design, fabrication, and characterization of graded ... · pdf fileaustenitic stainless steel...

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL 295-s

Introduction Fossil fuel-fired and nuclear powerplants use a combination of low-alloyCr-Mo ferritic steel and high-alloyaustenitic stainless steel piping. Thecheaper ferritic steels are used in thelower temperature segments while themore expensive austenitic stainlesssteels are used in the higher tempera-ture regions, thus requiring creep andcorrosion resistance (Refs. 1–3). This necessitates the use of dissimi-lar metal welds (DMWs) betweenthese alloys (Ref. 3). However, DMWsbring about sharp changes in chem-istry and coefficient of thermal expan-sion (CTE), which may lead to prema-ture failure of components and facilityshutdowns costing up to $850,000 per

day (Ref. 3). The DMWs are character-ized by two interfaces on either side ofthe fusion zone, one with austenitic al-loy and another with ferritic alloy. Ingeneral, the interface that separatesthe ferritic steel and fusion is consid-ered to undergo premature failure(Refs. 3–5). This is related to two phe-nomena discussed briefly in the fol-lowing section.

Localized Thermal Stress

The large CTE mismatch betweenthe 2.25Cr-1Mo and the stainlesssteels (typically 316 or 347H) leads tolocalized stress concentration thatmay overlap on the applied stresses.Although this problem has been par-tially addressed by using a trimetallic

transition joint to weld the ferritic(CTE 14.0 m m-1K-1) and austeniticsteels (CTE 18.5 m m-1K-1) by usingan Inconel alloy (CTE 16.9 m m-1K-1)(Refs. 5, 6), the local stresses arisingfrom the large creep strength mis-match between the different regionscan still cause a significant stress accu-mulation at the interface (Ref. 6). Forinstance, a stress of 34 ksi has been re-ported at the root of a typical dissimi-lar metal weld (Ref. 7).

Carbon Migration

Differences in carbon chemical po-tential brought about by differences inCr concentration allows carbon diffu-sion from ferrite (BCC) regions towardthe austenitic (FCC) interface, whichin turn reduces the creep resistance ofthe ferrite region. The reason for thereduced chemical potential for C in theaustenitic side is the increased chromi-um levels in the stainless steel (Ref. 8).Due to the inability of the carbon todiffuse farther into the FCC (becausethe diffusivity of C in FCC is orders ofmagnitude lower than BCC), carbidesstart to nucleate at the interface be-tween the Cr-Mo steel and the stain-less steel interface. The most frequently reported typeof carbide morphology is the type Icarbide (Refs. 4, 9, and 10). These car-bides evolve with time in the sensethat they nucleate with a sphericalmorphology but gradually acquire alenticular shape (Refs. 4, 9, and 10).These type I carbides have been hy-pothesized to be the source for the nu-cleation of creep voids (Refs. 3, 4). Following the nucleation of carbides,the chromium in the solution reduces,thus affecting the diffusivity of C. Thiscreates a C-denuded ferritic side and a

Design, Fabrication, and Characterization of Graded Transition Joints

The susceptibility of hot cracking in the graded transition region is evaluated

BY N. SRIDHARAN, E. CAKMAK, B. JORDAN, D. LEONARD, W. H. PETER, R. R. DEHOFF, D. GANDY, AND S. S. BABU

ABSTRACT Dissimilar welds between austenitic and ferritic steels suffer from premature failuredriven by interfacial stresses and material degradation brought about by a mismatch inthe coefficient of thermal expansion and carbon migration from ferritic steels to the interface, respectively. Trimetallic transition joints using graded composition between ferritic and austenitic alloys are considered a viable pathway to address this issue. However,hot cracking may occur when welding nickel alloys to stainless steel. This researchattempts to reduce the hot cracking susceptibility of Inconel82 alloys by functionallygrading them with 316L. Optical and electron microscopy showed extensive cracking inthe graded regions. Calculations using ScheilGulliver techniques attributed the crackingto the expansion in the solidification range of Inconel82. To circumvent solidificationcracking, another transition joint between SA 508 Grade 22 and SS 316L was designedand fabricated with coaxial powderblown additive manufacturing using an SS 410SS316L grading. After fabrication, the joint was characterized using optical and scanningelectron microscopy, wavelength dispersive spectroscopy, as well as electronbackscattered and xray diffraction techniques. Characterization showed a successfultransition joint with minor porosity. The measured composition gradients agreed withthe designed composition gradients. This study showed that 12Cr steels couldpotentially be used to fabricate transition joints without any hot cracking.

KEYWORDS • Graded Transition Joint • Laser Direct Metal Deposition • Characterization • Electron Backscatter Diffraction • Xray Diffraction

Sridharan 2016114.qxp_Layout 1 7/11/17 4:09 PM Page 295

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2017, VOL. 96296-s

C-enriched interface. The C-denuded re-gion, coupled with the stresses due tothe CTE mismatch, can create a signifi-cantly weakened zone and act to reducethe creep life of the material. According-ly, it may be concluded that the mi-crostructural differences play a majorrole in determining the creep life of thedissimilar metal welds.

Existing Engineering Solutions:Trimetallic Joints

While the carbon migration can bereduced by using an Inconel filler met-al for welding, it causes a semicontinu-ous network of carbides to develop atthe interface (Ref. 11). The dissimilarwelds between austenitic stainless andferritic steels made using an Inconelfiller metal fail in about one-third oftheir expected lifetime, while the Ni-based trimetallic transition joints failin about half of their expected lifetime(Ref. 5). While using the trimetallic

transition joint can increase the over-all creep life, it is challenging to fabri-cate defect-free welds at the stainlesssteel Inconel interface. Sireesha et al. performed detailedhot cracking studies on various fillermetals to weld SS 316L to Alloy 800H tofabricate a trimetallic transition joint(Refs. 12, 13). The results showed thatthe filler 16-8-2 displayed the highestresistance to hot cracking, and Inconel-82 (In-82) displayed the least resistance(Ref. 13). The cracking and microfissur-ing was attributed to the formation ofthe Nb-rich Laves phase resulting inweld metal liquation. However, In-82 isthe material of choice as a welding con-sumable due to the enhanced thermalstability, impact resistance, and tensilestrength (Refs. 12, 13).

