Mo in Stainless Steels Welds

12
WELDING RESEARCH -s 281 WELDING JOURNAL ABSTRACT. Superaustenitic stainless steel alloys can often pose difficulties dur- ing fusion welding due to the unavoidable microsegregation of Mo and tramp ele- ments, which lead to the loss of corrosion resistance and solidification cracking, re- spectively. A method of producing austenitic welds is proposed that can po- tentially circumvent these issues by de- signing fusion zone compositions that ex- hibit a primary ferrite solidification mode and subsequent solid-state transformation of ferrite to austenite. The ferritic solidifi- cation mode will minimize microsegrega- tion during solidification due to elevated diffusion rates, while a subsequent solid- state transformation of ferrite into austen- ite will create the austenitic matrix that is desired for good toughness. Thermody- namic calculations were used to isolate the range of compositions over which this phase transformation sequence can occur in Fe-Ni-Cr-Mo alloys. Experimental stainless steel alloys with a wide range in Ni, Cr, and Mo concentrations were then prepared with an arc button melting tech- nique to observe the microstructures and validate the thermodynamic diagrams. Four solidification modes (A, AF, FA, F) and three solid-state transformations (δ→ γ, δ→ (σ+γ), and γ→ martensite) were observed in this alloy system that pro- duced a wide variety of microstructures. Good agreement was shown between ex- periment and thermodynamic calcula- tions in the prediction of solidification mode. The amount of ferrite was also de- termined in each alloy via magnetic mea- surements. Empirical relations were as- sessed that relate the ferrite content to alloy composition, and the data were used for comparison with several weld constitu- tion diagrams. Several alloys were identi- fied that exhibited the desired transfor- mation sequence. Electron probe microanalysis measurements on these al- loys confirmed that Mo was more uni- formly distributed compared to alloys that solidified as austenite. Laser beam welds were also deposited on the surfaces of the button melts in order to observe the influ- ence of higher cooling rates. While no so- lidification mode shifts occurred (with the possible exception of one alloy), the high cooling rate inherent to laser welding caused a δ→γ massive transformation. This massive structure exhibited an en- tirely austenitic microstructure with a uni- form distribution of Mo at the nominal concentration. Introduction An advanced class of materials known as superaustenitic stainless steel (SASS) has recently fallen into wider use due to its excellent corrosion resistance and tough- ness. The decreased maintenance costs as- sociated with these materials have made them attractive to the pharmaceutical, food processing, and desalinization indus- tries. One difficulty associated with the in- corporation of SASS alloys has been the fabrication step of fusion welding. The melting and resolidification of these alloys has a considerable effect on mechanical and physical properties. Fusion welding is typically conducted on SASS alloys using high-Mo nickel-based filler metal alloys such as IN622 and IN625 (Ref. 1). The mi- crosegregation of Mo solute produces Mo-depleted dendrite cores that show a greater susceptibility to localized corro- sion (Ref. 2); the Ni-based filler metals contain high levels of Mo in order to help compensate for this detrimental effect. Welding parameters have been shown to exhibit a significant effect on weld metal composition, segregation behavior, corro- sion resistance, and weldability. There- fore, the processing parameters used dur- ing welding must be carefully controlled in order to ensure the deposition of welds that do not corrode easily or experience solidification cracking. Several of the ef- fects that welding parameters and filler metal composition have on SASS fusion welds have been studied extensively and recently published (Refs. 1, 3, 4). From this research, it was shown that a number of practical problems accompany the use of the high-Mo, nickel-based filler metals that solidify in the fully austenitic mode, including dendritic core corrosion brought on by the unavoidable microseg- regation of Mo during primary austenite solidification; relatively high solidification cracking susceptibility; high levels of pre- cision are required of the welding para- meters in order to attain the required weld composition; and a unit cost for the filler metal far beyond that of the stainless steel base metal, owing to the disproportionate levels of Ni and Mo present. Dendrite tip undercooling associated with high-energy-density welding processes can potentially minimize or eliminate Mo microsegregation and the associated accelerated corrosion of the Mo-depleted dendrite cores. Recent re- search has shown that the corrosion resis- tance will be enhanced through the use of laser welding (Ref. 5). The microstructure that produces this effect can be attained through proper control of the laser pro- cessing parameters. Despite this improved corrosion resistance, the laser welding process is not without its practical limita- tions, which include the following: laser welds are not accommodating of all joint designs, particularly fillet welds; achieving the desired microstructure requires pre- cise control of laser processing parameters and will depend on the plate thickness and joint design; implementing laser weld KEYWORDS GTAW Molybdenum Superaustenitic Stainless Steel Solidification Cracking Cracking T. D. ANDERSON, J. N. DUPONT, and A.R. MARDER are with Lehigh University, Bethlehem, Pa. M. J. PERRICONE is with RJ Lee Group, Inc., Pittsburgh, Pa. The Influence of Molybdenum on Stainless Steel Weld Microstructures A method of producing austenitic welds is proposed that can potentially circumvent the difficulties normally encountered when fusion welding superaustenitic stainless steels BY T. D. ANDERSON, M. J. PERRICONE, J. N. DUPONT, and A. R. MARDER

description

Mo in Stainless Steels Welds

Transcript of Mo in Stainless Steels Welds

Page 1: Mo in Stainless Steels Welds

WELDING RESEARCH

-s281WELDING JOURNAL

ABSTRACT. Superaustenitic stainlesssteel alloys can often pose difficulties dur-ing fusion welding due to the unavoidablemicrosegregation of Mo and tramp ele-ments, which lead to the loss of corrosionresistance and solidification cracking, re-spectively. A method of producingaustenitic welds is proposed that can po-tentially circumvent these issues by de-signing fusion zone compositions that ex-hibit a primary ferrite solidification modeand subsequent solid-state transformationof ferrite to austenite. The ferritic solidifi-cation mode will minimize microsegrega-tion during solidification due to elevateddiffusion rates, while a subsequent solid-state transformation of ferrite into austen-ite will create the austenitic matrix that isdesired for good toughness. Thermody-namic calculations were used to isolate therange of compositions over which thisphase transformation sequence can occurin Fe-Ni-Cr-Mo alloys. Experimentalstainless steel alloys with a wide range inNi, Cr, and Mo concentrations were thenprepared with an arc button melting tech-nique to observe the microstructures andvalidate the thermodynamic diagrams.Four solidification modes (A, AF, FA, F)and three solid-state transformations (δ→γ, δ → (σ+γ), and γ → martensite) wereobserved in this alloy system that pro-duced a wide variety of microstructures.Good agreement was shown between ex-periment and thermodynamic calcula-tions in the prediction of solidificationmode. The amount of ferrite was also de-termined in each alloy via magnetic mea-surements. Empirical relations were as-sessed that relate the ferrite content toalloy composition, and the data were usedfor comparison with several weld constitu-tion diagrams. Several alloys were identi-fied that exhibited the desired transfor-mation sequence. Electron probe

microanalysis measurements on these al-loys confirmed that Mo was more uni-formly distributed compared to alloys thatsolidified as austenite. Laser beam weldswere also deposited on the surfaces of thebutton melts in order to observe the influ-ence of higher cooling rates. While no so-lidification mode shifts occurred (with thepossible exception of one alloy), the highcooling rate inherent to laser weldingcaused a δ → γ massive transformation.This massive structure exhibited an en-tirely austenitic microstructure with a uni-form distribution of Mo at the nominalconcentration.

Introduction

An advanced class of materials knownas superaustenitic stainless steel (SASS)has recently fallen into wider use due to itsexcellent corrosion resistance and tough-ness. The decreased maintenance costs as-sociated with these materials have madethem attractive to the pharmaceutical,food processing, and desalinization indus-tries. One difficulty associated with the in-corporation of SASS alloys has been thefabrication step of fusion welding. Themelting and resolidification of these alloyshas a considerable effect on mechanicaland physical properties. Fusion welding istypically conducted on SASS alloys usinghigh-Mo nickel-based filler metal alloyssuch as IN622 and IN625 (Ref. 1). The mi-crosegregation of Mo solute producesMo-depleted dendrite cores that show agreater susceptibility to localized corro-sion (Ref. 2); the Ni-based filler metalscontain high levels of Mo in order to help

compensate for this detrimental effect.Welding parameters have been shown toexhibit a significant effect on weld metalcomposition, segregation behavior, corro-sion resistance, and weldability. There-fore, the processing parameters used dur-ing welding must be carefully controlled inorder to ensure the deposition of weldsthat do not corrode easily or experiencesolidification cracking. Several of the ef-fects that welding parameters and fillermetal composition have on SASS fusionwelds have been studied extensively andrecently published (Refs. 1, 3, 4). Fromthis research, it was shown that a numberof practical problems accompany the useof the high-Mo, nickel-based filler metalsthat solidify in the fully austenitic mode,including dendritic core corrosionbrought on by the unavoidable microseg-regation of Mo during primary austenitesolidification; relatively high solidificationcracking susceptibility; high levels of pre-cision are required of the welding para-meters in order to attain the required weldcomposition; and a unit cost for the fillermetal far beyond that of the stainless steelbase metal, owing to the disproportionatelevels of Ni and Mo present.

