pearlitic steel wasselected because it is austenitic at thesolidus temperature, providing anaustenite substrate at the fusion bound-ary. The weld microstructure generatedwith each of the base metals in combi-nation with Monel was fully austenitic.
In the Type 409/Monel system, therewas no evidence of epitaxial nucleationand growth as normally observed in ho-
mogenous weld metal combinations.The fusion boundary in this system ex-hibited random grain boundary misori-entations between the heat-affected zone(HAZ) and weld metal grains. In the1080/Monel system, evidence of normalepitaxial growth was observed at the fu-sion boundary, where solidification andHAZ grain boundaries converged. Thefusion boundary morphologies are a re-sult of the crystal structure present alongthe fusion boundary during the initialstages of solidification. Based on the re-sults of this investigation, a model for het-erogeneous nucleation along the fusionboundary is proposed when the base andweld metals exhibit ferritic (BCC) andFCC crystal structures, respectively.
Introduction
The importance of joining andcladding of dissimilar metals has in-creased substantially in all aspects ofmanufacturing over the past twodecades. Applications of dissimilar-metalwelds (DMWs) include cladding for cor-rosion resistance and joining base metalsthat exhibit large differences in structureand properties, i.e., plain carbon or low-alloy steels to austenitic stainless steels.This trend can be expected to continuewith the increasing use of duplex andsuper-austenitic stainless steels. Crackingassociated with DMWs has been a per-sistent problem, resulting in significanteconomic loss over the past severaldecades. Cracking in DMWs typicallyoccurs near the fusion boundary eitheralong the martensitic transition immedi-ately adjacent to the fusion boundary oralong the Type II boundary in the weldmetal. This Type II boundary parallels thefusion boundary typically less than 100µm away in the weld metal.
The evolution, nature and role of weldmetal interfaces in promoting or mitigat-ing weld-related cracking are not wellunderstood. The implications of bound-aries and structures with regards to crackgrowth rates, fatigue, stress corrosioncracking, etc., have been researched ex-tensively in the materials science arena.In spite of the recurring problems andeconomic losses, the role of boundaries
WELDING RESEARCH SUPPLEMENT | 329-s
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
WELDING RESEARCHSUPPLEMENT TO THE WELDING JOURNAL, OCTOBER 1999Sponsored by the American Welding Society and the Welding Research Council
Nature and Evolution of the Fusion Boundary inFerritic-Austenitic Dissimilar Weld Metals,
Part 1 — Nucleation and Growth
BY T. W. NELSON, J. C. LIPPOLD AND M. J. MILLS
A model for heterogeneous nucleation along the fusion boundary is proposed
KEY WORDS
Dissimilar-Metal WeldsEpitaxial GrowthFusion BoundaryHAZMicrostructureMonelPearlitic SteelT .W. NELSON, formerly with The Ohio State
University, is now with Brigham Young Uni-versity, Provo, Utah. J. C. LIPPOLD and M. J.MILLS are with The Ohio State University,Columbus, Ohio.
and structures in promoting or mitigatingweld-related cracking remains in dis-pute. Therefore, a fundamental investi-gation was undertaken to study the na-ture and character of those boundariesand structures near the fusion boundaryin dissimilar-metal welds.
The results of this study will be pre-sented in two parts. The first part ad-dresses the nature and character of the el-evated-temperature fusion boundary anddescribes mechanisms of nucleation andgrowth at the fusion boundary in DMWs.Although difficult to characterize, it isimportant to understand the nature of thefusion boundary at elevated temperaturein order to formulate the mechanismsthat explain the evolution of those struc-tures and boundaries observed along theroom-temperature fusion boundary. Thesecond part of this investigation, whichwill be reported in a subsequent article,will detail the effects of on-cooling trans-formations on the evolution of the fusionboundary and microstructure observedat room temperature.
Nucleation and Growth of SolidWeld Metal
Solidification is a process that occursby the nucleation and growth of a newphase (solid) at an advancing solid-liquidinterface (Ref. 1). This phenomenon hasbeen the subject of considerable researchover the past five decades in an attemptto better understand, model and predictnucleation, growth and segregation invarious types of casting applications(Refs. 2–3).
For many years, solidifying weld met-als have often been characterized assmall castings. In this manner, many ofthe fundamental principles of solidifica-tion in larger castings have been appliedto weld metal solidification phenomena,i.e., nucleation and growth, segregation,
interface stability,etc. Although thereare many similari-ties, there are someaxiomatic differ-ences between castand weld solidifica-tion. Those of great-est concern in thepresent investigationare nucleation and growth at the fusionboundary.
