corrosion behavior of welded stainless steel

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

SUPPLEMENT TO THE WELDING JOURNAL, MAY 1996 Sponsored by the American Welding Society and the Welding Research Council

Corrosion Behavior of Welded Stainless Steel

Corrosion resistance of stainless steel weld metal depends on an understanding of the nature of the protective passive film

that gives the steel its "stainless" characteristic

BY T. G. G O O C H

ABSTRACT. The corrosion resistance of stainless steels depends on a protective, passive film being fornled at the steel surface on exposure to the service environment. The use of fusion welding for fabrication leads to local compositional variations within the material, which may significantly alter the stability of the passive layer and hence the corrosion behavior. The paper reviews published work and practical experience on such effects of a weld thermal cycle.

Compositional heterogeneity and lo- cally reduced passive film stability can stem from three main causes. Attention is given first to the consequences of alloy element segregation during weld metal solidification and to the formation of a fusion boundary unmixed zone. Element partitioning as a result of solid-state phase change is then considered, with particular reference to duplex ferritic/austenitic materials. Finally, the effects are described of precipitation, both of carbide (or carbonitride) particles and of intermetallic phases, with resultant development of alloy depleted regions. Throughout, recognition is made of the different grades of stainless steel employed by industry in terms of overall composition and microstructure.

1 G. GOOCH is with T~/Vh Abin~4ton Hall, Abington, Cambrid,~e, U.K. Paper presented as the Comfort A. Adams Lecture at the AVVS 76th Annual Meetin,G Apri l 3-7, 1995, Cleve- land, Ohio.

Study remains necessary to ensure that corrosion resistance of welded joints is adequate for service in a range of media. However, controlling factors have in large part been defined for both established and recently developed steels, and quantitative recommendations on preferred welding procedures can generally be given.

Introduction

There can be no doubt that the development of stainless steel represents one of the major technical achievements of the 20th century. The demonstration and understanding that the addition of chromium to steels endowed thenl with the property of passivity led to the development of materials with remarkable

KEY WORDS

Stainless Steel Austenitic Martensitic Ferritic Duplex Welding Corrosion

corrosion resistance, coupled with ex- cellent mechanical properties (Ref. 1). Hence, "stainless steels" have reached a dominant position for a very wide range of applications. They give resistance to aggressive media, provide an inert surface such that a product stream is not contaminated, and can offer high strength and toughness for service at ex- tremes of temperature. The alloys are employed in chemical all([ process plants, for power generation and other energy industry equipment, for the food industry, and for medical duties.

The fact that passivity can be induced in an iron-based system is extrenlely for- tunate, since, by appropriate adjustment of elements other than chronlium, ferritic, martensitic or austenitic structures can be produced singly or in any combi- nation. The result is that alloys can be tai- lored to produce specific properties, depending on practical requirements and costs of alternative materials.

Welding, especially by a fusion process, plays an essential role in fabrication of stainless steel products because of the economy and flexibility afforded to design and manufacturing. A continuous item is produced, facilitating direct trans- mission of stresses through the stru(ture, and avoiding through-wall perforations and potential leakage resultant on other methods of joining. However, welding necessarily involves a thermal cycle with localized heating and cooling, and expansion and contraction. Regardless of

WELDING RESEARCH SUPPLEMENT I 135-s

'ZII lie

I

E I

W I ]

W I E l

i

I

L I 0 1

a l

iilUiii!iI ~iEi!i!ill

: I

1 1 |

: )

|

w I

:):?: , /t0: I J;:l

: U,i; I

I

0.50

~. -0 .25

! 1 I I

m

30~ ~0 MO

I I I I 10 10 z 10 ~ 10 ~ 105

Currenf densify, p.A/cm z

Fig. I - - Potentiostatic polar izat ion curves for 12% Cr (410), 17% Cr (430) and 18% Cr (304) stainless steels in 1N H2SO 4 + 1N NaCI at room temperature (Ref. 2).

-0.1 ~ Z Effect of chromium: Cr.8%Ni steel- I

-Ok

I Effect of nickeh 18%Cr-Ni steel

-~ - o . 2 1 ~

-OI,

-0 . I Effect of molybdenum: 18%Cr-lO%Ni steel I

- 0 3 o Mo

- 0 4 ' , 0 1 2 3

Currenf densih/,mA/cm z

Fig. 2 - - Effects o f varying Cr, N i and M e on active range o f stainless steels in 10% H2SO 4 at 20°C (Ref. 3).

1.0

0.8

~s 0,6

~ o,4

02

, , ' r = j . . . ~ ~ - ~ / ~

÷lOOmVsc e • •

O, I I I I 0 20 40 50 80 100

Steel Cr content, af %

Fig. 3 - - Relationship between Cr content and Fe-Cr base metal. A - - Cr cation fraction in surface f i lm; B - - atomic fraction in under ly ing material: I M H2SO 4.

the grade of stainless steel, and whether it be wrought or cast, the base material will be supplied following careful processing to achieve optimum properties. The additional welding cycle will differ appreciably from that imposed during material manufac- ture, and can markedly change the properties around the joint. Specif- ically, this is the case with corrosion resistance, and the present paper is intended to review the effects of welding as a method of fabrication on the service behavior of stainless steels in aggressive media.

It is generally considered that the development of passivity in iron-chromium alloys occurs virtually as a step change as chromium level is increased: below some 10% Cr, the material remains active in normal aqueous media, while above 12% Cr, a passive film develops and corrosion rate is greatly decreased (Ref. 1 ). While this is essentially true, the influence of the environment must be clearly recognized in terms of redox potential and content of specific anions such as chlorides (Ref. 2). Under aggressive conditions, a chromium level well in excess of 12% will be needed to maintain a stable passive film - - Fig. 1. There are, therefore, varying degrees of "stainlessness" (Fig. 2) (Ref. 3), and a major effect of welding is that the associated metallurgical changes may lead to some parts of the material having lower levels of chromium (or other elements which enhance corrosion resistance, primarily molybdenum and nitrogen) than elsewhere on the metal surface.

This situation can stem from segregation of alloy elements during solidification, from element partitioning during phase changes, or from precipitation of second phase particles. These aspects are considered below.

Passivity of Stainless Steels

Passive Film Formation

The nature of the passive film formed on stainless steels has been researched for many years (Refs. 1,4, 5). Although commonly, and conveniently, regarded as an oxide layer, it is in fact considerably more complex. Simple application of a Pourbaix diagram, for example, would not predict the observed fact of passivation in acid media on the basis of a thermodynamically stable oxide (or hydrox- ide) layer. The constitution of the passive film has not been precisely defined, but the development of x-ray photoelectron spectroscopy (XPS) has enabled study to be carried out on stainless steels surfaces exposed to various environmental conditions. The approach is in principle more direct than interpretation of electrochemical data or examination of film stripped by aggressive etchants, although the possibility of some change between removal from the test environment and XPS analysis must be recognized.

Accepting this limitation, Asami, e ta l .

(Ref. 4), working with a range of Fe-Cr alloys, found that the surface film formed after exposure to 1M H2SO 4 in the passive range became chromium-rich immediately after the substrate chromium content exceeded 12%, the minimum level commonly associated with passivity and hence "stainless" steels-- Fig. 3. Despite the high chromium content in the film, the composition of the metal surface corresponded to that of the bulk alloy, indicating that the film resulted from preferential dissolution of iron.

Chromium enrichment in the surface film has been shown (Ref. 5) for a range of potential, and in both chloride-free and chloride-containing media, but only under passive conditions. Below the passive/active transition, the surface film an- alyzed reflected the Cr and Fe contents of the base metal. Hashimoto, e ta l . (Ref. 5), concluded that the passive film was essentially a hydrated chromium oxy-hy- droxide (CrOx(OH)3_2x-nH20), with a small amount of hexavalent molybdenum if the substrate steel contained this element. Molybdenum was found to be beneficial primarily by suppressing active sites via formation of an oxy-hydrox- ide or molybdate, and this was further indicated by Halada, et al. (Ref. 6).

1 3 6 - s I M A Y 1 9 9 6

200 1 [ I I , ,

0

A lime sec

b . t 2 " i~.;l'z " , ,~,,

I r

?°

. ,'o '

Fig. 4 - - Current transients on austenitic stainless steel in chlor ide media (Ref. 11). A - - Metastable pitt ing; B - - stable pitting.

The beneficial effect of nitrogen has also been explored (Refs. 6, 7). Available evidence has indicated that surface enrichment of nitrogen takes place at breaks in the passive film, thereby blocking active dissolution. This view is consistent with the observed synergistic effect between molybdenum and nitrogen (Ref. 6). It has also been suggested that nitrogen encourages iron dissolution, thus en- hancing repassivation by the resultant chromium enrichment (Ref. 8), and the two mechanisms may be complementary.

Passive Film Breakdown

Stainless steels display passivity only over a certain range of potential (Ref. 1). At low potentials, the passive film cannot form and the material corrodes in the active state (Fig. 1), whereas under high potential conditions (in chloride-free media), the passive film is oxidized to form solu-

Fig. 5 - - Segregation and pitt ing in ful ly austenitic weld metal (125X).

ble hexavalent chromium ions. Neverthe- less, the passive region is sufficiently extensive for the materials to be employed in a wide variety of service media.

However, stainless steels are sensitive to local passive film breakdown and attack, most notably by chloride ions (Ref. 1 ). This limitation is well recognized, and has led to significant developments in material composition to achieve resistance to chloride pitting (and crevice attack), exemplified by the increased Mo and N levels in high-alloy "super-

austenitic" and "superduplex" grades marketed over the last decade or so. Pit- ting occurs at potentials corresponding to general passivity, becoming more pronounced at higher chloride levels and applied potential. The formation of a pit is preceded by electrochemical noise, and, from measurement of small corrosion current transients (typically less than lnA to over 101aA), the concept has emerged of the passive layer being frequently penetrated by chlorides (or other sufficiently mobile anions), but with re-

i|O l l I I w ! I

ou , [.

Fig. 6 - - Crit ical pit t ing temperature data from ferric chlor ide tests on stainless steel base metal and weld metals (Ref. 17).


