280-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

ABSTRACT. The weld microstructuralevolution, mechanical properties and so-lidification cracking susceptibility ofthree precipitation-strengthened marten-sitic stainless steels — PH 13-8 Mo, Cus-tom 450 and 15-5 PH — were investi-gated. Liquid tin quenching of gastungsten arc welds revealed that all threewelds solidified as single-phase ferritewith a high degree of microsegregation.However, during further solidificationand cooling almost complete homoge-nization occurred as a result of solid-statediffusion. The welds in all three alloys ex-hibited good resistance to solidificationcracking and generally exhibited tensileand impact properties similar to those ofthe base metal. However, in almost allcases, the weld Charpy impact energieswere somewhat less than those of thebase metals. The cracking behavior andmechanical properties are discussed interms of microstructural evolution.

Introduction

The precipitation-strengthenedmartensitic stainless steels are used fortheir high strength, higher than that of theaustenitic stainless steels, and good cor-rosion resistance, better than the harden-able 400 series stainless steels (Ref. 1).These alloys form a low-carbon marten-site with Mf temperatures just aboveroom temperature, and are precipitationstrengthened by one or more of the fol-lowing elements: Cu, Al, Ti, Nb and Mo(Refs. 2–4). In general, these materials ex-hibit good weldability; however, they ex-hibit an embrittlement coincident withthe onset of precipitation hardening that

has precluded the use of some alloys ofthis type in applications requiring hightoughness (Ref. 5).

Often, the evolution of the weld mi-crostructure is not well understood orconsidered by the welding engineer ormetallurgist. Rather, emphasis is di-rected at developing a defect-free weldthat meets minimum property require-ments. Although the engineering appli-cation is of primary importance, only byunderstanding the evolution of the mi-crostructure can one expect to readilyoptimize welding processes and subse-quent weld properties. The evolution ofweld microstructures of these alloys hasnever been studied in detail. This inves-tigation studies three such alloys: PH 13-8 Mo (USN S13800), strengthened byNiAl; 15-5 PH (USN S15500), strength-ened by particles of elemental Cu; andCustom 450 (USN S45000), strength-ened with particles of elemental Cu andpossibly a Nb- and Mo-containing Lavesphase. The solidification cracking sus-ceptibility and weld tensile and impactproperties were investigated and are dis-cussed in terms of microstructural evo-lution revealed with the use of liquid tinquenching.

Experimental

Compositions of the three martensiticstainless steels are shown in Table 1. Allthree alloys were from commercial heats.The VAR PH 13-8 was obtained fromArmco and the ESR Custom 450 was ob-tained from Carpenter Technology. Bothalloys were in the form of an 8-in. roundcorner square bar. The 15-5 PH was ob-tained from G. O. Carlson in the form of1⁄2-in. plate. The PH 13-8 and Custom 450were upset and cross forged at 1200°C(2192°F) to plate 0.6 and 0.4 in. thick, 14in. wide and approximately 3 ft long.Smooth axisymmetric tensile propertiesand the Charpy impact energy were de-termined for both welds and base metalsin the unaged condition and as a functionof aging temperature. The mechanicalproperties of the base plates were ob-tained using specimens that were austen-itized, oil quenched and then aged. Re-frigeration to liquid nitrogen temperaturesubsequent to oil quenching and prior toaging had negligible effects on thestrength and Charpy impact properties ofCustom 450 and 15-5 PH when aged be-tween 450 and 550°C (842–1022°F) for 1h. Under the same conditions, the impacttoughness of PH 13-8 was actually re-duced ~20% when aged at 450°C and re-duced ~30% when aged at 500°C, whilethe yield strength increased ~3 and 6%,respectively, as a result of refrigeration.When aged at 550°C, the properties wereunaffected as a result of refrigeration priorto aging. Consequently, refrigeration wasnot used on any of the specimens in thisstudy. Weld plates 6.4 x 30.5 cm (2.5 x12 in.) were austenitized for 1 h and aircooled prior to welding. An austenitizingtemperature of 927°C (1700°F) was usedfor PH 13-8 Mo while that for Custom450 and 15-5 PH was 1038°C (1900°F).

Mechanical test specimens were cutfrom GTA cold wire feed welds with the

Weld Microstructure Development andProperties of Precipitation-Strengthened

Martensitic Stainless Steels

BY J. A. BROOKS AND W. M. GARRISON, Jr.

The cracking behavior and mechanical properties are discussed in terms ofmicrostructural evolution

KEY WORDS

AusteniteCharpy Impact EnergyFerriteMicrostructurePrecipitation HardeningSolidificationTin QuenchingVarestraint TestingJ. A. BROOKS is with Sandia National Labo-

ratories, Livermore, Calif. W. R. GARRISON,Jr., is with Carnegie-Mellon University, Pitts-burgh, Pa.

sample length transverse to the weld di-rection. The weld fusion zone wasplaced in the center of the specimenswith the notch of the Charpy specimensat the weld root. Tensile specimens, 5mm in diameter and 20 mm long (0.2 x0.79 in.), contained a composite of thefusion zone (~50%), HAZ and basemetal. The tensile specimens deformeduniformly when tested, indicating no onezone dominated the tensile response;specimens also fractured near the weldcenter. The nominal weld parametersand plate thicknesses corresponding tothe test results are shown in Table 2. A U-groove joint preparation and 1.14-mm(0.045-in.) welding wire were used withan interpass temperature of ~65°C(~150°F) for both plate thicknesses.

The solidification cracking susceptibil-ity was evaluated in the unaged conditionusing sub-size Varestraint testing (Ref. 6).For comparison, tests were conducted ontwo commercial vacuum-inductionmelted/vacuum-arc remelted heats of304L (USN S30403): one that solidified asprimary ferrite, designated 304L-F, and theother that solidified as primary austenite,designated 304L-A. The compositions ofthese two heats of 304L are shown inTable 1. Welding parameters used duringVarestraint testing are given in Table 2.

