Heat-Affected Zone Infiltration By Dissimilar

Liquid Weld Metal

A qualitative study of several heterogeneous gas metal-arc
welds shows that the action of the liquid metal on the base
metal is an important aspect of dissimilar fusion welding
BY S. J. M A T T H E W S AND W. F. S A V A G E

ABSTRACT. An investigation was under- high iron concentrations and a corre- dissimilar arc welding is the interac-
taken to explore the mechanism by spondingly lower amount of penetration tion of the molten weld puddle with
which heat-affected zone cracks are was observed in base metals of higher the base metal, or more specifically,
backfilled with liquid weld metal in nickel concentrations. the occurrence of grain boundary pen-
heterogeneous consumable electrode fu- Liquid copper-nickel weld metal does
sion welding processes. not tend to penetrate base metals etration by a chemical specie associ-
In order to establish trends which may which possess a body-centered cubic ated with the weld metal. This prob-
be useful in predicting the occurrence of structure at elevated temperatures, even lem is usually maximized in the case
this phenomenon, several different combi- under the application of a considerable of the consumable electrode arc weld-
nations of heterogeneous weldments were amount of augmented strain. ing processes, wherein during the
made and evaluated using metallographic In general, the backfilling capability of transfer of filler metal across an ion-
techniques. In addition, a complete series liquid weld metal in dissimilar metal ized, gas-filled electric arc, tremendous
of "Varestraint" tests was conducted on weldments appears to obey the character- superheat is generated in the liquid
two heats of HY-80 steel welded both istic response of liquid metal embrittle-
autogenously without filler additions, and ment couples, in that penetration is more system.
heterogeneously using copper-nickel con- prevalent in cases where there is rela- The aforementioned weld metal-
sumable electrode material. tively low mutual solid solubility and an base metal interaction becomes a sig-
The infiltration of the grain boundaries absence of intermetallic compound for- nificant problem when it occurs and
in HY-80 steel by liquid copper-nickel mation between the two dissimilar spe- should be taken into consideration
weld metal under an applied strain field cies. when fabricating dissimilar metal
is postulated to follow a 3 step sequence An "unmixed" region was observed to parts by arc welding for the following
of events: occur in dissimilar weldments when the
1. Rapid grain boundary attack by melting range of the heterogeneous filler reasons:
copper atoms through the heat-affected metal was approximately equal to or 1. Liquid weld metal entering the
zone which is rendered temporarily aus- greater than the solidus of the base base metal via the grain boundaries
tenitic during the peak portion of the metal. It is postulated that the "un- may "heal up" portions of grain
weld thermal cycle. mixed" region may interfere with the boundary fissures created during arc
2. Significant reduction in grain infiltration of the weld metal species into welding. The unhealed portion of the
boundary cohesive strength resulting in the heat-affected zone grain boundaries.
severe fissuring under augmented strain crack may then go undetected upon
conditions. Introduction subsequent nondestructive testing.
3. The backfilling of those grain The demand for an increased num- 2. Weld metal entering the base
boundary crevices which intersect the ber of dissimilar metal weldments has metal via the grain boundaries may
molten weld puddle with nickel-depleted alter the grain boundary chemical and
weld metal driven by favorable interfa- become greater in recent years, pri-
marily due to the more complex de- mechanical properties sufficiently to
cial energy conditions. Backfilling is halt-
ed at a position in the base metal sign requirements encountered in the render the region incapable of with-
where the peak temperature imposed dur- nuclear and aerospace industries. The standing the design service conditions
ing the welding cycle equals the solidus majority of published technical pa- of stress and/or corrosion.
of the iron-diluted copper-nickel alloy. pers, in which the problems of dissimi- 3. Weld metal entering the grain
A significant amount of base metal lar welding are described, have out- boundaries of the base metal may, in
infiltration by liquid copper-nickel weld lined a number of metallurgical and some dissimilar couples, instantane-
metal was observed in weldments in design factors which must be reviewed ously embrittle the matrix causing a
which the base metal was either face- in order to predict how each joint will
centered cubic or experienced an allotrop- failure commonly observed in classic
ic transformation to an austenitic phase react under specific environmental cases of liquid metal embrittlement.
during welding. Also, more penetration stresses and conditions. Those factors,
was observed in those base metals of which have been emphasized to date, Background Information
are differences in coefficients of ther- Terminology
mal expansion, creation of galvanic Throughout this paper, several
S. J. MATTHEWS is with the Stellite Div., corrosion couples and, in the case of
Cabot Corp., Kokomo, Ind. and W. F. terms are used to describe the infiltra-
SAVAGE is Director of Welding Research, arc welding, control of filler metal tion of a heterogeneous weld metal
Rensselaer Polytechnic Institute, Troy, N.Y. dilution in order to prevent fissuring in phase into the adjacent base metal. In
Paper to be presented at AWS 52nd An- the as-deposited weld metal. order to avoid confusion with conflict-
nual Meeting in San Francisco, Calif, dur-
ing April 26-29, 1971. One factor generally neglected in ing connotations which may exist, and