Dendrite tip undercooling associatedwith high-energy-density weldingprocesses can potentially minimize oreliminate Mo microsegregation and theassociated accelerated corrosion of theMo-depleted dendrite cores. Recent re-search has shown that the corrosion resis-tance will be enhanced through the use oflaser welding (Ref. 5). The microstructurethat produces this effect can be attainedthrough proper control of the laser pro-cessing parameters. Despite this improvedcorrosion resistance, the laser weldingprocess is not without its practical limita-tions, which include the following: laserwelds are not accommodating of all jointdesigns, particularly fillet welds; achievingthe desired microstructure requires pre-cise control of laser processing parametersand will depend on the plate thickness andjoint design; implementing laser weld

KEYWORDS

GTAWMolybdenumSuperaustenitic Stainless SteelSolidification CrackingCracking

T. D. ANDERSON, J. N. DUPONT, and A.R.MARDER are with Lehigh University, Bethlehem,Pa. M. J. PERRICONE is with RJ Lee Group,Inc., Pittsburgh, Pa.

The Influence of Molybdenum on Stainless Steel Weld Microstructures

A method of producing austenitic welds is proposed that can potentially circumvent the difficulties normally encountered

when fusion welding superaustenitic stainless steels

BY T. D. ANDERSON, M. J. PERRICONE, J. N. DUPONT, and A. R. MARDER

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:21 PM Page 281

Page 2: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s282

tooling requires a large capital invest-ment; and the high travel speeds requiredto induce dendrite tip undercooling andreduced levels of microsegregation alsotypically produce very small welds, and arethus not amenable to many practical jointdesigns and base metal thicknesses.Clearly, each of the current solutions forjoining SASS alloys exhibit practical diffi-culties that can hinder their use in appli-cations where good corrosion resistanceof a welded assembly is critical.

One potential approach for reducing theproblems of Mo microsegregation and so-lidification cracking in Mo-bearing SASSalloys is to design an alloy or filler metalthat exhibits a primary ferrite solidifica-tion mode and subsequently transforms toaustenite by a solid-state reaction. Such al-loys are known to exist in the simple Fe-

Ni-Cr ternary system (Ref. 6). This solidi-fication sequence provides two primaryadvantages. First, the primary ferrite so-lidification mode does not segregate Moto the liquid. Rather, Thermo-Calc calcu-lations show that ferrite is expected to beenriched in Mo, since the partition coeffi-cient of Mo in ferrite is slightly greaterthan unity (Refs. 7, 8). The diffusion ratesof substitutional alloying elements such asMo and Cr are typically two orders of mag-nitude higher in ferrite than in austenite(Ref. 9). As a result, significant diffusionof Mo can occur during primary ferrite so-lidification and reduce or eliminate anyMo concentration gradient. The high dif-fusion rates will also permit the backdiffu-sion of nickel, which is segregated to theliquid by the solidifying ferrite. Second,ferrite is known (Refs. 10, 11) to exhibit

much higher solubility of the tramp ele-ments P and S. These elements are knownto cause solidification cracking due to theformation of P- and S-rich liquid films inthe interdendritic and grain boundary re-gions at the end of solidification. Alloysthat solidify in the primary ferrite modeare known (Ref. 10) to have significantlyreduced susceptibility to solidificationcracking due to the higher solubility of Pand S and concomitant reduced amount ofsolute-rich liquid films.

The eventual development of alloys re-quires a sound understanding of the influ-ence of composition on the phase trans-formation sequences and resultantmicrostructure. Thus, the objective of thisresearch is to develop a basic understand-ing of microstructural evolution in Fe-Ni-Cr-Mo alloys that will serve as the basis for

Fig. 1 — Separate views of the Fe-Ni-Cr-6Mo phase diagram.A — Vertical isopleth at 69 wt-% Fe; B — phase stability dia-gram highlighting Fe-Ni-Cr compositions with 6 wt-% Mo.The compositional zone of interest is highlighted in both diagrams.

Fig. 2 — Multicomponent phase stability diagrams of the Fe-Ni-Cr-Mo quaternary system,for each level of Mo content studied. The data points in each diagram represent experi-mental alloy compositions, with the shape of the data point signifying the observed solid-ification mode. The measured wt-% ferrite for each composition is included alongside therespective data point.

A

B

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:22 PM Page 282

Page 3: Mo in Stainless Steels Welds

WELDING RESEARCH

-s283WELDING JOURNAL

development of SASS alloys with im-proved corrosion resistance and weldabil-ity. This paper describes the overall influ-ence of alloy composition on the resultingmicrostructure. Details on various phasetransformations observed from this re-search can be found in Refs. 12 and 13.

Experimental Procedure

Numerous vertical isopleths, such asthat seen in Fig. 1A, were constructed ofthe Fe-Ni-Cr-Mo system using the CAL-PHAD software Thermo-Calc (Ref. 7) inconjunction with the Iron Alloy Database(Ref. 8). While calculations were per-formed with Mo additions of 0, 2, 4, 6, 8,and 10 wt-%, the number of possiblephases present was limited to three: liquid,ferrite, and austenite. The only solidphases included in the calculations wereferrite and austenite, as these are the onlyprimary solidification phases present inthe composition space considered here.The σ-phase was included in later calcula-tions in order to ascertain the develop-ment behavior of this phase, but was notincluded in the generation of phase stabil-ity diagrams in the following discussion.These calculations were used to generatecomposite phase stability diagrams thatcould be used to estimate the influence ofalloy composition on the solidificationmode, possible solid-state transformation,and resultant microstructure. Develop-ment and use of these diagrams is dis-cussed in the next section.

Laboratory studies were conducted inorder to verify the accuracy of the dia-grams in predicting microstructural devel-opment and identify alloys that exhibitedthe desired transformation sequence de-scribed above. A total of 95 alloys with sys-tematic variations in Ni and Cr across bothsides of the eutectic line were prepared foreach of the six nominal Mo compositionssimulated using Thermo-Calc. A masteralloy of Fe-28Cr together with virgin Fe,Ni, and Mo were combined using an arc-button melter (ABM), supplied by Ther-mal Technologies, Inc. The bell jar of theABM was evacuated and backfilled withAr shielding gas. A manually controlledgas tungsten arc welding (GTAW) torchrunning at 300 A and 10 V inside thechamber was manipulated to melt the ele-mental components on a water-cooled Cuhearth. The target alloy compositions areshown in Table 1. The ~50-g buttons weresubject to metallographic preparation andsubsequent microstructural analysis inorder to reveal and identify the particularsolidification mode and solid-state trans-formation mechanisms. The relative loca-tions and shape of the ferrite and austen-ite phases were used as indications of thesolidification mode and morphological

Table 1 — Nominal Compositions of Experimental Alloys, as Measured by Wet-Chemical Analysis(a)

Target Fe Ni Cr Mo SM wt-% Ferrite

0Mo-10Cr-10Ni 79.86 10.26 9.74 0 M —0Mo-12Cr-8Ni 80.16 8.12 11.56 0 M —0Mo-14Cr-6Ni 80.37 6.01 13.45 0 M —0Mo-16Cr-4Ni 80.23 3.97 15.59 0 M —0Mo-13Cr-12Ni 75.07 12.16 12.6 0 M —0Mo-15Cr-10Ni 75.27 10.13 14.4 0 M —0Mo-17Cr-8Ni 75.71 7.97 16.1 0 M —-0Mo-19Cr-6Ni 75.64 6.01 18.09 0.01 M —0Mo-16Cr-14Ni 70.61 13.85 15.34 0.01 A 00Mo-18Cr-12Ni 70.86 11.45 17.43 0.02 AF 2.2890Mo-20Cr-10Ni 70.14 10.02 19.59 0.02 F 10.530Mo-22Cr-8Ni 70.77 7.89 21.06 0.01 F 55.130Mo-18Cr-17Ni 65.69 17.01 16.98 0.01 A 00Mo-20Cr-15Ni 65.1 15.3 19.32 0.01 AF 0.21750Mo-22Cr-13Ni 66 12.55 21.18 0.01 FA 9.2050Mo-24Cr-11Ni 65.87 10.87 22.94 0.01 F 19.22