Heterogeneous nucleation is the mostprevalent form of nucleation outside ofthe controlled research solidification ex-periment. It requires much smaller un-dercooling, ∆T, or driving force, than ho-mogeneous nucleation. In practice, it iswell known that metals and most otherliquids rarely undercool by more than afew degrees before beginning to crystal-lize (Refs. 2–3). The reason they rarelyundercool is because the numerous nu-cleating agents, such as impurities, inoc-ulants, mold walls, etc., act as catalystsites for nuclei, thereby reducing the free-energy barrier or critical radius of a nu-clei (Refs. 1–3). This reduction in free en-ergy is brought about by a reduction inthe interfacial energy, which is effectivelyachieved if the solid forms on anothersurface.
When a mold wall is used as a nucle-ating substrate, the free energy requiredis a function of the wetting angle (θ) be-tween the substrate and solid formed.This is typically the case in a weld wherethe partially melted substrate at the fu-sion boundary acts as a mold wall to themolten weld metal. The relationship be-tween wetting angle and interfacial ener-gies is illustrated in Fig. 1 and Equations1–3. In welds of similar base and fillermetal compositions (homogeneouswelds), the liquid weld metal completelywets the partially melted substrate at thefusion boundary. As a result, the contact
or wetting angle, θ, is zero (Equation 3),and so the free energy, ∆G*, is also zero(Equation 2), and growth of the solid ini-tiates at the fusion boundary without dif-ficulty (Ref. 18). It is generally acceptedthere is little or no nucleation barrier insolidifying weld metals of homogeneouswelds.
(1)
(2)
where
(3)
Growth of the solid in a weld gener-ally initiates by epitaxial growth from thesubstrate and proceeds by competitivegrowth toward the centerline of the weld(Refs. 4–5). Because the weld metal com-pletely wets the substrate at the fusionboundary, grain growth is initiated in theweld by arranging atoms in the liquidphase on the existing crystalline sub-strate, thereby extending the solid with-
f θθ θ( ) =
+( ) −( )
2 1
4
2cos cos
∆
∆
G r G r f
G f
het V SL*
hom*
= − ∆ +
( ) =
( )
43
43 2π π γ θ
θ
γ γ γ θ
θγ γ
γ
LM SM SL
m sm
s
= +
=−( )
cos ,
cos
or
1
1
330-s | OCTOBER 1999
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
Fig. 1 — Schematic representation of a heterogeneous nucleation ofa spherical cap on a solid substrate.
Fig. 2 — Illustration showing the epitaxial nucleation and competi-tive growth in the weld fusion zone.
out altering the crystallographic orienta-tion (Refs. 6–10). In homogeneous welds,this type of solidification growth resultsin nearly complete continuity of the crys-tallographic orientation and grainboundary misorientation between baseand weld metal grains adjacent to the fu-sion boundary (Refs. 8–9). Once initi-ated, solidification proceeds toward theweld centerline in a competitive-typegrowth mechanism, as illustrated in Fig.2. That is, grains with their easy growthdirection oriented most preferentiallyalong the heat flow direction gradienttend to crowd out those grains whoseeasy growth directions are not as suitablyoriented, as illustrated in Fig. 2. The easygrowth direction in BCC and FCC mate-rials is the <100>. Competitive growth isthe predominant form of solidificationgrowth in welds, producing a cube-on-cube (<100>//<100>, {100}//{100}) ori-entation relationship between the HAZand weld metal grains along the fusionboundary and a weak <100> fiber tex-ture toward the centerline of the welds incubic metals. This relationship was firstconfirmed by Savage and Aronson (Ref.8) and Samuels (Ref. 9).
Although competitive growth in theweld fusion zone is the predominantmechanism of solidification in welds,there are others that can interrupt and/ordominate the solidification structure.Other possible mechanisms of grain for-mation include 1) dendrite fragmenta-tion, 2) grain detachment, 3) heteroge-neous nucleation and 4) surfacenucleation. The specific mechanisms ofthese have been discussed in detail byKou (Ref. 4). Although unlikely to occurin most welds, it is pointed out here to il-lustrate the various possibilities that
could contribute tothe weld metal mi-crostructures oftenobserved in ferritic-austenitic DMWsystems.
The nature ofepitaxial growth inweld metals de-scribed above pro-duces a network ofgrain boundaries inthe fusion zone. Asshown in Fig. 2, thesolidification grainboundary (SGB) net-work in the fusionzone is an extensionof the HAZ grainboundaries at the fu-sion boundary. Lip-pold, et al. (Ref. 10),defined the nature ofthe various bound-aries observed in single-phase weldmetal as solidification grain boundaries(SGB), solidification subgrain boundaries(SSGB) and migrated grain boundaries(MGB).
Interfaces and Grain Boundaries
Although there are many aspects toboundaries in general, the nature of in-terphase boundaries is of particular rele-vance to the present investigation. Inter-phase boundaries exist where the grainson either side of a boundary have differ-ent crystal structures, lattice parametersand/or compositions, such as exists inprecipitation or overgrowth. These inher-ent differences between the phases re-duce the likelihood the two lattices will
match perfectly along the interface, i.e.,spacing between atomic planes and di-rections are roughly equal. It is likelythese characteristic differences may existin ferritic-austenitic weld systems wherethe addition of filler metal changes thecomposition, crystal structure and latticeparameter of the weld metal relative tothe HAZ grains at the fusion boundary.The nature and relationship betweenphases across such boundaries are gov-erned by these differences.