Role of Additive Manufacturing

Additive manufacturing is a prom-ising candidate due to its ability to

form functionally graded materials, al-lowing combinations of various differ-ent alloys (Ref. 14). Extensive workhas been done by researchers atLehigh University, Bethlehem, Pa., todevelop functionally graded transitionjoints (GTJ) using the laser-directedenergy deposition (Ref. 15) and dual-wire gas tungsten arc welding (GTAW)processes (Ref. 1). Both processes haveshown reasonable promise to fabricatetransition joints. In graded transition joints, themetal matrix composition is to begradually changed from a 100% ferriticto a 100% austenitic microstructure(Refs. 1–3, 5). Using this approach,prefabricated transition blocks withmatching compositions to the low-and high-alloy steels on each respec-tive end can be inserted between theferritic and austenitic steel pipes. Thiscan then be welded together with twosimilar welds. Though desired gradi-ents can be obtained, hot cracking wasa major concern and was reported tooccur at the interfaces between the2.25Cr-1Mo and Alloy 800 (Ref. 1) aswell as in the stainless steel regions(Ref. 15) where the weld metal solidi-fied by nonequilibrium austeniticmode. The objectives of the current paperare as follows: 1. Evaluate the susceptibility of hotcracking in the graded transition region. 2. Develop a method to design, fab-ricate, and characterize a graded tran-sition joint using the directed energydeposition.

Experimental Details

GTJ Fabrication with the DirectMetal Deposition Process

The powders (supplied by CarpenterAlloys) were manufactured by the gasatomization process, which led to apowder size range between 44–120m. The GTJ block was made usingDMD-103D, a commercially availablecoaxial powder-blown laser (a diodelaser operating at 910 nm) depositiontechnique at ORNL’s ManufacturingDemonstration Facility (MDF), Tenn.A laser power of 400 W, a constantlaser travel speed of 600 mm/min witha powder feed rate of 5 g/min, and an

Fig. 1 — A — Carbon chemical potential gradient in the sample; B — gradient in the coefficient of thermal expansion in the sample.

Fig. 2 — Schematic illustration: A — Transition joint; B — tool path used for fabrication ofthe transition joint.

A B

A B

Sridharan 2016114.qxp_Layout 1 7/11/17 4:09 PM 96

WELDING RESEARCH


average build-layer thickness of 0.5mm were used during the build. Prealloyed powders supplied byCarpenter Powder Products Inc. wasused for the fabrication of the transi-tion joints. Initially, In-82 to SS 316Lreference builds were fabricated. Fol-lowing this experiment, SS 410 to SS316L transition joints were also fabri-cated after designing the composition-al gradients. The composition in thegraded region was optimized by con-trolling the powder flow from eachhopper. A detailed explanation of thedesign methodologies is discussed lat-er. The compositional gradients wereachieved by gradual variation of themass flow rates of the powders to ar-rive at the required grading ratios.

Compositional Design

Design methodology focuses onminimizing thermal stresses due tothe CTE mismatch and the gradientsin the carbon chemical potential. Inthis research paper, calculations fo-cused only on SS 410 to 316L-gradedtransition joints. The carbon chemicalpotential was calculated using Thermo-Calc, a commercial softwarethat uses the Calphad approach tomodel phase transformations in met-als (Ref. 16). To calculate the carbon chemical po-tential for the ferritic region, the matrixwas assumed to be 100% BCC without

any carbides, in agreement with the as-welded condition. For mixed BCC + FCCregions, the carbon chemical potentialwas calculated by setting the phase frac-tions of BCC and FCC to equilibriumvalues at 450°C. This may not have beena valid assumption because both weld-ing and additive manufacturing leads tononequilibrium solidification (Ref. 17).However, this assumption was consid-ered the first step to develop a designmethodology, due to lack of detailed ki-netic models that consider nonequilibri-um solidification (Ref. 18) as well as sta-bility of the same for complex micro-structural morphologies (Ref. 19). After calculating the compositionsthat minimize the carbon chemical po-tential, the corresponding CTE mis-match was calculated using isoexpan-sion contours developed by Elmer etal. (Ref. 20). These two calculationswere done iteratively until a minimumgradient in the carbon chemical poten-tial and CTE mismatch were obtained.The carbon-chemical potential gradi-ent and the CTE values from the de-sign are presented in Fig. 1. Based on the design, the followingdeposition strategy was used to fabri-cate the SS 410 to SS 316L transitionjoint. Deposits were made on a SA508 Grade 22 substrate. The first fourlayers were deposited as 100% SS 410followed by two layers of 80% SS 410/20% SS 316L, 60% SS 410/40% SS316L, 40% SS 410/60% SS 316L, and

20% SS 410/80% SS 316L each. Thesewere followed by 15 layers of 100% SS316L. A schematic illustration of thecompositions used and the deposit isillustrated in Fig. 2A. A preheat of300°C was maintained throughoutthe build to prevent the SS 410 steelfrom cracking.

Microstructure Characterization

The samples were sectioned usingelectric discharge machining. The spec-imens were then mounted in cold-setting epoxy and ground/polished us-ing standard metallography tech-niques. The samples were given a col-loidal silica (0.04 micron) final polishfor the EBSD measurements. The sam-ples were etched using a glyceregia so-lution. Optical micrographs were ob-tained using a Leica DMI 5000M opti-cal microscope with a motorized andautomated sample stage. Microhard-ness line scans were performed using aLECO TM103D microhardness testerwith a 200-g load. Elemental composition changesalong the graded block were investigat-ed using a JEOL JXA-8200X electronmicroprobe analyzer (EPMA) instru-ment equipped with five crystal-focusing spectrometers for wavelengthdispersive x-ray spectroscopy (WDS).Quantitative line scans were acquiredutilizing a 15-kV accelerating voltageand an electron beam current of 100nA. Additionally, elemental standardswere used to compare experimentallyacquired intensities to known stan-dard intensities. A 200-m step sizewas used for the scans, starting fromthe base plate and ending at the 100%316L side of the deposit. The XRD phase identification ex-periments were performed using aPANalytical X’Pert Pro diffractometerwith Mo K radiation ( = 0.709319Å). Continuous –2 scans were per-formed from nominally 17 to 42° 2.For these measurements, regions ofinterest along the build direction wereisolated using zero-background platesto achieve spatial resolution. The sam-pling was done with finer steps (~2mm) through the graded section to ob-tain the optimum balance between in-cident/diffracted intensity and spatialresolution in this zone. For quantitative phase analyses, Rietveld refinements were performed

Fig. 3 — A — Micrograph of the In82 to SS 316L transition. The cracks are marked usingblack arrows; B — the electron micrograph adjacent to the cracks showing the absenceof any liquidating phases; C — an EBSD micrograph of the cracked region. Note that thecracking intergranular proceeds along the grain boundaries, showing that the crackingmechanism is by solidification cracking.