174-s [ A P R I L 1971
to make the discussion more clear, the favorable interfacial surface energies actual fabrication conditions. The
following definitions are established: which promote grain boundary wet- weldments were then examined metal-
1. Backfilling—The act of liquid ting. The dihedral angle, a character- lographically for evidence of base
weld metal entering the base metal via istic of a liquid phase when it meets metal impregnation. Since the pol-
cracks at the fusion boundary which two crystals of another phase, varies ished surface of a dissimilar metal
intersect the molten weld puddle. It depending upon the magnitude of the couple can react differently to the
will be illustrated later that back-filled interfacial surface energies. When the attack of the proper chemical reagent,
cracks are usually characterized by an interfacial energy of the liquid phase a selective etching technique was em-
unfilled portion of the crack beyond is less than one-half the grain bound- ployed to detect any penetration of
the segment of crack containing dis- ary interfacial energy, the dihedral liquid weld metal into the base mate-
similar weld metal. angle is zero and the liquid phase rial.
2. Grain boundary wetting—The tends to establish itself continuously Table 1 summarizes the materials
act of liquid weld metal entering the along the grain boundary. A schemat- that were available throughout this
base metal by a surface-energy-con- ic representation of the energy condi- program. While every possible filler
trolled mechanism. In this case, the tions required for grain boundary wet- metal and base metal combination
weld metal usually establishes itself ting is given in Fig. 1. It is believed was not explored, a sufficient range of
continuously along a grain boundary that this action can be accelerated by combinations were welded in order to
without the presence of a crack in stress to the point where penetration generate meaningful qualitative data.
advance of the weld metal phase. can result even when the dihedral
angle is somewhat greater than zero. 2 Deposition of Heterogeneous
3. Penetration—This term is nor- Weld Beads
A majority of the knowledge of
mally used to describe the depth of All filler metal was deposited on the
liquid metal embrittlement has been
the weld bead below the plate surface. base metal using the consumable elec-
derived from extensive experimental
However, throughout this investiga- trode gas metal-arc process and pure
observation. In a survey of some 50
tion, the term is used to refer to the argon as the shielding gas. The weld-
liquid metal-solid metal combinations,
distance that the dissimilar weld metal
Rostocker 3 observed that in all cases ing current was held constant
infiltrates into the base metal by either throughout this investigation at an
the embrittlement couples were char-
one of the two preceding mechanisms amperage high enough to produce an
acterized by a low mutual solubility of
or by a combination of both. electrode transfer mode in the spray
the liquid metal and the solid metal, in
Liquid Metal Embrittlement addition to the absence of intermetal- arc region.
Since the characteristic nature of lic compound formation. Introduction of Strain to the
liquid metal embrittlement may be Heterogeneous Weldments
related to the penetration of dissimilar Object The RPI Varestraint apparatus was
liquid weld metal into the adjacent employed in order to introduce aug-
The object of this investigation was
base metal during arc welding, it mented strain to the dissimilar metal
twofold and can be summarized as
would be worthwhile to at least define weldments. The introduction of strain
follows:
this phenomenon. was desired in order to produce as
1. To study weld metal-base metal
Liquid metal embrittlement is de- much cracking in the base metal as
interactions of some commercially
scribed as a process wherein ductility possible. A complete description of
available high strength alloys fusion
or fracture stress of a solid metal is the Varestraint apparatus has been
welded using a heterogeneous filler
reduced by exposure of the surface to
metal. provided by Savage and Lundin, 4 and
a liquid metal. When elaborating on this equipment has been successfully
the fundamental mechanisms, howev- 2. To establish trends which may be
used in numerous hot-cracking investi-
er, there are two categories of liquid useful in predicting the penetrability
gations over the past 5 years.
metal embrittlement which must be of a base metal by a molten filler
metal during a dissimilar arc welding A V 4 in. groove was machined
identified and handled separately: 1 down the length of each test plate
process.
1. Instantaneous failure of a stressed prior to welding in order to receive
metal when liquid is applied due to the consumable electrode material
Experimental Procedure
reduction of solid metal cohesive en- with a minimum of as-deposited weld
ergy after adsorption of a liquid specie The experimental procedure em- reinforcement. After the Varestraint
preferentially at sites of high stress ployed in this investigation was essen- testing of each specimen, a top-view
concentration. tially metallographically oriented. section was removed from the
2. Grain boundary penetration of a Heterogeneous welds were made in a weldment containing the location of
solid metal by a liquid such that the simple bead-on-plate fashion, usually the molten weld pool at the instant of
material fails intergranularly. This accompanied by the application of an application of augmented strain. This
process is diffusion controlled and has augmented strain in order to simulate section was then mounted in epoxy,
been treated theoretically by grain the joint restraint encountered under ground and polished using standard
boundary wetting concepts. techniques to reveal a plane about 10
The second category of liquid metal mils below and parallel to the original
embrittlement appears to be the one top surface of the base plate.
most closely allied to the heat-affected Numerous chemical reagents were
zone penetration and crack backfilling explored for each specimen until an
encountered under the appropriate adequate selective etching effect could
welding conditions. A complete treat- be achieved.
ment of the mechanisms governing As a method of comparing the rela-
this form of liquid metal embrittle- SL < I tive extent of base metal penetration
2
ment has not yet been advanced or at 6ss produced in each combination, the
least has not yet achieved universal Fig. 1—Schematic representation of en- total lengths of all wetted grain
agreement. One popular theory is that ergy conditions required for grain boundaries and backfilled heat-
intercrystalline attack is influenced by boundary wetting affected zone cracks visible within a

WELDING RESEARCH SUPPLEMENT | 175-s

 70/30
Table 1—Weld Filler Metals and Base Metals COPPER
NICKEL
10 1
i.
Material Base Major a l l o y i n g O
Materials form element elements £ 8
UJ
Weld 70/30 Vis d i a m . wire Copper Nickel (30%) Z
o
filler Copper-nickel 1
6
Q
metals Inconel 62 ' { 6 d i a m . wire Nickel C h r o m i u m (16%) LU
r-
Iron (8%) O
1 LU
2319 a l u m i n u m {ts d i a m . wire Aluminum Copper (6%) Lu 1