2Mo-8Cr-12Ni 78.36 11.77 7.96 1.78 M —2Mo-10Cr-10Ni 77.39 10.33 10.27 1.86 M —2Mo-12Cr-8Ni 73.52 9.7 14.55 2.05 M —2Mo-16Cr-4Ni 78.96 3.85 15.31 1.66 M —2Mo-11Cr-14Ni 72.78 14.16 11.1 1.79 A 02Mo-13Cr-12Ni 73.35 11.96 12.71 1.78 M —2Mo-15Cr-10Ni 73.54 10.04 14.8 1.4 M —2Mo-17Cr-8Ni 73.11 7.88 17.06 1.71 M —2Mo-14Cr-16Ni 69.45 15.49 13.19 1.57 A 02Mo-16Cr-14Ni 68.24 13.76 15.58 2.19 AF 0.6172Mo-18Cr-12Ni 66.31 12.6 18.55 2.22 FA 7.9252Mo-20Cr-10Ni 66.91 10.26 20.92 1.6 F 15.332Mo-18Cr-17Ni 62.14 17.3 17.86 2.39 AF 0.1782Mo-20Cr-15Ni 62.17 15.02 20.33 2.12 FA 5.3952Mo-22Cr-13Ni 62.44 12.65 22.6 1.97 F 14.172Mo-24Cr-11Ni 63.75 10.85 23.3 1.78 F 50.255

4Mo-6Cr-14Ni 75.73 14.59 6.18 3.37 M —4Mo-8Cr-12Ni 76.77 11.81 7.92 3.36 M —4Mo-12Cr-8Ni 77.33 7.95 11.64 2.9 M —4Mo-14Cr-6Ni 76.61 6.06 13.85 3.28 M —4Mo-9Cr-16Ni 71.31 15.99 9 3.55 A 04Mo-11Cr-14Ni 71.08 14.21 11.06 3.46 A 04Mo-13Cr-12Ni 72.58 11.99 12.55 2.68 AF 1.3974Mo-17Cr-8Ni 72.99 7.8 15.83 3.13 F 50.524Mo-13Cr-17Ni 67.28 16.82 12.43 3.27 A 04Mo-15Cr-15Ni 67.9 14.83 14.46 2.58 AF 0.3754Mo-17Cr-13Ni 66.59 12.51 16.57 3.95 FA 7.314Mo-19Cr-11Ni 66.42 10.88 18.98 3.42 F 17.8354Mo-16Cr-19Ni 61.08 19.05 15.68 3.95 A 04Mo-18Cr-17Ni 61.58 16.88 17.6 3.67 AF 0.72454Mo-20Cr-15Ni 59.91 14.75 20.63 4.26 FA 9.8154Mo-22Cr-13Ni 62.91 12.31 20.62 3.79 F 17.42

6Mo-4Cr-16Ni 73.57 16.2 4.02 5.98 M —6Mo-8Cr-12Ni 73.88 12.02 8.01 5.84 M —6Mo-10Cr-10Ni 73.6 10.21 10.64 5.27 M —6Mo-14Cr-6Ni 74.41 5.95 13.59 5.72 M —6Mo-9Cr-16Ni 68.73 16.33 9.18 5.58 A 06Mo-11Cr-14Ni 68.66 14.57 11.2 5.29 AF 0.1846Mo-13Cr-12Ni 68.06 12.23 13.53 5.87 FA 5.336Mo-15Cr-10Ni 69.01 10.31 14.83 5.48 F 13.0256Mo-12Cr-18Ni 63.37 18.64 12.2 5.48 A 06Mo-14Cr-16Ni 63.55 16.4 14.37 5.36 AF 0.1486Mo-16Cr-14Ni 62.25 14.37 17.29 5.75 FA 6.4756Mo-18Cr-12Ni 62.93 12.31 18.97 5.44 F 14.3856Mo-16Cr-19Ni 58.52 19.31 16.08 5.74 AF 06Mo-18Cr-17Ni 57.39 17.57 19.26 5.41 AF 0.51256Mo-20Cr-15Ni 58.77 15.18 19.7 5.95 F 12.186Mo-22Cr-13Ni 57.05 13.41 23.17 5.94 F 38.92

(a) All values are expressed in weight percent. The solidification mode and measured weight percentage of ferrite of the alloy is listed,unless the alloy was found to contain martensite. (A=austenite, AF=austenite-ferrite, FA=ferrite-austenite, F=ferrite, M=martensite.) — continued on page 285-s

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:25 PM Page 283

Page 4: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s284

type in a manner previously described byElmer (Ref. 14). A Feritscope®, manu-factured by Fischer Technology, Inc., wasused to determine the specific weight per-centage of ferrite within each alloy button.In addition, the Ferrite Number (FN) wasmeasured from the set of samples knownto solidify as primary ferrite for the pur-poses of comparison with the WRC-1992weld constitution diagram. Precision wasmaintained for both sets of measurementsthrough the use of calibrated standards ofknown ferrite content.

Experiments were also performed tostudy the influence of increased coolingrates on the type and degree of solid-statetransformations that would occur throughthe use of high-energy-density (HED)welds. Autogenous laser welds were con-ducted using a 700-W Nd-YAG laser(measured at 370 W at the weld surface)at travel speeds of 10, 50, and 100 in./min(4.2, 21, and 42 mm/s, respectively). The

laser weld passes introduced significantlyhigher cooling rates relative to that in theABM method. The following sectionsfocus entirely on results accumulatedfrom the arc button microstructures, whilethe results of microstructural analysis ofthe laser welds is discussed in the last section.

Dendrite arm spacing measurementswere used to estimate the cooling rate as-sociated with the ABM and laser weldingconditions. The relationship between den-drite arm spacing λ and cooling rate ε hasbeen determined (Ref. 15) for 310 stain-less steel and is given by the following:

λ = 80ε–0.3

Several alloys were selected with simi-lar composition and identical solidifica-tion mode (austenite) as 310 stainless steelin order to use the expression above to es-timate the cooling rate. Although

austenitic stainless steels possess lowerthermal conductivities than ferritic stain-less steels, it is assumed there was no largechange in thermal properties over therange of composition studied.

Electron probe microanalysis (EPMA)was conducted on select alloys for each ofthe observed solidification modes in orderto measure the distribution of each elementwithin the microstructure. Raw data wereconverted to concentration values using thecorrection scheme devised by Heinrich(Ref. 16). EPMA was also conducted onseveral of the HED weld zones to deter-mine elemental distribution. Positive phaseidentification was carried out usingbackscattered electron Kikuchi pattern(BEKP) analysis on the microstructures ofboth the ABM button and the laser weldpasses. This technique captures an image ofthe diffraction pattern of backscatteredelectrons while the sample undergoes SEMphotomicrography. Indexing of the pattern

A B

C D

Fig. 3 — Representative microstructures of the four solidification modes observed in the set of Fe-Ni-Cr-Mo alloys generated in this study. A — A mode: 0Mo-18Cr-17Ni; B — AF mode: 0Mo-18Cr-12Ni; C — FA mode: 2Mo-18Cr-12Ni; D — F mode: 2Mo-17Cr-8Ni.

Anderson Supp Sept 2007 corr:Layout 1 8/9/07 8:52 AM Page 284

Page 5: Mo in Stainless Steels Welds

WELDING RESEARCH

-s285WELDING JOURNAL

reveals the crystal structure characteristicof the phase (fcc: γ-austenite, bcc: δ-ferrite,tetragonal: σ-phase).