A

B C


WELDING RESEARCH


on the obtained patterns using thehigh-score plus program from PANa-lytical. To perform the electronbackscatter diffraction (EBSD) analy-sis, a JEOL 6500F field emission gun-SEM (from JEOL USA Inc., Peabody,Mass.) equipped with an EDAX Apollosilicon drift detector and Hikari EBSDcamera were used. The electron gunwas set to an accelerating voltage of 20kV and a tip current of 4 nA for col-lecting the EBSD data. The postpro-cessing of the collected data was per-formed using the TSL OIM Analysissoftware to obtain the phase and in-verse pole figure (IPF) maps.

Results and Discussion

Rationalization of Cracking In82 and SS 316L ReferenceBuilds

As stated previously, the primaryaim behind fabricating the In-82 to SS316L transition joint was to evaluatethe susceptibility of the microstructureto weld solidification cracking. The opti-cal and electron microscopy shown inFig. 3A–D displayed extensive crackingin the graded region. The EBSD shownin Fig. 3C displayed that the cracking oc-curred primarily along the crystallo-graphic grain boundaries. This meansthat the cracks were

solidification cracks(Refs. 21–23). Thewidely accepted theoryfor hot cracking/solidi-fication cracking isthat, during solidifica-tion, the liquid film al-ways wets high-anglegrain boundaries, em-brittling them (Ref.24). This, coupled withthe thermal stresses, resulted in crack-ing along the grain boundaries. Solidi-fication cracking susceptibility is usu-ally related to a large solidificationtemperature range. To evaluate thecracking susceptibility of various re-gions of the build, a compositionanalysis was performed using electronprobe microanalysis in the cracked re-gion adjacent to the cracks. Scheil so-lidification simulations were per-formed using Thermo-Calc to estimatethe nonequilibrium solidificationrange as a function of compositionsmeasured along the graded region. Theresults are presented in Fig. 4A–C. Figure 4C clearly shows the increasein the solidification range with dilu-tion of In-82 with Fe. This proves thatsignificant changes to the compositionof In-82 must be made to eliminatehot cracking in the build. For instance,EPRI has developed a new weldingwire, EPRI P87, to fabricate Grade 91

to 347H steels by optimizing the hotcracking resistance with creep ruptureproperties (Ref. 25). In this work, wewere restricted to searching for an al-ternate GTJ configuration that was re-sistant to weld solidification cracking.

Design and Evaluation of a NewGTJ with 12Cr Steel as a Transition

Design Strategy. To minimize solid-ification cracking, we considered 12Crmartensitic stainless steel (SS 410) asa transition layer between 2.25Cr-1Moand SS 316L. For comparison, typicalelemental composition ranges of SS316L, SS 410, and SA 508 Grade 22are presented in Table 1. The SS 410 was the material ofchoice since its Cr content is betweenthe 2.25% Cr in the base Cr-Mo steeland 18% in the austenitic stainlesssteel. The material can provide an ex-

Fig. 4 — A and B — The Scheil solidification simulations performed for a 100% In82 and an alloy with 40% In82 and 60% SS316L compositions. Note the expansion of the solidification range;C — the solidification ranges for the various regions in the alloycalculated using the nonequilibrium solidification model.

A B

C


WELDING RESEARCH


cellent compatibility with the CTE ofthe 2.25Cr-1Mo steel and SS 316L ma-terial. The use of SS 410 steel is alsoadvantageous since the cost benefitsare significant compared to theiraustenitic counterparts (Ref. 26). Thechanges in chemical composition (Fe,

Ni, and Cr) from the microprobeanalyses are shown in Fig. 5. Althoughthe trace elements Mo and Mn weremeasured, they were not presenteddue to the scatter in the data. Feasibility of Achieving GradedComposition by the DMD Process. The

experimentally measured elementalcompositions of the SA 508 Grade 22base plate, 100% SS 410, and 100% SS316L zones in Fig. 5 were observed tobe within the elemental compositionlimits of the commercial alloys whencompared to the values in Table 1. Alsopresented in Fig. 5 are the expectedaverage elemental compositions(weighted averages from Table 1 basedon the grading design — Fig. 2A) over-laid (see dashed lines) on the experi-mentally measured compositions.Such a comparison was omitted for thetrace elements (Mo and Mn) becauseonly the maximum possible valueswere listed in the composition datasheets (Table 1). Overall, a good agreement betweenthe expected and measured composi-tions was observed. Such an agree-ment suggested that a good mixing be-

Table 1 — Typical Compositions of the Alloys Used in This Work in wt%

Alloy C Mn P S Si Cr Ni Mo N Fe

SS 316L 0.03 2.00 0.045 0.03 0.75 16.0– 10.0– 2.0– 0.1 Bal. (max) (max) (max) (max) (max) 18.0 14.0 3.0 (max)

SS 410 0.15 1.00 0.04 0.03 1 11.0– — — — Bal. (max) (max) (max) (max) (max) 13.5

SA508 Gr 0.11– 0.3– — — — 2.0– — 0.9– — Bal. 22 0.15 0.6 2.5 1.1

Fig. 5 — The chemical compositions in the graded regionsmeasured using wavelength dispersive spectroscopy. Thedashed lines indicate the expected compositions according tothe design. The lines with the markers show the measuredcomposition across the transition joint.

Fig. 6 — The Schaeffler diagram showing various solidificationmodes as a function of the Nieq and Creq values. Overlaid onthe diagram are data points from the graded joint based onweighted average compositions.

Fig. 7 — Optical micrograph montage of the transition joint. Various zones such as the SA508 Grade 22 base plate, 100% SS 410, 410/316L graded zone, and 100% SS 316L zoneare marked with arrows. Also marked with small yellow arrows on the micrograph arethe pores observed in the graded zone.


