Corrosion of Weldments
A. Wahid, D.L. Olson, and D.K. Matlock, Colorado School of Mines
C.E. Cross, Martin Marietta Astronautics Group
WELDMENTS exhibit special microstruc-
tural features that need to be recognized and
understood in order to predict acceptable corro-
sion service life of welded structures (Ref 1).
This article describes some of the general char-
acteristics associated with the corrosion of weld-
ments. The role of macrocompositional and mi-
croeompositional variations, a feature common
to weldments, is emphasized in this article to
bring out differences that need to be realized in
comparing corrosion of weldments to that of
wrought materials. A more extensive presenta-
tion, with data for specific alloys, is given in
Volume 13 of the ASM Handbook (Ref 2).
Weldments inherently possess compositional
and microstructural heterogeneities, which can
be classified by dimensional scale. On the larg-
est scale, a weldment consists of a transition
from wrought base metal through a heat-affected
zone and into solidified weld metal and includes
five microstructurally distinct regions normally
identified (Ref 3) as the fusion zone, the un-
mixed region, the partially melted region, the
heat-affected zone, and the base metal. This mi-
crostructural transition is illustrated in Fig. 1.
The unmixed region is part of the fusion zone,
and the partially melted region is part of the
heat-affected zone, as described below. Not all
five zones are present in any given weldment.
For example, autogenous (that is, no filler
metal) welds do not have an unmixed zone.
The fusion zone is the result of melting
which fuses the base metal and filler metal to
produce a zone with a composition that is most
Weld nugget
Weld interface / Unmixed region
P a r t i a l l y melted ~ /
region
Unaffected base metal
Schematic showing the regions of a heteroge-
Fig. 1 neousweld. Source: Ref3
often different from that of the base metal. This
compositional difference produces a galvanic
couple, which can influence the corrosion proc-
ess in the vicinity of the weld. This dissimilar-
metal couple can produce macroscopic galvanic
corrosion.
The fusion zone itself offers a microscopic
galvanic effect due to microstructural segrega-
tion resulting from solidification (Ref 4). The
fusion zone also has a thin region adjacent to the
fusion line, known as the unmixed (chilled) re-
gion, where the base metal is melted and then
quickly solidified to produce a composition sim-
ilar to the base metal (Ref 5). For example,
when type 304 stainless steel is welded using a
filler metal with high chromium-nickel content,
steep concentration gradients of chromium and
nickel are found in the fusion zone, whereas the
unmixed zone has a composition similar to the
base metal (Fig. 2).
Heat-Affected Zone. Every position in the
heat-affected zone relative to the fusion line ex-
periences a unique thermal experience during
welding, in terms of both maximum temperature
and cooling rate. Thus, each position has its own
microstructural features and corrosion suscepti-
bility.
The partially melted region is usually one or
two grains into the heat-affected zone relative to
the fusion line. It is characterized by grain
boundary liquation, which may result in liqua-
tion cracking. These cracks, which are found in
the grain boundaries one or two grains below the
30
D 28
2 4
0 22
2O
16
. . . . . . +.". +oo i=:01 ,:z HAz°°°I . . . . . . . . .
CHROMIUM
I I
20
18 '
16 ~
14 z
12 o
NICKEL 10 m
I 8
0 500 1000 1500
DISTANCE(MICRONS)
Concentration profile of chromium and
F ig . 2 nickel across the weld fusion boundary re-
gion of type 304 stainless steel. Source: Ref 5
fusion line, have been identified as potential ini-
tiation sites for hydrogen-promoted underbead
cracking in high-strength steel.
Microstructural Gradients. On a fine scale,
microstructural gradients exist within the heat-
affected zone due to different time-temperature
cycles experienced by each element of material.
Gradients on a similar scale exist within solidi-
fied multi-pass weld metal due to bead-to-bead
variations in thermal experience. Compositional
gradients on the scale of a few microns, referred
to as microsegregation, exist within individual
weld beads due to segregation of major and trace
elements during solidification (Ref 4).
Forms of Weld Corrosion
Weldments can experience all the classical
forms of corrosion, but they are particularly sus-
ceptible to those affected by variations in micro-
structure and composition. Specifically, gal-
vanic corrosion, pitting, stress corrosion,
intergranular corrosion, and hydrogen cracking
must be considered when designing welded
structures.
Galvanic Couples. Although some alloys can
be autogenously welded, filler metals are more
commonly used. The use of filler metals with
compositions different from the base material
may produce an electrochemical potential differ-
ence that makes some regions of the weldment
more active. For example, Fig. 3 depicts weld
metal deposits that have different corrosion be-
havior from the base metal in three aluminum
alloys (Ref 6).