Results and Discussion

Phase Stability Diagrams

Specific points of interest in the calcu-lated vertical isopleths were used to gener-ate the phase stability diagrams. As shownin Fig. 1A, the compositions of the eutecticpoint and the maximum solubility of Cr inaustenite were taken from pseudo-binaryphase diagram at various concentrations ofFe. Alloy compositions to the right of theeutectic point will solidify as ferrite (assum-ing negligible dendrite tip undercooling),while compositions to the left of the maxi-mum solid solubility of Cr in austenite havethe thermodynamic potential for the ferrite-to-austenite transformation to go to com-pletion. Thus, alloys between these two keypoints represent the range of compositionsthat can potentially exhibit the desiredphase transformation sequence. Whenplotted on the axes of Ni and Cr concentra-tion, similar to a liquidus projection, thesetwo points form lines that bound the rangeof desired compositions. An example of amulticomponent phase stability diagram isshown in Fig. 1B. The band across the cen-ter describes the three vertices of the eutec-tic triangle. The dashed line that intersectsthis band represents the maximum solubil-ity of Cr in austenite, which curves below theeutectic triangle at high Cr and Ni concen-trations. This computational approach canbe used to bound a wide range of composi-tions that have the thermodynamic poten-tial to develop an austenitic matrix that ispreceded by ferrite solidification. Since thediagrams take into account only thermody-namic considerations, they are not profi-cient at predicting the mechanism oramount of transformation that will occur.This factor can only be found through ananalysis of the kinetics of the system, whichconsiders the cooling conditions present.Therefore, the diagrams represent a neces-sary, but not sufficient, condition for attain-ing an austenitic matrix alloy derived fromferritic solidification.

Compositions that meet both require-ments have been shaded for emphasis inboth Fig. 1A and B. The martensite linederived from the Schaeffler diagram isalso included, the basis for which is de-scribed below. The range of compositionsthat can be viewed simultaneously isgreater in this multicomponent phase sta-bility diagram than in the vertical iso-pleths. Figure 2 shows the entire array ofphase stability diagrams generated in thisstudy for each Mo content. The datapoints on each represent the experimentalalloys, the characterization results of

which are discussed in more detail below.Successively higher additions of Mo causethe eutectic triangle not only to widen, butalso to shift its location toward higher Nicontents. The latter effect is a result of theferrite-stabilizing effect of Mo compara-ble to that of Cr. In addition, the contin-ued addition of Mo widens the space be-tween the eutectic and the boundaryrepresenting the maximum solubility of Crin austenite, thereby broadening the rangeof compositions with the potential to ex-hibit the target behavior. The phase sta-bility diagrams calculated in this researchwere validated using metallographic ob-servations of each alloy. The results of theexperimental alloy data are placed on thephase stability diagram with the nearestMo content at the Cr and Ni contentsmeasured via wet chemical techniques.The type of solidification mode is denotedby the shape of the data point. The target

compositions for the alloys, the actualmeasured compositions, the resulting so-lidification mode, and the measuredweight-percent ferrite can be seen in Table1. Alloys containing martensite are notedas such in place of the solidification mode,and the corresponding ferrite contents areomitted since the measurements experi-enced interference from the magnetic sig-nature of martensite.

The calculated eutectic lines were fairlyaccurate in separating alloys between theaustenitic and ferritic primary solidifica-tion modes, especially in the lower Mo di-agrams. Although the eutectic line is seento curve, it runs nearly parallel to a Creq/Niratio of 1.5 over the compositional rangeof the experimental alloys. Nearly a thirdof the samples showed evidence of themartensite phase, as predicted by theSchaeffler diagram. The first indication ofthe presence of martensite was given by

Table 1 — Continued

Target Fe Ni Cr Mo SM wt-% Ferrite

8Mo-2Cr-18Ni 71.31 18.34 1.96 8.29 M —8Mo-6Cr-14Ni 72.11 13.72 5.75 8.27 M —8Mo-8Cr-12Ni 71.94 11.99 7.91 7.99 M —8Mo-12Cr-8Ni — — — — — —8Mo-7Cr-18Ni 66.76 18.02 6.91 8.15 A 08Mo-9Cr-16Ni 67.03 15.75 8.82 8.22 AF 08Mo-13Cr-12Ni 68.7 11.49 12.24 7.12 F 3.038Mo-15Cr-10Ni 69.12 9.3 13.71 7.61 F 26.898Mo-11Cr-19Ni 62.57 18.66 10.39 8.18 AF 08Mo-13Cr-17Ni 63.86 16.86 12.06 6.99 AF 08Mo-15Cr-15Ni 60.53 14.58 16.11 8.24 FA 1.928Mo-17Cr-13Ni 61.3 13.01 17.38 7.83 F 14.518Mo-14Cr-21Ni 58.1 20.3 13.21 8.14 AF 08Mo-17Cr-18Ni 57.71 17.88 17.03 7.08 AF 08Mo-20Cr-15Ni 58.1 14.32 19.4 7.85 F 13.178Mo-22Cr-13Ni 58.94 12.22 20.97 7.52 F 48.15

10Mo-2Cr-18Ni 69.44 18.24 1.92 10.29 M —10Mo-4Cr-16Ni 70.48 16.59 4.06 8.73 M —10Mo-8Cr-12Ni 70.36 11.69 7.71 10.06 F 8.02510Mo-12Cr-8Ni 69.79 7.78 11.83 10.37 F 65.7610Mo-5Cr-20Ni 64.88 20.08 4.93 9.96 AF 010Mo-9Cr-16Ni 65.3 16.14 8.98 9.39 AF 010Mo-13Cr-12Ni 65.69 11.76 12.56 9.42 F 4.810Mo-15Cr-10Ni 66.14 10.05 14.55 8.94 F 40.6310Mo-10Cr-20Ni 62.92 20.02 9.63 7.22 AF 010Mo-12Cr-18Ni 62.21 17.92 12.1 7.71 AF 010Mo-14Cr-16Ni 61.1 14.82 13.59 9.86 FA 0.399510Mo-16Cr-14Ni 61.29 14.1 16.57 8.55 F 8.9710Mo-14Cr-21Ni 55.73 20.9 13.37 9.74 AF 010Mo-16Cr-19Ni 54.8 18.88 15.84 10.18 AF 010Mo-18Cr-17Ni 53.08 16.84 18.44 10.95 FA 010Mo-20Cr-15Ni 56.66 14.69 19.72 8.51 F 8.355

Table 2 — Cell Spacing Measurements Used to Calculate Cooling Conditions in the Arc-Melt andHED Weld Conditions

Melt Travel Speed (mm/s) λ (μm) St. Dev. (μm) ε (K/s)

Button — 27 7 30LW1 4.2 2.6 0.3 3 × 104

LW2 21 1.8 0.2 1 × 105

LW3 42 1.4 0.2 2 × 105

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:32 PM Page 285

Page 6: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s286

the abnormally high magnetic signal inthese alloys, which could be attributed tothe magnetic properties of martensite;metallographic observations confirmedthe presence of this phase. After marten-site was identified, the Schaeffler marten-site line was added to the phase stabilitydiagrams to further confine the range ofdesired compositions. This boundary wasfound to provide a good estimation ofcompositions that are expected to formmartensite. Since the martensite phasewas not desirable in this particular appli-cation, microstructural analysis of these 28alloys is omitted in order to maintain focuson alloys composed predominately ofaustenite. A separate article (Ref. 12) dis-cusses the myriad microstructural devel-opment sequences involving martensitethat are possible in this alloy system. Ofthe 67 alloys that contained no martensite,only 6 were found to disagree with the pri-mary solidification modes predicted by the

phase stability diagrams. In all cases, theirnominal compositions were very close tothe calculated eutectic lines.

Weld Metal Microstructures

A detailed description of the widerange of transformation sequences and re-sultant microstructures of the experimen-tal alloys is presented elsewhere (Ref. 12),and only a summary of the pertinent re-sults are provided here. The matrix of al-loys was shown to contain each of the fourcommon solidification modes exhibited bystainless steels: 1) austenitic (A); 2)austenitic-ferritic (AF); 3) ferritic-austenitic (FA); and 4) ferritic (F). A rep-resentative microstructure for each ofthese modes can be seen in Fig. 3. The firsttwo images in Fig. 3A and B show alloyswith austenitic primary solidificationmodes. The A mode alloy in Fig. 3A is en-tirely austenite, while the AF mode alloy

in Fig. 3B contains interdendritic ferrite asa result of eutectic solidification. Solidifi-cation mode determination of primaryaustenite alloys can be difficult, as eutec-tic ferrite in AF mode alloys may bepurged by subsequent solid-state transfor-mation. In this work, only those alloys thatemitted a magnetic signature (measuredwith the ferrite detector) or otherwiseshowed microstructural evidence of fer-rite (see below) were designated with theAF mode.