WELDING RESEARCH


tween the alloying powders, withclose-to-expected ratios, were ob-tained for each layer within the build,and the desired compositions can beachieved through careful control of thepowder flow rates. Finally, there appeared to be a dilut-ed zone between the base metal andthe 100% SS 410 (with the composi-tion gradually changing from the basemetal to SS 410, indicated with blackarrows in Fig. 5). However, this dilu-tion was not quantified, and work wasin progress as to how to calculate thedilution a priori. The Schaeffler diagram (Ref. 27), aconstitution diagram for stainlesssteel weld metals, is widely used topredict the final microstructure ofwelds based on elemental composi-tion. By using the elemental composi-tions determined from the World DataSystem (WDS), the expected solidifica-tion microstructures at different sec-tions of the graded joint can be esti-mated with the Schaeffler diagram. For this purpose, the expected com-positional changes in each zone werecalculated based on weighted averages(based on the grading design — Fig.2A) using the elemental compositionsof 100% SS 410 and SS 316L as deter-mined from the WDS analyses. The re-sultant average compositions, overlaidon the Schaeffler diagram, are present-ed in Fig. 6. According to the diagram in Fig. 6,the 100% SS 410 and the 80% SS 410-20% SS 316L compositions remain inthe M + F solidification regime, where-as the 60% SS 410-40% SS 316L and40% SS 410-60% SS 316L fit into the A+ M + F region. Finally, the 20% SS410-80% SS 316L and the 100% 316Lfall near the border of the A + F side,close to the fully austenitic region. Itshould also be emphasized that Fig. 6is used as a guideline. Nonequilibriumsolidification conditions taking placeduring additive manufacturing mayyield different microstructures thanshown on Fig. 4. Microstructure and Hardness inGTJs. A montage image of the opticalmicrographs obtained from the gradedblock in its as-built state is presentedin Fig. 7. Due to the unique etchingcharacteristics of each zone (i.e., SA508 Grade 22 base plate, 100% SS 410,SS 410-SS 316L grading, and 100% SS316L), the different zones can be clear-

ly distinguished in the micrograph (seearrows in Fig. 7). The optical micrographs demon-strate a successful graded transitionjoint without any cracking. Minorporosity in the graded zone (highlight-ed with the yellow arrows on Fig. 7)was observed. Porosity formation inadditively manufactured materials iscurrently a topic under research, andmore fundamental research on theprocess parameter optimization iswarranted to eliminate the porosityand fabricate a fully dense material

(Ref. 14). The as-built structure can befurther improved by eliminating theporosity either through optimizing thebuild parameters (Ref. 28) and/or per-forming a postbuild hot isostaticpressing treatment. Furthermore, thedeposit exhibited a good bonding withthe SA 508 Grade 22 base plate andbetween the layers. Finally, a heat-affected zone was observed in the baseplate close to the weld interface linewith the 100% SS 410 region. Representative optical microscopyimages were obtained from the base

Fig. 8 — The optical micrographs: A — Base material; B — interface between the basematerial and the SS 410 region; C — 80% SS 410 region; D — 60% SS 410 region, wherelocalized pools of austenite start to form; E — 20% SS 410 region, where the microstructure is predominantly austenite; F — 100% SS 316L, showing a 100% austentic microstructure.

Fig. 9 — A — Hardness map from a section of the transition joint. Note the presence of asoft zone in the SS 410 region, which is characterized by the high hardness. The differentzones can be characterized by the changes in the hardness with the highest hardness corresponding to the SS 410 region, the lowest hardness corresponding to the base material,and the intermediate hardness corresponding to the graded region; B — the line scanacross the whole joint where the various regions are marked out, which can be used to interpret the hardness maps.

A

A

D E F

B

B

C


WELDING RESEARCH


plate, SS 410, grading, and SS 316Lzones, which are presented in Fig.8A–F, respectively. The base metal tothe SS 410 interface is shown in Fig.8B and is 100% martensitic, as expect-ed. The WDS line scans showed signifi-cant dilution (Fig. 5), which increasedthe hardenability of this region. The optical micrographs reveal thatthe 100% SS 410 zone showed a com-pletely martensitic microstructure asshown in Fig. 8B. However, the Schaef-fler diagram predicts a martensite +ferrite microstructure, meaning thatthe presence of delta ferrite can be ex-pected. Hardness maps were generatedto complement the optical microscopyresults — Fig. 9. The unexpected hard-ness dropped in the transition regionand indicated the possibility of theformation of delta ferrite in the mi-crostructure. However, this is not sup-ported by the microstructural analysis,so the reason for the drops in hardnesswarrants further investigation. Discussion on Microstructure Evolution in GTJs. In 9–12% Cr steels, the

preferred transformation path is nor-mally as follows:L L + Fd Fd Fd + A A Martensite (Ref. 29). The retention of delta ferrite (Fd) atroom temperature occurs if it is signif-icantly enriched in Cr and other fer-rite-forming elements, thus stabilizingit at room temperature. The presenceof delta ferrite in these martensiticsteels was not desirable due to the em-brittling effect (Ref. 26). This made itmandatory to investigate if residualdelta ferrite was present in the steelafter welding. Along the transition region, the op-tical micrographs show a predomi-nantly martensitic microstructure atlocations where the dilution of SS 410was less than 40% — Fig. 8C and D.Austenite pools began to show up atdilution levels above 40% and close to20% — Fig. 8E and F. The austenite phase was preferen-tially attacked by the etchant due tothe lower chromium content in theaustenite. This created a contrast

where the austenite appeared darker.It was necessary to understand thepercentage of phase fractions andcompare it with the fractions predict-ed by the Schaeffler diagram. This wasimportant because the calculations forCTE were based on the fact that theexact phase fractions could be predict-ed from the Schaeffler diagram. For in-stance, an increase in the percentageof ferrite could decrease CTE (Ref. 20)and provide a “diffusion highway” forcarbon to migrate (Ref. 30). Hence, x-ray and electron backscatter diffrac-tion measurements were performed toquantify the phase fractions and thedistribution of the phases in the ma-trix. The results will be presented laterin this paper. The SS 316L zone showed a mixtureof columnar and cellular dendritic mi-crostructure without any delta ferrite.The absence of cracks was also surpris-ing. It has often been suggested in theliterature that stainless steels that so-lidify in the fully austenitic conditionare more susceptible to hot cracking

Fig. 10 — A — The xray diffraction patterns as a function of position (see position [mm] in red on the righthand side of the graph) alongthe build direction starting from the SA 508 Grade 22 base plate and ending at the 100% SS 316L zone; B — the changes in the FCC(austenite) and BCC (ferrite/martensite) weight fractions as a function of position along the build direction. The dotted lines correspond tothe borders of the spatially distributed xray measurement zones. The data points correspond to the midpoints of each measurement zone;C — detailed view of the (111)FCC and (110)BCC peaks from select zones: 5.9, 8.3, and 10.4 mm (midpoints of measured zones) from the bottom of the base plate.