<
r- INCONEL
HY-80 steel J/2 i n . plate Iron Nickel (2.5%) LU 62
x
Incoloy Yi i n . plate Iron Nickel (34%) 2
LU 2319
C h r o m i u m (21%) I
ALUMINUM
Haynes 718 Y% i n . plate Nickel C h r o m i u m (20%) ALLOY
o
Iron (12%)
WELD FILLER WIRE
Hastelloy X % i n . plate Nickel C h r o m i u m (22%)
Iron (18%) Fig. 2—Relative amount of backfilling
Base Haynes R-41 % i n . plate Nickel C h r o m i u m (20%) produced when three different filler
metals Cobalt (12%) metals are deposited on HY-80 steel,
Hastelloy B Y% i n . plate Nickel M o l y b d e n u m (28%) welded under constant conditions
Iron (6%)
l
Nickel (99.58%) /i i n . plate Nickel None
430 s t a i n l e s s x
/i i n . plate Iron C h r o m i u m (17%) shown in each of the photomicro-
steel graphs deserves the following discus-
Ti-3AI-14V-llCr H i n . plate Titanium V a n a d i u m (14%) sion:
t i t a n i u m alloy C h r o m i u m (11%) Copper-Nickel Deposited on HY-80
Under 4% Augmented Strain (Fig.
3). The copper-nickel weld fusion
V 4 in. long region along the fusion affected zone crack backfilling en- zone is visible in the upper portion of
line were recorded. This procedure countered when 70/30 copper-nickel, the photomicrographic montage of
was adopted since it minimized the Inconel 62 and Type 2319 aluminum Fig. 3 as light etching material. Note
tedious nature of the measurements filler metal were deposited on HY-80 that a phase which etches in a similar
and still provided adequate data for steel under the application of 4 % manner exists not only in the heat-
comparative purposes only. augmented strain. The distinct differ- affected zone cracks as backfilled met-
ence in backfilling is also illustrated in al, but also appears as a penetrating
Evaluation of the Effect of Strain on phase along a considerable number of
Figs. 3-5 which are photomicrographs
Copper-Nickel-HY-80 Weldments HY-80 grain boundaries. It is also
of copper-nickel on HY-80, Inconel
The evaluation of the effect of 62 on HY-80, and aluminum alloy on important to note in Fig. 3 that a
strain was accomplished by preparing HY-80 specimens respectively. significant amount of cracking extends
a series of 5 weldments tested in the well beyond the maximum extent of
The copper-nickel filler metal re-
Varestraint device using die blocks of the backfilled regions.
sulted in the greatest amount of heat-
various radii to produce augmented The unique characteristics exhibited
affected zone crack backfilling. Incon-
strains of 0, V 4 , 1, 2, and 4 % . An in Fig. 3 may be accounted for in the
el 62 filler metal produced only a
identical series of Varestraint tests following way:
slight amount of backfilling and the
were conducted on the same heat of Regions adjacent to the molten
aluminum alloy filler metal generated
HY-80 steel using the autogenous gas weld puddle experience elevated tem-
no base metal penetration. The mor-
tungsten-arc process for comparison peratures causing a transformation to
phology of the resulting interactions
with the heterogeneous cracking re- face-centered-cubic (fee) austenite.
between weld metal and base metal
sults. During the application of strain by the
The parameters employed for the Varestraint device, atoms of an ele-
autogenous process were adjusted to ment present in the weld puddle com-
produce approximately the same ener- position rapidly attack the austenite
gy input as the gas metal-arc process grain boundaries by means of an ac-
used to make the heterogeneous celerated diffusion mechanism. The
welds. The autogenous specimens atomic specie is most likely copper
were examined metallographically and rather than nickel since the bulk dif-
the crack lengths existing in the heat- fusion coefficient of copper in iron is
affected zone were recorded. In this several orders of magnitude greater
fashion, the effect of the copper- than the bulk diffusion coefficient of
nickel filler metal on the heat-affected nickel in iron. The cohesive strength
zone hot cracking of HY-80 steel of the austenite grain boundaries is
could be compared to a reference sufficiently altered by the copper to
amount of hot cracking produced un- cause severe cracking under the appli-
der identical conditions without the cation of augmented strain. As an
presence of a foreign filler metal. open crack intersects the molten weld
puddle, the crack is backfilled with
Results and Discussion liquid copper-nickel filler metal alloy
Fig. 3—Photomicrograph of 70/30 cop-
Effects of Depositing Three Different quite readily because of favorable sur-
per-nickel deposited on HY-80 steel at
Filler Metals on HY-80 face energy conditions. Based upon
4% augmented strain—top view section.
Figure 2 is a bar chart which sum- Etchant: ammonium persulfate. X60 (re- the grain boundary wetting visible in
marizes the relative amount of heat- duced 35% on reproduction) Fig. 3, the interfacial energy of cop-

176-s | A P R I L 1971
 HY-80
10
I i
— ro
INC0L0Y
h- X 8
s: —
o V/
m w U A C T F I I nv HASTELLOY
& N 6 HAYNES HASTELLOY B
x
718 HAYNES
I- O
•z: w
§i2 4r
it<
R-41

n
LU I
Fig. 4—Photomicrograph of Inconel 62
deposited on HY-80 steel at 4% aug-
mented strain—top view section. Etch- 430 TITANIUM
ant: ammonium persulfate. XlOO (re- NICKEL STAINLESS ALLOY
duced 23% on reproduction) 'A
CATEGORY I BASE MATERIALS (F.C.C.I -CATEGORY H
(B.C.C.)
Fig. 6—Relative amount of backfilling produced when 70/30 copper-
nickel is deposited on various base metals