As there are varying degrees of differ-ence between phases, interphase bound-aries in solids are typically divided intothree classes: coherent, semicoherentand incoherent (Ref. 11). These describethe degree to which the opposing crys-talline matrices match across the inter-
WELDING RESEARCH SUPPLEMENT | 331-s
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
Fig. 3 — Fusion boundary microstructure in Type 409/Monel base and weld metals. A — Transverse photomicrograph showing representative fu-sion boundary and weld metal microstructures; B — scanning electron micrograph of the fusion boundary in Type 409/Monel system. Note thenumerous weld metal grains along the fusion boundary (black arrows).
A B
Fig. 4 — Transverse section of fusion boundary microstructure in Type409/Monel system. Note how the HAZ boundaries and SGBs are notcontinuous across the fusion boundary.
phase boundary. For example, a coherentinterface results when two opposing lat-tices match perfectly across the bound-ary interface. As the degree of lattice mis-match increases between the phases orgrains, the boundary becomes increas-ingly incoherent. Likewise, the strain en-ergy at the interface increases as the twolattices attempt to maintain a match ofatomic planes across the boundary. Asthe incoherency or strain energy in-creases, it may be more energetically fa-vorable for one phase to adopt a new ordifferent orientation relative to the otherin order to reduce the lattice mismatchbetween phases. Typically, nucleatingphases will orient themselves so that cer-tain atomic planes and directions in thegrowing phase are parallel (or nearly par-allel) to specific planes and directions inthe matrix crystal, referred to as an ori-entation relationship (OR).
There are specific ORs that are mostoften observed in FCC/BCC systems,which include the Bain, Kurdjumov-Sachs (K-S) and Nishyama-Wasserman(N-W) ORs. These ORs tend to orient
close-packed planes(CPP) parallel toCPP and/or close-packed directions(CPD) parallel toCPDs in order tominimize the inter-facial strain energybetween the phases.These can be de-fined by specifyingthe parallel planeand directions in therespective phases orby an angle and axisof rotation neces-sary to align theseplanes and direc-tions. Examples ofboth of these meth-ods are listed belowfor the Bain, K-S andN-W ORs.
OR Parallel Planes and Angle/AxisDirections
Bain {011}BCC//{001}FCC: 45 deg/100<111>BCC//<110>FCC
K-S {011}BCC//{111} FCC: 35.3 deg/110<111>BCC//<110>FCC
N-W {011}BCC//{111}FCC: 45 deg/110<100>BCC//<110>FCC
In the present investigation, misorienta-tions between the HAZ and weld metalat the fusion boundary will be presentedby the angle/axis of misorientation.
Experimental Approach
Material Selection
Two base metals, Type 409 stainlesssteel and AISI 1080, were selected to pro-duce ferrite and austenite phases, re-spectively, in the HAZ at elevated tem-peratures. Monel, a 70Ni-30Cusingle-phase (FCC) binary alloy with low
impurity content, was selected as thefiller metal. The chemical compositionsof the base and filler metals are listed inTable 1. Type 409 is a ferritic stainlesssteel (procured as 3-mm [0.12-in.] thicksheet) and was selected since it is fullyferritic from room temperature to thesolidus temperature, providing a BCCsubstrate at the fusion boundary. The se-lection of this alloy also precluded anyon-cooling transformations in the HAZthat may eliminate or disguise the struc-tures and boundaries present along thefusion boundary at the onset of solidifi-cation. On-cooling transformations iniron or steels make it difficult to charac-terize the relationship between base(BCC) and weld (FCC) metals at the fu-sion boundary. Therefore, by eliminatingthese on-cooling transformations, it wasanticipated that the nature and characterof the fusion boundary would be retainedto room temperature for analysis. Suchan analysis would provide valuable in-formation regarding the nature of epitax-ial growth between an FCC weld metaland a BCC substrate.
A pearlitic steel, AISI 1080, was se-lected because it is austenitic at thesolidus temperature. This material wasprocured as 6.24- by 51-mm (0.25- by 2-in.) bar stock. Although the compositionof 1080 is significantly different from theweld metal, the 1080/Monel combina-tion resulted in the same crystal structureat the solidification temperature, i.e.,both base and weld metals would beaustenitic (FCC).