Metallographic samples of tinquenched and Varestraint specimens

were prepared using standard metallo-graphic techniques with the final polishmade with a 0.05-micron silica aqueoussolution. A Vilella’s etch was used to re-veal the microstructural features. Micro-probe analysis for compositional deter-mination was conducted using a Joel8900 microprobe with a 1-micron beamsize operated at 15 and 25 KEV, de-pending upon feature size, and a 25-nAbeam current. Samples for TEM analysiswere electropolished and examined on aJoel 200 CX microscope. Fracture sur-faces of Charpy samples were examinedusing SEM.

Microstructural Evolution

In this investigation, GTA welds werequenched with liquid tin during weldingin an attempt to capture insitu the evolu-tion of the weld microstructure from thefirst stages of solidification to that ob-served at room temperature. Figure 1

shows results from a Custom 450 weldobtained using this technique. The mi-crostructure is evolving from left to rightwith the quenched weld pool and solidi-fying dendrite tips shown in the left sideof the micrograph. Microsegregation oc-curring from elemental partitioning be-tween the solid and liquid during weld so-lidification is evident in the interdendriticregions of the cellular/cellular-dendriticsolidifying structure. However, it appearsthat homogenization occurs during fur-ther solidification and cooling (movingfrom left to right in the micrograph). At alower temperature (in the right-hand sideof the micrograph), the ferrite has startedto transform to austenite by a diffusion-controlled process. When cooled belowthe Ms (in this case, assisted by the tinquenching), the austenite transforms tomartensite. Thus, the total evolution ofthe weld microstructure can be observedin one micrograph.

Electron microprobe analysis was

WELDING RESEARCH SUPPLEMENT | 281-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

50 µm

Fig. 1 — Montage of liquid tin-quenched Custom 450 GTA weld showing the evolution of the weld microstructure from the dendrite tips in theweld pool to the region of solid-state transformation of ferrite to austenite. Captions show sequence of transformations.

conducted to determine elemental parti-tioning and microsegregation in thequenched Custom 450 weld sampleshown in Fig. 1. An analysis using a 1.75-micron spacing was taken across thedendritic structure near the solidificationfront in the region shown at higher mag-nification — Fig. 2A. The results areshown for Ni, Nb, Cu, Mo and Si in Fig.2B, C. It can be seen the partitioning co-efficients (k defined as Cs/Cl, where Cs =concentration of the solid and Cl = con-centration in liquid) are less than 1 for allthese elements. That is, the intracellularregions, the first to solidify, are depletedin these elements and the cell bound-aries, the last regions to solidify, are en-riched. Partitioning coefficients near theliquidus, calculated using microprobeanalysis of the cell tips and the bulk com-position given in Table 1, are shown inTable 3. Although carbon was not mea-sured, it is known that carbon also parti-tions to the liquid, k<1.

The interdendritic region near the so-lidification front is shown at higher mag-nification in the SEM micrograph of Fig.3. As discussed below in the section onsolidification cracking, the second phaseis composed primarily of NbC. It is im-portant to note in the optical micrographof Fig. 1 that, at a distance of about 400microns (0.016 in.) from the solidifica-tion front, little evidence of the secondphase exists.

An electron microprobe analysis,again using a spacing of 1.75 microns(6.9 x 10-5 in.), was conducted transverseto the weld along a line 400 micronsfrom the weld front where in Fig. 1 thereis no evidence of microsegregation. Themicroprobe analysis shown in Fig. 4 forNi, Nb and Cu verified the material inthis region is very homogeneous com-pared to the region near the solidificationfront shown in Fig. 2A. The Ni variationwas reduced from ~2.5 wt-% at the so-lidification front to only ~0.7 wt-% at adistance 400 microns from the solidifica-tion front. Similarly, variations were re-duced in Cu from 0.65 to 0.2 wt-%, in Nbfrom 1.2 to 0.2 wt-%, in Si from 0.35 to0.04 wt-% and in Mo from 0.4 to 0.1 wt-%.

Welds of PH 13-8 were also liquid tinquenched. This alloy solidified in thesame fashion as Custom 450, that is, assingle-phase ferrite. The microstructuresin Fig. 5A taken near the solidificationfront clearly show evidence of alloy par-titioning that occurred during weld solid-ification. Microprobe analysis was con-ducted in regions of the cell core and cellboundaries. The average of a number ofanalyses in the two regions were the fol-lowing: Ni core = 7.6 wt-%, boundary=10.5 wt-%; Mo core = 1.9 wt-%,

282-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

Fig. 2 — SEM micrograph (A) showing solidification front of sample in Fig. 1. The line indicatesregion of microprobe analysis shown in B and C.

A

B

C

WELDING RESEARCH SUPPLEMENT | 283-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

boundary = 2.6 wt-%; and Al core = 1.2wt-% and boundary = 1.4 wt-%. How-ever, it can be seen in this sample the verytips of the solidifying dendrites have notbeen captured. Thus, it must be expectedthat, as a result of some solid-state diffu-sion occurring during solidification, theconcentrations measured at the dendritecores are somewhat higher than whatwould exist at the dendrite tips. There-fore, the initial partitioning coefficientsshown for PH 13-8 in Table 3 are likelysomewhat high.

The region of the solid-state transfor-mation of ferrite to austenite is shown inFig. 5B ~1000 microns (~0.039 in.) fromthe weld front. Note that in the region ofthe quenched ferrite no evidence of seg-regation exists. Similar to Custom 450,homogenization of the PH 13-8 weldsalso occurred during solidification andcooling. As a result of elemental parti-tioning during the ferrite-to-austenitetransformation, a small amount of ferriteis retained along the Widmanstattenaustenite boundaries. In this weld, theferrite fraction appeared to be 1–2%.