Table 1 - - EDX Analyses (wt-%) on GTA Root Passes in $31254 Pipe Butt Joint Welds

Dendrite inter- Filler Element Bulk Core dendritic

Autogenous Mo 6.1 4.2 6.9 Cr 20.2 19.3 20.5 Ni 17.6 17.6 19.2

ERNiCrMo-3 Mo 7.6 5.8 10.5 Cr 21.4 20.7 20.9 Ni 47.2 47.7 43.6

ERNiCrMo-4 Mo 11.3 9.8 14.4 Cr 17.3 17.2 18.1 Ni 40.7 43.1 39.2

pair rapidly taking place after dissolution of a small volume of metal (Refs. 9-11). As chloride activity or applied potential is increased, the volume of metal dissolved is sufficient that local hydrolysis and acidification prevent repassivation, and a stable pit is formed which wi l l then propagate - - Fig. 4. The critical volume of metal constituting the pit nucleus depends on the environmental conditions, but has been estimated as a hemisphere from about 10 nm to 11Jm diameter (Ref. 9). In this model, the initial passive fi lm rupture is considered to take place at

Fig. 7 - - Depen- dence of stainless

steel weld metal passivation on in-

terclendritic Mo segregation: 30%

H2SO 4 with 0.1 g/L NH4CNS at 25°C (Ref. 15).

.! ~. 0

0.5

I I I

/ I I !

1.0 1 .5 2 0

Molybdenum segmc~hon ratio 2.5

Fig. 8 - - Relation- ship between element segregation,

partitioning and bulk composition and solidification

mode (Ref. 15).

¢3

I I I I I

.~l id i fication mode P f f m a r y [ Mixed P r i m a r y ous fen i fe

~ ~ I 2.S " I

b - - - - - o ~ ~, 8 o - o " - - o - - - o 1.0 1- ii

0 l , I , ~ ' , . ~ 1 I I , ~ I 0,4 0.5 0.6 0.7 0.8 0,9 1.0 1.1 1.6

' w / P # c r ~

weak points. Mattin associated rupture events with inclusions in the steel surface (Ref. 9), but in principle any local alloy element depletion would be expected to constitute a weak point and thus a pit initiation site.

Sol id i f icat ion

Alloy Element Segregation

A weld metal effectively constitutes a chil l casting and is subject to the normal segregation of alloy elements during so- l idif ication of the molten weld pool (Fig. 5) (Refs. 12, 13) in the first instance, the degree of segregation wi l l depend upon the equi l ibr ium partition coefficient of individual elements between the solid and liquid states, and this can vary considerably depending on the element and the primary sol id i f icat ion phase. Chromium and molybdenum are re- jected from the solid to a greater extent with primary austenite solidification than in weld metals freezing to ferrite (Ref. 14), whilst the reverse is true for nickel. Nitrogen displays a somewhat unusual behavior in that, thermodynamically, it is more soluble in solid austenite than in l iquid steel, and thus tends to segregate in the reverse sense to other elements (Ref. 1 5).

Because of segregation, the corrosion resistance of weld metal is inferior to that of base metal of identical bulk composition (Refs. 15-17), In effect, the behavior of the weld deposit is determined by the minimum content of "passivating" elements at a point w i th in the sol idi f ied weld metal, generally the dendrite centers since these are the first to solidify. The consequences of segregation have long been recognized, and consumable specif ications for common grades of stainless steel are generally slightly overalloyed primarily in Cr and Mo, so that even depleted regions retain sufficient levels of these elements for satisfactory corrosion resistance to be obtained.

Segregation is not commonly seen as a practical problem with ferritic alloys. The liquid/solid partition coefficients for chromium and molybdenum approach unity under solidification to ferrite, whi le high-temperature diffusion in the bcc ferrite lattice is very rapid, so that some homogenization takes place in the solid state during the weld cooling cycle. Ni- trogen might segregate strongly, but this element is usually held at a very low level in ferritic stainless steels. In contrast, segregation becomes much more significant in high-alloy superaustenitic grades. The materials typical ly contain 20-25%Cr, 20-25%Ni, 6%Mo and 0.2%N, and seg-

138 -s l MAY 1996

regation of molybdenum can be sufficiently marked that some regions of the weld metal contain only about 4% (Ref. 15), leading to an appreciable loss in corrosion resistance - - Table 1. This has most frequently been highlighted with reference to chloride pitting attack, and indeed any reduction in passive film stabil ity would be expected to have a pronounced effect on pitting behavior; given passive film breakdown at alloy-depleted regions and stable pit initiation, local hydrolysis and acidif ication wi l l render subsequent pit growth autocatalytic - - Fig. 5. Over the last decade, it has become common practice to describe the pitting resistance of stainless steels by an empirical compositional parameter (Ref. 17), variously termed a "pitting index" or "pitt ing resistance equivalent" (PRE), such as (Ref. 18)

PRE = %Cr + 3.3%Cr + 16%N (1)

A direct equivalence in pitting behavior can be drawn between the minimum alloy level in the weld metal solidification structure and a base metal of similar "lean" bulk composition - - Fig. 6. At the same time, the adverse influence of segregation on passive f i lm stabil i ty has been demonstrated also in other media, notably sulphuric acid (Ref. 15) in which general rather than pitting corrosion wi l l occur Fig.7.

Various factors for nitrogen in Equa- tion 1 have been proposed. While Suu- tala and Kurkela used 13 (Ref. 17), 16 is most commonly employed (Ref. 18), although a higher value of 30 has been rec- ommended for high-alloy superduplex and superaustenitic steels (Refs. 18, 19). Recognizing that the PRE concept is empirical, Walker and Gooch showed that the "optimum" compositional relationship, in fact, depends on the environmental conditions causing attack (Ref. 20).

Effect of C o m p o s i t i o n

The degree of segregation is influenced by the total composition of the material, and tends to be more pronounced in iron-based than in nickel-based systems (Refs. 15, 21 ). As illustrated in Fig. 8, segregation is likely to be most marked in high-alloy austenitic steels containing around 20-25%Ni, decreasing at higher and lower Ni levels. Figure 9 shows the correlation between weld metal pitting resistance and a PRE parameter incorpo- rating an empirical factor to recognize the overall composition (Ref. 15), derived from studies of stainless steel solidification. From Figs. 8 and 9, it appears

800

600

400

200

0 0

I

n

I I I I I I 'j" I / I - I I J I

20 4O 50

Cr+ 3,S Mo- 14N÷ (1.T Mo, 23N) - -

I 8O

L.j 0

60

40

20

D IQ! i N # H # O ~

. 4r/3N z

Ar

I I I 0.5 20 1.5

Arc ene~jy, k J / ~

o,24

o z2

0.20

.•E 0"18

0.16

0.11,

0.12

0.~

I i J I t .

I I I I I 2 4 6 8 10 Nitrogen in A r sh ie ld ,%

Fig. 9 - - Relation- ship between weld metal pit t ing resistance and modi f ied pit t ing index (MPI) (Re£ 15). Potentio- dynamic passive flint in 3% NaCI at 50°C.

Fig. 10 - - Effect o f arc energy and shielding gas on FeCI 3 pit t ing resistance o f GTA welds in $31254 steel (Ref. 24).

Fig. 11 - - Effect o f nitrogen level in Ar shielding gas on 22% Cr duplex weld metal nitrogen content (Ref. 25).


Fig. 12 - - Variat ion o f compos i t i on across an interdendritic volume

element al lowing com- plete mixing in the

l iquid (Re£ 13).

z

c. I C O

!

Distance ---~.

that segregation and the effect on passive film stability is greatest in weld metals of composition such that the primary solidification phase is austenite, but in which the Cr and Mo rejection is sufficient for the interdendrit ic regions to approach final solidification to ferrite.

Experience wi th plant made from 316L-type steels and producing urea or some organic acids, such as pteraph- thalic acid, has indicated that preferential corrosion can occur in weld metals conta in ing a small amount of ferrite (Refs. 22, 23). Consequently, it is normal to specify low or zero ferrite contents, with attendant sensitivity to solidification cracking and microfissuring. Almost certainly, these media result in a metal/environment potential close to the active/passive transition, depending on the oxygen level of the process stream. Passive film

!

f~ • ~ 20

b~ 10

0 I I I I i i i i 0 0.2 0.4 0.6 08

A NOrl~lised d~-h~nce, r/R

&

Ni

I I I I I I I I

f=O.18 sec ~

,1 tO

EE

m

~3 C O

kco

D m ~ n ~ ~

Fig. 13 - - Calculated interdendritic microsegregation patterns for GTA welds in stainless steel (Ref. 28). A 21% Cr, 14% Ni, 65% Fe alloy, solidifying to primary austenite; B - - 23% Cr, 12% Ni, 65% Fe alloy, solidifying to primary ferrite.

1.0

0.9

0.8

0.7

O6

0.5

04

0310.~

, , /I-~......--~ , , / cont,

- - . ~ A z i z

m ~ W o o d I , | .........

10 "1 I 10 102 Dimensionless velocify

Fig. 14 - - Calculated variation o f partition coefficient with velocity using various models (Rei~ 13).

80

70

6O ~Z

50

40

I I I

Base O - -

~Itzl

,/.

• QO /" I0 100 I000 10000

Laser veloc #y , mm /min

Fig. 15 - - Effect o f laser travel speed on ferric chloride critical pitt ing temperature o f surface melted UNS $31254 steel (Ref. 31).

140-s M A Y 1 9 9 6

stability is therefore marginal, and it is probable that the observed corrosion, which tends to be initiated at ferrite- austenite interfaces, is due not to the presence of ferrite per se, but rather to the associated segregation and alloy depletion. Interfacial effects, such as residual element segregation, may also be con- tributory. Comparable restriction on weld metal ferrite content has seldom been found necessary in other environments, and is apparently a consequence of the specific cathodic corrosion reactions in the organic media concerned. It seems that improved behavior could equally well be obtained with weld metals of composition giving either fully austenitic or primary ferritic solidification. In the latter regard, satisfactory service has been reported to TWl of materials such as 309LMo or 22% Cr duplex steel which solidify initially to ferrite, although these are also somewhat more highly alloyed than the 316L base metals generally used.

A further point regarding weld metal corrosion behavior is that the possibility must be recognized of compositional changes during welding. With flux processes, factors such as element transfer efficiencies, alloying addition particle sizing, etc., are well understood by consumable manufacturers, and the required deposit composition can normally be obtained for a wide range of welding conditions. However, recent years have seen increased use of nitrogen as an alloying addition to enhance corrosion resistance, and, when GTA welding high nitrogen steels with a conventional inert gas shield, this element tends to be lost from the molten pool because of the relatively low partial pressure in the surrounding at- mosphere. In consequence, shielding gases are finding favor that contain a few percent nitrogen to counter the nitrogen loss, and even to promote pickup by the weld metal and improve corrosion behavior (Fig. 10) (Refs. 24, 25). The per- missible nitrogen content in the shielding gas depends on the solubility in the molten pool, and hence on the composition, especially Cr and Mn levels. Exces- sive shielding gas nitrogen levels may lead to porosity, and to evolution during solidification, causing "spitting" and degradation of the tungsten electrode. At least for duplex ferritic-austenitic grades, up to some 2% N 2 in the gas shield is normally acceptable for a 22%Cr steel, and possibly 3% for a 25%Cr alloy - - Fig. 11. Higher nitrogen levels may be tolerable with the plasma arc process, used in key- hole mode (Ref. 26).