:' Steel (Fig. 4 ) . The Inconel 62 weld backfilling, provided both filler metals
fusion zone is visible in the upper were deposited using the same weld
portion of the photomicrographic parameters.
montage of Fig. 4 as a light etching 2319 Aluminum Filler Metal De-
material. The backfilling by light posited on HY-80 Steel (Fig. 5 ) . No
etching material into the base metal is penetration of aluminum alloy filler
also apparent, but occurred to a sig- metal is apparent into the HY-80
nificantly lesser degree than that ob- steel, visible in the lower portion on
Fig. 5—Photomicrograph of 2319 alu- served for the copper-nickel on HY- the photomicrograph in Fig. 5.
minum alloy deposited on HY-80 steel 80 welds made under the same condi-
at 4% augmented strain—top view sec-
Aluminum atoms are apparently
tions. unable to diffuse down the austenite
tion. Etchant: 5% hydrofluoric acid and
ammonium persulfate. X200 (reduced In this combination, the region ad- grain boundaries of the steel readily
23% on reproduction) jacent to the molten weld puddle is since no weakening or resulting crack-
also rendered temporarily austenitic ing of the heat-affected zone was ob-
per-nickel against austenitic steel during the peak portions of the weld served. This can be explained by not-
should be nearly equal to one-half the thermal cycle. Diffusion of an atomic ing that aluminum and iron form a
interfacial energy of a low alloy steel element unique to the Inconel weld series of very stable intermetallic
grain boundary in the austenitic state. puddle (probably nickel) occurred at compounds. The formation of these
Van Vlack 5 reports that the inter- a much slower rate than for that intermetallic compounds impedes the
facial energy of pure copper against observed for copper. Moreover, the diffusion process since the mobility of
gamma iron is 430 ergs/cm- at 1105° presence of an element such as nickel aluminum atoms in iron is sacrificed
C (2021° F) and the interfacial ener- at the austenite grain boundaries may as a result of the reactivity of the
gy of gamma iron grain boundaries is not adversely alter the cohesive aluminum specie.
850 ergs/cm- at 1105° C. While the strength. Those heat-affected zone It should be mentioned that, while
interfacial energy of liquid copper is cracks which do form and intersect the calculated heat input was the same
obviously altered by the presence of the weld puddle are almost completely when all three filler metals were de-
nickel, the possibility of good grain backfilled. However, the lengths of the posited on HY-80, the aluminum filler
boundary wetting by 70 Cu-30 Ni is backfilled regions are relatively small. metal has a much higher coefficient of
justifiable based upon the criteria set An explanation of the above condi- thermal conductivity and a significant-
forth in Fig. 1 and by the evidence tions can be rationalized in the follow- ly lower melting point than copper-
provided by Van Vlack. ing way: nickel or Inconel. The result is that
Backfilling of the heat-affected zone 1. The interfacial energy of Inconel much of the heat may have been
cracks proceeded as far into the open against steel is probably greater than carried off through the trailing solid-
crevices as permitted by favorable one-half of the interfacial energy of ified weld bead rather than through
surface energy conditions but was the austenite grain boundaries. This is the surrounding base metal. The ad-
halted at a point in the base metal substantiated by the significantly large jacent HY-80 heat-affected zone,
where the peak temperature of the dihedral angles at the tips of some of therefore, may not have experienced
weld thermal cycle is less than the the backfilled cracks visible in Fig. 4 as high a peak temperature as in the
solidus temperature of the copper- and by the limited extent of grain other two cases, which could also
nickel-iron filler metal alloy backfilling boundary wetting. account for the absence of cracking.
the crack. 2. The maximum peak temperature In general, however, it is interesting to
Since a separate program was un- distribution is known to fall off rapidly note that the relative difference in
dertaken to investigate the copper- away from the fusion line. Since the penetration by the three weld filler
nickel-HY-80 heterogeneous combina- solidus temperature of Inconel 62 is metals into the HY-80 steel grain
tion in greater detail, further discus- several hundred degrees higher than boundaries agrees surprisingly well
sion of this interaction is continued in 70Cu-30Ni, it is reasonable to expect with the observations made by Ros-
later sections. that the depth of Inconel backfilling tocker:i for liquid metal embrittle-
Inconel 62 Deposited on HY-80 should be less than copper-nickel ment couples.

WELDING RESEARCH SUPPLEMENT! 177-s

 According to Rostocker, liquid met- by Salter 0 in which increased nickel The response of the iron-base filler
al embrittlement is not observed in additions apparently decreased cop- metals to copper-nickel weld metal is
systems which form intermetallic per-induced hot-shortness in mild visible in Fig. 7 for Incoloy and is
compounds. This agrees with the ab- steel. also visible in Fig. 3 for HY-80 steel.
sence of base metal cracking in the Category II filler metals are com- The microstructures in both cases are
aluminum filler metal on HY-80 pletely body-centered-cubic at ele- characterized by severe hot-cracking,
weldment. Rostocker observed that vated temperatures. Cracking and a considerable backfilling of liquid weld
embrittlement by a liquid is more resulting backfilling action was virtu- metal, and an aggressive spreading
likely to occur when there is low ally non-existent in these materials nature of the copper-nickel through-
mutual solubility in the system. This welded with copper-nickel. Grain out the grain boundaries of the heat-
agrees with the observed difference in boundary wetting by liquid weld met- affected zone.
infiltration of copper-nickel and In- al was also generally absent. In Hastelloy X, a nickel-base super-
conel filler metals into HY-80 steel alloy shown in Fig. 8, the extent of
Discussion of Category I Materials.
since copper is relatively insoluble in copper-nickel infiltration is much
Figures 7-9 are photomicrographs of
iron, whereas nickel and iron can smaller than in the iron-base alloys.
heterogeneous weldments representa-
achieve complete solid solubility. Also the spreading of copper-nickel
tive of the materials in Category I.
The attack of copper-nickel weld met- into the grain boundaries appears to
The Effect of Depositing 70/30 Copper-
al is shown on Incoloy, Hastelloy X be less aggressive.
Nickel Filler Metal on a Variety of
Different Base Metals and pure nickel, respectively, in the In the copper-nickel on pure nickel
Figure 6 is a bar chart which sum- above figures. In each photomicro- weldment illustrated in Fig. 9, no
marizes the relative amount of graph, the copper-nickel fusion zone is observable grain boundary penetration
backfilling created when copper-nickel visible as the material along the upper was detected by metallographic tech-
filler metal was deposited on various portion of the figure. niques. Preferential grain boundary
base metals. The base metals are di- diffusion of copper into the pure nick-
vided into two categories depending el is probably minimized since the two
on the crystal structure of the alloy at elements are completely soluble in one
an elevated temperature. another which indicates that lattice
diffusion may be a competitive proc-
The base metals in Category I are ess. Furthermore, because of the
arranged from left to right in order of complete miscibility, the existence of
decreasing iron content and increasing copper, if present in the pure nickel
nickel content. Category I filler metals grain boundaries, could well serve to
are either austenitic (fee) or undergo strengthen the regions instead of em-
an allotropic transformation to a face- brittling them as postulated for the
centered-cubic structure in the vicinity copper-iron case. The absence of any
of the molten weld pool during weld- liquid penetration again agrees with
ing. It is interesting to note that as the Rostocker's'1 observation that liquid
iron decreases and the nickel increases metal embrittlement couples are sel-
in each Category I filler metal, there dom observed in metal combinations
tends to be a lesser amount of copper- which exhibit complete solid solubil-
nickel infiltration. This trend is similar ity.
to a series of experiments performed
Figure 10 reinforces the observa-
tion that nickel tends to impede grain
boundary penetration when using cop-
per-nickel filler metal. Figure 10 is a
photomicrograph of a side branching
region which is visible in the lower
center of Fig. 3. Here the copper-
r/-- ' 'j nickel has left the central backfilled
region of an HY-80 heat-affected zone