Discussion of Microstructure Evolution

In Custom 450, as well as the other twoalloys studied, little evidence of microseg-regation exists in the room-temperatureweld microstructures. Thus, one may haveexpected that partitioning coefficientscould be close to 1. However, in the tin-quenched samples it was found a high de-gree of alloy partitioning does occur dur-ing solidification. The experimentallymeasured partitioning coefficients shownin Table 3 vary from ~0.85 for Ni to ~0.36for Nb. It is evident during solidificationand cooling, even at the fairly rapid cool-ing rates of welding, a large degree of ho-mogenization can occur. This observedbehavior is in agreement with the weld so-lidification model developed by Brooksand Baskes (Ref. 7) in which they incor-porated solid-state diffusion. In their mod-els, they showed, for example, that Nb inFe, with a partitioning coefficient of~0.25, rapidly diffuses during weld solid-ification, resulting in a nearly homoge-neous composition. However, more re-

cently they have shown that even in GTAwelds undercooling can result in higherNb concentrations of the dendrite tip, butthis effect is diminished during further so-lidification and cooling as a result of solid-state diffusion (Ref. 8). It must be expectedthat, in the welds in this study, some tip un-dercooling also occurred that slightly in-creased the initial dendrite tip concentra-tion above the equilibrium value forelements with partitioning coefficientsless than 1.

The quenching results shown for Cus-tom 450 and PH 13-8 provide insight tothe microstructural evolution of othermartensitic stainless steels. In both Cus-tom 450 and PH 13-8, the welds solidi-fied as single-phase ferrite with no sec-ondary solidification of austenite. Onecan predict the solidification behavior ofstainless steels using Creq/Nieq ratios (Ref.9) from a variety of diagrams, such as theWRC-1992 diagram (Ref. 10). For thecomposition of Custom 450 used here,the Creq/Nieq ratio, using the WRC, is1.87, above the minimum value of ~1.7predicting solidification occurring com-

Fig. 3 — SEM micrograph showing at higher magnification thedendritic region near the tin-quenched solidification front of theCustom 450 weld.

Fig. 4 — Microprobe analysis taken across the dendrite structure ~400 micronsfrom the solidification front of the sample shown in Fig. 1.

Fig. 5 — A — Solidification structure near the weld pool (to the right) of a liquid tin-quenched PH 13-8 Mo GTA weld showing evidence of alloysegregation; B — region of the decomposition of ferrite to austenite ~1 mm from the solidification front.

A B

284-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

pletely as ferrite. This prediction of ferritesolidification is consistent with the be-havior shown in Fig. 1. The Creq/Nieq ratiofor 15-5 PH (using N=0.03) and PH 13-8are 2.24 and 1.60, respectively. Onewould then predict that 15-5 PH wouldalso solidify as single-phase ferrite andexhibit similar behavior to that shown inFig. 1. However, it may have been ex-pected that the heat of PH 13-8 with aCreq/Nieq ratio of 1.60 used in this studywould have solidified as primary ferritewith the peritectic solidification ofaustenite. The tin-quenched welds of thePH 13-8 alloy clearly showed that solid-ification occurred as single-phase ferrite,with the solid-state transformation takingplace at a temperature considerablylower than that of final solidification.However, as discussed by Cieslak, et al.(Ref. 11), if Al is included in the Creq with

a value of 2.48, the Creq/Nieq is 1.90, wellwithin the single-phase ferrite solidifica-tion region.

It is interesting to note in Fig. 1 amartensitic structure is evident through-out the microstructure. In the lower tem-perature regions of the micrograph (rightside), the ferrite that transformed via a dif-fusional process to austenite transformedto martensite when cooled to a littleabove room temperature, Ms~120°C(250°F). The microstructure in the leftside of the micrograph is different thanthat on the right side. It is proposed that,at the higher temperatures near the weldfront, rapid cooling rates during quench-ing suppressed the diffusion-controlledtransformation of ferrite to austenite, al-lowing the ferrite to transform massively(Ref. 12) to austenite, as has been re-ported in some austenitic stainless steels

(Refs. 13, 14). It has also been suggesteda massive transformation can occur dur-ing water quenching thin sheets of PH13-8 (Ref. 13). Upon further cooling, thisaustenite also transformed to martensite.This later sequence of transformations isa manifestation of the tin quench and isnot representative of the microstructuralevolution of the GTA welds. However,this sequence of transformations couldoccur under the rapid cooling rates of thehigh-energy-density welding processes.

Mechanical Properties

The tensile properties of both weldsand base metal were obtained for the un-aged condition and as a function ofaging temperature for an aging time of3.2 h. An aging time of 3.2 h was chosenfor direct comparisons with base metal

Fig. 6 — Transverse weld tensile yield strengths (0.2% offset) of thethree alloys studied.

Fig. 7 — Yield strengths of weld and base metals of Custom 450 inthe unaged condition and as a function of aging temperature.

Fig. 8 — Yield strengths of weld and base metals of PH 13-8 Mo inthe unaged condition and as a function of aging temperature.

Fig. 9 — Yield strengths of weld and base metals of 15-5 PH in theunaged condition and as a function of aging temperature.

WELDING RESEARCH SUPPLEMENT | 285-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

properties for which aging times of up to100 h were studied (1, 3.2, 10, 32 and100 h). The 3.2 h is close to the conven-tional aging time of 4 h for these materi-als. The tensile data reported are the av-erages of two tests in which the variationin the two test results were typicallywithin several ksi for both the welds andbase metals. The weld tensile yieldstrengths are shown in Fig. 6. The agingresponse of the welds of all alloys is sim-ilar in that the yield strength reaches amaximum at intermediate aging temper-atures and then decreases with furtherincreases in aging temperature. Peakyield strengths for the 3.2-h age wereabout 195 ksi for PH 13-8 and about 180ksi for Custom 450 and 15-5 PH. Thepeak strengths of Custom 450 and 15-5PH are reached at an aging temperature

50–75°C less than the aging temperatureat which the peak yield strength wasachieved for PH 13-8, ~450 vs. 525°C(842 vs. 977°F).