Welding Conditions
Gas tungsten arc welding (GTAW)using a cold wire feed was used for pro-ducing single-pass bead-on-plate welds.Wire feed speed was used as the primaryvariable for controlling and changing thebase metal dilution (BMD), where BMDdefines the percent base metal compos-ing the weld metal. Current, voltage andtravel speed were held constant at 130 A,10.0 V and 6 in./min (2.5 mm/s), respec-tively, for Type 409/Monel welds. Wirefeed rates ranged from 20 to 40 in./min(0.85 to 1.7 cm/s), producing BMDsranging from 30 to 50%. Current, voltageand travel speed were constant at 250 A,11.0 V and 6 in./min (2.5 mm/s), respec-tively, for welds produced on Type 1080steel. Wire feed rate ranged from 40 to 70in./min (1.7 to 3.0 cm/s), producingBMDs of 50–70%.
Microstructural Characterization
Numerous transverse and plan viewsamples were removed from welds ofvarious BMDs for metallographic analy-
332-s | OCTOBER 1999
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
Fig. 5 — Fusion boundary microstructure in 1080/Monel combina-tion. Note the evidence of epitaxial growth as HAZ and SGBs con-verge at the fusion boundary, and there is no evidence of Type IIboundaries.
sis. Standard metallographic techniqueswere used to prepare optical and SEMsamples. Because of the different charac-teristics of the base and weld metal, twoetchants were used for metallographicpreparation of samples. These includedthe following: 1) 4% nital, which etchesthe pearlitic steel; and 2) an electrolyticsolution of 5 g Fe3Cl, 2 mL of HCL and99 mL of methanol for the weld metal.Optical microscopy and scanning elec-tron microscopy (SEM) were used for mi-crostructure characterization of struc-tures and boundaries along the fusionboundary. Optical metallography wasperformed at magnifications up to 400Xand SEM up to 1600X.
Electron Microscopy Analysis
Electron backscatter diffraction analy-ses were performed on both TEM thinfoils and bulk metallographically pre-pared samples. Samples were loaded inthe SEM at an angle of 70 deg from theincident beam toward a phosphor detec-tor. The electron beam was then rasteredacross the sample in a hexagonal grid atspecified increments. Kikuchi signalswere automatically analyzed usingOIM™ software (Refs. 12–14). This soft-ware calculates the Euler angles of theelectron backscatter diffraction (EBSD)patterns with reference to the samplenormal, then stores the position and an-gles of each pattern. These data werethen used for various grain boundary andtexture analysis.
Results
Optical Microscopy
The resulting microstructures pro-duced in the Type 409/Monel system areshown in Figs. 3 and 4. Over the range ofBMDs investigated, all welds exhibitedroughly the same microstructures alongthe fusion boundary. In general, the HAZexhibited a large, fully ferritic grain struc-ture typical of fully ferritic materials withinsufficient secondary phases present tohinder grain growth. The weld metal ex-hibited a fully FCC cellular orcellular/dendritic microstructure with noevidence of epitaxial growth observed atany of the BMDs produced. This is veryevident in Fig. 4, where the HAZ andSGBs at the fusion boundary are not con-tinuous; instead, they intersect the fusionboundary at roughly mid-grain of one an-other. In homogeneous metal welds, theHAZ grain boundary and SGB would becontinuous across the fusion boundary,and the misorientation across theseboundaries would be the same.
Another interesting microstructural
WELDING RESEARCH SUPPLEMENT | 333-s
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
Fig. 6 — Image quality grain map of fusion boundary region in Type 409/Monel base and fillermetal.
Fig. 7 —Discrete pole figures showing individual orientations by color from grain map shownin Fig. 6. Base metal grains are represented by the large blue and green spots while the weldmetal grains are represented by small red spots.
feature in Type 409/Monel system was thepresence of numerous small equiaxedgrains along the fusion boundary, asshown in Fig. 3. This type morphologywas observed in the majority of samplesevaluated in the Type 409/Monel system.The morphology of these grains resem-bles the appearance of a “chill zone” in alarge casting where equiaxed grains nu-cleate randomly along the mold wall.
Pearlitic steel (AISI 1080) was selectedas a base metal because its primary solidi-fication phase is austenite. As a result, so-lidification of an FCC weld metal (Ni rich)would initiate epitaxially on an austeniticsubstrate, reducing the lattice mismatchbetween base and weld metal crystals; itwould also eliminate the delta-to-gammaphase transformation in the HAZ. All ofthese may influence the subsequent evo-lution of various boundaries and structuresnear the fusion boundary in DMWs. Themicrostructure in Fig. 5 is representative ofall the HAZ and weld metal microstruc-tures observed in welds on the 1080 steel.The weld metal microstructures exhibit afully austenitic solidification structure withinterdendritic eutectic. Weld metal epi-taxy, as observed in homogeneous welds,is evident at the fusion boundary as HAZboundaries are continuous with SGBs.This observation is important with respectto the effects of crystal structure and latticeparameter on the evolution of the fusionboundary in DMWs.