A

C

B


WELDING RESEARCH


(Refs. 22, 31, and 32). During rapid so-lidification of stainless steel welds, ashift in the primary phase selectionmay occur depending on the thermalgradients and liquid solid interface ve-locities. Several theories based on parti-tionless solidification (Ref. 33) and Azizsolute trapping (Ref. 34) have been pro-posed to rationalize this change in thesolidification mode. However, later re-search has shown that the shift in thesolidification mode can be attributed todendrite tip undercooling, which wasproposed by Fukumoto and Kurz (Ref.35) as well as shown experimentally byBabu et al. (Ref. 36). Knowledge of the solidificationmode of austenitic alloys is importantfrom the standpoint of resistance tohot cracking during welding. Despitethe fully austenitic mode of solidifica-tion, the reason for the high crackingresistance in this case was not fully un-derstood. It could also be possible thatthe melt solidified in a ferritic deltamode, and the ferrite transformed toaustenite via a massive transformation,resulting in a 100% austenitic mi-crostructure (Ref. 32). However, such atransformation should not show anysolidification subgrain boundaries. Ad-ditionally, the presence of solidificationsubgrain boundaries suggests other-wise (Ref. 32). The dendritic solidification mi-crostructure observed in the 100% SS316L zone suggested that completeaustenitic solidification took place inthis zone. On the other hand, thechanges in the substructure from cel-lular to columnar dendrites could berelated to the degree of constitutionalundercooling where formation ofcolumnar dendrites was favored as theundercooling was increased (Ref. 37). Validation of Phase Evolution inGTJs. The XRD patterns obtained fromvarious isolated zones of the transi-tion joint and the changes in the vol-ume fractions of the FCC (austenite)and BCC (ferrite/martensite) phasesas a function of position along thebuild direction are presented in Fig.10A and B, respectively. Figure 10A shows the obtained dif-fraction patterns starting from thebase plate, passing through the 100%SS 410 and the graded zone, ending atthe 100% SS 316L zone. The totalheight of the transition joint was ~22mm and the spatial positions from

where the patterns were obtained arepresented next to each pattern (see inred), with the zero-point correspon-ding to the bottom of the base plate.Here, such positions correspond to themidpoints of each isolated zone sincethe diffraction data was averaged overthe measured zone. Additionally, onlytwo regions from the 100% SS 316Lwere presented to save space becauseno major changes were recorded up-wards of these zones. It should also benoted that due to having the samecrystal structure (BCC) with similarlattice parameters, it was not possibleto differentiate the ferrite and marten-

site phases, particularly in the transi-tion zone from the SA 508 Grade 22base plate to the 100% SS 410 zone. In Fig. 10A, the strongest peaks forthe FCC (austenite) and the BCC (fer-rite/martensite) phases were observedto belong to the (111) and (110) reflec-tions, respectively (see highlightedwith arrows). Thus, these peaks wereselected for pointing out the qualita-tive changes in the FCC and BCC phas-es. The XRD patterns revealed a fullyferritic structure in the SA508 Grade22 base plate. According to the Schaeffler diagramshown in Fig. 6, the SS 410-rich side of

Fig. 11 — The changes in the lattice parameters of the FCC and the BCC phases as a function of position along the build direction.

Fig. 12 — The inverse pole figures: A — 100% SS 316L showing a 100% austenitic microstructure; B — 20% SS 410 region where the microstructure is predominantlyaustenite; C — 40% SS 410 region, where localized pools of austenite start to form; D— 60% SS 410 region, where localized pools of austenite start to form; E — 80% SS 410region, where localized pools of austenite start to form; F — 100% SS 410 region,where localized pools of austenite start to form.

A D

E F G

B C


WELDING RESEARCH


the grading can be expected to solidifywith a mixture of the ferrite andmartensite phases. Therefore, in thesubsequent patterns through gradingthe BCC peaks from the pure ferritewere replaced with the BCC peaksfrom the mixture of ferrite andmartensite. Additionally, the presenceof the FCC phase (austenite) started tobe visible in the base plate to the grad-ing transition zone. It then becamemore prominent further into the grad-ing, eventually turning into the onlypresent phase in the 100% SS 316Lzone. The presence of austenite as thesingle phase in the 100% SS 316L zonealso revealed that a fully austenitic so-lidification occurred without the pres-ence of a ferritic phase as discussed inthe microstructure analysis section. Furthermore, it is important tonote that broad BCC peaks were ob-served in the graded zone (indicated inFig. 10A and emphasized in Fig. 10C)as opposed to the sharper peaks in thebase metal, which may suggest thepresence of either closely overlappingpeaks of ferrite and martensite phasesor a lattice distortion, or both. Latticedistortion could indicate the forma-tion of fresh martensite (i.e., solidifi-cation from liquid occurred as austen-ite transformed into martensite dur-ing cooling). Martensitic transforma-tions occurred with an increased dislo-cation density and twinning, whichcan also cause significant lattice dis-tortion/peak broadening.

Figure 10B shows the phase frac-tion data points (obtained from theRietveld refinements), which corre-spond to the average weight fractionsof the FCC and the BCC phases overthe zones that the measurements wereperformed. Also shown in the graphare the borders of each measured zone(separated with dotted lines) as well asfits to the phase fraction data. Thephase fraction showed a sigmoidaltype of distribution with a relativelysharp change over a short distance,which corresponded to the gradedzone, as expected. The midpoint of thegraded zone should ideally correspondto the crossing point of the fits to thephase fractions (i.e., 50% of both FCCand BCC). Figure 10B shows that thiscrossing point corresponds to a dis-tance of ~10 mm from the bottom ofthe base plate, which approximatelycorresponds to the midpoint of thegraded zone based on optical observations. The evolution of the lattice parame-ters of both the FCC and the BCCphases are presented in Fig. 11. It isinteresting to note the increase in thelattice parameter of martensite and adecrease in the lattice parameter ofthe austenite in the graded regionscompared to the regions that were100% SS 410 and SS 316L, respectively. At present, the reason for the in-crease in the lattice parameter of themartensite is not yet clear. The de-