E f f e c t o f W e l d i n g C o n d i t i o n s

The residual segregation in a weld

q

I I I

I / I

I 50 !

I " !

50 ~l Ilr~ / 6 0 ........................................................ 4 5

4 # i I I

30 40 50 60 Pl = %Er÷ 93°/ot.fo+16°/oN

Fig. 16 - - Relationship between FeCI 3 pitt ing resistance of welded joints in UNS $31254 steel and composition (Ref. 32). A - - Weld metal bulk composition: O@ = GTA weld metal; O~D = SMA weld metal. Hol low symbols = 15% Mo filler metal; solid symbols = autogenous or 9% Mo filler metal; numbers = % dilut ion; B - - min imum alloying level in solidification structure: A = base metal; 17 = autogenous weld metal; Ci = GTA unmixed zone; l l = SMA unmixed zone.

• 7T - ) . % " "~'~ "'" ~.~x;- - - . • <~ ' > >~>.," .,/., - J f ' L ~ - : ~ . ' . > ' /7). _ . . . . . . . - , ~ _ . • ,

V >r-';.-<<-.<,-d-.:~-,<I~J,4tD>.Q t \ ' ~ , .~ :~ ' ~ t ,.1~I~!

, ~ ~ -z :<.',.,~., . . - J ~.. 'W'"R~EVZ:z:

' , ~ , - t . . . . / ~ O 0 1 x m . < t ~- ' "..' . . . . ,

"\ L

+t

Fig. 1 7 - Unmixed zone at weld interface o f welds in 531254 steel: N i -20Cr -9Mo fil ler metal. A - - SMA, 50)0

(_b) ~ . ~ B - - center o f A, 320X;

c - - GTA root toe, 160X.


Fig. 18 - - Pitting attack initi- tared at U M Z in GTA weld in

$31254 steel: ERNiCrMo-3 fi l ler metal, 25X.

metal at room temperature will depend on element rejection during solidification, on the extent of intermixing in the liquid phase ahead of the liquid/solid interface, and on the degree of diffusion and homogenization in the solid phase during cooling and any reheating by further weld runs. The last few years have seen considerable advances in understanding the dependence of the solidification se- quence in stainless steels on the bulk composition (Ref. 27). This has been associated with evaluation of element segregation behavior, as reviewed by David

and Vitek (Ref. 13). In large part, the Scheil hypothesis has been followed, viz.

C s = k C o (1 - f s ) k-1 (2)

where C s = composition of solid; C o = initial alloy composition; fs = volume fraction of solid; k = partition coefficient.

This assumes that pool stirring by con- vection and electromagnetic and surface tension forces is sufficient to give com- plete mixing in the liquid phase - - Fig: 12. The approach has been modified to

Fig. 19 - - EDX scans across weld

interface o f SMA weld in UNS

531254 steel using ENiCrMo-3 elec-

trodes (Ref. 32). Oisfonce, ~m

Fig. 20 - - H A Z microstructures in duplex ferritic-austenitic stainless steels, 40X. A - - Predominant ly ferritic; B - - austenite reformation.

allow for solid state diffusion after solidification, and found (Ref. 28) to give good qualitative prediction of final segregation gradients in stainless steels - - Fig. 13. A problem arises, however, in that no direct allowance is made for a time factor during the solidification process, with the resultant implication that the degree of segregation is unaffected by welding procedure. From studies on castings, segregation is known to depend upon solidification rate because this influences element redistribution apart from diffusion and intermixing in liquid and solid phases. Most alloy weld metals in which the effects of segregation are greatest solidify primarily by dendritic growth, and various models for solute redistribution during such solidification have been derived (Ref. 13). These have been based mainly on the assumption of segregation under equilibrium conditions, and this is unlikely to be the case over the range of solidification rates normally encom- passed by fusion welding.

Accordingly, models have been developed in which the segregation coefficient varies between the equilibrium value and unity as growth rate is increased from low to high velocities (Fig.14) (Refs. 13, 29). On this basis, element segregation will be reduced and weld metal corrosion resistance improved as solidification rate is increased. This has been experimentally demonstrated by Nakao and Nishimoto (Ref. 30) and by Woollin (Ref. 31), for the laser process - - Fig. 15. With laser welding, the solidification rate will be high, although rather lower than the travel speed since solidification will occur epitaxially from the base metal, with preferential growth parallel to the maximum temperature gradient, and perpendicular to the travel direction. For travel speeds associated with more common arc welding procedures, growth rate will be relatively low, and will vary around the periphery of the weld pool, so that a lesser effect of solidification rate might be anticipated.

Nonetheless, im- provement in corrosion resistance has been found at low arc energy (Ref. 24), with the rider that, unfortunately, the necessary arc energy limits to achieve practical benefit for high-alloy austenitic deposits may be so low as to be unacceptable from the productivity view- po in t - - Fig. 10.

142-s I MAY 1996

Unmixed Zone

Because of the problem of segregation in high-alloy austenitic steels, it is normal industrial practice to weld these materials using nonmatching nickel-based filler metals such as ENiCrMo-3 (Ref. 19). These were selected originally simply by virtue of their high molybdenum content, and although segregation still occurred, even the dendrite cores had an alloy content and corrosion resistance equivalent to that of the base metal (Table 1 ). The approach is now almost invariably employed for fabrication. Seam welded tube may be welded autogenously (at, it may be noted, high speed), but this product form will be given a high-temperature homogenization treatment by the tube- maker prior to delivery. Use of nonmatching consumables requires close attention to welding procedure to obtain adequate filler metal addition (Ref. 32). Dilution by base metal can be tolerated, up to perhaps 60% (illustrated by the hatched area in Fig. 16), but this still means that a root opening is preferable, to ensure that filler metal addition takes place during welding.

The application of "overalloyed" consumables has enabled satisfactory service to be obtained from high-alloy austenitic types in a range of media, but does not give weldment corrosion resistance completely matching the base metal. It has been known for many years that, during arc welding, a stagnant zone can exist at the weld interface (Fig. 17), constituted by melted base metal that has not mixed into the bulk weld metal (Ref. 33, 34). Following solidification, the region displays segregation analogous to that in an autogenous weld metal and is thus sensitive to localized attack in ap- propriately aggressive media (Refs. 32, 34, 35). This effect was identified originally in welds of ordinary 300 series stainless steels (Ref. 36), causing a minor increase in the active/passive transition potential, but it becomes much .~.'-..,~ more significant ~ , ~ ~ with high-alloy grades to the extent that it is lim- ~ ' - ~ i iting on corrosion resistance of joints made with overalloyed consumables. With ~ ~ increasing environmental corro- sivity, for example higher (a) exposure temperatures in an ASTM

~o, I o

G48 ferric chloride pitting test, attack generally takes place first on the unmixed zone (Fig. 18), resulting in a weldment critical pitting temperature (CPT) some 15-20°C (27°-36°F) below that of the base metal. In fact, it is the presence of the unmixed zone (UMZ) that renders fairly high dilution levels acceptable in joints made with nickel-based filler metals (Fig. 16), and further means that there is no practical advantage in employing even more highly alloyed filler metal types for improved weldment corrosion behavior (Ref. 32)

Available corrosion test data indicate that the UMZ is not as detrimental as conventional autogenous weld metal. This is indicated by the dotted line in Fig.16 (Ref. 32). Solidification in the UMZ may be by cellular as opposed to dendritic growth, but microanalyses have not shown any marked difference in minimum Cr and Mo levels from segregation in the UMZ (Fig. 19) and those in bulk autogenous weld metal• The reason for the disparate corrosion behavior has not been established, but it may be that nitrogen loss from the UMZ is less than from an autogenous weld bead. Cer- tainly, pitting resistance of autogenous GTA weld metal can be increased by use of an Ar/5% N 2 shielding gas to promote

Fig. 21 - - Calcu- lated PRE values of ferrite and austenite in alloys with 25%Cr and 4%Mo. Ni was varied to keep a constant ferrite content. PRE = Cr + 3.3%Mo + 16%N (Ref. 18)•

net nitrogen pickup in the bulk weld metal (Fig. 10), and weldment pit initiation then takes place preferentially at the UMZ (Ref. 24).

The unmixed zone may be essentially stagnant when molten, or may display re- stricted laminar flow (Ref. 33). In either case, its thickness is not uniform around the weld. This is partly because fluid flow in the molten pool can lead to swirls of base metal becoming detached but not fully mixed into the bulk weld deposit, but in addition there will be an effect of the drag induced by the surface adjacent to the molten metal. This drag is less at the pool/gas interface than at the pool/base metal junction, and in general the stagnant zone tends to be thinner to- ward the surface of the material as opposed to the mid-depth of the weld pool (Fig. 17A)(Ref. 32).

This may not be the case actually at the toe of a root run, where a more pronounced UMZ has been noted for GTA than SMA welds, presumably reflecting a local stagnant region - - Fig. 17C. Un- derstanding of weld pool flow and the relative effects of buoyancy, electromagnetic and surface tension forces has increased substantially (Reg. 37, 38), but modeling remains a complex problem

' ~ ,?

a ~ ' ~ ~ ~ ' 4 ' ~ 0Y875

Fig. 22 Preterential weM metal attack in FeCI 3. A - - Austenite, 70X; B - - ferrite, 250X.


, /~ • ~ • ,; ?- I s

.

Fig. 23 - - Secondary austenite in superduplex weld metal, 1000X.

[ ~~. 10 ' z ' I

.

Ferri~ content, %

Fig. 24 - - Effect of ferrite-austenite balance on pitting resistance of 22%Cr- 0.12%N GTA weld metal (Ref. 49).

especially when filler metal additions are made. In this respect, there will be some disruption with GTA welding, as filler metal is added, but the resultant stirring will normally be low compared to gas metal arc processes, whether metal transfer takes place by short circuiting or under droplet spray conditions. The situation thus exists that, although GTA welding may be preferred for control of bead shape in root runs, this is the process in which the unmixed zone is likely to be most marked. In principle, the width of the UMZ may be reduced via welding conditions giving steep thermal gradients (Refs. 33, 39). There may also be benefit from use of electromagnetic stirring, either from external induction coils, or perhaps by high current, high travel speed conditions to increase the Lorentz force in the pool. Further work is

80

6O

P

20

I I I

-

J I I l l l f

. , d r l I 1 % 1 . ~ r l i i I J F _/

~ l l l ~

i ' ~ I// q -.- /

f ---/

0 I I I 30 34 38 42

PI= %Cr+~3(% Mo+OSx% W~ 16x% N

Fig. 25 - - Relationship between duplex steel and weld metal composition and critical pitt ing temperature in EeCI 3 (Ref 46).

necessary, and this should recognize also the possible influence of relative base steel and weld metal melting points.