Electron Microanalysis
Results obtained from OIM analysisalong the fusion boundary in the Type409/Monel system are shown in Figs. 6and 7. Figure 6 shows a simple grain mapplotted from OIM analysis with the FCCweld metal at top and the BCC substrateon the bottom. The black region alongthe fusion boundary separating weld andbase metal exhibited a martensitic struc-ture. Analysis using the OIM softwareonly allows two-phase analyses (FCCand BCC in this investigation); therefore,patterns and crystal orientation informa-tion from this martensitic region were notindexed. As a result, the software assignsthese patterns a low value on a grayscale, indicated by the black coloring.
In Fig. 6, there is no evidence of thetypical epitaxial growth observed in ho-mogeneous welds as the HAZ grainboundary is not continuous across the fu-sion boundary. Also note the majority ofsmall grains along the fusion boundary,also observed optically in Fig. 3, exhibitlow-angle boundaries between manyneighboring grains. This is evident by thethinner grain boundary lines represent-ing misorientations between 5 and 15deg, or changes in shading representingmisorientations less than 5 deg. Likewise,there is no indication that the HAZ grainboundary extends across the fusionboundary, becoming an SGB, as ob-
served in homogeneous welds. Misori-entation measurements of both HAZ(blue) and SGBs (yellow), as shown inFig. 6 are listed in Table 2. As observedin this table, none of the SGBs (yellow)exhibits similar misorientations to that ofthe HAZ (blue) grain boundary. The HAZgrain boundary observed in Fig. 6 is ahigh-angle grain boundary exhibiting thefollowing misorientation:
θmis = 32.3 degUVW ≈ [18 7 12]
where θmis is the angle of misorientationand UVW is the misorientation axis. Thismisorientation was resolved from Eulerangles (ϕ1, Φ, ϕ2) determined from OIManalysis, as shown in Table 2.
Misorientation analyses between HAZand weld metal grains along the fusionboundary (blue) exhibit no evidence oftypical weld metal epitaxial growth (Table3). Unlike homogeneous welds, wherethere is little or no misorientation be-tween base and weld metal grains at thefusion boundary, there is a random distri-bution of large-angle grain boundary mis-orientations in the Type 409/Monel sys-tem. Likewise, these results exhibit notrend in preferred misorientation betweenHAZ and weld metal grains, i.e., K-S, N-W or Bain relationships.
Orientation distribution analyses re-veal some interesting trends in the weld
334-s | OCTOBER 1999
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
microstructure. Individual orientationdata of weld metal grains in Fig. 6 indi-cate a weak <100> fiber texture, asshown in Fig. 7. To separate base andweld metal, individual weld metal grainswere highlighted in red, while the twobase metal grains were highlighted inblue and green. Individual weld metalgrain orientations, shown in Fig. 7, ex-hibit a weak [100] fiber texture with the[100] direction oriented toward the topof Fig. 6, i.e., toward the center and topof the weld. The <100> fiber texture is in-dicative of the solidification directionnormally observed in cubic metals.
Although the weld metal exhibits a[100] fiber texture toward the top of Fig.6, this orientation is not shared with ei-ther of the two base metal HAZ grains.Base metal grain orientations highlightedin blue and green in Fig. 7 show no cor-relation with the weld metal texture. Al-though the HAZ grains exhibit orienta-tions close to that of the weld metal<100> fiber texture, they are rotatedroughly 10–15 deg from the weld metalclustering in the <100> pole figure in Fig.7. In homogeneous welds in cubic mate-rials, solidification initiates from the par-tially melted HAZ grain in a preferred[100] direction with little or no misori-entation across the fusion boundary.
Discussion
Nucleation of Solid at the Fusion Boundary
Nucleation events, other than epitax-ial nucleation, that initiate the formationof solid nuclei in welds are relatively in-frequent given the ideal substrate pro-vided by the partially melted HAZ grainsat the fusion boundary (Refs. 7, 18, 19).In homogeneous welds, where the base
and weld metals are of similar composi-tion, structure and lattice parameter,there is little or no free-energy barrier tothe phase change from liquid to solid.This is a result of the complete wetting ofthe substrate by the molten weld metal,and the nearly ideal substrate/interfaceprovided by the partially melted HAZgrains at the fusion boundary from whichthe solid can nucleate and grow. Asstated previously, it is generally acceptedthat growth of a solid in the molten weldpool evolves as atoms in the molten weldpool arrange themselves on the partiallymelted HAZ grains at the fusion bound-ary (Refs. 7, 18, 19). In this manner, HAZgrains at the fusion boundary are ex-tended at the expense of the liquid in theweld pool as the weld cools. In homoge-neous welds, the crystallographic struc-ture and orientation of the HAZ grains ismaintained without alteration across thefusion boundary. However, in DMWs,where base and weld metal may exhibitlarge differences in both compositionand structure, nucleation events requir-ing small undercoolings may be neces-sary to initiate growth of the solid at thefusion boundary. The nature and charac-ter of these events will be governed bydifferences in composition, crystal struc-ture and lattice parameter between thebase and weld metal adjacent to the fu-sion boundary. As these become increas-ingly different, the nature and characterof the fusion boundary in DMWs may de-viate from those observed in homoge-neous welds.