crease in the lattice parameter of theaustenite can be explained as follows.Martensite formation can induce acompressive residual stress due to thevolume expansion associated with thetransformation. This could be mani-fested as a reduction in the lattice pa-rameter of austenite. Finally, the dropin the lattice parameter of the BCCphase near the 12-mm mark could bedue to the delta ferrite formation.However, it should be noted that thisdecrease could also be due to the errorof the fit since the BCC phase fractionin this zone was considerably low —Fig. 10B. Morphology and Crystallography ofthe GTJ Microstructure. The EBSDmeasurements were performed tocomplement the optical microscopy in-vestigations. EBSD was performed pri-marily to study the microstructures inthe graded region. It may be recalledthat one of the primary aims of fabri-cating the transition joint is to arrestthe carbon migration. The idea behindusing a BCC + FCC graded region is toprevent a continuous network of BCCin a FCC matrix. It is hypothesizedthat a discontinuous BCC phase willnot provide a diffusion path for thecarbon atoms to diffuse. The EBSD in-verse pole figures and phase maps ob-tained from the base material, gradedtransition region, and the 100% SS316L zone are shown in Figs. 12 and13A–G, respectively. The phase maps in Fig. 13 show thedistribution of the FCC (austenite)

Fig. 13 — The phase maps corresponding to the previous inverse pole figures: A —100% SS 316L showing a 100% austenitic microstructure; B — 20% SS 410 region, wherethe microstructure is predominantly austenite; C — 40% SS 410 region, where localizedpools of austenite start to form; D — 60% SS 410 region, where localized pools ofaustenite start to form from this zone and gradually increases as the percentage of SS410 decreases; E — 80% SS 410 region, where localized pools of austenite start to formfrom this zone; F — 100% SS 410 region; G — base material.

Fig. 14 — The martensite and austeniteboundaries obey the KurdjimovSachs orientation relationship (marked using bluelines), proving that the BCC phase is a result of a solidstate phase transformationfrom the hightemperature austenite.

A D

E F G

B C


WELDING RESEARCH


and the BCC (ferrite/martensite) phas-es in the material, which are repre-sented with the green and red colors,respectively. Additionally, in agree-ment with the microscopy and XRD re-sults, the structure in the 100% SS316L zone was observed to be fullyaustenitic. Figures 12 and 13B revealthat the BCC (martensite/ferrite)phase formed a semicontinuous net-work inside a continuous FCC (austen-ite) matrix. This observation may sug-gest that the initial solidification tookplace in austenitic mode, and themartensite/ferrite phase was formedby a solid-state transformation duringcooling. Figure 12E–G shows that the mi-crostructure was 100% martensitic atthe dilution zone of the 2.25Cr-1Mosteel (SA 508 Grade 22 base plate)100% SS 410, and 80% SS 410-20% SS316L regions. We arrived at this con-clusion based on the heavily dislocatedstructure of martensite that manifest-ed itself on the image quality of theKikuchi patterns (Ref. 38). These observations were also foundto be in agreement with the Schaefflerdiagram (Fig. 6), which predicted a M + F type structure for these compo-sition ranges. However, as the percent-age of alloying with SS 316L increased,the fraction of austenite phases steadi-ly increased — Fig. 13B–D. The 60% SS410-40% SS 316L and the 40% SS 410-60% SS 316L agreed with the expectedA + M + F solidification structure fromthe Schaeffler diagram, with a moredominant presence of the BCC phases(red) compared to the FCC phase(green). It is also interesting to notethat the austenite and the martensitephases did not form a continuous ma-trix but were in discrete packets or locations. Thus, it is hypothesized that themicrostructure was more efficient inblocking carbon migration. However,more experiments involving postweldheat treatments are necessary to vali-date the hypothesis. Finally, the microstructure ob-served in the 20% SS 410-80% SS316L zone in Fig. 13B also appeared toagree with the expected A + F struc-ture from the Schaeffler diagram (Fig.6), with the more dominant presenceof the FCC phase (green). However, atpresent, it is not clear whether theBCC phase in this region was marten-

site or ferrite (formed due to solidifica-tion in the FA mode). The decrease in the lattice parame-ter may also support the fact that thiscould have been transformed ferriteand not martensite. To further under-stand if the ferrite is a result of solidi-fication in the FA mode, the grainboundaries between the austenite andmartensite must be analyzed. Never-theless, the fact that the observed mi-crostructures (Fig. 13) agreed withthose expected from the Schaeffler di-agram (Fig. 6), and that the martensitephase transformed from austenite, canstrongly suggest that a chemical mix-ing between the two alloys occurredduring deposition. In the case of FCC to BCC transfor-mations, there often exists a distinctorientation relationship (OR) betweenthe parent and the product phases. Awell-known commonality of the ORs isthat the close packed planes tend to beparallel to minimize the interfacial en-ergy (Ref. 39). The most common ORsin steel are the Kurdjimov-Sachs,Nishiyama Wasserman, and the Bainorientation relationships, though oth-er less widely known ORs also exist,such as Greninger and Troiano (Ref.39). The Kurdjumov-Sachs (K-S) ORrequires the following condition be-tween the two phases: {111} FCC//{110} BCC, <110> FCC//<111> BCC. How-ever, slight deviations from this exactrelation are also possible (Ref. 39). Fig-ure 14 presents EBSD micrographs(obtained at 3500) showing a moredetailed view of the BCC phase insidethe austenite grains. To test the presence of a K-S OR,the phase boundaries satisfying the{111} FCC//{110} BCC, <110> FCC//<111> BCC relation within 5 deg weredetermined using the TSL OIM soft-ware and highlighted with blue on thephase map in Fig. 14. It shows that thebase and product phase boundariessatisfy the K-S OR within a 5-deg tol-erance. It can further be observed inFig. 14 that not all of the austenite-martensite phase boundaries fit thisrelation. Several of these boundariesare pointed out with black arrows inthe phase map. However, at this stage,it is not possible to clearly state if theBCC phase is delta ferrite or marten-site. Further investigation is warrant-ed in this area. Future work would aimat evaluating the mechanical proper-

ties of the transition joint in the as-de-posited condition and after subjectingit to long-term aging to characterizethe creep response.