Solid State Transformation

Given the possible coexistence of ferrite, martensite and austenite, consideration must be given to the effects of elemental partitioning between the phases. Since martensite formation is essentially diffusionless, concern attaches primarily to the ferrite and austenite phases, and particularly to recently produced ferritic/ austenitic grades.

These alloys were developed primarily to avoid the problem of chloride stress corrosion cracking associated with fully austenitic steels. Originally, the composition was designed to achieve a two- phase structure that could be accommo-

dated by minimum changes to existing rolling schedules to produce a wrought product. This approach did not adequately allow for subsequent transformation during a welding cycle. The annealed base metal might have a 50/50

. phase balance, but weld metals displayed primary ferritic solidification, while the high-temperature heat- affected zone (HAZ) transformed to ferrite close to the solidus. Under the

I fairly rapid cooling associ- /'6 ated with welding, trans-

formation to austenite was suppressed, leading to a predominantly ferritic weld area structure (Fig. 20) with

reduction in corrosion resistance and mechanical properties (Ref. 40). Hence, it is now normal for welding consumables to be slightly overalloyed in nickel to promote austenite reformation, while nitrogen levels have been increased in base and weld metals to promote transformation to austenite and obtain a more equal ferrite/austenite balance, apart from a beneficial effect on corrosion resistance (Refs. 8, 41 ).

The two phases in duplex steels can differ appreciably in composition, Cr and Mo partitioning to ferrite and nitrogen and nickel to austenite. Accordingly, a thermodynamic approach to alloy design has been pursued by Berhardsson, so that, by appropriate control ofCr, Mo and N levels, the compositions of both phases following conventional annealing are likely to lead to similar corrosion resistance (Fig. 21) (Ref. 15). This does not, however, avoid problems at welded joints. Equilibrium partitioning of elements between the phases during welding is most unlikely to be achieved, and, even under the range of welding conditions appropriate to normal industrial practice, the compositions of ferrite and austenite can differ markedly.

The ferrite-austenite transformation during welding has been modeled by Hertzman, et al. (Ref. 42), using a thermodynamic approach based on the Ther- mocalc system, with diffusional growth. Viewed more simply, the mean diffusion distance, x, of any element from random walk theory is roughly (Ref. 43)

x = 2~/Dt (3)

where D is the diffusion coefficient and t is the time. Integrating the diffusion dis-

144-s MAY 1996

J . , ,q

"4/

lOOp, m { • 0 T 2 5 8 2 I

Fig. 26 - - Intergranu/ar po/yth ionic acid cracking in sensitized 304 stainless steel, 125X.

tance (Ref. 43) over a cooling cycle, for say a 1 kJ/mm (25 kJ/in.) GTA weld run in 10-mm plate (0.4 in.), substitutional elements Cr and Ni would diffuse in ferrite (Ref. 44) during cooling some 0.5 and 2 pm, respectively. As an interstitial element, nitrogen would diffuse roughly 50-100 pm. The resultant austenite laths are perhaps 4 lam thick, and, in this example, austenite formation could be due to substitutional element diffusion. Cer- tainly, at higher arc energy and slower cooling, appreciable redistribution of Cr and Ni between the ferrite and austenite would be expected. Conversely, with lower heat input, austenite reformation would be controlled by nitrogen move- ment, as was assumed by Hertzman (Ref. 42). Such differences in ferrite and austenite composition have been observed over the range of cooling rate appropriate to normal welding practice (Refs. 45, 46), and are illustrated in Table 2. Consequently, the relative corrosion resistance of the two phases can vary considerably, depending on the specific steel and welding conditions entailed - - Fig. 22.

The situation is further complicated in that a single weld thermal cycle is most unlikely to result in the equilibrium austenite content being obtained, so that additional transformation will occur on deposition of subsequent weld runs. [n part, this will involve growth of existing transformed units, but it is possible also for secondary austenite to be nucleated and grow within the prior ferrite grains on reheating (Fig. 23) (Ref. 47). The nitrogen content of the ferrite matrix is fairly low and, following diffusion of the nitrogen

/200

1000

6OO

400 0

I I I I I

- - - - - - 17%Cr: 0.05%C - . , - ~ 18%Cr: 8%Ni: 0 .05%C . - . - - . 18%Cr: 9%Ni: 0.06%C: 0 .64%Nb . . . . . 18%Cr: 8%Ni: 0.02%C

0,01 0.1 1,0 10 100 1000 Ageing time, hrs

Fig. 2 7 - - Effect of steel composition on sensitization behavior. Strauss testing, sensitized within bounding lines (Ref. 54).

Table 2 - -EDX Phase Analysis (wt-%) on SMA Welds in $31803 Steel Made with Overalloyed Filler Metal under Varying Arc Energy Conditions

Element, wt-%

Weld region Phase Cr Mo Ni

As-deposited ferrite 23.9 3.0 7.5 root (a) austenite 23.7 2.8 7.8

Reheated ferrite 23.7 3.0 7.2 root TM austenite 23.4 2.7 7.7

Reheated ferrite 25.1 3.7 6.2 root (b) austenite 22.7 2.5 8.6

Base ferrite 23.2 3.3 4.1 metal austenite 19.5 2.4 7.0

(a) Root and ill[ passes at 0.7 kJ/mm. (b) Root at 0.5 kJ/mm, fill passes at 3.2 kJ/mm.

and substitution elements, the final PRE of the secondary austenite will be below the surrounding material (Table 3), leading to appreciable reduction in corrosion resistance (Ref. 48).

From the metallurgical viewpoint, the prime objective in formulating a welding procedure for duplex steels should be the attainment of a satisfactory phase balance in the weld locale. It is generally accepted that austenite contents between about 30 and 70% are preferred for optimum overall properties. In particular, high ferrite contents can lead to reduced resistance to pitting corrosion (Fig. 24) (Ref. 49), and to stress corrosion, whether in hot chloride media (Ref. 50) or under sour, H2S conditions (Ref. 51). Such an effect of high ferrite levels may be exac- erbated by intragranular precipitation of chromium nitrides, as a result of nitrogen supersaturation in the ferrite at peak temperatures during welding (Ref. 52).

However, even with control of material composition and weld cooling rate to obtain 50/50 phase balance, corrosion resistance will not necessarily equal the base metal - - Fig. 25. The adverse effects of nonequilibrium partitioning tend to be most marked in weld metal, and, since they cannot yet be predicted reliably, the

use of overalloyed filler metals has been propounded (Ref. 20). Commercial consumables for the established 22%Cr grades are commonly slightly enriched in Cr, Mo or N relative to base metal, while even more highly alloyed 25%Cr superduplex filler metals may be used for root runs exposed to the service environment. Experience has shown that analogous overalloying is problematic with superduplex base metal, because the weld metal may be unacceptably sensitive to precipitation of intermetallic phases, which reduce corrosion resistance and toughness (Ref. 53).

The last decade has seen considerable attention paid to the chloride pitting resistance of stainless steels, both

Table 3 - - EDX and WDX Analyses on 26Cr/9Ni/4Mo/0.26N Superduplex Weld Metal (Ref. 47) (PRE = Cr + 3.3Mo + 16N)

Element, wt-%

Region Cr Mo N PRE

Primary austenite 26.6 3.3 0.52 45.8 Ferrite 27.4 4.0 0.07 41.7 Secondary austenite 24.3 3.4 0.24 39.4


Table 4 - - PRE Relationships Derived from Polarization Tests on Duplex Weld Metals in CO2-Saturated 3%NaCI: (FeCI3 data in brackets) (ref. 20)

Pitting Criterion Specific PRE Equation

Pitting potential at 80°C Cr + 1.9Mo + 0.55N - 0.27Ni CPT at 700mVscE Cr + 2.9Mo + 21N - 0.25Ni (CPT in FeCI3 Cr + 2.1Mo + 40N - 1.2Ni)

austenitic and duplex grades, for marine and oil and gas service. The PRE concept of Equation 1 was developed originally from laboratory pitting tests, especially in ferric chloride. Clearly, from Tables 2 and 3, the PRE values of ferrite and austenite can differ relatively, and the net pitting resistance of a weld wil l depend on the minimum PRE in any one region. At the same time, studies on duplex steels have shown that caution is necessary in ap- plying the PRE approach directly to practice. As indicated in Table 4, the relationship between resistance to chloride pitting and material composition depends critically on the environmental criterion adopted to describe corrosion behavior (Ref. 20).

Precipi tat ion of Second Phases

Carbide Formation

Austenitic Steels

All grades of stainless steel are subject to precipitation of a range of second- phase particles at intermediate temperatures. The effect is particularly associated with austenitic alloys, in which "weld decay" constitutes the classic case of the corrosion resistance of a material being adversely affected by fusion welding (Ref. 1). Grain boundary precipitation of chromium-rich M23C 6 carbides takes place at temperatures between about

500°-900°C (932°-1652°F), as inevitably experienced in part of the heat- affected zone of a fusion weld. Because the chromium diffusion rate is appreciably less than that of carbon, chromium is removed from the surrounding matrix as the particles form, leaving a depleted area of lower passivity than the grain centers - - Fig. 26. Historically, the problem was overcome (Refs. 1,54) by addition of "stabilizing" elements, principally niobium or titanium to the material to form stable carbides and getter the carbon from the matrix, or by reduction in material carbon content via the use of low- carbon ferro-chrome in s t e e l m a k i n g - Fig. 27. With the introduction of AOD or VOD processing in steelmaking, carbon levels in commercial products have fallen dramatically, typically to 0.05% or below, so that the practical problem is greatly reduced, and weld decay is now rarely encountered in plant used in the as-welded condition (Ref. 55). This is the case even without going to extra-low- carbon grades (i.e., less than 0.03% carbon), although postweld heat treatment (PWHT) or service within the "sensitizing" temperature range can still induce susceptibility to corrosion.