The yield strengths of the weld andbase metal of the three alloys studied areplotted in Figs. 7–9 as a function of agingtemperature. In all the alloys, the natureof the aging response is similar, thoughvarying slightly in detail. Therefore, theirgeneral behavior can be discussed to-gether. In the unaged condition, thewelds have a higher yield strength thanthe base metals. This trend also exists forthe lower aging temperatures, with thepeak strengths of the welds and base met-als being very similar. Except for 15-5PH, the peak strength of the welds isreached at an aging temperature ~50°Clower than that of the base metals. For

aging temperatures greater than that pro-ducing peak strength, the base metal ofboth Custom 450 and PH 13-8 hasslightly higher strengths than the weldmetal. However, in all three alloy sys-tems the welds have a higher strengththan the base metal when aged at 600°C(1112°F). Thus, the results suggest thatweld microstructures age harden moreintensely at lower aging temperaturesand the weld microstructures overagemore slowly than the base metals.

The Charpy impact energies of Cus-tom 450 and PH 13-8 welds and basemetals were determined for the unagedconditions and as a function of agingtemperature for an aging time of 3.2 h.As shown in Fig. 10, the Charpy impactenergy of the Custom 450 base metal is

Fig. 10 — Charpy impact energies of the weld and base metals forCustom 450 in the unaged condition and as a function of aging tem-perature.

Fig. 11 — Charpy impact energies of the weld and base metals for PH13-8 Mo in the unaged condition and as a function of aging temper-ature.

Fig. 12 — TEM dark field micrograph of PH 13-8 Mo weld fu-sion zone aged for 3.2 h at 550°C showing distribution of NiAl.

Fig. 13 — Charpy impact energy and yield strength of PH 13-8 Mo basemetal plotted as a function of aging temperature.

286-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

125 ft-lb in the unaged condition and de-creases with increasing aging tempera-ture to a minimum value of 24 ft-lb onaging at 425°C (797°F). On aging above425°C, the impact energy increases withincreasing aging temperature. TheCharpy impact energies obtained fromthe two weld plates of Custom 450 arequite similar, as shown in Fig. 10. In theunaged condition, they are ~25% lowerthat that of the base metal. At an agingtemperature of 400°C (752°F), the im-pact energy drops drastically to only~35–40% of that of the base metal. Ataging temperatures from 450 to 550°C(842 to 1022°F), the Charpy impact en-ergy of the weld metal is still less thanthat of the base metal, even though in

this range the Charpy impact energy ofboth the weld and base metal increasewith increasing aging temperature.

The Charpy impact energies of theweld and base metals for PH 13-8 areshown in Fig. 11. In the unaged condi-tion, the impact energy of the base metalis quite high, at 161 ft-lb, but decreasesto 25 ft-lb at aging temperatures between450 and 525°C. For aging temperaturesgreater than 525°C, the Charpy impactenergy increases rapidly with increasingaging temperature. Thus, for the basemetal, there is an embrittlement of thealloy that coincides with the onset of sig-nificant age hardening, and the mini-mum in Charpy impact energy is ob-tained at an aging temperature about

50°C lower than the aging temperature atwhich the maximum yield strength is ob-tained. Both welds exhibit behavior qual-itatively similar to that of the base metal,but the levels of toughness achieved bythe two welds are very different. Weld 2has a Charpy impact energy in the un-aged condition of 105 ft-lb, while theCharpy impact energy of weld 1 in theunaged condition is only 56 ft-lb. Whileboth welds exhibit an embrittlementtrough in the same range of aging tem-peratures as exhibited by the base metal,material from Weld 1 is, at every agingtemperature, of lower toughness thanmaterial taken from Weld 2.

Fig. 14 — Fracture surfaces of Charpy impact specimens of PH 13-8 Mo base metal aged at: A — 450°C; B — 500°C; C — 550°C (all same mag-nification).

Fig. 15 — Cross sections of the welds of: A — Custom 450; B — PH 13-8 Mo.

A

A B

B C

Discussion of MechanicalProperties

In using these materials, one mustchoose the aging temperature that pro-vides the best combination of strengthand toughness for the particular engi-neering application. The differences inweld and base metal aging response mustthen also be considered. Qualitatively,for a given alloy type, the base metalsand welds exhibited similar age-harden-ing responses in that similar peak yieldstrengths were obtained and the peakyield strengths were obtained at similaraging temperatures. However, there areconsistent differences.

The primary difference between thebase and weld metal aging responses wasthat the weld metals exhibited higherstrengths than did the base metals in theunaged condition and at the lower aging

temperatures. This effect is believed to bedue to the thermal cycles experienced byweld deposits from subsequent weldpasses. After each weld pass, the weldplates were allowed to cool to belowabout 65°C, less than the martensite starttemperatures for PH 13-8, Custom 450and 15-5 PH, which are 120, 118 and132°C (248, 244 and 270°F), respec-tively. Therefore, one would expect anew weld pass to have the effect of heat-ing some region of the previously de-posited weld to temperatures sufficient toresult in particle precipitation in themartensite. While the times at these tem-peratures may be relatively brief and thetemperature excursions themselves maynot result in marked age hardening of themartensite, they could have the effect ofaltering particle precipitation kinetics onsubsequent aging of the weld. This isconsistent with the yield strengths of the

welds being greater than the yieldstrengths of the base metals in the un-aged conditions, the welds having higheryield strengths than the base metals afteraging at 400°C (752°F) and the weld met-als achieving their maximum yieldstrength at lower aging temperatures thanthose of the base metals.