Implications of the CurrentResults As previously pointed out, the ideaof a graded transition joint is very re-cent but not new. The major problemthat has been reported in the litera-ture is the dilution of Ni-based alloyswith Fe leading to extensive solidifica-tion cracking. This paper proposessome design rules where SS 410 hasbeen used instead of In-82 to preventhot cracking. Based on the extensivecharacterization results, the presentresults indicate the feasibility of usingthe proposed design rules and currentfabrication technique to fabricate acrack-free, functionally graded transi-tion joint. The microstructure in thegraded region did not have a continu-ous matrix but a discrete mix of a FCCand BCC structure. The diffusivity of Cin BCC is orders of magnitude higherthan that of FCC, resulting in a discon-tinuous network of BCC that couldhelp prevent carbon migrating fromthe 21⁄4Cr-1Mo side to the austeniticside. However, before deployment ofthese joints, it is necessary to do thefollowing: 1. Validate the hypothesis that amixed microstructure leads to reducedcarbon migration due to the removalof the highways for C migration byperforming long-term aging treatments. 2. Evaluate the substitute SS 410instead of In-82 since the effect ofsuch a substitution on the creep rup-ture properties is not yet known. 3. Understand the unknown mech-anism of strain partitioning betweenthe FCC and BCC in the graded regionduring loading.

Summary and Conclusions The conclusions from this work areas follows: 1. Hot cracking was a major prob-lem while fabricating In-82 and SS316L transition joints. Hence, 12-Crsteels were used and characterized in-stead of In-82. 2. The composition gradients across


the build measured using a wavelengthdispersive spectroscopy (WDS) dis-played good agreement with the builddesign, showing that good control ofpowder-mixing ratios was achieved. 3. Optical micrographs demonstrat-ed that a good bonding was achievedbetween the base plate and the depositas well as between the subsequent lay-ers. While cracking was not observed,some level of porosity was detected,particularly in the graded zone. 4. Through scanning electron mi-croscopy and electron backscatter dif-fraction measurements, it was de-duced that the 100% SS 316L solidi-fied in the fully austenitic mode. How-ever, despite the fact that the austen-ite solidified by a primary austenite so-lidification, there was no cracking de-tected. The reason is under investiga-tion. The graded regions showed thatthe FCC and BCC phases in the gradedregions were not continuous, and thiswas hypothesized to be effective in ar-resting the C migration during service. 5. Extensive peak broadening wasobserved in the BCC phases in thegraded region, suggesting the forma-tion of martensite. The volume expan-sion of the martensite resulted in acompressive residual stress in theaustenite grains characterized by a de-crease in the lattice spacing in thegraded region.

The authors gratefully acknowledgeElectric Power and Research Institutefor funding this work. Research at theManufacturing Demonstration Facili-ty, Oak Ridge National Laboratory,was sponsored by the U.S. Departmentof Energy, Office of Energy Efficiencyand Renewable Energy, AdvancedManufacturing Office, under contractDE-AC05-00OR22725 with UT-Battelle LLC. The authors would alsolike to thank Tom Geer for metallo-graphic sample preparation and opticalmicroscopy measurements.

1. Brentrup, G. J., and DuPont, J. N.2013. Fabrication and characterization ofgraded transition joints for welding dissimi-lar alloys. Welding Journal 92(3): 72-s to 79-s.

2. Brentrup, G. J., Snowden, B. S., Du-Pont, J. N., and Grenestedt, J. L. 2012. De-sign considerations of graded transitionjoints for welding dissimilar alloys. WeldingJournal 91(9): 252-s to 259-s. 3. DuPont, J. N. 2012. Microstructuralevolution and high temperature failure offerritic to austenitic dissimilar welds. Inter-national Materials Reviews 57(4): 208–234. 4. Parker, J. D., and Stratford, G. C.1999. Review of factors affecting conditionassessment of nickel based transitionjoints. Science and Technology of Welding &Joining 4(1): 29–39. 5. Bhaduri, A. K., Venkadesan, S., Rod-riguez, P., and Mukunda, P. G. 1994. Tran-sition metal joints for steam generators —An overview. International Journal of Pres-sure Vessels and Piping 58(3): 251–265. 6. Bhaduri, A. K., Gowrisankar, I.,Seetharaman, V., Venkadesan, S., and Rod-riguez, P. 1988. Development of transitionmetal joint for steam generator circuit ofprototype fast breeder reactor. MaterialsScience and Technology 4(11): 1020–1029. 7. Dalcher, A., Yang, T., and Chu, C.1977. High temperature thermal-elasticanalysis of dissimilar metal transitionjoints. Journal of Engineering Materials andTechnology 99(1): 65–69. 8. Sudha, C., Terrance, A. L. E., Albert,S. K., and Vijayalakshmi, M. 2002. System-atic study of formation of soft and hardzones in the dissimilar weldments of Cr-Mo steels. Journal of Nuclear Materials302(2): 193–205. 9. Parker, J. D., and Stratford, G. C.2001. The high-temperature performanceof nickel-based transition joints: II. Frac-ture behaviour. Materials Science and Engi-neering A 299(1): 174–184. 10. Parker, J. D., and Stratford, G. C.2000. Characterisation of microstructuresin nickel based transition joints. Journal ofMaterials Science 35(16): 4099–4107. 11. Klueh, R. L., and King, J. F. 1982.Austenitic stainless steel-ferritic steel weldjoint failures. Welding Journal 61(9): 302-sto 311-s. 12. Sireesha, M., Albert, S. K., Shantar,V., and Sundaresan, S. 2000. A comparativeevaluation of welding consumables for dis-similar welds between 316LN austeniticstainless steel and Alloy 800. Journal of Nu-clear Materials 279(1): 65–76. 13. Sireesha, M., Shankar, V., Albert, S.K., and Sundaresan, S. 2000. Microstruc-tural features of dissimilar welds between316LN austenitic stainless steel and alloy800. Materials Science and Engineering A292(1): 74–82. 14. Gu, D., Meiners, W., Wissenbach, K.,and Poprawe, R. 2012. Laser additive man-ufacturing of metallic components: Materi-als, processes, and mechanisms. Interna-tional Materials Reviews 57(3): 133–164. 15. Farren, J. D., DuPont, J. N., and