Increasing time in the sensitizing temperature range leads to greater width of the depleted zone and lower local matrix chromium content. The situation is illustrated by the unidirectional analysis of Stawstr6m and Hillert (Ref. 56): assuming a thin continuous carbide film,

Fig. 28 - - Flow diagram for prediction of sensitization in austenitic

steels (Re£ 59)

I Kinetic parameters diffusivities

Temperature (t) Strain (t)

Input information Ef Alloy composition d Alloy condition

Initial DOS N Prior work

TM history TA history Strain history

I

Effective Cr diffusivity

Nucleation kinetics

Minimum Cr at interface

Thermodynamic parameters Activity coefficients

Composite Cr C solubility

Equilibrium constant Carbide concentration

Cr depletion width

I I Degree of

sensitisation

[ I IGSCC

susceptibility

~ - ( Cr~ - Cri I m= 2~/Dtl ~ ) (4)

where m is the depleted zone width, while Cr o, Cr i and Crp are the chromium levels in the bulk material, at the carbide interface, and necessary for passive film stability, respectively. Since this study, the development of depletion has been wel l modeled, both under isothermal conditions (Refs. 57, 58) and during a weld thermal cycle (Fig. 28) (Refs. 59, 60). Using a short-term electrochemical potentiokinetic reactivation (EPR) test, Bruemmer was able to obtain good correlation between the degree of sensitization and material variables, composition, heat treatment conditions, and thermo- mechanical history (Fig. 29) (Ref. 57). It has been assumed that Crp in Equation 4 is about 13% (Refs. 56, 57), i.e., the chromium content necessary to render iron alloys capable of forming a passive film. In fact, any reduction in chromium below the matrix content wil l expand the potential range over which the material is in the active state (Fig. 2) (Ref. 3), and wil l thus lead to some susceptibility to preferential intercrystall ine corrosion (Ref. 61 ). Whether or not an attack occurs then depends on the service redox potential relative to the grain boundary active range, i.e., the risk of weld decay depends on the specific environmental conditions, as well as on the steel composition and welding procedure. A small reduction in grain boundary chromium content wil l be significant only in media close to the active/passive transition, whereas extensive carbide formation can lead to intercrystalline attack over the en- tire potential range giving a stable passive fi lm on the grain centers (Fig. 30) (Ref. 61 ). Further to this thermodynamic effect on passive fi lm stability, the carbide particles also act to accelerate passive fi lm breakdown (Fig. 31 ), increasing the chance of attack under transient plant conditions associated, for example, with changes in oxygen activity.

Given that carbide precipitation is time dependent, as illustrated by Equa- tion 4, reduction in arc energy, and the time spent in the sensitizing temperature range, is advisable (Ref. 55). This is commonly recognized in welding procedures for austenitic steels which stipulate maximum levels of arc energy and interpass temperature. The effect of varying cooling time was modeled by Solomon (Ref. 60), who demonstrated the relationship between increased cooling time or carbon level and enhanced susceptibility to intercrystalline corrosion. The work further showed the environmental dependence of attack, and much more rapid cooling was necessary to avoid attack in the electrolytic oxalic acid test (ASTM

146-s I MAY 1996

A262A) than in acid CuSO 4 (ASTM A262E). Two caveats to this approach must be added. First, in an actual weld, concern attaches to the maximum sensitizing time experienced, and this wi l l occur at some distance from the fusion boundary where the peak temperature is high enough to enter the sensitizing range, but not so high that the region is in effect solution treated, i.e., sensitization takes place on heating as well as on cooling. Second, the conjoint strain from local expansion and contraction during welding can act to accelerate the sensitisation process, as illustrated by Atteridge and his coworkers (Fig. 32) (Ref. 59),

Although the general reduction in carbon content from AOD/VOD steelmaking has diminished the practical problem of weld decay, experience in pipework for light water reactor power generation plants has indicated that the consequences of precipitation and sensitization can be greatly increased by residual welding stresses, and intercrystalline cracking has been experienced in a number of units. A considerable amount of work has been carried out, therefore, to address residual stress patterns at welds. Techniques such as "heat sink" welding, with or without local induction heating, have been employed to induce compres- sive stresses on the inner wall of pipe, thereby avoiding cracking (Ref. 62). At the same time, the problem is especially associated with the environmental conditions encountered in some light water systems, and these measures would not normally be employed for the typical welded chemical plant, for example.

Again, in the context of light water reactor power plants, it is recognized that initiation of grain boundary carbide precipitation in the HAZ can take place very rapidly. Hence, although a controlled weld thermal cycle may not induce sufficient precipitation to cause sensitization directly, the nucleated carbides might grow in subsequent long-term service at lower temperatures, in this case roughly 300°C (572°F). Indeed, from evaluation of the activation energy as determined by slow strain rate tests, Povich and Rao demonstrated that some mea- sure of "low-temperature sensitization" of welded Type 304 pipe could be anticipated after, say, 10 years exposure to 300°C (Ref. 63).

Most environmental conditions in which stainless steels are used corre- spond to a moderately low redox potential. Under highly oxidizing conditions, exemplified by a nitric acid plant, the effect of chromium depletion can be very marked, while it is possible also for the carbides to be dissolved. As a result, these media are very searching and either

100

/6)

1.0

0.1

0.01

! !

m

=m|

001 01

•0

• 0 0

I

°

' , 3

,4

L~. " '1" ° •

° •

I I 1 10

l'fe.asured hme, h 100

0.50

~ O25

o

"-0.25

-0.50

! !

0.15%C / ./ / i 0.osO/oi

l ) l i , i l l i l ~,, i , i N m ~m, • im qim ° i m , ~ m

).Ol 0.1 1 Ageing hme,hrs

/0

0.6

0.4,

02

0

-02

"" -04

-0.6 -0.8

0

i ! HAZ of electron beam weld (0.16kJ/mm

" m e = HAZ of friction weld . . . . HAZ of S M A weld (1 kJ /mm) - - - - - - HAZ of G M A weld (0.6 kJ/mm) ~ - - - Solution treated material

, ,4##amnt Fl~e peter,~l

I - ~ I " ~ . A p p a r e n f F l a d e "

"" ,I ! • #

0.1 1 10 100 Time,rain

Fig. 29 - - Measured and predicted sensitization times for 304 and 316 steels. Isothermal heat treatment at 600°-700°C.

Fig. 30 - - Relation- ship between potential necessary Ior grain boundary passivation and minimum aging time between 500 ° and 850°C, 18%Cr- 10%Ni austenitic steel 20% H2SO 4 with 0. l g/L NH4CNS at room temperature, general passive/active transition for grain centers below-0.48 VSH E (Re£ 61).

Fi~¢. 31 - - Decay of HAZ passive film tbllowing anodic polarization at 0.4VsH E tbr 10 rain, 20% HjSO 4 with 0.1 g/L NH4CNS, 18%Cr- 10%Ni- 0.15%C austenitic steel.

WELDING RESEARCH SUPPLEMENT[ 147-s

k j

40

30

10

0

! I I I

Thermal only Pass by pass strain

0 0.06

• Cumulat ive strain • ,/" . .~pv .... i 1 ~ . , , , , . ~ 1 0 I I

0.02 0.03 0.04 0.05 Carbon contenf , vf %

Fig. 32 - - Predicted sensitization in a 304 steel pipe weld, assessed by EPR test. Effect o f local strain (Ref. 59).

I

i ~ ,, 0

/

,0.2mm, AF1035

Fig. 33 - - Intercrystalline attack at weld in $40900 ferritic steel, 50X.

Fig. 34 - - Fracture face o f intercrystalline attack in low-carbon 12% Cr steel. Corrosion took place at ferrite/ferrite and ferrite/martensite boundaries, leaving martensite units unattacked, 200X.

an extremely low carbon content or effective stabiliza- tion is essential for satisfactory service to be obtained. Fur- ther, boron added to the steel for improved hot worka- bi l i ty can stabilize the carbides to higher temperatures, and this has led to the development of low-boron "nitric acid grade" (NAG) alloys, in which silicon and phosphorus are also controlled to mini- mize adverse effects of grain boundary segregation (Ref. 64). Even with stabilized steels, there is an advantage in re- ducing carbon for nitric acid duty. When making a weld, the material immediately adjacent to the fusion boundary is effec- t ively solution treated at high temperature, and some carbon is thus re- leased into solid solution to form M23C 6 on cooling. Intercrystalline corrosion in this region in a nitric acid plant is known as "knifeline attack"(Refs. 54, 65), and experience has indicated that niobium is more effective at gettering carbon in the high- temperature HAZ than titanium.

The effect of carbide formation depends upon the density of precipitation at a single point, and thus on the grain size or total boundary area. With castings and weld metals that contain some ferrite, precipitation occurs init ial ly on the ferrite/austenite boundaries (Refs. 58, 66, 67). The initial development of depletion depends

on the supply of carbon from the austenite and of chromium from the ferrite, but overall the effects of precipitation are reduced, both by the high phase boundary area available and also by the higher chromium content in the ferrite phase. Thus, materials containing appreciable amounts of ferrite are resistant to sensitization (Ref. 66), although they are by no means immune to the problem given pro- longed exposure to elevated temperatures for postweld heat treatment or in service (Ref. 67). As a consequence, weld metals can contain higher carbon contents than base metal (Ref. 55), and this permits the use of shielding gases containing some CO 2, provided that a sufficient ferrite level is maintained. For gas metal arc welding, an addition of CO 2 may improve metal transfer characteris- tics across the arc, and reduce porosity from shielding gas entrapment.

Martensitic and Ferritic Steels

Carbide formation and associated chromium depletion as a result of welding is by no means confined to austenitic grades but can develop also with martensitic and ferritic alloys (Ref. 54). With martensitic steels, the problem is primarily associated with medium- to high-carbon contents (e.g., 0.2-0.5%) as may be employed for surfacing applications. The loss in corrosion resistance takes place not only at prior austenite grain boundaries but also within the martensitic structure of the matrix (Ref. 68). The effect is again controlled by chromium diffusion, and, since this is more rapid in a martensitic or ferritic matrix than in fcc austenite, depletion can occur in regions heated to a peak temperature rather lower than the austenitic case, typically 450°-500°C (842°-932°F). Low arc energy is preferred, but the situation is complicated by the need for fairly high preheat levels to avoid hydrogen cracking on cooling to room temperature. In most applications, martensitic stainless steels will receive a postweld heat treatment to temper the material. This wil l lead to healing of any chromium-depleted areas, and from the practical standpoint, the problem is encountered only at weldments put into service in the as-welded condition or tempered at very low temperatures to maintain high material hard- ness.

Because of the high chromium diffusion rate and low carbon solubility, sensitization develops rapidly in ferritic steels (Refs. 54, 68, 69). The critical temperature range is similar to that for austenitic alloys, but sensitization in ferritic grades requires only a fairly short exposure time even at temperatures below

148-s I MAY 1996

say 500°C (Fig. 27). It can therefore be difficult or impossible to avoid sensitization conlpletely by use of low arc: energy, especially if multipass welding is in- volved. This is the case for virtually all ferritic stainless steels, from lean alloy 11-12%Cr grades (Ref. 70) up to "super- ferritic" materials containing 25 30%Cr (Ref. 71). In ferritic materials, nitrogen has an analogous effect to carbon, although probably less severe on a wt-% basis. Attempts have been made to avoid the problem by steel processing procedures giving exceptional ly low total (C+N) levels, but in general practice, it has been found more satisfactory to em- ploy stabil ization by t i tanium or niobium. It is remarked that nitrogen is not recognized in standards such as UNS $40900, yet preferential intercrystalline attack has been observed in practice at welded joints in such stainless steel com- plying with specification limits on carbon and titanium but having relatively high nitrogen levels - - Fig. 33.