For the majority of aluminum alloys, the weld
metal and the heat-affected zone become more
noble relative to the base metal, as demonstrated
in Fig. 3(a) and (b) for a saltwater environment
(Ref 6). Certain aluminum alloys, however,
form narrow anodic regions in the heat-affected
zone and are prone to localized attack. Alloys
7005 and 7039 are particularly susceptible to
this problem (Fig. 3c).
There are a number of other common weld
deposit/base metal combinations that are known
to form galvanic couples. It is common practice
to use austenitic stainless steel welding consum-
ables for field repair of heavy machinery, partic-
ularly those fabricated from high-strength low-
ASM Handbook, Volume 6: Welding, Brazing, and Soldering
D.L. Olson, T.A. Siewert, S. Liu, and G.R. Edwards, editors, p 1065-1069
Copyright © 1993 ASM International®
All rights reserved.
www.asminternational.org
1 0 6 6 / Special W e l d i n g and Joining Topics
Distance from weld centerline, in.
o ~ 1 2 3 4 o 900' I Edg~ of wel~l bead ' =
,
I Hardness
i=-f - - " ~-E > 750 ~I~ I ~Co~oslon potential
70C I I
0 25 50 50 1 O0
Distance from weld centerllne, mm
65 =
55 -r
45 ,~
2 5
Distance from weld centertine, in.
O 1 2 3 4 800
=-~mm o 750 llLII [Edgel of weld = bead ' l 1 65 m
e= (n Hardness :=
¢~ 600 If t I I I I 25
0 25 50 50 1 O0
Distance from weld centerline, mm
Distance from weld centerllne, in.
1 2 3 4 uo 10500 , ,
[ IEdoe°'w'd'"ad' '1 '° w 1 DO0 ~ Hardness m~T: / , . . . . . . . . . . . 80 n-
o = 950 70 ==
i T 900 Corrosion potential
5O
o 8 5 0 I J I
0 25 50 50 100
Distance from weld centerline, mm
(a) (b) (c)
Effect of welding heat on microstructure, hardness, and corrosion potential of three aluminum alloy welded assemblies. (a) Alloy 5456-H321 base metal with alloy
Fig. 3 5556 filler. (b) Alloy 2219-T87 base metal with alloy 2319 filler. (c) Alloy 7039-T651 base metal with alloy 5183 filler. Source: Ref 6
i n t r ag ranu la r~ grain boundary
~ E 2 , , ~ ~ d e p l e t i o n
~3EI ~
El: potential for ppt. boundary
E2: potential for depleted z o n e precipitate
E3: potential for matrix
Depleted regions adjacent to precipitates. r l a
4 ='~" These regions cause an electrochemical po-
tential (E) difference that can promote localized corro-
sion at the microstructural level.
alloy steel. This practice leaves a cathodic
stainless steel weld deposit in electrical contact
with the steel. In the presence of corrosive envi-
ronments, hydrogen is generated at the austen-
itic weld metal cathode, which is capable of
maintaining a high hydrogen content without
cracking. However, the cathodic behavior of the
austenitic weld deposit may increase the suscep-
tibility for stress-corrosion cracking in the heat-
affected zone of the high-strength steel. A 40%
thermal expansion mismatch between the auste-
nitic stainless steel and ferritic base metal pro-
duces a significant residual stress field in the
weldment; this residual stress field also contrib-
utes to cracking susceptibility. A similar, but
more localized, behavior may explain the corre-
lation between stress-corrosion cracking suscep-
tibility and the presence of retained austenite in
high-strength steel weld deposits.
Another common dissimilar metal combina-
tion involves the use of high-nickel alloys for
weld repair of cast iron. Fe-55Ni welding elec-
trodes are used to make weld deposits that can
hold in solid solution many of the alloying ele-
ments common to cast iron. Furthermore, weld
deposits made with Fe-55Ni welding consum-
ables have an acceptable thermal expansion
match to the cast iron. Because cast iron is an-
odic to the high-nickel weld deposit, corrosive
attack occurs in the cast iron adjacent to the weld
deposit. It is suggested that cast iron welds made
with high-nickel deposits be coated (painted) to
reduce the susceptibility to selective corrosion
attack.
Plain carbon steel weldments can also exhibit
galvanic attack. For example, the E6013 weld-
ing electrode is known to be highly anodic to
A285 base metal in a seawater environment (Ref
7). It is important to select a suitable filler metal
when an application involves a harsh environ-
ment.