The primary ferrite solidificationmodes were of greatest importance in thisstudy, since they fulfill the requirementsset forth previously. The difference be-tween the two modes is the existence ofaustenite at the end of solidification. FAmode alloys exhibit this interdendriticaustenite, from which the solid-statetransformation of ferrite into austenitemay ensue epitaxially (Ref. 14). A repre-sentative microstructure of the FA mode

Fig. 4 — Arc-melt microstructures of secondary solid-state transformations that occurred in the set of Fe-Ni-Cr-Mo alloys: A — A mode with martensite: 2Mo-8Cr-12Ni; B — AF mode with martensite and γ/σ eutectoid: 8Mo-13Cr-17Ni; C — FA mode with γ/σ eutectoid: 10Mo-18Cr-17Ni; D — F mode with γ/σ eu-tectoid: 10Mo-18Cr-17Ni.

A B

C D

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:34 PM Page 286

Page 7: Mo in Stainless Steels Welds

WELDING RESEARCH

-s287WELDING JOURNAL

is shown in Fig. 3C. The residual ferriteseen in the structure is located at whatwere formerly the dendrite cores. Accord-ingly, the centerline of the austenite re-gions represents the former location of in-terdendritic regions. The compositions ofall FA mode alloys were found to fall in theprescribed region of the phase stability di-agram between the eutectic line and themaximum solubility of Cr in austenite.

The rest of the primary ferrite alloysbelonged to the F mode group. An exam-ple of an F mode structure is shown in Fig.3D. Fully ferritic grains were consumed byaustenitic allotriomorphs at the grainboundaries, from which austenitic Wid-manstatten side-platelets grew. Results offerrite measurements indicate that thistransformation mechanism is not as profi-cient as the FA mode at reducing theamount of residual ferrite. This is due to alack of solidified ferrite in the structure, so

nucleation of thisphase is first required.Eleven alloys were ob-served to solidify in theFA mode composi-tions; the ferrite con-tents were nearly allbelow that measured inthe F mode samples.

Of special note isthe reduction in ferritecontent in primary fer-rite samples as the Mocontent is increased.Indeed, several 10 wt-% Mo F-mode samplescontain less than 10 wt-% ferrite, and one FAalloy with 10 wt-% Moproduced a reading of0 wt-% ferrite. Mi-crostructural examina-

Fig. 5 — The effect of Creq/Ni ratio on the measured weight percent of ferrite.The data points are separated by the type of solidification mode.

Fig. 7 — The Schaeffler 1949 weld constitution diagram, with the entire set ofexperimental alloys plotted by nominal composition.

Fig. 8 — Experimental Fe-Ni-Cr-Mo alloys plotted on the WRC-1992 weldconstitution diagram. The martensite boundary previously proposed byKotecki (Ref. 20) is included. A new boundary is also given that divides Aand AF mode alloys that contain significant amounts of Mo.

Fig. 6 — Microstructural map of alloy solidification mode based on Creq/Niratio and Mo content. Boundaries are included to separate regions by so-lidification mode and σ-content.

Fig. 9 — Comparison of experimental Ferrite Number (FN) measured viaFeritscope® and the FN predicted by the WRC-1992 diagram.

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:35 PM Page 287

Page 8: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s288

tion of these regions revealedthe presence of anotherphase: the high-Mo inter-metallic σ-phase. Figure 4shows light optical photomi-crographs of microstructuresassociated with the forma-tion of both sigma (σ) andmartensite. Positive phaseidentification of the σ-phasewas made through BEKP re-sults (Refs. 12, 13). Previousresearch has shown that a eu-tectoid σ/γ constituent maynucleate and grow from theparent ferrite phase (Refs.17, 18). This eutectoid struc-ture is observed in several so-lidification modes: AF (Fig.4B), FA (Fig. 4C), and F (Fig.4D). Such a structure is typi-cally associated with anneal-ing heat treatments that ex-pose stainless steel ferrite toelevated temperatures forlong durations. However, thehigh Mo contents of these al-loys coupled with the coolingconditions of the ABM weresufficient for this eutectoidtransformation to occurunder the conditions studiedhere. The dissolution of theeutectoid constituent in theHAZ of laser welds and itspresence in the residual fer-rite of primary ferrite solidi-fication microstructures indi-cated that it was the productof a solid-state transforma-

tion, and not as a solidification product. Amore detailed description of the δ → γ +σ transformation observed in these alloysis available elsewhere (Ref. 13). The pres-ence of the eutectoid constituent was alsoused as a secondary criterion for identify-ing AF mode alloys, as this transformationrequires a ferritic parent phase. There-fore, several alloys containing 0.0 wt-%ferrite were labeled as AF mode due to thetransformation of the interdendritic fer-rite into the σ+γ eutectoid constituent.

The austenitic cell spacing measure-ments are summarized in Table 2. It wasobserved that the cell spacing increasedthrough the thickness of the button. Theregion closest to the Cu hearth experi-enced the fastest cooling, which led to thesmallest dendrites observed in the button.Accordingly, all data were acquired fromthe central band of the button in order toreport the average cooling rate within theentire button. The photomicrographsgiven in Figs. 3 and 4 are from the same re-gion of the arc-melt button. The data in-dicate an average cell spacing of 27 μm,which represents an average cooling rateof approximately 30 K/s. This figure iswithin the range of casting and high-heat-input arc welding conditions (Ref. 14).This relatively low cooling rate would fa-cilitate the nucleation and growth of theγ/σ eutectoid structure.

Effects of Composition on Microstructureand Ferrite Content

As would be expected, the weight per-centage of ferrite measured in the alloysincreased as a function of Creq/Ni (Creq =

Fig. 10 — EPMA linescan data of the following: A — Solidifiedaustenite cells with interdendritic eutectic ferrite and eutectoid σ-phases produced by AF solidification mode (10Mo-12Cr-18Ni); B— solid-state transformation product austenite with eutectoid σ-phase in the residual ferrite, produced by FA solidification mode(10Mo-14Cr-16Ni); C — austenitic allotriomorph and Wid-manstatten platelets bordering a ferrite grain (6Mo-22Cr-13Ni).

A B

C

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:35 PM Page 288

Page 9: Mo in Stainless Steels Welds

WELDING RESEARCH

-s289WELDING JOURNAL

wt-% Cr + wt-% Mo). These data areplotted in Fig. 5 with the alloys differenti-ated according to their solidificationmode. These data indicate that each so-lidification mode belongs to a specificregime of Creq/Ni ratio. The boundary be-tween AF and FA mode alloys that isfound at 1.5 Creq/Ni is in good agreementwith previous findings (Ref. 10). Anothersignificant boundary can be seen betweenFA and F mode alloys at ~1.7 Creq/Ni.This value is slightly lower than the 1.9Cr/Ni limit that was reported for Mo-freealloys by Elmer et al. (Ref. 14). The solid-ification mode boundaries in the literaturewere developed with chemically complexalloys, often containing C, Si, or Mn. Thealloys studied herein represent simplifiedapproximations of stainless steel composi-tions, which should be noted when makingcomparisons. The plot also shows that theFA mode was far more proficient at re-ducing the amount of the primary solidi-fied ferrite, as the ferrite content in Fmode alloys ranged between 3 and 55 wt-% while the ferrite content of FA alloysnever surpassed 10 wt-%. A polynomialtrendline has been included in this figurethat relates the ferrite content to theCreq/Ni ratio. Only those alloys with a de-tectable amount of ferrite were includedin this relation.