Time risk of sensitization of low-carbon 12% Cr alloys depends also on the weld area phase balance. Fully ferritic materials are appreciably more prone to the problem than ferritic/martensitic alloys, the adverse effect of precipitation being increased by the local grain growth in- herent in welding these steels. It has been suggested that the formation of grain boundary austenite, and hence martensite (Refs. 54, 69), during a weld thermal cycle has a deleterious effect either by leading to increased strain in the surrounding lattice, or because the austenite has a lower chromium content than the matrix. The latter mechanism is not valid, since, with normal welding conditions and cooling rates, EDX analysis at TWl has shown the final martensite and ferrite to have vir tual ly the same chromium level. Moreover, welds which have suffered intergranular attack display no evidence of preferential corrosion of time martensite - - Fig. 34. Overall, there is no doubt that high niartensite contents are beneficial from the corrosion viewpoint, and apart from the high interphase boundary area, this is probably associated with the higher solubility of carbon and nitrogen in austenite than in ferrite, together with the fact that, when nlartensite formation occurs, it takes place at a temperature below the sensitizing range. Further, if precipitation then occurs on reheating in multipass welds, the "density" wil l be reduced by the high number of nucleation sites within the martensite and at the martensite/ferrite boundaries. The extent of transformation to martensite at welds is therefore important, and it must be recognized that, even though a

/00

80

50

4,0

20

0

5 10 15 20 " 2F 30 Keltmt~user ferrite ~dor

Fig. 35 - - Eftect o f c o m p o s i t i o n on .structure o f 12% Cr ter r i t ic /mar tens i t ic steels, x = base meta l (Ret: 27); ha tched area = H A Z data ob ta i ned at TWI. Ferri te l~ctor = Cr + 6 % S i + *%Ti + 4 % M o + 2 % A I + 4 % N b - 2 % M n 4 % N i - 40(C + N).

12%Cr steel may be two phase (ferrite and tempered nlartensite) in the annealed condition (Ref. 72), rapid cooling during welding can lead to retention of the ferrite formed at high temperature (Fig. 35) (Ref. 73). Direct analogy can be drawn with duplex ferrite/austenitic material (Ref. 40), and, in both cases, development of a predominantly ferritic structure at welds increases the risk of intergranular corrosion in service.

The coarse grain structure normally produced in ferritic stainless steel weld metals can lead to unacceptable loss of toughness, as well as sensitization, and the materials are therefore sometimes welded with austenitic consumables. The activity of carbon and nitrogen wil l depend upon the levels of chromiun] and other elements, but if carbon and nitrogen contents in an austenitic weld metal are appreciably higher than in the base metal, these elements wil l diffuse into the high-temperature HAZ (Ref. 74). This can promote precipitation of chromium carbides or nitrides on cooling and thus loss of corrosion resistance in this area, and must be recognized by minimizing heat input or ensuring that the base metal is "overstabilized."

Intermetallic Phases

In the past, intermetallic formation in conventional austenitic stainless steels has been associated primarily with ex- tended cycles at between perhaps 500 ° 900°C, during PWHT or service (Ref. 54). More recently, it has become evident that significant precipitation of

intermetallics can occur in the course of welding, especially ill high-alloy austenitic and duplex steels, leading to loss of corrosion resistance. A number of intermetallic types has been identified, and, since they are highly alloyed and presumably resistant to direct corrosion in a wide range of media, it would seem that the precipitation has an adverse effect by a depletion mechanism, analogous to that with carbide formation.

Intermetall ic precipitation in austenitic grades can be enhanced in weld metal by alloy element segregation, possibly contributing to the decreased corrosion resistance of autogenous weld metals (Ref. 32). Intermetallics can form also when high-molybdenum nickel- based filler metals are used, but joint behavior remains controlled by the UMZ. Grain boundary intermetallic precipitation has been further observed in the HAZ of superaustenitic weldments (Ref. 75), although reduced corrosion resistance in this region is apparent only when solidif ication conditions have been such that the UMZ is m in im ized- - Fig. 36. Nonetheless, the effect renders it difficult to achieve weldment corrosion properties matching those of the base steel unless some form of PWHT is carried out for homogenization. Nitrogen retards intermetallic formation (Refs. 21, 76), and the problem may be ameliorated in new superaustenitic steels containing very high levels of nitrogen (circa 0.4-0.5%) for improved resistance to pitting and crevice corrosion in chloride media (Ref. 77).

The consequences of intermetall ic


1

I

Fig. 36 - - FIAZ attack at weld in $31254 steel, 5X.

formation are very much more severe in duplex and especially superduplex grades (Ref. 78). The rapid diffusion in ferrite means that precipitate nucleation and growth can be extremely rapid (Figs. 37 and 38), and the concomitant depletion in chromium and molybdenum is sufficient to cause drastic loss in pitting resistance (Fig. 39), although with somewhat less severe effect on general acid corrosion resistance (Ref. 79).

Intermetallic formation has been ex- tensively studied in the context of welding superduplex stainless steels since it is seen as a major hindrance in obtaining optimum joint properties. Its reliable avoidance necessitates use of low arc energy and very low interpass temperatures, which severely restricts overall joint completion rate (Ref. 46). Only a small amount of intermetallic phase can diminish pitting resistance (Ref. 80) - -

1000

1 • UNS $32304 | [] UNS $31803

1 UNS $32550 [] Modified UNS $32550: 3.8%Mo, 0.26%N

. ,800

S 600

D

2O0

0 10 "100 1000 10000

Time,see

Fig. 37 - - Time-temperature transformation curves for duplex stainless steels (Ref. 78). Intermetal l ic formation, 550 ° - 1000°C; (z', 300°-550°C.

Fig. 40. Indeed there is debate as to whether any intermetallic formation at welds in superduplex steels is tolerable, yet it is difficult to see how it can be avoided entirely. Certainly, the rapid- ity of intermetallic formation (Fig. 37) makes it problematic to design welding consumables for superduplex steels that are overalloyed to a significant degree to mitigate the effects of partitioning during the ferrite/austenite

transformation (Ref. 53). Accordingly, nonmatching nickel-based fillers are being explored (Ref. 81), and, provided that adequate joint strength can be achieved, these may well prove of practical advantage for superduplex steels.

The effect of intermetallic formation when welding duplex/superduplex steels has received particular attention with respect to pitting attack in chloride media. From data such as in Fig. 40, the need has been expressed for definition of a maximum tolerable intermetallic volume fraction. This, however, is not the critical parameter. The form of Equation 4 is of general application to describe the depleted zone round a precipitate particle. Both the width of a depleted region and the local reduction in alloy content will increase with increasing time in the precipitation range, and pitting resistance

will thus depend on the maximum size of an intermetallic particle, and not the volume fraction.

Intermetallic formation in HAZs or weld metals of superduplex grades may readily reduce the CPT in a ferric chloride test by some 20°C (36°F), equivalent to a drop in PRE of about 7 units in the adjacent material (Fig. 6). Typical compositions of intermetallic phases are given in Table 5 (Refs. 82-85), and enrichment of a sigma particle in Cr and Mo would re-

Table 5 - - Reported Analyses (wt-%) for Intermetallic Phases in Duplex Stainless Steels (Refs. 80-82)

Phase Cr Mo Ni

Sigma 30 6 5 34 7 4 30 10 4

Chi 27 12 5 28 18 3

R 20 39 3 26 25 4

quire diffusion from a radius roughly two times that of the particle, to decrease the PRE by 7 units, i.e., a reduction of 2.3% Cr and 1.4% Mo in the ferrite. This assumes that sigma formation is controlled by diffusion in ferrite and the ferrite composition. The fall in CPT in Fig. 40 is associated with intermetallic particles some 0.5 to 1.0 IJm in diameter. The associated volume of alloy depleted material of 1-2 pm radius is certainly large enough to constitute a stable pit nucleus (Ref. 9) and the observed effect on pitting resistance would be expected.

This represents an extreme case, since such a particle size requires a fairly ex- tended time in the precipitating temperature range. Diffusion of Mo and Cr (Ref. 44) over 1 pm would involve several min- utes exposure at, say, 800°C (1472°F), equivalent to an arc energy of well over 2 kJ/mm (51 kJ/in.) on 12-ram (0.47-in.) plate, given room temperature "preheat." This arc energy would now be regarded as excessive, following procedural development trials over the last few years. From TTT curves such as in Fig. 37, intermetallic formation may be more rapid than indicated by bulk diffusion data, and it is likely that growth is influenced by faster phase boundary diffusion, which wil l modify the elemental depletion pattern and the ease of initiation of stable pitting. Nevertheless, a particle size of 0.1-0.2 lam could be attained with welding conditions deemed entirely acceptable for 22% Cr alloys, the concomitant alloy depletion being sufficient to reduce the CPT by perhaps 10°C (18°F). This particle width is at the limit of resolution by opti- cal microscopy, while, assuming continuous phase boundary precipitation and a ferrite/austenite lamellar width in the sub- critical HAZ of say 20-40 pm, the volume fraction would be less than 0.5%. Clearly, a weld procedure qualification specification requiring intermetallic phases to be below, say, 1% would not offer a safe- guard against loss of pitting resistance due to intermetallic precipitation during welding, and it is better to place reliance on actual corrosion testing.

150-s I MAY 1996

Discussion

The need to consider weldability is well recognized by steelmakers and has appreciably influenced alloy design in recent years, for example in the development of NAG austenitic steels and of current production duplex alloys. The industrial importance of welding has further led to significant study of the fundamental changes that take place during a welding cycle, with regard to solidification, partitioning and precipitation. Equally, factors contributing to passive film stability and breakdown have been increasingly researched, and in combi- nation the two lines of investigation have greatly clarified the practical effects of fusion welding on corrosion behavior of stainless steels.

Perhaps inevitably, the greater part of research directly aimed at corrosion properties of weldments has been in re- sponse to industrial problems. This has normally entailed identification of controlling metallurgical or process factors, followed by trials to optimize the key variables. This is well illustrated by the extensive work in the last two decades on the problem of sensitization of stainless steels and nickel alloys for nuclear power plants, which has led to a good understanding of the interplay between material, welding and environmental factors and to demonstration of the effects of applied stress upon passive film breakdown at chromium depleted areas (Ref. 62). The work has heightened awareness of the importance of the metal/environment potential in service relative to a passive/active transition, and this concept is being increasingly employed in other industrial regimes (Ref. 86),

Much of the recent work on corrosion of welds in stainless steels has entailed newly developed alloys, both superaustenitic and duplex grades. In the for- mer context, the practical significance of weld metal segregation and the fusion boundary UMZ is now well recognized. It remains to be seen if the latter can be controlled by design of filler metal to obtain a preferred freezing temperature range or by welding techniques giving more rapid pool stirring. It may be also that new high-nitrogen steels (circa 0.5%N) (Refs. 76, 77) in which segregation of alloy elements may be rather different (Ref. 15) will be less sensitive to the problem. If the unmixed zone can be negated, then, in high-alloy austenitic steels at least, weldment corrosion resistance is likely to be limited by intermetallic formation. As a precipitation process, this should be controllable by use of suitable low arc energy and total

thermal cycle, albeit at some cost in productivity. One area which does not seem to have received attention is the consequence of the base metal solution treatment procedure. If the final anneal is set at too low a temperature, residual intermetallic nuclei may remain, which could act to form particles during a weld thermal cycle, analogous to the problem of low-temperature sensitization of 300 series alloys.