J crack and has wandered through the

r< ^ \
f A
V- C^7T-~/
base metal in a direction parallel with
the weld fusion line. In Fig. 10, a

7 > Q p £ • ~]
- m}

jr 17

<
Fig. 7—Photomicrograph of 70/30 cop- Fig. 8—Photomicrograph of 70/30 cop- Fig. 9—Photomicrograph of 70/30 cop-
per-nickel deposited on Incoloy at 4% per-nickel deposited on Hastelloy X at per-nickel deposited on nickel at 4%
augmented strain—top view section. 3% augmented strain—top view section. augmented strain—top view section.
Etchant: Flat's reagent. XlOO (reduced Etchant: 10% phosphoric acid. X200 Etchant: modified Wazau's reagent.
38% on reproduction) (reduced 17% on reproduction) XlOO (reduced 31% on reproduction)

178-s I A P R I L 1971
 , •rWB*-_->

7f^j^yrf7^7. • .:-.A*,:!-7r-
«"—••'*yAW"••»•<' . i i l i — j

', ' I • r\£, >

Fig. 10—Photograph of magnified por- Fig. 11—Photomicrograph of 70/30 cop-

tion of Fig. 4, showing copper-nickel per-nickel deposited on Type 430 stain-
side branching in HY-80 steel. 3-stage less steel at 4% augmented s t r a i n -
etchant: 1% picric-1% nital-3% sodium top view section. Etchant: modified
bisulfite. X200 (reduced 43% on re- Wazau's reagent. XlOO (reduced 35%
production) on reproduction)
" " • • * • • . 1

Fig. 12—Photomicrograph of 70/30 cop-

selective etching reagent was em- better the metals will wet each other. per-nickel deposited on Ti-3Al-14V-
ployed which is known to reveal While this hypothesis conflicts with HCr titanium alloy at 4% augmented
regions of nickel segregation in HY-80 Rostocker's observation of liquid met- strain—top view section. Etchant:
steels.7 The dark etching regions are al embrittlement couples, it is useful in Kroll's and modified Wazau's reagents.
depleted in nickel and the light etching showing that copper should tend to X150 (reduced 1 / % on reproduction)
regions are nickel enriched. The alter- wet fee iron better than bec iron. The
nating bands exhibiting nickel segrega- maximum solubility of copper in fee fee austenite grain boundaries is fa-
tion were created within the micro- gamma iron is approximately 9% vored because lattice diffusion is not a
structure during the processing of the while the solubility of copper in bec competitive process.
material from ingot to the final hot alpha iron is about 2 % . Thus, copper Another factor to be considered is
rolled plate. would be expected to wet austenite that alloying elements present in a
Attention is directed to the fact grain boundaries temporarily present base metal can markedly influence the
that the light etching copper-rich liq- in the low-alloy steels better _than it diffusion of a foreign element. Thus,
uid tends to penetrate along a dark- could wet the ferritic grain boundaries the high chromium concentrations in
etching (nickel-depleted) band in the of the iron-chrome steel. the ferritic stainless steel may affect
center of Fig. 10 rather than through Taylor 0 has calculated that the the rate of copper penetration into the
a light-etching (nickel enriched) mean interfacial energy of fee gamma ferrous grain boundaries.
region. This further indicates a dis- iron against liquid copper is some 35 Before leaving the subject, one last
tinct tendency for copper to penetrate ergs/cm 2 higher than the mean inter- attempt to explain the difference in
the nickel depleted regions in the ma- facial energy between bec alpha iron penetration exhibited by the two fer-
trix rather than nickel enriched re- and liquid copper. The system with rous alloys can be based upon charac-
gions, even when the difference in the lower interfacial energy is the one teristics associated with a form of
concentrations is relatively small. more favored to produce extensive liquid-metal embrittlement described
Discussion of Category II Materi- grain boundary wetting. However, it earlier. The penetration of the copper-
als. Figures 11 and 12 are photomi- could be noted that interfacial ener- nickel weld metal into the HY-80
crographs of heterogeneous weld- gies are altered significantly by alloy steel was postulated to begin with a
ments in which copper-nickel was de- additions. Thus, for the present, it rapid attack by copper atoms along
posited on Type 430 ferritic stain- must be assumed that the high chrome the austenite grain boundaries to
less steel and a titanium alloy respec- content in the ferritic stainless steel or cause the catastropic failure of these
tively. A very slight amount of grain the nickel content in the copper-rich stressed regions. This action is essen-
boundary penetration is visible in Fig. weld metal may serve to increase the tially a form of liquid metal embrittle-
12 for the copper-nickel on titanium interfacial energy values even further, ment. It is significant to note that
alloy combination. Absolutely no pen- thereby lowering the possibility of increasing the number of slip systems
etration was detected by metallogra- grain boundary wetting in the Type
in the solid material is reported to
phic techniques in the case of the 430 stainless steel.
decrease the severity of this form of
copper nickel on Type 430 stainless embrittlement. 1
However, the minor differences in
steel illustrated in Fig. 11.
wetting capabilities of copper on the The bec structure of the ferritic
It was discussed previously that a two crystal structures does not seem steel is known to possess 48 independ-
liquid copper-base alloy appears to to be significant enough to explain the ent slip systems. The fee austenite
have a special affinity for the grain dramatic differences in grain bound- present at high temperatures in the
boundaries of iron-base alloys. If this ary penetration of the two ferrous HY-80 steel is known to possess only
is the case, then a definite anomaly alloys completely. The rapid diffusion- 12 independent slip systems. Thus, the
exists since the grain boundaries of the al mobility of the attacking weld met- increase in slip systems present in the
ferritic stainless steel, with approxi- al specie should also be explored as a ferritic steel may improve its resist-
mately 84% iron, was not penetrated possible factor. ance to embrittlement by the liquid
by the copper-nickel weld metal. It is significant to note that there is copper-nickel weld metal.
Wassink, 8 in his paper describing less preferential grain boundary dif- The above attempts to account for
the wetting ability of liquid metals, fusion in the less densely packed bec the effect of crystal structure on the
states that the greater is the solubility materials than in the more closely infiltration of copper-rich liquid into
of metals in the solid state, the lower packed fee materials. This implies that ferrous base metal are by no means
will be the interfacial energy and the the mobility of copper atoms along conclusive, and further investigation