The solidification mode and presenceof σ-phase are mapped out as a functionof Mo concentration and Creq/Ni ratio inFig. 6. As shown in Fig. 5, a switch in so-lidification mode from AF to FA occurs ata Creq/Ni ratio of ~ 1.5. Also, the switchfrom FA to F mode is found at a Creq/Niratio of ~ 1.7. Whereas the previous fig-ure plotted all alloys by one compositionalparameter (Creq/Ni), Fig. 6 shows that thepreviously observed boundaries that sepa-rate solidification modes do not change as

the Mo content is altered. Thesefindings seem to indicate that thelong-standing assumption that Mopossesses a ferrite-stabilizingstrength equal to that of Cr is validfor the high-Mo alloys. Bound-aries are also included that sepa-rate σ-containing samples from σ-free samples. Figure 6 thusrepresents a map by which mi-crostructures may be predictedbased on composition for a widerange of Mo-bearing stainlesssteels prepared under similar con-ditions. Of particular interest arethe near-eutectic alloys with 4–6wt-% Mo, characteristic of manycommercially available alloys. Al-loys that solidified in the AF modeshowed substantial amounts of σ-phase that formed when the inter-dendritic ferrite decomposed to σ-phase via the δ → (σ + γ)eutectoid-type reaction. (Detailsof this transformation are pro-vided in Ref. 13.) On the otherside of the eutectic, FA mode al-loys showed no evidence of σ-phase up to6 wt-% Mo, save for a few isolated parti-cles in the 6Mo-16Cr-14Ni sample.

The lack of σ-phase in these 0–6 wt-%Mo alloys implies that the Mo contentwas more uniformly distributed through-out the microstructure as opposed to

Table 3 — Six Candidate Mo-Bearing Stainless Steel Compositions, in wt-%, for the SASS FillerMetal Application. (All contain <10 wt-% ferrite and contain little to no σ-phase.)

Designation Fe Ni Cr Mo (Cr+Mo)/Ni wt-% Ferrite

2Mo-18Cr-12Ni 66.31 12.6 18.55 2.22 1.648 7.932Mo-20Cr-15Ni 62.17 15.02 20.33 2.12 1.495 5.44Mo-17Cr-13Ni 66.59 12.51 16.57 3.95 1.640 7.314Mo-20Cr-15Ni 59.91 14.75 20.63 4.26 1.687 9.826Mo-13Cr-12Ni 68.06 12.23 13.53 5.87 1.586 5.336Mo-16Cr-14Ni 62.25 14.37 17.29 5.75 1.603 6.48

Fig. 11 — EPMA trace of the Mo content of a single austenite region. The AFtrace (6Mo-16Cr-14Ni) consists entirely of solidified austenite, while the FAtrace (6Mo-16Cr-14Ni) measures the Mo in both interdendritically solidifiedaustenite and the austenite that grew epitaxially from it. The nominal Mo con-centrations are also given for comparison.

Fig. 13 — Microstructural map showing the incidence of themassive transformation as a function of composition and so-lidification mode. Alloys designated as experiencing “some”transformation contained massive product only in laser weldsat 21 and 42 mm/s.

Fig. 12 — Representative microstructure and accompanyingEPMA trace of 100 in./min HED weld in 10Mo-14Cr-16Nisample.

Anderson Supp Sept 2007 corr:Layout 1 8/9/07 8:52 AM Page 289

Page 10: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s290

residing in the interdendritic regionsforming σ-phase. This is confirmed quantitatively in a later section.

Comparison with Stainless Steel WeldConstitution Diagrams

Stainless steel constitution diagramssuch as the Schaeffler diagram or WRC-1992 diagram are empirically derived toolsfor predicting the phase balance and fer-rite content of stainless steel alloys basedon the alloy composition, which is reducedto the simplified Cr and Ni equivalencies.The large number of alloys presentedherein is useful in assessing the accuracyof these diagrams in predicting the mi-crostructure of high (4–10 wt-%) Mostainless steels.

The most modern constitution dia-gram is the WRC-1992 diagram (Ref. 19).The advantage of this diagram is its inclu-sion of boundaries that demarcate the var-ious solidification modes available in therange of industrial stainless steels. Al-though previously included in the Schaef-fler and DeLong diagrams, the WRC-1992diagram does not include the martensitephase. The martensite boundary devisedby Schaeffler in his diagram was superim-posed on the stability diagrams shown inFig. 2. This line was quite effective at sep-arating martensite-free alloys from theirmartensite-containing counterparts, asshown in Fig. 7, which shows the entire setof experimental alloys plotted on theSchaeffler 1949 diagram. Numerous AFmode samples can be seen in the region ofthe Schaeffler diagram labeled as A (i.e.,fully austenitic). This designation de-scribes only the predicted microstructure— it is not a description of solidificationmode as in the WRC-1992 diagram. A ma-jority of these AF alloys were high-Mosamples that contained σ-phase. If the Mowas not present to enable the eutectoidtransformation, it is possible that the in-terdendritic ferrite would have otherwisetransformed into austenite, in effect cre-ating the fully austenitic structure pre-dicted by the diagram. However, the lowpartition coefficient of Mo causes it to behighly segregated to the liquid (Ref. 3).This in turn leads to a greater likelihood offorming interdendritic ferrite as com-pared to samples with equivalent amountsof Cr.

This effect can be seen as well in theWRC-1992 diagram, which is shown inFig. 8. Plotting the samples according tosolidification mode on the WRC-1992 di-agram shows poor agreement between thepredictions of the diagram and the exper-imentally observed solidification mode forprimary austenite alloy compositions. Theeffects of Mo microsegregation are likelynot accounted for in the WRC-1992 dia-gram, because the body of samples used in

the initial design suffers from a lack ofhigh-Mo compositions (only four samplescontained greater than 5 wt-% Mo, thehighest reaching 6.85 wt-% Mo). Most ofthe AF alloys that disagree with the pre-dictions of the WRC-1992 contain highconcentrations of Mo. Thus, a new bound-ary is given in Fig. 8 to distinguish A andAF mode alloys with high (4–10 wt-%) Mocontents. Good agreement is shown, how-ever, for the primary ferrite alloys. The di-agram successfully predicted the solidifi-cation mode of these alloys since the bccferrite lattice is less likely to produce a seg-regated microstructure.

Recent research efforts have sought tolocate a martensite line within the WRC-1992 diagram. Kotecki has proposed amartensite boundary (Ref. 20) (shown inFig. 8) based mainly on the results of bendtests. The alloys that were found to con-tain martensite, based on both magneticmeasurements and metallography, can beused to validate the boundary proposed byKotecki. Five of the martensitic samplesfrom this study are actually located abovethis boundary. All of these samples con-tain high amounts of Mo additions (>8 wt-%) and experienced austenitic solidifica-tion modes. The disagreement betweenKotecki’s boundary and these alloys canagain be explained by the high potentialfor microsegregation of Mo. Visual cuesgiven in the microstructure show thatmartensite rarely occurs in the interden-dritic regions, and is typically localized inthe dendrite cores (see Fig. 4A). Becauseof microsegregation, the Mo concentra-tion at the core is actually less than thenominal composition. Thus, the incidenceof martensite in these samples does notnecessarily disagree with Kotecki’smartensite boundary, since the true Moconcentration in the primary dendritewould shift the Cr equivalency of theseareas to the left of the points demarcatedin Fig. 8. Similarly, this explains the loca-tion of the four martensitic samples thatexceed Schaeffler’s martensite boundary,as seen in Fig. 7.

Another function of the WRC-1992weld constitution diagram is its ability toestimate the ferrite content of a stainlesssteel alloy based on its composition. It usesthe magnetically derived Ferrite Number(FN) value instead of weight-percent ferrite, as this parameter wasfound to be more reproducible. The FN ofall primary ferrite, martensite-free alloyswas measured for the purposes of com-parison with the diagram. The results aredisplayed in Fig. 9. A fair amount of agree-ment is shown between the experimentalFN and the predictions of the WRC-1992.By separating the data into σ-free and σ-containing alloys, the effects of the δ→ (σ+ γ) eutectoid transformation are high-

lighted. The measured ferrite content inσ-containing alloys was always less thanthat predicted by the model, which cor-roborates a solid-state transformation ofthe residual ferrite into the eutectoid con-stituent. Since the ferrite phase is con-sumed by this eutectoid transformation, itwould logically follow that the measuredFN values for these alloys would be lessthan predicted by the diagram.

Solute Redistribution

An example of the difference in soluteprofiles between alloys with similar Moconcentrations that form by AF and FAsolidification are shown in Fig. 10A, B.The Mo profile shows a distinct increasein concentration across the distance be-tween a cell core and the neighboring in-tercellular region produced by AF solidi-fication. This was caused by the solidifyingaustenite cell rejecting Mo across thesolid/liquid interface into the liquid. Thesolute accumulates in the intercellular re-gion, since it is the last to solidify. Duringprimary ferrite solidification, Mo is notsegregated to the liquid. The austenitic re-gions produced by solid-state transforma-tion in Fig. 10B show a more uniform con-centration of Mo. Since the central axis ofthe austenite regions represents the for-mer intercellular regions of the solidifiedferrite structure, this observation is con-sistent with the lack of Mo segregation.The regions of elevated Mo concentrationobserved in the FA alloy shown are associ-ated with the presence of σ-phase in thisparticular alloy.