The exact nature of the fusion bound-ary at the onset of solidification inDMWs, or any weld, is difficult to ascer-tain. The fact that it is difficult to analyzestructures and boundaries at these hightemperatures, coupled with the nonequi-
librium conditions associated with weld-ing and the on-cooling transformations iniron and steels, make this particular taskvery difficult. In the present investigation,information regarding the nature of thefusion boundary was obtained usingType 409 ferritic stainless and AISI 1080pearlitic steels as base metal substrates.The results presented from these twobase metals provided information aboutthe nature of the elevated-temperaturefusion boundary between dissimilar met-als.
The fusion boundary morphologiesobserved and analyses presented in thepresent investigation are in contrast tothe traditional weld metal epitaxialgrowth morphologies observed in homo-geneous weld metal microstructures. Inthe photomicrographs (Figs. 4, 5) andEBSP analyses (Figs. 6, 7) presented, sev-eral unique characteristics are observedwith respect to the fusion boundary. Inthe Type 409/Monel system (Fig. 3 and 4),these would include 1) the lack of corre-lation or continuity of HAZ grain bound-aries and SGBs at the fusion boundaryand 2) the presence of numerous smallequiaxed weld metal grains growing epi-taxially from individual HAZ grains. Inthe 1080/Monel system, despite the dra-matic difference in composition betweenbase and weld metals, evidence of epi-taxial growth is observed along the fusionboundary.
It seems intuitive that any lack of cor-relation between HAZ grain boundariesand SGBs in DMWs on iron or steels is aresult of the on-cooling transformations.As these transformations move throughthe microstructure and across the fusionboundary, the original higher-tempera-ture fusion boundary morphologies aredisguised. As the weld cools, the delta-
WELDING RESEARCH SUPPLEMENT | 335-s
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
ferrite-to-austenite and austenite-to-alpha-ferrite transformations wouldchange the morphology of the HAZ mi-crostructure, making it difficult to distin-guish any relationship between the orig-inal δ-ferrite HAZ grain boundaries andSGBs along the fusion boundary. How-ever, these transformations were not pre-sent in Type 409, and there is no evi-dence of the typical weld metal epitaxialgrowth, i.e., HAZ grain boundaries andSGBS are not continuous at the fusionboundary.
The following mechanism is pro-posed by which these morphologiesevolve. The heterogeneous nucleation ofsolid nuclei occurs in the molten weldpool along the partially melted HAZgrains, creating a weld “chill zone” sim-ilar to a casting. The grain boundary mor-phologies and orientation analyses willaid in the justification of such nucleationevents and the weld chill-zone-type mor-phologies so often observed in DMWs
It is widely agreed upon there is littleprobability of any significant undercool-ing necessary to initiate solidification inmost welds. This stems from the conceptthat the partially melted grains at the fu-sion boundary provide an almost idealsubstrate for the solid to grow from. How-ever, the possibility of nucleation eventsoccurring in DMWs should not be alto-gether disregarded. Two elements thatlend support to the heterogeneous nucle-ation mechanism are 1) the difference ininterfacial energies between base andweld metals as a result of differences incomposition and 2) the abundance of het-erogeneous sites provided by the partiallymelted grains at the fusion boundary.
As BMD decreases, creating a largercompositional difference between thebase and weld metals, the interfacial en-ergies between the base and weld metalsbecome increasingly different. InDMWs, the addition of alloying elementswill undoubtedly change the composi-tion and, thus, the interfacial energies be-tween substrate and weld metal. As thisoccurs, the wetting angle (θ) subtendedbetween base and weld metal increasesand the free energy of formation of solidincreases (Equations 2 and 3). As the con-tact angle increases and the liquid is un-able to completely wet the surface, thefree-energy barrier (that is, the necessaryundercooling for changing from a liquidto solid phase) will also increase. Thisdoes not imply that homogeneous nucle-ation must occur to form a solid as thiswould require large undercooling in themelt; rather, the typical epitaxial growthobserved in homogeneous welds maynot occur. Therefore, there must be othermeans of forming solid nuclei at the fu-sion boundary to initiate solidification in
the melt. The most likely mechanism isheterogeneous nucleation of solid nucleialong the fusion boundary initiating theonset of solidification in the weld pool.