WELDING RESEARCH S U P P L E M E N T | 179-s

 RELATIVE AMOUNT OF BACKFILLING EXPERIENCED WHEN
INCONEL 62 IS DEPOSITED ON VARIOUS BASE MATERIALS

<
CO
- -
o o
M HAYNES
718

< u. HY-80
Ii. STEEL
UJ <
HAYNES
> h-
h- < INCOLOY HASTELLOY
< LU
_l x B HASTELLOY
X
or z 1 1 1 1 , .
BASE MATERIAL
Fig. 13—Relative amount of backfilling experienced when Inconel 62 is
deposited on various base metals
Fig. 14—Photomicrograph of Inconel 62
deposited on Haynes R-41 at 3% aug-
of this effect is certainly in order. superheat associated with the weld mented strain—top view section. Etch-
Russian investigators have also iden- pool has nowhere to go except ant: 10% phosphoric acid. X200 (re-
tified this phenomenon; however, to through the colder base metal im- duced 32% on reproduction)
date, their attempts to explain this mediately surrounding the weld pud-
effect have not been successful either. dle. This would cause a zone with- region extends from left to right
in the base metal at the fusion bound- across the mid-portion of Fig 14. It is
The Effect of Depositing Iconel 62 ary where complete melting occurs
Filler Metal on Various Base Metals important to note that the solidifica-
provided its melting range was near tion subgrain boundaries of the un-
The preliminary Inconel 62 hetero- that of the heterogeneous filler metal. mixed region etch in a manner identi-
geneous weldments were not pursued This melted region of the base metal cal to that of the material in the
in detail during this investigation since does not mix with the weld metal backfilled grain-boundary crack in the
the amount of backfilling was small since, even in the presence of the base metal visible in the lower por-
and in some cases non-existent. How- stirring action of the arc, a stagnant tion of the photomicrograph.
ever, Fig. 13 is presented to summar- layer of liquid exists at the solid-liquid
ize the relative amount of backfilling interface. Figure 14 thus lends metallographic
observed when Inconel 62 was de- support to the effect of the unmixed
Since an unmixed region of finite
region proposed above. Since the un-
posited on various base metals, but no width was observed to exist in a ma-
mixed region is located between the
attempt is made to correlate the ob- jority of the Inconel weldments, which dissimilar filler metal and any cracks
served results. also displayed little or no base metal in the base metal, the liquid from the
However, one interesting feature of penetration, one might expect that the unmixed zone is more likely to
this series of welds is worth noting. An unmixed region may serve to interfere backfill the newly formed crack.
"unmixed" region was observed to with the penetration of the grain
occur in a great number of the dissim- boundaries by the mechanisms previ- Of course, the composition of the
ilar weldments in which Inconel 62 ously proposed and illustrated in the metal on the unmixed zone is un-
was employed as the filler metal. A copper-nickel-HY-80 system. This can doubtedly modified by the diffusion of
typical "unmixed" region is visible in be rationalized by considering that elements from the molten filler metal.
Fig. 14 as a banded portion of the preferential grain boundary diffusion However, to date, no microprobe data
fusion zone adjacent to the base metal of a weld-metal element into the base are available for the specimen shown
which etches somewhat like the base metal is not likely to occur if there is in Fig. 14.
metal but exhibits a definite solidifica- a region of molten or semi-molten
tion substructure. It is interesting to base metal surrounding the weld met- The Effect of Strain
note that, throughout this program, an al. This is because the solubility of Figure 15 summarizes the effect of
unmixed region was observed in most elements is greater in the liquid increasing the amount of augmented
heterogeneous couples where the melt- state than in the solid state. Thus, the strain imposed by the Varestraint
ing range of the consumable elec- unmixed regions of the base metal, device on heterogeneous welds in
trode approached or exceeded the initially lean in the preferentially dif- which copper-nickel was deposited on
melting range of the base metal. The fusing weld-metal element or elements, HY-80 steel.
"unmixed" region was observed with may serve to "soak up" the offending The total height of each bar indi-
the greatest frequency in the Inconel specie and minimize preferential grain cates the maximum crack length,
62 heterogeneous weldments, since In- boundary penetration of the solid base while the height of the crosshatched
conel 62 has a liquidus equal to or metal. portions of each bar indicate the max-
greater than that of most of the base Figure 14 is a photomicrograph of imum penetration by the backfilled
metals upon which it was deposited. a heterogeneous weldment between weld metal. The height of the dark,
The creation of an unmixed region Inconel 62 and Haynes R-41 base solid portions of the last three bars
in the fusion zone under these condi- metal subjected to 3 % augmented indicate the maximum crack length in
tions is not difficult to understand strain. The Inconel 62 fusion zone the heat-affected zone of the auto-
when one considers that the bulk of appears as a light etching phase along genous welds made on the same heat
the latent.heat of fusion of the weld the extreme top of the photomicro- of HY-80 using the same heat input,
metal in addition to the tremendous graph. The darker etching unmixed but without the addition of filler met-