While there were small but consistentdifferences between the aging responsesof weld and base metals for these alloys,the peak strengths obtained for the baseand weld metals after age hardeningwere similar. This similarity in strengthcan be related to the evolution of theweld microstructure, as observed in thetin-quenched samples. Here, it wasshown that considerable alloy partition-ing occurred during weld solidification,but solid-state diffusion was sufficientlyfast for a large degree of homogenization

WELDING RESEARCH SUPPLEMENT | 287-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

Fig. 16 — Fracture surfaces at two magnifications of Custom 450 weld Charpy impact specimen aged at 475°C. The fracture path is bottom to topin both fractographs.

Fig. 17 — Fracture surfaces of Custom 450 base metal Charpy impact specimens: A — Aged at 475°C for 3.2 h; B — aged at 450°C for 100 h.

A B

288-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

to occur during further solidification andcooling. This resulted in a uniform pre-cipitate distribution throughout the weldmicrostructure represented in the TEMmicrograph (Fig. 12) taken from the fu-sion zone of a PH 13-8 weld aged at550°C (1022°F) for 3.2 h.

This behavior can be contrasted to theprecipitation-strengthened austeniticstainless steels in which the high degreeof microsegregation is retained to roomtemperature due to the sluggish diffusionkinetics in the austenitic structure. Thisoften results in a nonuniform precipitatedistribution with precipitate-free den-drite cores and welds with inferior me-chanical properties (Refs. 15, 16). Thesimilar aging response of the weld andbase metal of the class of alloys studiedhere is an attractive engineering featureof these materials.

The Charpy impact energy results forthe base and weld metal displayed simi-lar behavior in that both exhibited a min-imum in Charpy impact energy at anaging temperature slightly less than theaging temperature at which the peakyield strength was obtained. However, atmost aging temperatures, the weld metalfor both Custom 450 and PH 13-8 did notachieve the same levels of Charpy impactenergy as did the base metals. There arealso significant differences between thelevels of toughness achieved by the twoPH 13-8 weld plates. These differences intoughnesses can be discussed best by firstreviewing the effect of aging temperatureon the toughness of PH 13-8.

The Charpy impact energy and yieldstrength of PH 13-8 aged for 1 h are plot-ted as a function of aging temperature inFig. 13. The room-temperature Charpy

impact energy decreases dramatically onaging at 450°C, has a minimum on agingat 500°C (932°F) and then increases withincreasing aging temperature. The mini-mum in Charpy impact energy occurs atan aging temperature 50°C less than thatresulting in peak yield strength. The em-brittlement of this alloy coincident withthe onset of precipitation strengthening isassociated with a change in fracturemode — Fig. 14. Even after aging at450°C, fracture is by microvoid coales-cence, but after aging at 500°C the frac-ture mode is primarily quasi-cleavage. Asthe aging temperature is increased be-yond 500°C, the fracture mode returns tomicrovoid coalescence. It may bethought that this embrittlement behavioris related to microstructural changes as-sociated with the decomposition of re-tained austenite, such as carbide precipi-tation at lath boundaries and/or changesin the mechanical stability of the retainedaustenite, or the precipitation of intra-lathparticles. However, there is no evidencethat in PH 13-8 there is any effect of agingtemperature on austenite content, me-chanical stability or morphology over therange of aging temperatures of interest(Ref. 5). It has also been found that in in-vestment-cast material, austenite rever-sion does not occur at aging temperaturesat or below 566°C (1051°F) (Ref. 18).Therefore, the embrittlement is ascribedto the precipitation of NiAl particles,which are coherent to the matrix andwhich would be expected to be shearedby dislocations at smaller particle sizes(Ref. 18). This shearing of particles is be-lieved to result in a local work softeningthat promotes the strain localization re-quired to initiate cleavage fracture.

The results for Custom 450 in Fig. 10show that the Charpy impact energies inthe unaged condition are less in the weldsthan in the base metal (~95 vs. 125 ft-lb).Three factors could be contributing to thisdifference in toughness. The first is thatsome precipitation of the strengtheningparticles is believed to have occurred inthe weld metal; this alone would be ex-pected to lead to reduced toughness in theweld metal. Second, the inclusion volumefraction and spacing in the weld metalcould be different from those in the basemetal. The particles in the base metal atwhich microvoids can initiate are primar-ily 5-micron-diameter Al- and Si-contain-ing oxides, ~2-micron-diameter Nb-en-riched carbonitrides and small~0.15-micron Mn-enriched inclusions.By both introducing small amounts ofoxygen and nitrogen during welding,which would increase the volume frac-tion of inclusions, and by refining particlesize and spacing during weld solidifica-tion and cooling, the toughness would de-crease (Refs. 19, 20). However, morework is needed to better characterize in-clusions in weld deposits in these materi-als. The third factor that could result inlower weld metal toughness is the largergrain size, which can influence toughnesswhen the fracture mode is microvoid co-alescence and the inclusion spacingsmall, as in Custom 450 (Ref. 19).

The Custom 450 weld macrostruc-ture, shown in cross section in Fig. 15, ischaracterized by a columnar grain struc-ture that persists through the entire weldthickness. This structure is a result of epi-taxial growth and preferred solidificationgrowth directions. A low magnificationfractograph of the Custom 450 weld aged

Fig. 18 — Fracture surface of PH 13-8 weld aged at 525°C. Note sec-ondary cracking associated with weld filler passes.

Fig. 19 — Subsize Varestraint test results showing solidification crack-ing behavior of the three martensitic stainless steels and two heats of304L: 304L-A, primary austenite solidified, and 304L-F, primary fer-rite solidified.