Weld Decay of Stainless Steel. During
welding of stainless steels, local sensitized
zones (i.e., regions susceptible to corrosion) of-
ten develop. Sensitization is due to the forma-
tion of chromium carbide along grain bound-
aries, resulting in depletion of chromium in the
region adjacent to the grain boundary (Ref
8-14). This chromium depletion produces very
localized galvanic cells (Fig. 4). If this depletion
drops the chromium content below the necessary
12 wt% that is required to maintain a protective
passive film, the region will become sensitized
to corrosion, resulting in intergranular attack.
This type of corrosion most often occurs in the
heat-affected zone. Intergranular corrosion
causes a loss of metal in a region that parallels
the weld deposit (Fig. 5). This corrosion behav-
ior is called weld decay (Ref 13).
The formation of sufficient chromium carbide
to cause sensitization can be described by the
C-shaped curves on the continuous cooling dia-
gram illustrated in Fig. 6. The figure shows sus-
ceptibility to sensitization as a function of tem-
perature, time, and carbon content (Ref 15). If
the cooling rate is sufficiently great (curve A in
Fig. 6), the cooling curve will not intersect the
given C-shaped curve for chromium carbide and
the stainless steel will not be sensitized. By de-
creasing the cooling rate, the cooling curve
(curve B) eventually intersects the C-shape nu-
cleation curve, indicating that sensitization may
occur. At very low cooling rates, the formation
of chromium carbide occurs at higher tempera-
ture and allows for more nucleation and growth,
resulting in a more extensive chromium-de-
pleted region.
The minimum time required for sensitization
as a function of carbon content in a typical stain-
less steel alloy is depicted in Fig. 7. Because the
WELD DECAY
Intergranular corrosion (weld decay) of stain-
Fig. 5 less steel weldments. FZ, fusion zone
normal welding thermal cycle is completed in
approximately two minutes, for this example the
carbon content must not exceed 0.07 wt% to
avoid sensitization. Notice that the carbide nu-
cleation curves of Fig. 6 move down and to
longer times with decreasing carbon content,
making it more difficult to form carbides for a
given cooling rate.
The control of stainless steel sensitization
may be achieved by using:
• A postweld high-temperature anneal and
quench to redissolve the chromium at grain
boundaries, and hinder chromium carbide for-
mation on cooling
• A low-carbon grade of stainless steel (e.g.,
304L or 316L) to avoid carbide formation
• A stabilized grade of stainless steel containing
titanium (alloy 321) or niobium (alloy 327),
which preferentially form carbides and leave
chromium in solution. (There is the possibil-
ity of knife-line attack in stabilized grades of
stainless steel.)
• A high-chromium alloy (e.g., alloy 310)
Role of Delta Ferrite in Stainless Steel
Weld Deposits. Austenitic weld deposits are
frequently used to join various ferrous alloys. It
has been well established that it is necessary to
have austenitic weld deposits solidify as primary
ferrite, also known as a ~ ferrite, if hot cracking is
to be minimized (Ref 16, 17). The amount and
C o r r o s i o n of W e l d m e n t s / 1 0 6 7
900
8O0
P 700
o .
E 600
500
400
10
~"o.oso ,t"
\ ( o0 0(
1 min 10 min 1 h 10 h
Time to sensitization
- 16oo
1400
1200 ~.
E
19% C - lOOO ~
800
100 h 1000 h
Time-temperature-sensitization curves for type 304 stainless steel in a mixture of CuSO4 and HSO 4
Fig. 6 containing copper. Source: Ref 15. Curves A and B indicate high and medium cooling rates, respec-
tively.
104
"~ 10 3
d
E
10 2 p,
. - 10
i -
\
0.1
0.01 0.03 0.05 0.07 0.09 0.11
Carbon content, wt%
Minimum sensitization time from a time-tem-
=::nr,~. 7 perature-sensitization diagram as a function
of carbon content for a typical 300-series stainless steel
alloy. Source: Ref 15
form of ferrite in the weld metal can be con-
trolled by selecting a filler metal with the appro-
priate chromium and nickel equivalent. A high
chromium/nickel ratio favors primary ferrite for-
mation, whereas a low ratio promotes primary
austenite (Fig. 8). An optimum condition can be
attained for ferrite contents between 3 and 8
vol% in the weld deposit. Ferrite contents above
3 vol% usually guarantee primary ferrite forma-
tion and thus reduce hot cracking susceptibility.
However, ferrite above 10 vol% can degrade
mechanical properties at low- or high-tempera-
ture service. At low temperatures, excess ferrite
can promote crack paths when the temperature is
below the ductile-brittle transition temperature.
At high temperatures, continuous brittle sigma
phase may form at the interface between the
austenite and the ferrite. The ferrite content can
be confirmed using magnetic measuring equip-
ment (Ref 16, 17).
