Professional Documents
Culture Documents
WJ 1996 10 s330 A.G.Fox
WJ 1996 10 s330 A.G.Fox
ABSTRACT. The effect of small differ- metal, the composition of the flux used the alloying elements such as carbon,
ences in flux basicity on the microstruc- during welding and the cooling rate of nickel, chromium, copper and niobium
ture and mechanical properties of sub- the weld metal during the transformation are present to give the required strength
merged arc weld metal of Navy HY-100 of the undercooled metastable austenite. levels, and a fortuitous outcome of their
steel has been investigated. It was found The strength and toughness of weld metal presence is a continuous cooling trans-
that an increase in the flux basicity index improves as the amount of acicular fer- formation (CCT) diagram with features
from 2.5 to 3.0 (calculated by the formula rite increases due to its fine "basket that mean that bainite (or, more correctly,
proposed by Tuliani, Boniszewski and weave"microstructure, and so it is im- fine mixtures of lath ferrite, lath marten-
Eaton, 1969, Weld. Met. and Fab. 37 (8): portant to understand the mechanism of site and both interlath and "blocky" re-
327-329) led to a decrease in the total its formation so that the volume fraction tained austenite, which are best de-
oxygen content of the weld metal from of acicular ferrite can be maximized. This scribed as granular ferrite in low-carbon
0.034 to 0.027 wt-%. This decrease in topic has been extensively studied. For steels since carbides characteristic of bai-
weld metal oxygen content led to in- recent reviews see Refs. 1-4. However, nite are usually absent) is the major trans-
creasing average nonmetallic inclusion as yet, the welding conditions appropri- formation product in the fusion zone dur-
size and an increased propensity for aci- ate to maximizing the amount of acicu- ing submerged arc welding of these
cular ferrite formation, which were found lar ferrite in Navy HY-100 steel during steels, provided suitable heat inputs and
to increase both the strength and tough- submerged arc welding have yet to be es- preheat/interpass temperatures are cho-
ness of the weld metal. These results sug- tablished. sen during multipass welding. Since aci-
gest that careful control of the chemistry The chemical composition of the filler cular ferrite is also nucleated at similar
of both the filler metal and flux is neces- metal principally determines the final weld metal cooling rates (Refs. 7, 8),
sary to produce consistently high composition of the weld metal, although some workers effectively describe it as
strength and toughness in weld metal the composition of the base plate is im- intragranularly nucleated bainite (Refs.
during the submerged arc welding of HY- portant because of dilution effects. Also, 1,9, 10). It would appear that controlling
100 steel. there may be transfer of elements from the weld thermal cycle so that conditions
the flux (Refs. 5, 6). In high-strength steels are favorable for acicular ferrite nucle-
Introduction ation is not a problem, although bainite
(granular ferrite) and martensite are usu-
The origin of acicular ferrite in the fu- ally also formed. This subject will be dis-
sion zone of submerged arc weldments KEY WORDS cussed again later.
on high-strength steels is very complex Once a suitable weld metal harden-
and depends upon the chemical compo- HY-100 ability and cooling rate have been estab-
sition of the steel base plate and filler Submerged Arc Welding lished, the amount of acicular ferrite nu-
CCT Diagram cleated will depend upon oxygen
A.G. FOX and M.W EAKESare with the Cen- Weld Metal Toughness content and the size, number, distribu-
ter for Materials Science and Engineering, tion and chemical composition of the
High-Strength Steel
United States Naval Postgraduate School,
nonmetallic inclusions present (Refs.
Flux Basicity
Monterey, Calif. G. L. FRANKE is with Code 1-4, 6-63, 69). Appropriate inclusions
Filler Metal
615 Welding Branch, Annapolis Detachment,
appear to be in the size range 0.2-2.0 pm
Acicular Ferrite
Carderock Division, Naval Surface Warfare
with a mean size of about 0.4 IJm being
an optimum value (Refs. 1,2, 34). These
Center, Annapolis, Md.
30- 30 -
tno i'o- 1
I I I I I I I I l I l I I -- illi |
0.3 0-~ 0.7 0.9 1.1 1.3 1.5 0.~ 0.5 0.7 0.9 1.1 1.3 1.5
Inclusion size (microns) Inclusion Size (microns)
40 40
F293 average size = 0.322 F295 average size = 0.323 pm
30 - 30 -
qJ 20 - ~ 20 -
10- lo-
0 , ~--~' T 0 , , , , '7 , ,
0 0.3 0.5 0.7 0.9 1.1 1.3 1.5 0.3 0.5 0.7 0.9 1.1 1.3 1.5
liinoooo
(SAW) flux should be such that the final reduce prior
weld metal total oxygen content is austenite grain
around 200 to 300 ppm if large amounts size so that grain
of acicular ferrite are to be nucleated boundary-nucle-
10-
(Refs. 28, 63). This, of course, can be cor- ated microcon-
related with the comments concerning stituents form
the size, distribution and composition of rather than acicu-
the oxide inclusions made previously. lar ferrite (Refs. 1, 0 , I ! I I
.
I
. .
I
. .
I
Lower levels of total oxygen (< 200 ppm) 2, 32). Total oxy- 0 0.3 0.5 0.7 0.9 1.1 1.3 1.5
mean that insufficient inclusion numbers gen contents
are generated for the formation of acicu- around 250 to 300 Inclusion size (microns)
lar ferrite so that other microconstituents ppm are routinely
such as bainite or Widmanst~tten ferrite achieved in sub-
are formed preferentially. Increasing total merged arc weld- Fi,~. 5 - - Inclusion size distributions for all samples.