WELDING RESEARCH SUPPLEMENT | 289-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

at 475°C is shown in Fig. 16A. The frac-tograph indicates vertical striations asso-ciated with the long columnar grains ob-served in Fig. 15. This fracture surface isseen in higher magnification in Fig. 16B.These fractures exhibit a cleavage frac-ture mode with cleavage facets both par-allel and perpendicular to the macro-scopic fracture surface. The facets tend tohave their longest dimension parallel tothe columnar grains. If the cleavagefacets are [100] planes, as is often ob-served in b.c.c. materials, the frac-tographs are consistent with the colum-nar grains having a [100] growthdirection. For comparison, the fracturesurface of Custom 450 base metal agedat 475°C (887°F) is shown in Fig. 17A.The fracture surface consists of bothquasi-cleavage and microvoid coales-cence, with cleavage facets 10–20 mi-crons in dimension. The fracture surfaceof Custom 450 base metal aged at 450°Cfor 100 h, which resulted in a Charpy im-pact energy of about 8 ft-lb, is shown inFig. 17B. The cleavage facets are moreeasily recognized on this fracture sur-face, but, again, they have dimensions ofabout 10 to 20 microns. Therefore, it issuggested that the lower toughness of theweld metal, particularly for aging tem-peratures of 450 through 500°C(842–1022°F) are strongly influenced bylarge and oriented grains, which promotecleavage fracture. However, after agingat 400°C (752°F), the strength of the weldmetal is considerably greater than thestrength of the base metal, as discussedpreviously. It is believed that the onset ofthe embrittlement at this temperature isprimarily due to the precipitation ofshearable particles.

In comparing the Charpy impact en-ergies of the PH 13-8 weld and base

metal, there are a number of points of in-terest. Again, the Charpy impact energiesof the welds are less than the impact en-ergy of the base metal for the unaged andaged (at 400°C) conditions. Microstruc-tural factors that could lead to the weldmetal having lower toughness than thebase metal for the unaged condition andafter aging at 400°C would be the sameas those presented in the discussion ofthe toughness of Custom 450. However,it is not clear why one of the PH 13-8welds had much higher toughness thanthe other, although it is believed to bedue to microstructural differences intro-duced by welding. It is possible that therewere slight undetected variations in thewelding procedure between the twoplates. The effects of different thermal cy-cles and interpass temperature of themultiple pass welds on the aging re-sponse, austenite reversion, carbide pre-cipitation, etc., can be very complex indetermining the final weld microstruc-ture. More work is needed to determinethe sensitivity of weld process variableson weld properties.

As shown in Fig. 14, the welds of PH13-8 are not characterized by the longcolumnar grains observed for Custom450. For PH 13-8, the regions associatedwith each weld pass appear to be clearlydelineated and large columnar grains donot appear to continue past theseboundaries, as they did in the Custom450 welds. The interfaces between weldpasses appear to be important to thefracture process. As shown in Fig. 18,the crack advancing from the base of thenotch appears to interact strongly withthese interfaces, and these interfaces areassociated with secondary cracks —perpendicular to the fracture surface —exhibiting a ductile fracture mode. The

energy absorbed by the secondarycracking processes could strongly influ-ence the Charpy impact energy. Thesesecondary cracks were not observed onthe fracture surfaces of the Custom 450welds that were examined. Given thehigh aluminum content of PH 13-8, thepossibility exists that the particles at theinterfaces between the weld passes inthe PH 13-8 welds, which lead to theductile secondary cracking along theseinterfaces, are aluminum oxide and alu-minum nitride particles.

Solidification Cracking Susceptibility

Weld solidification cracks may formduring the final stages of solidification ifsufficient stress or strain occurs to fracturethe solidifying structure. Cracking occursprimarily along grain boundaries con-taining low-melting liquids. The weld so-lidification cracking susceptibility of Cus-tom 450, PH 13-8 and 15-5 PH wasevaluated using sub-size Varestraint test-ing (Refs. 6, 21). The test results, plottedas total fusion zone crack length vs. aug-mented strain, are shown in Fig. 19. Forcomparison, results from two heats of304L, one that solidified as primary fer-rite, 304L-F, with a measured ferrite num-ber FN = 4.5, and the other that solidifiedas primary austenite, 304L-A with FN~1,are also plotted. The Varestraint test re-sults show that PH 13-8 is extremely re-sistant to solidification cracking; nocracking occurred at applied strains lessthan 2%, comparable to that of primaryferrite solidified 304L. Custom 450 and15-5 PH exhibited similar cracking be-havior and were more crack susceptiblethan PH 13-8. However, the behavior ofthe two alloys was much closer to the heat

Fig. 20 — A — Solidification crack tips region of a Custom 450 Varestraint sample showing second phases associated with cracking (15-5 PH ex-hibited very similar appearance); B — solidification crack tip region of PH 13-8 Mo Varestraint sample.

A B

304L-F solidifying as primary ferrite thanthat of primary austenite, 304L-A, espe-cially at the low strain levels. The maxi-mum crack lengths of the martensiticstainless steels for three different strainlevels are shown in Table 4. It can be seenthe maximum crack length for Custom450 and 15-5 PH are very similar and areabout twice as long as those of PH-13-8Mo. This behavior is consistent with thetotal crack length data shown in Fig 19.

Solidification cracks were analyzed todetermine the nature of the phases asso-

ciated with cracking that could be iden-tified at room temperature. Crack tips ofa Custom 450 Varestraint sample areshown in Fig. 20A. Little observable mi-crosegregation is evident except in theboundaries in front of the crack. Analyti-cal electron microscopy analysis of par-

ticles extracted from thegrain boundary regionsshowed the second phasecontained primarily Nband lighter elements. Theelectron diffraction pat-terns index as Nb(C,N).Microprobe analysis ofthe region containing thesecond phases alsoshowed enrichments of P.

The crack tip regionof a PH 13-8 is shown inFig. 20B. In these sam-ples, little evidence of mi-crosegregation exists ex-cept for a slightsegregation a few mi-crons ahead of the cracktip. This region containshigher Ni (~12%) vs. theaverage alloy composi-tion 8%Ni, and higherMo, ~5 vs. ~2.2% for thealloy average. The fewlight second-phase parti-cles in Fig. 20B are highin Mo and P but were notidentified. The lacy sec-ond phase to the bottomleft of the micrograph isdelta ferrite residual fromthe solid-state transfor-mation of ferrite toaustenite.