Figure 8 can be used to predict the type of
ferrite (primary or eutectic) and the ferrite con-
tent when a difference exists between the stain-
less steels being joined, such as when welding
type 304 to type 310 stainless steel (Ref 18).
This diagram shows the compositional range for
the desirable primary solidification mode. The
dotted lines on the diagram indicate the various
transitions in the primary solidification phase.
Because not all ferrite is primary ferrite (i.e.,
some is a phase component of a ferrite-austenite
eutectic), this diagram can be used to ensure that
ferrite is the first solid (primary) phase to form.
This condition occurs when the weld deposit has
a composition in the range labeled FA in Fig.
8. Because primary ferrite is the preferable mi-
crostructure, use of this diagram should re-
duce problems of hot cracking during welding.
Also, the corrosion behavior of stainless steel
weld deposits and castings is measurably dif-
ferent depending on whether the stainless steel
has a microstructure generated with primary
ferrite or primary austenite (Ref 19-25). Thus,
knowledge of the weld metal ferrite content
and form is necessary in order to be able to
properly characterize and predict corrosion be-
havior.
Pitting is a form of localized attack caused by
a breakdown in the thin passive oxide film that
protects material from the corrosion process.
Pits are commonly the result of a concentration
cell established by a variation in solution com-
position that is in contact with the alloy material.
Such compositional variations result when the
solution at a surface irregularity is different from
that of the bulk solution composition. Once a pit
has formed, it acts as an anode supported by
relatively large cathodic regions. Pitting has a
delay time prior to nucleation and growth, and
nucleation is very site-selective and microstruc-
ture-dependent. Pits are often initiated at spe-
cific microstructural features in the weld deposit
(Ref 26). Pitting occurs when the material/
solution combination achieves a potential that
exceeds a critical value, known as the pitting
potential. The tendency for a given alloy/
solution combination to pit can often be charac-
terized by critical potentials for pitting and
repassivation determined by a cyclic potentiody-
namic polarization technique.
Pits develop more readily in metallurgically
heterogeneous materials. For example, when
austenitic stainless steel is heated to tempera-
tures where sensitization takes place (Ref 26,
27), the resulting chromium-depleted region is
subject to pitting. Pits may also initiate at the
austenite-ferrite interfaces in stainless steel weld
metal.
Although weld metal has a higher probability
of being locally attacked because of microsegre-
gation in the dendritic structure, filler metals are
now available that have better pitting resistance
than their respective base metals; information
about these filler metals can be obtained from
consumable suppliers. However, even when the
proper filler metal is used, pitting may still occur
in the unmixed zone.
Duplex stainless steels, with ferrite contents
in the range of 40 to 50 vol%, are often used to
decrease the tendency of stress-corrosion crack-
ing in chromium-nickel high-alloy steels. The
welding practice for duplex stainless steels must
be given special attention (Ref 20-22, 24, 25,
27) to avoid reduction in corrosion resistance.
The combination of a low carbon content and a
carefully specified nitrogen addition have been
reported to improve resistance to pitting corro-
sion, stress-corrosion cracking, and intergranu-
lar corrosion in the as-welded condition. The
low carbon content helps avoid sensitization,
while the addition of nitrogen slows the precipi-
tation kinetics associated with the segregation of
chromium and molybdenum during the welding
process (Ref 1). On rapid cooling from high
temperature, nitrogen also has been reported to
form deleterious precipitates (for example,
Cr2N) in the ferrite, thus reducing the corrosion
resistance (Ref 28). Nitrogen also increases the
formation of austenite in the heat-affected zone
and weld metal during cooling. A minimum pit-
ting corrosion rate is achieved at a ferrite content
of about 50 vol%.
Stress-Corrosion Cracking. Weldments can
be susceptible to stress-corrosion cracking under
specific environmental conditions. This crack-
ing requires the proper combination of corrosive
media, susceptible microstructure, and tensile
stress. Welds are often loaded in tension (due to
residual stress) to a level approaching the yield
strength of the base metal. A weld, with its var-
ious heterogeneous microstructural features,
thus becomes an excellent candidate for stress-
corrosion cracking.
Stress-corrosion cracks have an anodic crack
tip and often leave apparent corrosion products
along the fracture. Cracking is often character-
ized by crack branching and usually has a delay
time prior to crack initiation, with initiation oc-
curring at corrosion pits. Increasing the ferrite
content in stainless steel weld metal reduces
stress-corrosion cracking susceptibility. Ap-
proximately 50 vol% ferrite gives optimum
stress-corrosion cracking resistance.
Welding parameters influence the amount and
distribution of residual stress, because the extent
o