In reality, there are a host of differentheterogeneous sites on any substratefrom which a solid can nucleate. As dis-cussed previously, the free-energy barrieris reduced dramatically when nucleiform at heterogeneous sites; thus, mostsolids nucleate in the liquid only after asmall degree of undercooling (Ref. 1). Itis a good assumption that the partiallymelted HAZ grains at the fusion bound-ary would provide numerous heteroge-neous nucleation sites, such as interfacesteps, particles, HAZ grain boundaryjunctions, inclusions, precipitates, etc.This would explain the fusion boundarymorphologies observed in the Type409/Monel system, where numerousweld metal grains are observed growingfrom individual base metal grains. Eachof these weld metal grains has nucleatedat some heterogeneity on the HAZ grainsat the weld periphery. Because the nucleiform on heterogeneous sites, the nucle-ation barrier is minimal, requiring little orno undercooling in the melt. Similar to acasting, the weld metal grains exhibit noorientation relationship with the sub-strate grains. This is evident in Fig. 7 andin Table 3, where the HAZ and weldmetal grains share no common OR. Like-wise, there is no grain boundary misori-entation correlation between HAZ andsolidification grain boundaries (Table 2).This produces the “chill zone” mi-crostructure described previously.
Once nuclei have formed at the fu-sion boundary, the weak [100] fiber tex-ture observed in Fig. 7 suggests solidifi-cation proceeds in a typical manner. Incubic materials, solidification generallyoccurs along an easy growth direction,i.e., <100>, in the direction of the steep-est temperature gradient, i.e., toward thetop of Fig. 6, which corresponds to thetop and center of the weld. Similar re-sults would be expected in a casting orin homogeneous welds where competi-tive growth leads toward a weak <100>fiber texture.
Even more perplexing is the observa-tion that an individual weld metal graincan grow epitaxially from two adjacentHAZ grains of large misorientation, as isevident in Fig. 4. In this case, the weldmetal grains may grow from a particularHAZ grain with which it shares a pre-ferred misorientation, but any relationwith an adjacent HAZ grain may be a“closest match” type boundary. This issimilar to the way in which ferrite nucle-ates at austenite grain boundaries (Ref.1). In fact, the SGBs in Fig. 4 intersect thefusion boundary in the center of two HAZ
grains, probably sharing a specific ORwith one grain, while adopting a closestmatch OR with the other grain.
AISI 1080/Monel System
Although the Type 409/Monel systemdemonstrated the effects of compositionand crystal structure on the nature andevolution of the fusion boundary, the re-sults obtained from the 1080/Monel sys-tem demonstrate which of these charac-teristics has the greatest effect. In the1080 pearlitic steel, δ-ferrite does notform at elevated temperature, therebyproviding an austenitic substrate fromwhich solidifying austenitic weld metalgrows epitaxially. Although the base andweld metal exhibit the same crystal struc-ture, the weld metal had been sufficientlyalloyed by the filler metal additions as toproduce dramatic differences betweenbase and weld metal composition. De-spite the dramatic difference in composi-tion, evidence of epitaxial growth frompartially melted HAZ grains was ob-served in the as-welded microstructuresproduced on 1080 pearlitic steel. Theseresults were presented in Fig. 5, wherethe HAZ grain boundaries (decorated bythe retained austenite and indicated bythe arrows) and SGBs are continuousalong the fusion boundary.
As the first solid forms on the partiallymelted HAZ grains, the crystal structuresof base and weld metal are the same.Likewise, lattice parameters between thebase and weld metals are presumablyvery similar, even though the compositionis vastly different. Once the latent heat offusion has been liberated, the atoms in theweld metal arrange themselves on thesubstrate provided by the partially meltedHAZ grains without altering the crystallo-graphic structure or orientation. Thus, theHAZ grains and grain boundaries are ex-tended across the fusion boundary with-out alteration, similar to that observed inhomogeneous welds.
Summary and Practical Implications
As discussed above, the nature of theelevated-temperature fusion boundary inDMWs depends on the characteristic dif-ferences between base and weld metals.Of these, the crystal structure is the pri-mary influence and lattice parameter asecondary influence governing the na-ture and evolution of the fusion bound-ary microstructure. To reduce the free-energy barrier of forming a solid in theweld pool of DMWs, nonepitaxial, het-erogeneous nucleation occurs at the sub-strate interface when the crystal struc-tures are dissimilar (FCC/BCC).Heterogeneous sites may be HAZ grain
336-s | OCTOBER 1999
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
boundary junctions, interfacial steps ordiscontinuities, secondary phases, etc.Because these nuclei form at heteroge-neous sites on the HAZ interface, theymay or may not exhibit those preferredorientation relationships observed inFCC/BCC systems. As a result, the fusionboundary is likely to exhibit random mis-orientations between base and weldmetal grains.
It was the intent of the present inves-tigation to gain a better understanding ofthe nature and evolution of the fusionboundary during the initial stages of so-lidification in DMWs. From this basis, itis possible to postulate the theories andmechanisms that explain the nature andevolution of those boundaries and theirsusceptibility to failure in DMWs in en-gineering applications, such as pressurevessel steels clad with austenitic stainlesssteels. A subsequent paper will explainthe nature and evolution of the fusionboundary during the cooling portion ofthe weld thermal cycle and the effects ofthe δ–γ and γ–α transformations. Like-wise, the nature and evolution of theType II boundary and its susceptibility tocracking will be explained.