Turning to the duplex grades, the advances in modeling alloy element partitioning (Refs. 42, 48) are extremely en- couraging. With current welding procedures, it would seem that the consequences of partitioning are most marked in consequence of secondary austenite formation (Ref. 47), and in many cases, the net corrosion resistance is dictated by the formation of this con- stituent. The general guidelines for welding are understood in that secondary austenite is particularly likely to develop when a low arc energy root pass is reheated by a high arc energy second run. This situation can be avoided by appropriate design of the welding procedure. However, the propensity for secondary austenite formation depends also on the bulk material composition, for example

Fig. 38 - - SEM backscat- tered electron image showing intermetallic formation in a superduplex steel (Ref. 80).

the presence or absence of tungsten (Ref. 47), and further effort is required to clar- ify this effect. In superduplex steels especially, the susceptibility to intermetallic formation is evident, and with present alloys, it must be accepted that low arc energy/interpass temperature conditions are required, although again with loss of productivity. A predictive model, as derived for sensitization of 300 series steels (Refs. 56-60), would be valuable, but would necessitate careful consideration of the range of intermetallic types and compositions formed in superduplex alloys. More detailed quantification of the effects of welding is needed, since the recommendations for avoiding secondary austenite and intermetallic phases are not entirely compatible. De- velopment of the equilibrium austenite content in a weld run so that there is lit- tle tendency for secondary austenite requires observance of a minimum arc energy, whereas suppression of intermetallics will entail a restriction to the usable heat input.

The increased use of duplex steels has had a major effect on weldment testing requirements. In the past, the principal problem stemming from welding was weld decay, and this was controlled by

Fig. 39 - - Chlo- ride pitting due to phase boundary intermetallic formation in superduplex steel (Re£ 80), 200X.

WELDING RESEARCH SUPPLEMFNTI 151-~

Fig. 40 - - Effect of intermetallic precipitation in

superduplex stainless steels

on reduction in FeCI ~ pitting

resistance (Re£ 80).

601

0

I

specifying that the base steel should pass on intercrystalline corrosion test as in ASTM A262, fol lowing a simple isothermal sensitizing heat treatment to simu- late a welding operation. This approach is not viable for avoidance of weld area attack in duplex steels, since both high and low heat input may be damaging, whi le it does not al low for loss of weld metal properties. Hence, it has now become the norm to include corrosion testing in weld procedure qualification, usually involving ferric chloride solution at one or more temperatures. The tendency has been to specify conservative acceptance criteria, and the need to pass such tests in practice had stimulated much of the basic work carried out on the welding behavior of duplex steels.

The phi losophy of incorporat ing a corrosion test in stainless steel weld procedure qual i f icat ion requirements is gaining wider acceptance. However, it means that study is needed of the practical consequences of a welding cycle on corrosion behavior, in terms of the relationship between short tests and longer term operation. To illustrate the point, whi le intermetallic precipitation in superduplex steels has an adverse effect in chloride media at fairly high potential (exemplified by the ferric chloride test method), the critical size of the pit nucleus to develop stable growth may well be higher in service media of lower oxidizing power, such as CO2-containing brines. If this is the case, then intermetallic formation wi l l be of less practical im- port than generally considered, provid- ing of course that it does not unacceptably reduce other service properties, notably toughness. Clearly, it must be recognized that imposition of unreal- istically demanding test procedures may severely restrict the appl icat ion of welded stainless steels and incur appreciable economic penalty.

Concluding Remarks

A review has been carried out of a re-

I I a

1

4 l n ~ l i c . %

m

5

cent study on the effects of fusion welding on the corrosion resistance of stainless steels. Reference has been made to ferritic, martensitic and austenitic alloys and to dual-phase grades. In all cases, concern attaches primarily to the stability of the passive film and the effects have been addressed of 1) segregation during sol idi f icat ion, 2) part i t ioning during phase changes, and 3) precipitation of second-phase particles. Advances have stemmed from fundamental metallurgical and corrosion studies, and from practical procedural development trials, especially on recently developed alloys.

Service behavior wi l l be determined also by factors such as the degree of residual oxide present on the joint on en- tering service (Ref. 46), and study remains necessary so that base metal corrosion resistance can be rel iably obtained at welded joints and unforeseen service failures are avoided. Nonethe- less, a good level of understanding has been established, and in most cases quantitative recommendations on welding procedures can be made to obtain optimum corrosion resistance at fusion weld joints.

Acknowledgments

The author thanks colleagues at TWl and associates throughout the world for their detailed study of stainless steels, for their published papers and for many in- formative discussions. Their work is gratefully acknowledged as the primary input of this paper.

Reteren( es

1. Uhlig, H. H., and Revie, R.W. 1985. Corrosion and Corrosion Control. New York, N.Y., Wiley.

2. Jackson, R. if, and van Rooyen, D. 1971. Electrochemical evaluation of resistance of stainless steels to chloride media. Corrosion 27(5): 203-210.

3. Shaw, D., and Edwards, A. M. 1965. A transistorised potentiostat system for corrosion

studies. Corr Sci 5:413-424. 4. Asami, K., et al. 1978. An XPS study of

the passivity of a series of iron-choromium alloys in sulphuric acid. Corrosion Science 18: 15]-160.

5. Hashimoto, K., et al. 1979. An x-ray photoelectron spectroscopic study on the role of molybdenum in increasing the corrosion resistance of ferritic stainless steels in HCI. Cor- rosion Science 19:3-14.

6. Halada, G. if, et al. 1995. Electrochem- ical and surface analytical studies of the inter- action of nitrogen with key alloying elements in stainless steels. Corrosion '95, Orlanclo, Fla., NACE, Paper 95531.

7. Palit, G. C., et al. 1993. Electrochemical investigations of pitting corrosion in nitrogen- bearing Type 316LN stainless steel. O)rrosion 49 (12):977 991.

8. Pawe[, S. J., etal. 1989. Role of nitrogen in the pitting resistance of cast duplex CF-l~/pe stainless steels. Corrosion 45(2): 125 133.

9. Mattin, S. ff 1994. Nucleation of corrosion pits on stainless steel. Ph.D. dissertation, Cambridge University, Cambridge U.K.

10. Galvele, ]. R. I981. Transport processes in passivity breakdown II - - full hydrolysis of the metal ions. Corrosion Sci. 21 (8): 551-579.

11. Pistorius, P., and Burstein, T. 1992. Metastable pitting corrosion of stainless steel and the transition to stability. Phil. Trans. Roy. So(. A341 : 531-559.

12. Easterling, K. 1983. Introduction to the Physical Metallurgy of Welding. London. But- ter~ :orths.

13. David, S. A., and Vitek, J. M. t989. Correlation between solidification parameters and weld microstructures. International Mate- rials Reviews 34(5): 213-245.

14. Hammar, O., and Svensson, U. 1979. Influence of steel composition on segregation and microstructure during solidification of stainless steels. Solidification and Casting of Metals. London, The Metals Society, Book 192: 401-410.

15. Marshall, P. I., and Gooch, T. G. 1993. Effect of composition on corrosion resistance of high-alloy austenitic stainless steel weld metals. Corrosion 49(6): 514-526.

16. Garner, A. 1979. The effect of autogenous welding on chloride pitting corrosion in austenitic stainless steels. Corrosion 35(3): 108-113.

17. Suutala, N., and Kurkela, M. 1984. Lo- calized corrosion resistance of high-alloy austenitic stainless steels and welds. Stainless Steels '84. Gothenburg, London, The Metals Society, pp. 240-247.

18. Bernhardsson, S.-O. 1991. The corrosion resistance of duplex stainless steels. Du- plex Stainless Steels '91. Beaune, France, Les Editions de Physique, pp. 185-210.

19. Rabensteiner, G. 1989. The welding of fully austenitic stainless steels with high molybdenum contents. Welding in the World 27(1/2): 2-13.

20. Walker, R. A., and Gooch, T. G. 1991. Pitting resistance of weld metal for 22Cr5Ni ferritic-austenitic stainless steels. Brit. Corr. J. 26(1 ): 51-59.

21. Koseki, T., and Ogawa, T. 1991. An investigation on the weld solidification of Cr-Ni- Fe-Mo alloys. Quarterly l. Jap. Weld. Soc. 9(1 ): 143-149.

22. Van der Horst, J. M. A., and Clark, G. A. 1974. Weld corrosion in urea synthesis.

152 -s l MAY 1996

Corrosion Science 14 (11/12): 631-634. 23. Gooch, T. G. 1975. Welding and the

corrosion resistance of austenitic stainless steels. The influence of welding and welds on the corrosion behavior of constructions. Tel Aviv, IIW: 1.53 1.75.

24. Ginn, B. J., and Gooch, T. G. 1993. Pit- ting resistance of autogenous welds in UNS $31254 high alloy austenitic stainless steel. 12th International Corrosion Congress, Hous- ton, Tex., NACE, 4: 2895-2906.

25. Gunn, R. N., and Anderson, P. C. J. 1994. Development of special Ar-He-N 2 gases for TIG welding of duplex and superduplex stainless steels. Duplex Stainless Steels '94. Glasgow, TWI, Paper 30.

26. Rogne, T., Fostervall, H., and R()rvik, G. 1991. Plasma arc welding of duplex stainless steels, vide ref 18: 421-429.

27. Brooks, I. A., and Thompson, A. W. 1991. Microstructural development and solidification cracking susceptibility of austenitic stainless steel welds. Int. Mat. Rev. 36 (1): 16-44.

28. Brooks, J. A. 1990. Weld microsegregation: modelling and segregation effects on weld performance. Weldability of Materials. Detroit, ASM International, pp. 41-47.

29. Aziz, M. K. 1982. Model for solute re- distr ibution during rapid sol idif ication. J. Appl. Phys. 53:1158.

30. Nakao, Y. and Nishimoto, K. 1992. Ef- fects of laser surface melting on corrosion resistance in stainless steel and nickel-base alloy clad layers of cast bi-metallic pipes. IIW Doc. IX-1666-92.