The weld solidifica-tion cracks in 15-5 PHwere very similar in ap-pearance to those shownin Fig. 20A for Custom450 and contain the sameNb(C,N) particles. The15-5 PH Varestraint testspecimens also exhibitedvery small shallow inter-

granular surface cracks in the weld HAZ.An example of these cracks in a metallo-graphically prepared sample is shown inFig. 21A. These cracks, shown at highermagnification in Fig. 21B, were associ-ated with Cu apparent as a solidified liq-uid in the grain boundary (note arrow inFig. 21B). These surface cracks are at-tributed to the vaporization of Cu underthe arc and its condensation on thecooler base metal ahead of the weld.During Varestraint testing, strains weresufficient at temperatures above the melt-

ing point of Cu for the fine surface cracksto form.

Discussion of Solidification Cracking

The extensive use of the austeniticstainless steel alloys and the knowledgeof their solidification mode and crackingsusceptibility provides a good basis forcomparison of the three martensitic al-loys. Welds in the 300 series alloys thatsolidify as austenite can be very suscep-tible to solidification cracking. Thecracking susceptibility increases with in-creasing levels of P, S and Si that promotethe formation of low-melting liquids.However, compositions that solidify asprimary ferrite with the peritectic solidi-fication of austenite, or in which thetransformation of ferrite to austenite startsto occur at a temperature close to thebulk solidus, are extremely resistant tocracking (Refs. 22–25), even with highlevels of P and S (Refs. 26, 27). In com-positions that solidify completely as fer-rite and the solid-state transformation offerrite to austenite does not occur untillower temperatures, cracking suscepti-bility can then again increase (Ref. 28).The results in Fig. 19 for 304L-A and304L-F are consistent with the known so-lidification behavior of 304L.

Figure 19 shows like primary ferrite so-lidified 304L (304L-F), the PH 13-8 wasvery resistant to solidification cracking.The cracks that did form were fairly bluntin nature, indicating that a high degree ofstrain was required for their initiation andpropagation. One of the reasons for thehigh cracking resistance is evident whenexamining regions of the cracks that didform when highly strained. Fig. 20Bshows extensive second phases were notassociated with the solidification cracksand only a small degree of microsegrega-tion is present. However, it is more infor-mative to examine the solidification frontof the tin-quenched welds in Fig. 5A,where the microsegregation and mi-crostructure is characteristic of that whenthe cracks form. Here, it is more evidentthat the composition of the liquid presentwhen cracking occurred is enriched in Niand Mo. Nevertheless, the degree of seg-regation is still relatively small due to theabsence of minor alloying elements, suchas Nb with small partitioning coefficients.In this alloy, Si and the impurities, P andS, which further promote low-melting liq-uids, are also very low. Thus, the absenceof reaction products of low-melting liq-uids in the solidifying microstructure isconsistent with the high solidificationcracking resistance and relative shortcrack lengths in the Varestraint tests.

290-s | AUGUST 1999

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

Fig. 21 — A — Shallow intergranular surface cracks in the weldHAZ of a 15-5 PH Varestraint test specimen; B — higher mag-nification of A showing Cu in the grain boundaries.

A

B

The alloys Custom 450 and 15-5 PHare more crack susceptible than the alloyPH 13-8, even though all three alloys so-lidified as single-phase ferrite. In these al-loys, the examination of Varestraint sam-ples (Fig. 19A) shows the second phasespresent that can be related to cracking.However, the tin-quenched welds (Fig. 3)further show the high degree of alloy par-titioning of a number of elements; thewelds also show the degree of microseg-regation occurring, which leads to theformation of low-melting liquid. Theseresults would suggest the synergistic ef-fects of a number of elements, though theexamination of Varestraint samples maysimply suggest the segregation of a Nb-containing liquid solidifying by a eutec-tic reaction involving NbC. It can also beexpected that the elements Si and P mayhave further suppressed the temperatureof final eutectic solidification.

The maximum crack lengths of theVarestraint tests are related to the tem-perature range of the cracking suscepti-ble region and the thermal gradient in theregion in which the crack forms. Sincethe low-temperature solidification reac-tions are similar for Custom 450 and 15-5 PH, it is not surprising that the maxi-mum crack lengths of the two alloys arealso similar. They are also much greaterthan those in PH-13-8 Mo, in which nosignificant secondary solidification reac-tions were observed.

Summary

The evolution of weld solidificationbehavior and microstructure develop-ment was clearly revealed using tin-quenching techniques. All three alloyswere shown to solidify completely as fer-rite, even though some existing diagramspredict that PH 13-8 would solidify asprimary ferrite with some secondary so-lidification of austenite. A high degree ofalloy partitioning occurs during solidifi-cation, but the welds are almost com-pletely homogenized during further so-lidification and cooling.

The strength and Charpy impact prop-erties of the weld and base metals werequalitatively similar for the aging condi-tions studied. This similarity in the prop-erties of the base metals and the weldswas attributed to homogenization occur-ring during solidification and cooling,which leads to very similar microstruc-ture and precipitate distribution. How-ever, there were systematic differencesbetween the aging responses of the weldand base metals; in particular, the weldshad, at lower aging temperatures, astronger aging response than did the basemetals. In addition, the toughness of the

welds exhibited an embrittlement atlower aging temperatures than did thebase metal.

Custom 450 and 15-5 PH would ap-pear from Varestraint tests to be fairly re-sistant to solidification cracking. In thesetests, PH 13-8 was shown to be extremelyresistant to solidification cracking. Thetin-quenched welds revealed the natureof microsegregation and structure char-acteristic of the state in which solidifica-tion cracking occurs. Cracking of bothCustom 450 and 15-5 PH was attributedprimarily to the segregation of Nb and theformation of a low-melting liquid, result-ing in the eutectic solidification involv-ing NbC. Segregation of Si and Mo mayfurther promote cracking susceptibility.