Conclusions
1) The effect of base metal and weldmetal microstructure at elevated temper-ature significantly influences the natureand evolution of the fusion boundary mi-crostructure in dissimilar metal welds.
2) When the base and weld metal ex-hibit different crystal structures(BCC/FCC) at the solidification tempera-ture, nucleation of solid weld metal oc-curs on heterogeneous sites on the par-tially melted HAZ grain at the fusionboundary.
3) When different crystal structuresare present, the fusion boundary exhibits
random misorientations between baseand weld metal grains as a result of het-erogeneous nucleation in the weld pool.The various misorientations along the fu-sion boundary may or may not exhibitsome of those preferred BCC/FCC rela-tionships, i.e., Bain, Kurdjimov-Sachs,and Nishiyama-Wasserman.
4) When the base and weld metal ex-hibit the same crystal structure at thesolidus temperature, standard weldmetal epitaxial growth may occur despitesignificant differences in composition.
5) Although the fusion boundary doesnot exhibit the typical cube-on-cube rela-tionship, weld solidification proceeds inthe normal <100> easy growth direction.
Acknowledgments
This work was supported by Dr. JohnC. Lippold at The Ohio State University.The authors wish to thank INCO Alloysfor providing the filler metal used in thisinvestigation. Technical support pro-vided by the Welding and Joining Metal-lurgy Group at The Ohio State Universitywas greatly appreciated.
References
1. Shewmon, P. G. 1983. Transformationsin Metals. Jenks, Okla.: J. Williams Book Co.
2. Chalmers, B. 1967. Principles of Solidi-fication. New York, N.Y.: John Wiley & Sons,Inc.
3. Flemings, M. C. 1974. SolidificationProcessing. New York, N.Y.: McGraw-HillPublishing Co.
4. Kou, S. 1987. Welding Metallurgy. NewYork, N.Y.: John Wiley & Sons, Wiley Inter-science.
5. Savage, W. F. 1980. Solidification, seg-regation and weld imperfections. Welding inthe World 18(5-6): 89–114.
6. Savage, W. F., and Hrubec, W. J. 1972.Synthesis of weld solidification using crys-talline organic materials. Welding Journal51(5): 260-s to 271-s.
7. Savage, W. F., Lundin, C. D., and Aron-son, A. H. 1965. Weld metal solidificationmechanisms. Welding Journal 44(4):175-s to181-s.
8. Savage, W. F., and Aronson, A. H. 1966.Preferred orientation in the weld fusion zone.Welding Journal 45(2):85-s to 89-s.
9. Samuel, J. 1979. Crystallography in fu-sion-weld-metal solidification mechanics.Ph.D. diss. Rensselaer Polytechnic Institute,Troy, N.Y.
10. Lippold, J. C., Clark, W. A. T., and Tu-muluru, M. 1992. An investigation of weldmetal interfaces. The Materials Science ofJoining. Edited by M. J. Cieslak, J. H. Perepe-zlo, and M. E. Glicksman, The Minerals, Met-als & Materials Society, pp. 141–145.
11. Porter, D. A., and Easterling, K. E. 1981.Phase Transformations in Metals and Alloys.Berkshire, England: Van Nostrand ReinholdInternational.
12. Adams, B. L. 1993. Orientation imag-ing microscopy: Application to the measure-ment of grain boundary structure. MaterialsScience and Engineering A166: 59–66.
13. Wright, S. I. 1993. A review of auto-mated orientation imaging microscopy (OIM).Journal of Computer-Assisted Microscopy5(3):207–221.
14. Adams, B. L., Wright, S. I., and Karsten,K. 1993. Orientation imaging: The emergenceof a new microscopy. Met. Trans. A 24A:819–831.
15. Jackson, K. A., Hunt, J. D., Uhlmann,D. R., and Seward, T.P., III. 1966. On the ori-gin of the equiaxed zone in castings. Trans.AIME 236(2):149–158.
16. Bower, T. F., and Flemings, M. C. 1967.Structure of dendrites at chill surfaces. Trans.AIME 239:1620–1625.
17. Hellawell, A., and Herbert, P. M. 1962.Proceeding of the Royal Society, A269.
18. Savage, W. F., Nippes, E. F., and Erick-son, J. S. 1976, Solidification mechanisms infusion welds. Welding Journal 55(8):213-s to221-s.
19. Davies, G. J., and Garland, J. G. 1975.Solidification structures and properties of fu-sion welds. International Metallurgical Re-views 20:83–106.
WELDING RESEARCH SUPPLEMENT | 337-s
RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
/RE
SE
AR
CH
/DE
VE
LO
PM
EN
T/R
ES
EA
RC
H/D
EV
EL
OP
ME
NT
To Order Custom Reprints ofArticles in the
Welding Journal
Call Denis Mulliganat (800) 259-0470
REPRINTS REPRINTS
REPRINTS REPRINTS