180-s I A P R I L 1971
 OPEN CRACKS IN HETEROGENEOUS
CU-NI-HY-80 WELDMENTS
^j
o
l~
<
rr
i-
O
X

Ld I I I
m WETTED GRAIN BOUNDARIES OR
BACKFILLED CRACKS

Z 7" OPEN CRACKS IN HY-80

HI
LL <)
KI AUTOGENOUS WELDS
? Q
~
2
> LU
O
X IH
li_
< LL
5
LL I -
<
u 5 *?
IN
T
o III
;<*
d 1
H- H-
(1)
Q ^ I WSJ

1/4 % I% 2% 4%
AUGMENTED STRAIN
Fig. 15—Distance of maximum penetration of grain boundary wetting, Fig. 16—Photomicrograph of 70/30 cop-
backfilling and cracking as a function of strain per-nickel deposited on HY-80 steel at
0% augmented strain—top view section.
Nital etch. X200 (reduced 29% on re-
al. The height of the crosshatched affected zone hot cracks in the hetero- production)
bars for the two lowest strains rep- geneous welds were approximately 10
resent the maximum extent of grain times the length of the longest heat- yond the maximum extent of liquid-
boundary wetting since hot cracking affected zone cracks in the autogenous metal backfilling.
was not induced at augmented strains welds. (Compare height to top of Figure 17 is included as a closing
below 1 %. open bars to height of solid bars in argument to support the hypothesis
A number of interesting features Fig. 15.) Figure 15 also shows that in that the high temperature embrittle-
associated with this bar chart should the heterogeneous welds subjected to ment and the backfilling of hot-cracks
be noted. Attention is drawn to the strains greater than 1%, the max- which occur when cupro-nickel filler
fact that significant amount of grain imum crack length extends well be- metal is deposited on HY-80 base
boundary wetting resulted even in the metal are separate phenomena. This
absence of applied augmented strain micrograph was taken of a Vare-
( 0 % strain bar in Fig. 15). Figure 16 straint specimen in which the molten
is a photomicrograph which shows the weld pool inadvertently wandered
ability of copper-nickel to wet the away from the side of the weld
grain boundaries of HY-80 steel weld- shown just prior to the application of
ed without the application of aug- the augmented strain.
mented strain. Note the thin light The fact that a hot-crack propa-
etching network along the grain gates across the HY-80 interface into
boundaries extending downward from the weld metal at the top proves that
the fusion line through the center of a layer of solid weld metal existed at
this figure. While shrinkage stresses this interface at the instant the aug-
were undoubtedly present in this spec- mented strain induced the cracking.
imen, they appear to have been of Had molten weld metal been in con-
insufficient magnitude to cause crack- tact with the HY-80 interface at this
ing of the wetted grain boundaries and instant:
subsequent backfilling by the liquid 1. The cupro-nickel from the mol-
weld metal. ten weld pool would have backfilled
As mechanical stresses of greater the cracks as shown previously in Fig.
magnitude were introduced to the out- 3.
er fibers of the plate by the Vare- 2. The hot cracking would have
straint apparatus, the amount of grain terminated at the HY-80 interface
boundary penetration increased. Ulti- rather than extending into the cupro-
mately, at strains of 1% or more, hot nickel weld metal. (In other words,
cracking of the copper-permeated one cannot crack a liquid.)
grain boundaries occurred. It is inter- Thus, the fact that the molten pool
esting to note that the threshold was not in contact with the HY-80 at
amount of augmented strain necessary the instant the cracks formed explains
to produce open hot-cracks in the the absence of backfilling in Fig. 17.
heterogeneous welds (open bars) was However, embrittlement of the
the same as the threshold strain re- HY-80 base metal by penetration of
quired to produce heat-affected zone Fig. 17—Photomicrograph of the heat
the austenite grain boundaries by
hot cracking in the autogenous welds affected zone of a Cu-Ni-HY-80 speci-
men in which the weld bead accident- some component derived from the
(solid bars). Once the augmented filler metal obviously occurred in the
strain was increased to 1 % or more, ally strayed to the opposite side of the
groove—top view section. Nital etch. sample shown in Fig. 17. The cracks
the maximum length of the heat- extend nearly 0.125 in. (about 6 in. at
X50 (reduced 29% on reproduction)