Acknowledgments

The microprobe analysis by P. Hlavaand N. Yang, metallography by A. D.Gardia and laboratory support by J. S.Krafcik, all of Sandia National Laborato-ries, are greatly appreciated. This workwas supported by the U. S. Departmentof Energy under Contract Number DE-AC04-76DP00789.

References

1. Pollard, B. 1993. Selection of wroughtprecipitation-hardening stainless steels. ASMHandbook 6: 482–494.

2. Pickering, F. B. 1963. Iron and Steel In-stitute, Pub. 114, London, U.K., p. 131.

3. Haudin, J. M., and Montheillet, F. 1978.Study by electron-microscope of hardeningprecipitation in delta ferrite of a stainless-steel(15%Cr,7%Ni, 2%MO): evidence of R-phaseprecipitation. Metallography 11(4): 391–438.

4. Pickering, F. B. 1978. The Physical Met-allurgy and the Design of Steels, Applied Sci-ence Publishers Ltd., London, U.K., p. 177.

5. Garrison, Jr., W. M., and Brooks, J. A.1991. The thermal and mechanical stability ofaustenite in the low carbon martensitic steelPH 13-8. Mater. Sci. Eng. A149: 65–72.

6. Campbell, R. D., and Walsh, D. W.1993. Weldability testing. ASM Handbook 6,603–613.

7. Brooks, J. A., and Baskes, M. I. 1986.Weld microsegregation characterization andmodeling. Conference on Trends in WeldingResearch. Advances in Welding Science andTechnolgoly, ed. S. A. David, ASM Interna-tional, Materials Park, Ohio, pp. 93–99.

8. Brooks, J. A., Li, M., Baskes, M. I., andYang, N. C. Y. 1997. Roles of dendrite tip un-dercooling and solid state diffusion on weldmicrosegregation. Science and Tech of Weld.and Joining 2(4): 160–166.

9. Olson, D. L. 1985. Prediction ofaustenitic weld metal microstructure andproperties. Welding Journal 64(10): 281-s to295-s.

10. Kotecki, D., and Siewert, T. A. 1992.WRC constitution diagram for stainless-steelweld metals: a modification of the WRC-1988diagram. Welding Journal 71(5): 171-s to 179-s.

11. Cieslak, M. J., Hills, C. R., Hlava, P. F.,and David, S. A. 1990. An investigation of the

high temperature and solidification mi-crostructures of PH 13-8 Mo stainless steel.Metall. Trans. A, 21A: 2465–2475.

12. Massalski, T. B. 1970. Phase Transfor-mations, ASM International, Materials Park,Ohio, pp. 433–486.

13. Brooks, J. A., Baskes, M. I., andGreulich, F. A. 1991. Solidification modelingand solid state transformations in high energydensity welds. Metall. Trans. A 22A: 915–925.

14. Elmer, J. W., Allen, S. M., and Eager, T.W. 1989. Microstructural development duringsolidification of stainless steel alloys. Metall.Trans. A 20A: 2117–2131.

15. Summers, L. T., and Morris, Jr., J. W.1985. Improvements in the Weldability of aSuperconductor Sheath Material, LawrenceLivermore National Laboratory, UCRL-92346.

16. Brooks, J. A. 1991. The role of solidifi-cation and microsegregation on weld perfor-mance — austenitic vs. martensitic PH stainlesssteels.The Metal Science of Joining, eds. M.Ceislak, J. Perepezko, K. Kang and M. Glicks-man, pp. 107–113. TMS, Warrendale, Pa.

17. Hochanadel, P. W., Robino, C. V., Ed-wards, G. R., and Cieslak, M. J. 1994. Heattreatment of investment cast PH 13-8 Mostainless steel: Part 1, mechanical propertiesand microstructure. Metall. Trans. A 25:789–798.

18. Taillard, R., and Pineau, A. 1982.Mater. Sci. Eng. 54: 209.

19. Wojcieszynski, A. L. 1993. Particlespacing and grain size effects, Ph.D. thesis,Carnegie-Mellon University, Pittsburgh, Pa.

20. Hale, G. E., and Nutting, J. 1984. In-ternational Metals Reviews 29: 273–298.

21. Lundin, C. W., Lingenfelter, A. C.,Grotke, G. E., and Lessmann, G. G. 1982. TheVarestraint test. Welding Research CouncilBulletin, No. 280.

22. Thier, H. 1976. DVS-Ber. 41:100–104.23. Brooks, J. A., and Thompson, A. W.

1991. Microstructural development and solid-ification cracking susceptibility of austeniticstainless steel. International Materials Review36(1): 16–44.

24. Lippold, J. C. 1982. Weld crackingmechanisms in austenitic stainless steels.Trends in Welding Research in the UnitedStates, ASM Conference Proceedings, ed. S.David, Materials Park, Ohio, pp. 209–241.

25. Kujanpaa, V. P., David, S. A., andWhite, C. L. 1986. Formation of hot cracks inaustenitic stainless steel welds. Welding Jour-nal 65(8): 202-s to 212-s.

26. Brooks, J. A., Thompson, A. W., andWilliams, J. C. 1984. A fundamental study onthe beneficial effects of delta ferrite in reduc-ing weld cracking. Welding Journal 63: 71-s to83-s.

27. Lundin, C. D., Lee, C. H., and Menon,R. 1988. Hot ductility and weldability of freemachining austenitic stainless steel. WeldingJournal 67(6): 119-s to 130-s.

28. Kujanpaa, V., Suutala, N., Takalo, T.,and Moisio,T. 1979. Correlation between so-lidification cracking and microstructure inaustenitic and austenitic-ferritic stainless steelwelds. Welding Research International 9(2):55–75.

WELDING RESEARCH SUPPLEMENT | 291-s

RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT

/RE

SE

AR

CH

/DE

VE

LO

PM

EN

T/R

ES

EA

RC

H/D

EV

EL

OP

ME

NT