Noecker, F. F. 2007. Fabrication of a carbonsteel-to-stainless steel transition joint us-ing direct laser deposition — A feasibilitystudy. Welding Journal 86(3): 55-s to 61-s. 16. Andersson, J.-O., Helander, T.,Höglund, L. Shi, P., and Sundman, B. 2002.Thermo-Calc & DICTRA, computationaltools for materials science. Calphad 26(2):273–312. 17. Babu, S. S. 2009. Thermodynamicand kinetic models for describing mi-crostructure evolution during joining ofmetals and alloys. International MaterialsReviews 54(6): 333–367. 18. Fukumoto, S., and Kurz, W. 1999.Solidification phase and microstructure se-lection maps for Fe-Cr-Ni alloys. ISIJ Inter-national 39(12): 1270–1279. 19. Vitek, J., Vitek, S., and David, S.1995. Numerical modeling of diffusion-controlled phase transformations in terna-ry systems and application to the ferrite/austenite transformation in the Fe-Cr-Nisystem. Metallurgical and Materials Transac-tions A 26(8): 2007–2025. 20. Elmer, J. W., Olson, D. L., and Mat-lock, D. K. 1982. Thermal expansion char-acteristics of stainless steel weld metal.Welding Journal 61(9): 293-s to 301-s. 21. David, S. A., Siefert, J. A., DuPont,J. N., and Shingledecker, J. P. 2015. Weld-ability and weld performance of candidatenickel base superalloys for advanced ultra-supercritical fossil power plants: Part I:Fundamentals. Science and Technology ofWelding and Joining 20(7): 532–552. 22. Siefert, J., and David, S. A. 2014.Weldability and weld performance of can-didate austenitic alloys for advanced ultra-supercritical fossil power plants. Scienceand Technology of Welding and Joining 19(4):271–294. 23. Siefert, J., Shingledecker, J. P.,DuPont, J. N., and David, S. A. 2015. Weld-ability and weld performance of candidatenickel based superalloys for advanced ul-trasupercritical fossil power plants: Part II:Weldability and cross-weld creep perform-ance. Science and Technology of Welding andJoining 21(5): 397–427. 24. Park, J.-W., Vitek, J. M., Babu S. S.,and David, S. A. 2004. Stray grain forma-tion, thermomechanical stress and solidifi-cation cracking in single crystal nickel basesuperalloy welds. Science and Technology ofWelding and Joining 9(6): 472–482. 25. Siefert, J., Sanders, J. M., Tanzosh,J. M., Newell, W. F., and Shingledecker, J. P.2010. Development of EPRI P87 solid wire.Materials at High Temperatures 27(3):243–252. 26. David, S. A., Siefert, J. A., and Feng,Z. 2013. Welding and weldability of candi-date ferritic alloys for future advanced ul-trasupercritical fossil power plants. Scienceand Technology of Welding and Joining 18(8):631–651.

WELDING RESEARCH

AUGUST 2017 / WELDING JOURNAL305-s

Acknowledgments

References


27. Schaeffler, A. L. 1949. Constitutiondiagram for stainless steel weld metal. Met-al Progress 56(11): 680–680B. 28. Prabhu, A. W. 2014. Improving Fa-tigue Life of LENS Deposited Ti-6Al-4Vthrough Microstructure and Process Con-trol. Master’s Thesis. The Ohio State University. 29. Lippold, J. C., and Kotecki, D. J.2005. Welding Metallurgy and Weldability ofStainless Steels. John Wiley and Sons Inc. 30. Helander, T., and Ågren, J. 1997.Computer simulation of multicomponentdiffusion in joints of dissimilar steels. Met-allurgical and Materials Transactions A28(2): 303–308. 31. Suutala, N. 1983. Effect of solidifi-cation conditions on the solidificationmode in austenitic stainless steels. Metal-

lurgical Transactions A 14(1): 191–197. 32. Lippold, J. C. 1994. Solidificationbehavior and cracking susceptibility ofpulsed-laser welds in austenitic stainlesssteels. Welding Journal 73(6): 129-s to 139-s. 33. Vitek, J., Dasgupta, A., and David,S. 1983. Microstructural modification ofaustenitic stainless steels by rapid solidifi-cation. Metallurgical Transactions A 14(9):1833–1841. 34. Bhadeshia, H., David, S., and Vitek,J. 1991. Solidification sequences in stain-less steel dissimilar alloy welds. MaterialsScience and Technology 7(1): 50–61. 35. Fukumoto, S., and Kurz, W. 1998.Prediction of the to transition inaustenitic stainless steels during lasertreatment. ISIJ international 38(1): 71–77.

36. Babu, S. S, David, S. A., Vitek, J. M.,Mundra, K., and DebRoy T. 2002. Time-resolved x-ray diffraction investigation ofprimary weld solidification in Fe-C-Al-Mnsteel welds. Acta Materialia 50(19):4763–4781. 37. Naffakh, H., Shamanian, M., andAshrafizadeh, F. 2009. Dissimilar weldingof AISI 310 austenitic stainless steel tonickel-based alloy Inconel 657. Journal ofMaterials Processing Technology 209(7):3628–3639. 38. Wilson, A., Madison, J., and Spanos,G. 2001. Determining phase volume fractionin steels by electron backscattered diffrac-tion. Scripta Materialia 45(12): 1335–1340. 39. Dahmen, U. 1982. Orientation rela-tionships in precipitation systems. ActaMetallurgica 30(1): 63–73.

WELDING RESEARCH


NIYANTH SRIDHARAN ([email protected]) and SUDARSANAM SURESH BABU are with the Department of Mechanical, Aerospace, and BiomedicalEngineering, University of Tennessee, Knoxville, Tenn. SRIDHARAN and BABU, along with BRIAN JORDAN, WILLIAM H. PETER, and RYAN R. DEHOFF are also withthe Manufacturing Demonstration Facility, Oak Ridge National Laboratory, Oak Ridge, Tenn. ERCAN CAKMAK and DONOVAN LEONARD are with the MaterialScience and Technology Division, Oak Ridge Laboratory, Oak Ridge, Tenn. DAVID GANDY is with the Electric Power Research Institute, Charlotte, N.C.

Authors: Submit Research Papers Online

Peer review of research papers is now managed through an online system using EditorialManager software. Papers can be submitted into the system directly from the Welding Jour-nal page on the AWS website (aws.org) by clicking on “submit papers.” You can also accessthe new site directly at editorialmanager.com/wj/. Follow the instructions to register or login. This online system streamlines the review process, and makes it easier to submit papersand track their progress. By publishing in the Welding Journal, more than 70,000 memberswill receive the results of your research.

Additionally, your full paper is posted on the American Welding Society website for FREEaccess around the globe. There are no page charges, and articles are published in full color.By far, the most people, at the least cost, will recognize your research when you publish inthe world-respected Welding Journal.


design, fabrication, and characterization of graded ... · pdf fileaustenitic stainless steel...

Documents