The significant backdiffusion of solutein the ferrite phase is also verified in Fig.10C, which shows the microstructure pro-duced by F mode solidification. The centralregion of this solute profile shows the uni-form distribution of all measured elementsacross a region of untransformed ferritethat encompasses several cell widths. Thesubsequent solid-state transformationfrom ferrite to allotriomorphic austenitecaused a difference in Mo concentrationbetween the austenite and residual ferrite.The partitioning of Mo during the δ → γtransformation is also apparent in Fig. 10Bby the high levels of Mo seen in the γ/σ-phase dual structure (recall that the γ/σconstituent forms form preexisting δ-fer-rite). As shown by peaks in the solute pro-file, the elements Cr and Mo, both stabiliz-ers of the ferrite phase, were rejected intothe residual ferrite during the δ → γ solid-state transformation process (again, this δ-ferrite later transforms to γ/σ). The austen-ite matrix resulting from the FAsolidification mode contains slightly lessthan nominal, yet essentially uniform, con-centrations of these solute atoms. Con-versely, austenite-stabilizing Ni was drawn

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:38 PM Page 290

Page 11: Mo in Stainless Steels Welds

WELDING RESEARCH

-s291WELDING JOURNAL

into the newly created austenite, resultingin a uniform profile of Ni within the austen-ite that was greater than the nominal com-position. As seen in Fig. 10B, the regionsoccupied by the eutectoid product are thusdepleted of Ni.

Solute partitioning for a singleaustenitic transformation product can beexamined using the grain boundary al-lotriomorph contained in Fig. 10C. Whilethe concentrations of solute are relativelyuniform within the allotriomorphic trans-formation product, there exist distinct dif-ferences in concentration between it andthe neighboring ferrite grain. The amountof Mo and Cr are lower in the allotri-omorph, while the amount of Ni has in-creased as a result of transformation. Theferrite nearest to the allotriomorph evenshows elevated concentrations of Mo andCr, the result of being partitioned to theparent ferrite. The general upheaval ofconcentration distribution brought aboutby solid-state transformations can also beseen on the right-hand side of the micro-graph. The production of Widmanstattenaustenite platelets has caused the uniformdistribution of solute in the parent ferriteto be disrupted. Although the platelets aretoo small to be individually measured, thepartitioning of solute is apparent in themeasured chemical profile.

Solute partitioning resulting from the δ→ γ solid-state transformation does not,however, cause a chemical gradient ashigh as that produced by microsegregationassociated with austenitic (AF) solidifica-tion. Figure 11 compares the profile of Mosolute between a single solidified austen-ite cell brought about by AF solidificationwith a region of solid-state transformationaustenite derived by the FA solidificationmode. While neither austenite regionreaches the nominal concentration of ~6wt-% Mo, the concentration in the FAaustenite is certainly closer to 6 wt-%across its entire width. The difference be-tween the nominal and the core composi-tions is 1.74 wt-% for AF solidification,but a difference of only 0.88 wt-% Mo ex-ists for the FA solidification mode.

Candidate Alloys

It is worth noting that several alloys in-vestigated in this research exhibited thedesired FA solidification mode, a low fer-rite content, and no σ-phase. Of the 67 al-loys created that did not contain marten-site, nine Mo-bearing alloys wereidentified that exhibited the FA solidifica-tion mode. Of these nine alloys, one con-tained 8 wt-% Mo and two contained 10wt-% Mo. However, these three alloys allcontained σ-phase. Since σ-phase isknown for its unfavorable effects on me-chanical properties and corrosion resis-

tance (Ref. 21), these three alloy compo-sitions may not be optimal. The remainingsix alloys, which are summarized in Table3, contain only 5–10 wt-% of the ferritephase, as measured by magnetic instru-mentation, and provide a basis for promis-ing candidate compositions for furtherstudy.

Massive Transformation Structures

Several laser welds were deposited oneach experimental alloy, with the travelspeed being adjusted from 10 to 50 in./minto 100 in order to affect the cooling rate.As shown in Table 2, measurements ofdendrite arm spacings in primary austen-ite samples indicated that cooling rateswithin the laser weld zones ranged from~3 × 104 to 2 × 105 °C/s, increasing with thetravel speed. Each of the four solidifica-tion modes observed in the melt buttonsamples were found in the laser welds. Innearly all cases, the solidification mode ofthe weld microstructure matched that ofthe button, indicating that the solidifica-tion velocities used here were not highenough to induce shifts in the solidifica-tion mode, with the possible exception ofone alloy discussed below. More impor-tantly, though, microstructural character-ization revealed a subset of the weld struc-tures that was found to experience amassive transformation of ferrite toaustenite. This type of transformation,known to occur specifically at high coolingrates, causes no solute partitioning be-tween the parent ferrite and productaustenite phases and, under the correct setof processing conditions, can producefully austenitic microstructures. Similarresults have been observed in ternary Fe-Ni-Cr alloys (Refs. 6, 22). As shown in Fig.12, a patchy morphology with irregulargrain boundaries was observed that ischaracteristic of the massive product. Pos-itive phase identification using backscat-tered electron Kikuchi (BEKP) patterns,the results of which can be found else-where (Ref. 23), confirmed that this wasindeed the austenite phase. EPMA analy-ses were also performed on laser weldscontaining this patchy morphology. Asshown by the example in Fig. 12, the mas-sive product displays a uniform distribu-tion of all alloying elements at the nomi-nal levels. This segregation-free, fullyaustenitic structure should exhibit excel-lent corrosion resistance and toughness.

The solidification modes of the pri-mary ferrite laser weld microstructures,which are susceptible to the massive trans-formation, have been plotted to show thecompositional range over which the mas-sive transformation was initiated in thelaser welds. Figure 13 shows that the mas-sive transformation is more likely to occur

in alloy compositions within a shortCreq/Ni range from the eutectic located atCreq/Ni = 1.5. This appears to be associ-ated with the compositional dependenceof the To temperature, which is defined asthe temperature at which both phases in atwo-phase region of the phase diagrampossess equivalent free energies. It is com-monly held (Ref. 24) that, for the massivetransformation to occur, this temperaturemust be surpassed during cooling beforecompeting long-range diffusional trans-formations can occur. The compositionaldependence of this value has a shape sim-ilar to the δ + γ phase boundaries seen inFig. 1A. Further discussion of the thermo-dynamics of the massive transformation asthey pertain to these alloys can be foundelsewhere (Ref. 23).

Three samples, labeled as “Some Mas-sive” in Fig. 13, did not show evidence ofthe massive transformation until the laserweld travel speed was raised to 50 in./min(21 mm/s). The remainder of samplesshowed evidence of massive product in themicrostructure of welds produced at allthree travel speeds (labeled “All Massive”in Fig. 13). Interestingly, the FA modealloy located directly on the eutectic line(2Mo-20Cr-15Ni) does not present anymassive product. This is believed to be aconsequence of a possible kinetic shiftduring solidification. While the button mi-crostructure solidified as the FA mode, theelevated cooling rates in the HED weldzones resulted in dendrite tip undercool-ing that was sufficient to cause primaryaustenite solidification. The growth veloc-ities present in these welds are similar tothose found to cause kinetic shifts byFukumoto and Kurz (Refs. 25–27). How-ever, only the 2Mo-20Cr-15Ni composi-tion was located close enough to the eu-tectic so as to be susceptible to a kineticshift. Work performed by Fukumoto andKurz on the Fe-Ni-Cr-C alloy system con-cluded that elements with a low partitioncoefficient (e.g., C and Mo) could alter thesolidification behavior in near-eutectic al-loys at high growth velocities (Ref. 26).Since the structure of this button was al-ready austenite, there was no parent phasethat could experience the massive trans-formation. Further investigations of thisweld microstructure are necessary to de-termine whether this behavior can be at-tributed to a kinetic shift in solidificationmode or localized compositional variationin a near-eutectic alloy such that AF so-lidification is favored.