]1. Woollin, P. 1995. laser beam surface melting of high alloy austenitic stainless steel. 4th International Conference on Trends in Welding Research. Gatlinbu rg, Ten n., ASM In- ternational, Session V.

32. Gooch, T. G., and Elbro, A. C. 1995. Welding corrosion-resistant high-al loy austenitic stainless steel. Corrosion in natural and industrial environments: problems and solutions. Grado, NACE Italia, Paper 13.

33. Savage, W. F., Nippes, E. F., and Szek- eres, E. S. 1976. A study of weld interface phenomena in a low-alloy steel. Welding Journal 55(9): 260-s to 268-s.

34. Baeslack, W. A., Lippold, J. C., and Savage, W. F. 1979. Unmixed zone formation in austenitic stainless steel weldments. Weld- ing Journal 58(6): 168-s to 176-s.

35. Flasche, L. H., and Ahluwalia, H. S. 1993. Localized corrosion of the unmixed zone in nickel-base alloy weldments, vide ref 24: 2907-2925.

36. Gooch, T. G., Honeycombe, J., and Walker, P. 1970. Potentiostatic study of the corrosion behavior of austenitic stainless steel weld metal. Brit. Corr. J. 6:148-154.

37. Oreper, G. M., and Szekely, I. 1984. Heat and f lu id- f low phenomena in weld pools. J. Fluid Mech. 147: 53-79.

38. Choo, R. T. C., and Szekely, J. 1994. The possible role of turbulence in GTA weld pool behavior. Welding Journal 73(2): 25-s to 31-s.

39. Hibner, E. L., Hinshaw, E. B., and Lamb, S. 1992. Weld fabrication of a 6% molybdenum al loy to avoid corrosion in bleach plant service. Proc. TAPPI Engineering Conference. Nashvil le, Tenn., TAPPI, reprinted by Nickel Development Institute, No. 14 020.

40. Gooch, T. G. 1982. Weldability of duplex ferritic-austenitic stainless steels. Duplex

Stainless Steels "82. St. Louis, Mo. ASM, pp. 573-602.

41. Combrade, P., and Audouard, I.-P. 1991. Duplex stainless steels and localized corrosion resistance, vide ref 18: 257-281.

42. Hertznlan, S., Roberts, W., and Lin- denmo, M. 1986. Microstructure and properties of nitrogen alloyed duplex stainless steel after welding treatments. Duplex Stainless Steels '86. The Hague, Nederlands Instituut voor Lastechniek, pp. 257-267.

43. Leone, G. L., and Kerr, H. 1982. The ferrite to austenite transformation in stainless steels. Welding Iourna161 (1): 13-s to 21 -s.

44. Brandes, E. A., and Brook, G. B., ed. 1992. Smithells Reference Book. Oxford,U. K. Butterworth-Heinemann, Chapter 13.

45. Baeslack, W. A., and Lippold, I. C. 1988. Phase transformation behavior in duplex stainless steel weldments. Met. Con. 20 (1): 26R-31R.

46. Gooch, T. G. 1991. Corrosion resistance of welds in duplex stainless steels, vide ref 18: 325-346.

47. Nilsson, J.-O., Jonsson, P., and Wilson, A. 1994. Formation of secondary austenite in superduplex stainless steel weld metal and its dependence on chemical composition, vide ref 25, Paper 39.

48. Hertzman, S., and Jargelius-Petterson, R. 1994. vide ref 25, Paper 1.

49. Ogawa, T., and Koseki, T. 1989. Effect of composition profiles on metallurgy and (or- rosion behavior of duplex stainless steel weld metals. Welding Journal 68(5): 181 -s to 191 -s.

50. Gooch, T. G. 1986. A review of stress corrosion cracking of welded duplex ferritic/austenitic stainless steels. Welding in the World 24 (7/8): 148 167.

51. Kudo, T., Tsuge, H. and Moroishi, T. 1989. Stress corrosion cracking resistance of 22%Cr duplex stainless steel in sinlulated sour environments. Corrosion 45(10): 831-838.

52. Hoffmeister, H., and Lothongkum, G. 1994. Quantitative effects of nitrogen contents and cool ing cycles on 6-y transformation, chromium nitride precipitat ion and pitt ing corrosion after weld simulation of duplex stainless steels, vide ref 25, Paper 55.

53. Fager, S-A. 1991. Design of consumables for the welding of superduplex stainless steel, vide ref 18, pp. 403-411.

54. Folkhard, E. 1988. Welding Metallurgy of Stainless Steels. Vienna, Austria, Springer- Verlag.

55. Gooch, T. G., and Willingham, D. C. 1975. Weld Decay in Austenit ic Stainless Steels. Abington, Cambridge, TWl.

56. Stawstr6m, C., and Hillert, M. 1969. An improved depleted-zone theory of intergranular corrosion of 18-8 stainless steel. JISI, 207: 77-85.

57. Bruemmer, S. M. 1990. Quantitative model l ing of sensitization development in austenitic stainless steel. Corrosion 46(9): 698-709.

58. Nakao, Y., et al. 1992. Effect of delta ferrite on sensitization of austenitic stainless steel. Welding International 6 (7): 523-530.

59. Atteridge, D. G., et al. 1992. Model predictions of HAZ (heat-affected zone) grain boundary chromium depletion development for austenitic stainless steel. 3rd International Conference on Trends in Welding Research. Gatlinburg, Tenn. ASM International, pp. 681-685.

60. Solomon, H. D. 1984. Influence of

composition on continuous cooling sensitization of Type 304 stainless steel. Corrosion 40(2): 51-60.

61. Gooch, T. G. 1968. Corrosion of AISI type 304 austeniti( stainless steel. Brit. Weld- ingJ. 15(7): 345-357.

62. Danko, J. C. 1992. Stress corrosion cracking of weldments in boiling water reactor service. Stress Corrosion Cracking, R. H. Jones, ed., ASM International, Materials Park, Ohio, Chapter 15, pp. ~45-354.

63. Povi(h, M. J., and Rao, P. 1978. Low temperature sensitization of welded ]~/pe 304 stain less steel. Corrosion 34(8): 269-275.

64. Beckitt, E. R., Bastow, B., and Glad- man, 1. 1987. Effects of residual element contents on corrosion resistance of Type 304L stainless steels in boiling nitric acid. Stainless Steels "87, York, U.K., Institute of Metals, pp. 234-246.

65. Ikawa, H., Nakao, Y, and Nishimoto, K. 1977. Study on the knifeline attack phe- nomenon in stabil ized austenitic stainless steels. Trans. Jap. Weld. Soc. 8(1 ): 918.

66. Lundin, C. D. 1991. Repair welding of austenitic stainless steel castings preservation of corrosion resistance. Proceedin~,s Mainte- nance and Repair Welding in Power Plants, Orlando, Fla, AWS, pp. 46-56.

67. Devine, T. M. 1980. Influence of carbon content on the ferrite morphology and the sensitization of duplex stainless steel. EPRI Workshop Report, WS 79-174, Vol. 1, Paper 5.

68. Truman, J. E. 1994. Stainless steels. Corrosion: Metal/Environment Reactions, 3rd edition, ed. by Shriver, L. L., Jarman, R. A. and Burstein, G. T., Oxford, U.K., Butterworth- Heinemann, 3:34 3:77.

69. Bond, A. ff 1969. Mechanisms of intergranular corrosion in ferritie stainless steels. Trans. Met. Soc. of AIME 245(8): 2127-2134.

70. Devine, T. M., and Ritter, A. M. 1983. Sensitization of 12 wt-pct chromium, titanium-stabil ized ferritic stainless steel. Met. Trans (4). 14(A) (8): 1721-1728.

71. Nichol, T. J., and Davis, I. A. 1978. In- tergranular corrosion testing and sensitization of two high chromium ferritic stainless steels. ASTM STP 656: 179-196.

72. Hewitt, J. 1992. High chromium controlled hardenability steels. Ist International Chromium Steels and Alloys Congress, Cape Town, South Africa, SAIMM, Vol. 2, pp. 71-88.

73. Gooch, T. G., and Ginn, B. J. 1990. Heat-affected zone toughness of SMA welded 12%Cr martensitic-ferrit ic steels. Welding Journal 69(11 ): 431 -s to 440-s.

74. Honeycombe, J., and Gooch, T. G. 1983. Corrosion and stress corrosion of arc welds in 18% chromium-2% molybdenum-titanium stabilized stainless steel. Brit. Corr. J. 18(1): 25-34.

75. Liljas, M., Holmberg, B., and Ulander, A. 1984. Welding of a h igh-molybdenum austenitic stainless steel, vide ref 17: 323-329.

76. Charles, J., etal. 1994. A new high nitrogen austenitic stainless steel with improved structure stability and corrosion resistance properties. 10th Anniversary C~;nference: Cor- rosion and Materials Offshore. Oslo, NITO/NKF.

77. Liljas, M., and Stenwall, P. 199~. Weld- ing of UNS $32654 - - corrosion properties and metallurgical aspects, vide ref 24: 2882- 2894.

78. Charles, J. 1991. Superduplex stainless

WFI DINIG R F ~ F A R C H ~1 ]PPl FMFNIT I 1 ~'~-~

steel - - structure and properties, vide ref 18: 3-48.

79. Potgieter, J. H. 1992. Influence of sigma phase on general and pitting corrosion resistance of SAF 2205 duplex stainless steel. Brit. Corr. J. 27(3): 219-223.

80. Gunn, R. N. 1993. Unpublished work at TWl, Abington, U.K.

81. Karlsson, L., Andersson, S. L., and Huhtala, T. 1994. New Ni-base consumables for welding of duplex and superduplex stainless steels, vide ref 25, Paper 42.

82. Chance, J., Coop, W., and Gradwell, K.

J. 1982. Structure property relationships in a 25Cr-7Ni-2Mo duplex stainless steel casting alloy, vide ref 40, 371-398.

83. Karlsson, L., Ryen, L., and Pak, S. 1995. Precipitation of intermetallic phases in 22%Cr duplex stainless steel weld metals. Welding Journal 74(1 ): 28-s to 40-s.

84. Huhtala, T., etal. 1994. Influence of W and Cu on structural stability in superduplex weld metals, vide ref 25, Paper 43.

85. Kondo, K., et al. 1993. Precipitation behavior of sigma-related phases in 25%Cr based superduplex stainless steels. Innovation

Stainless Steel. Florence, Italy, AIM, 2.191-2.195.

86. Johnsen, R., and Olsen, S. 1992. Expe- rience with the use of UNS $31254 in seawa- ter systems-- case histories from the field. Cor- rosion "92, Nashville, Tenn., NACE, Paper 397.

154-sl MAY 1996

corrosion behavior of welded stainless steel

Documents