WELDING RESEARCH SUPPLEMENT I 181-s

X50 magnification) from the fusion upper and lower critical temperatures 3. The application of external ten-
line. This is approximately a 10 fold a mixed structure of ferrite and aus- sile stresses increased the tendency for
increase in crack length as compared tenite coexist, while above the upper the liquid copper base alloy to pene-
with a Varestraint specimen subjected critical temperature the steel is fully trate the grain boundaries of ferrous
to the same strain but made with a gas austenitic. Thus, it seems reasonable alloys.
tungsten-arc bead-on-plate weld with that the penetration of the grain 4. Heterogeneous welds made with
no filler metal addition. boundaries extended from the fusion 70 Cu-30 Ni filler metal on HY-80
It is postulated, therefore, that cop- line to a point in the inter-critical steel result in significantly more heat-
per from the filler metal selectively region of the heat-affected zone where affected zone fissuring than autogen-
diffuses from the weld pool into the significant amounts of untransformed ous welds made under the same condi-
austenite grain boundaries in the heat- ferrite remained at the peak tempera- tions without the addition of filler
affected zone of the HY-80, causing ture. metal. The fissuring is due to the pres-
significant reduction of the cohesive ence of copper atoms and is not di-
strength of these boundaries. The ap- Conclusions rectly caused by the backfilling action
plication of the augmented strain then A number of conclusions can be of liquid metal.
causes hot cracking to occur along the drawn based upon the results of this 5. The backfilling mechanism is a
weakened grain boundaries. Although investigation. three step process which follows a
the exact mechanism remains in The following conclusions pertain definite sequence:
doubt, Fig. 17 demonstrates conclu- generally to the backfilling phenomena (a) Rapid attack of copper atoms
sively that extensive impairment of encountered in heterogeneous con- down the base metal grain
the elevated temperature strength of sumable-electrode arc welding: boundaries.
the austenite boundaries occurs during 1. The backfilling mechanism ap- (b) A decrease in the grain
the brief interval of contact with the pears to be inhibited by the presence boundary cohesive strength
molten pool prior to the application of of an unmixed zone at the edge of the resulting in the formation of
the augmented strain. The observed fusion zone which is produced when hot cracks in a strain field.
behavior proves that backfilling by the melting range of the consumable (c) The backfilling of these cracks
molten metal from the weld pool must electrode is equal to or greater than by liquid metal from the weld
occur after the hot cracks are formed, the melting range of the base metal. puddle, providing favorable
while the reduction in the cohesive 2. A lesser amount of backfilling of interfacial energy conditions
strength of the boundaries must occur heat-affected zone cracks is produced exist.
prior to the hot cracking. Thus, as by filler metals of higher melting References
postulated, the two are separate points than by filler metals of lower
phenomena. 1. Stoloff, N. S., "Liquid Metal Embrit-
melting point, all else being equal. tlement," Reprinted Irom Surfaces <£ Inter-
Finally, it should be noted that the 3. A dissimilar weld filler metal is faces, Vol. II, Syracuse University Press,
Syracuse, N. Y., 1968.
point in the heat-affected zone of the more likely to penetrate the grain 2. Smith, C. S., "Grains, Phases, and
HY-80 at which the high temperature boundaries of a base metal if the two Interface: An Interpretation of Microstruc-
embrittlement ceases corresponds ture," Trans. AIME, 175, 1948.
major elements have low mutual solid 3. Rostocker, W.. McCawghey, J. M. and
closely to the point where the peak solubility and do not form any inter- Markus, H.. Embrittlement by Liquid
temperature corresponded to the up- metallic compounds. Metals, Reinhold, New York (1960).
4. Savage, W. F., and Lundin, C. D.,
per critical temperature. Note in Fig. The following conclusions pertain "The Varestraint Test," WELDING JOURNAL,
17 that the dark etching heat-affected specifically to the infiltration of cop- Research Suppl., pp. 433-s to 442-s (1965).
5. Van Vlack. L. H., "Intergranular En-
zone is bordered by a lighter etching per-base weld filler metal into the ergy of Fe and Some Fe Alloys," Journal
band of base metal near the bottom of surrounding base metal: of Metals, 3 (3), March 1951.
the photomicrograph. Metallograph- 6. Salter. W. J. M., "Effects of Alloying
1. Grain boundary wetting by a Elements on Solubility and Surface Energy
ic examination revealed that this liquid copper base alloy tends to occur of Copper on Mild Steel." Journal of the
change in etching behavior corre- Iron and Steel Institute, 204 (5). May 1966.
in alloys which are austenitic or tem- 7. Szekeres. E. S., "A Study of Weld-
sponds to the locus of points where porarily austenitic during welding. Interface Phenomena and Associated Crack
the maximum temperature just The higher the iron content and the Initiation in a Low-Alloy Steel," Ph.D.
Thesis, Department of Materials Engineer-
reached the lower critical (or Ax) lower the nickel content, the more ing, Rensselaer Polytechnic Institute, Troy.
temperature. Note that the cracks likely is grain boundary penetration. New York, February 1968.
stop just short of this line of demarca- 8. Klein Wassink, R. J.. "Wetting of
2. A liquid copper-base alloy does Solid-Metal Surfaces by Molten Metals,"
tion. This is consistent with the previ- not appear to penetrate a body- Journal of the Institute of Metals, 95
ous observations regarding the relative (1967).
centered-cubic crystal structure in the 9. Taylor, J. W., "An Evaluation of In-
ease of penetration of ferrite and aus- region adjacent to the weld puddle terface Energies in Metallic Systems,"
tenite grain boundaries. Between the Journal of the Institute of Metals, 86 (10),
during welding. June 1958.

Order now-slip cases for your Welding Journals •

• Each case holds 12 issues (yearly volume of the Journal). • Made from finest quality binders board—covered with washable
Stands upright. Journal issues slip in and out easily. simulated leather.

• Black sides, with back in Decorator's red. Title and AWS symbol • Available from Welding Journal at $3.50 each. (Price outside
imprinted in 23K gold. Gold foil provided to enable user to in- USA or its possessions—$4.50 each. Add 6 % sales tax on
sert year and volume number within seconds. New York City orders. Allow 3 to 4 weeks for delivery.)

182-s | APRIL 1971