Conclusions

The phase transformations and resul-tant microstructures that form in Mo-bearing stainless steels have been evalu-ated over a wide range of Ni, Cr, and Mo

Anderson Supp Sept 2007 corr:Layout 1 8/9/07 8:53 AM Page 291

Page 12: Mo in Stainless Steels Welds

WELDING RESEARCH

SEPTEMBER 2007, VOL. 86-s292

concentrations. The following conclusionscan be drawn from this research:

1. The phase stability diagrams con-structed from thermodynamic calcula-tions provided a good predictive tool ofsolidification modes of Mo-bearing stain-less steel alloys.

2. The ferrite content was shown to bea function of not only Creq/Ni ratio, butalso Mo concentration, specifically due tothe transformation of ferrite into eutec-toid γ + σ in high-Mo alloys.

3. The FA solidification mode pro-duced a more uniform distribution of Mosolute in the transformed austenite phasethan observed in the cores of solidifiedaustenite cells produced by the AF mode.

4. The Schaeffler weld constitution di-agram accurately predicted the presenceof martensite in low-Mo alloys. The highmicrosegregation of Mo led to martensiteforming in high-Mo alloys that exceededSchaeffler’s martensite boundary. For thesame reasoning, several high-Mo alloyscontaining martensite exceed the marten-site boundary proposed by Kotecki on theWRC-1992 diagram. Otherwise, these di-agrams provide good accuracy of marten-site prediction for high Mo alloys.

5. The location of numerous high-MoAF mode alloys in the A mode region ofthe WRC-1992 diagram was attributed tomicrosegregation of Mo during solidifica-tion. The segregation of Mo solute to theliquid led to a larger amount of interden-dritic ferrite. However, good agreementwas shown for alloys that experience pri-mary ferrite solidification.

6. The incidence of σ-phase formationin the high-Mo samples also led to inaccu-rate prediction of ferrite content by theWRC-1992 diagram for many of these alloys.

7. A massive transformation of δ → γcan produce fully austenitic microstruc-tures with uniform distributions of Mo atthe nominal concentration. The transfor-mation is seen to occur in primary ferritealloys that are near the eutectic composition.

8. Several alloys have been identifiedthat solidify in the FA mode and have anaustenite matrix with a more uniform dis-tribution of Mo than that exhibited by pri-mary austenite solidification. These alloysshould have improved resistance to solid-ification cracking and localized corrosion.

Acknowledgments

The authors gratefully acknowledge fi-nancial support of this research by the Of-fice of Naval Research, under ContractNo. N00014-00-1-0448. Generous finan-cial support was also provided by the AWSFoundation and the Navy Joining Centerthrough an AWS Graduate Research Fel-

lowship. Some phase identification wasperformed at Sandia National Laborato-ries. Sandia National Laboratories is amultiprogram laboratory operated bySandia Corporation, a Lockheed MartinCompany, for the United States Depart-ment of Energy’s National Nuclear Secu-rity Administration under contract DE-AC04-94AL85000.

References

1. Banovic, S., DuPont, J., and Marder A.2001. Dilution control in gas tungsten arc weldsinvolving superaustenitic stainless steels andnickel based alloys. Metallurgical and MaterialTransactions 32B: 1171–1176.

2. DuPont, J., Friedersdorf, L., Marder, A.,and Banovic, S. 2001. Weldability and Corro-sion Performance of Welds in AL-6XN Super-austenitic Stainless Steel. Lehigh UniversityATLSS Report No. 01-03. 2001.

3. Banovic, S., DuPont, J., and Marder, A.Dilution and microsegregation in dissimilarmetal welds between superaustenitic stainlesssteel and nickel base alloys. Science and Tech-nology of Welding and Joining (UK) 7: 374–383.

4. Banovic, S., DuPont, J., and Marder, A.2003. Microstructural evolution and weldabilityof dissimilar welds between a superausteniticstainless steel and nickel-based alloys. WeldingJournal 82(6): 125–135.

5. Perricone, M., and DuPont, J. 2003. Laserwelding of superaustenitic stainless steel.Trends in Welding Research: Proceedings of the6th International Conference. Materials Park,Ohio: ASM International. pp. 47–52.

6. Brooks, J., Baskes, M., and Greulich, F.1991. Solidification modeling and solid-statetransformations in high-energy density stainlesssteel welds. Metallurgical and Material Transac-tions 22A: 915–926.

7. Sundman, B. 2001. Thermo-Calc. S-10044[[N]]. Stockholm, Sweden, Department ofMaterials Science and Engineering, KTH.

8. Saunders, N. 2001. Fe-Data Thermody-namic Database. [[3.0]]. The Surrey ResearchPark, Guildford, UK, Thermotech, Ltd.

9. Brooks, J., and Baskes, M. 1989. Mi-crosegregation modeling and transformation inrapidly solidified austenite stainless steel welds.Proceedings of the Second International Confer-ence on Trends in Welding Research, 153–158.

10. Brooks, J., and Thompson, A. 1991. Mi-crostructural development and solidificationcracking susceptibility of austenitic stainlesssteel welds. International Materials Reviews 36,16–44.

11. Kou, S. 2003. Welding Metallurgy. Hobo-ken, N.J.: John Wiley & Sons, Inc.

12. Anderson, T., DuPont, J., Perricone, M.,and Marder, A. 2007. Phase transformationsand microstructural evolution of Mo-bearingstainless steels. Metallurgical and MaterialsTransactions 38A: 86–99.

13. Perricone, M., Anderson, T., Robino, C.,DuPont, J., and Michael, J. 2007. PredictingSigma Formation during Welding of Mo-Bear-ing Stainless Steels. Rio Grande Symposium onAdvanced Materials '06.

14. Elmer, J., Allen, S., and Eagar T. 1989.Microstructural development during solidifica-tion of stainless steel alloys. Metallurgical and

Material Transactions 20A: 2117–2131.15. Katayama, S., and Matsunawa, A. 1984.

Proceedings of ICALEO. pp. 60–67.16. Heinrich, K., Mykleburst, R. L.,

Yakowitz, H., and Rasberry, S. D. 1972. SimpleCorrection Procedure for Quantitative Elec-tron-Probe Microanalysis. NBS Technical Note719, 1–31.

17. Reick, W., Pohl, M., and Dapilha, A.1998. Recrystallization-transformation com-bined reactions during annealing of a coldrolled ferritic-austenitic duplex stainless steel.ISIJ International (Japan) 38, 567–571.

18. Shek, C., Shen, G., Lai, J., and Duggan,B. 1994. Early stages of decomposition of fer-rite in duplex stainless steel. Materials Scienceand Technology 10, 306–311.

19. Kotecki, D., and Siewart, T. 1992. WRC-1992 constitution diagram for stainless steelweld metals: a modification of the WRC-1988diagram. Welding Journal 71(5): 171–178.

20. Kotecki, D. 2000. A martensite bound-ary on the WRC-1992 diagram. Welding Journal78(5): 180–192.

21. Koseki, T., and Ogawa, T. 1992. An in-vestigation of weld solidification in Cr-Ni-Fe-Mo alloys. Welding International 6, 516–522.

22. Brooks, J., Headley, T., and Robino, C.2000. Microstructure design for laser deposited304L stainless steel. Solid Freeform and AdditiveFabrication: A Materials Research Society Sym-posium. San Francisco, Calif.

23. Perricone, M., Anderson, T., DuPont, J.,and Robino, C. 2006. On the massive transfor-mation from ferrite to austenite in laser weldedMo-bearing stainless steels. Materials Science &Technology ’06, Cincinnatti, Ohio.

24. Aaronson, H. 2002. Mechanisms of themassive transformation. Metallurgical and Ma-terials Transactions 33A: 2285–2297.

25. Fukumoto, S., and Kurz, W. 1997. Thedelta to gamma transition in Fe-Cr-Ni alloysduring laser treatment. ISIJ International(Japan) 37, 677–684.

26. Fukumoto, S., and Kurz, W. 1998. Pre-diction of the delta to gamma transition inaustenitic stainless steels during laser treat-ment. ISIJ International (Japan) 38, 71–77.

27. Fukumoto, S., and Kurz, W. 1999. Solid-ification phase and microstructure selectionmaps for Fe-Ni-Cr alloys. ISIJ International(Japan) 39, 1270–1279.

Anderson Supp Sept 2007 corr:Layout 1 8/8/07 1:37 PM Page 292