334-s I O C T O B E R 1 9 9 6
0.45
0.425 U
0
0.4-
o I O
i-
0.375 -
N
°~
= 0.35
.o
0.35 -
0.325
0 0 []
0.3 , , ,
0.3 I I I I 0 500 1000 15J00 2000 2500
250 275 300 325 350 375
Weld metal oxygen content (ppm) Weld oxygen/Inclusion volume fraction (ppm x I00)
Fi~. 6 - - Inclusion size vs. weld metal oxy,~en. The correlatk~n coet L Fi,~. 7 - - Inclusion size vs. normalized weld metal oxy,~ten (weld metal
ficient, t, tor the best linear fit is indicated (0.356) and it is c/ear that oxy,~en/inclusion vc~lume #action).
there is no linear conelation between these two variables, as discussed
in the text.
him and for each weldment two tensile Table 6 - - Charpy V-Notch Fracture Appearance Data for the Five Weld Metals, the
tests were performed using a Instron ten- Percentage Brittle Fracture (%BF) and the Temperatures of Testing
sile testing nlachine. Standard (10 mm 2
cross-section) Charpy samples were cut F289 F292 F293 F295 F296
so that the 2 - m m - d e e p 45-deg notch
w o u k i progress through the center of the %BF at 95 100 100 100 95
-151 °C
weld metal in the welding direction dur- %BF at 90 100 100 90 80
ing testing and three tests were per- _118oc
formed at each of the temperatures %BF at 67.5 87.5 83.3 63.3 60
shown in Table 5. The results of the room 84°C
temperature tensile testing (averaged for %BF at 44 46 54 53.3 30
the two tests for each sample) are shown -51 °C
in Table 4 and the Charpy impact testing %BF at 30 14 28 36.7 10
(averaged for the three tests at each tenl- -18°C
perature) in Table 5 together with the mil- %BF at 5 (1 12.5 10 3
16°C
itary specifications for the mechanical %BF at 1.67 0 1.67 1.67 2.5
properties of Navy HY-100 steel. The 49oC
fracture appearance data resulting from %8F at 0 0 0 0 0
the Charpy tests is given in Table 6 (per- 82°C
cent brittle fracture) and the fracture ap- %BF at 0 0 0 0 0
pearance transition temperatures (FATT) 116°C
(50% brittle fracture) and the ductile-to-
brittle transition temperatures (DBTT)
( m i d p o i n t between upper and l o w e r Table 7 J Charpy V-Notch Transition Temperature Data for the Five Weld Metals Calculated
from the Data in Tables 5 and 6
shelves) deduced from Tables 5 and 6 are
presented in Table 7.
F289 F292 F293 F295 F296
Cross-section samples of each weld-
ment were cut and polished down to one DBTT (C) -64 -65 -62 -42 -78
I]m diamond and areas corresponding to FATT (C) -54 -54 -49 -47 -73
the positions from which the mechanical
properties test pieces were machined
were identified. For subsequent micro-
scopical examination every attempt was Table 8 - - Average Inclusion Composition (atomic %) for each Sample Determined by Energy
made to examine those regions where Dispersive X-ray Analysis in the Transmission Electron Microscope
the Charpy notch was likely to have been
(i.e., in the center of the weld metal). F289 F292 F293 F295 F296
These areas were examined in the un-
Mn 38.40 25.05 36.95 46.48 38.42
etched condition in a Cambridge $200 Ti 8.41 12.16 5.68 5.42 7.96
s(anning ele{tron microscope (SEM) op- AI 28.69 50.87 25.15 20.29 29.77
erating in the backscattered imaging Si 23.08 9.77 30.54 26.87 21.9(1
Zr 1.41 2.18 1.69 0.94 1.95
2.8 - 2.8"
t~
2.7 - 2."/-
[] []
[]
2.6 . , , , 2.6
250 275 300 325 350 375 0 $00 1000 15100 2(;00 2500
Weld oxygen/Inclusion volume fraction (ppm x 100)
Weld metal oxygen content (ppm)
Fig. 8 - - Basicity index vs. weld metal oxygen. There is no linear cor- Fig. 9 - - Basicity Index vs. normalized weld metal oxygen.
relation between these two variables (r = 0.294), as discussed in the text.
mode at 20 kV in order to determine the content for each weld metal is given in examined in a JEOL 100 CX transmission
size distribution and volume fraction of Table 3. The balance of the microstruc- electron microscope (TEM) equipped
the nonmetallic inclusions. This was ac- ture in each case was found to be pre- with a Kevex Quantum light element x-
complished by counting and sizing the dominantly bainite, which is consistent ray detector. This allowed a quantitative
inclusions in 100 random fields observed with other microstructural observations chemical analysis of the elements with
at a magnification of 7040 X with a work- on high-strength steels containing acicu- atomic number greater than 11 in the in-
ing distance of 9 mm. Use of the SEM in lar ferrite (Refs. 1, 43, 50). Attempts to clusions and a qualitative appraisal of the
this way allowed inclusions as small as modernize this procedure using ad- light elements present. For each sample,
50 nm to be included in the count. A typ- vanced image processing software forty inclusions over the whole size range
ical inclusion field is shown in Fig. 1. The proved difficult since the boundaries be- were chemically analyzed and very little
samples were then etched with 5% Nital tween the microstructural constituents variation in the chemical compositions
and optical micrographs were taken from were not well defined. After optical ex- within each set of forty was found.
the same areas from which the inclusion amination, the etched samples were
counts were taken - - Fig. 2. From these coated with about 20 nm of carbon in a Results and Discussion
the amount of acicular ferrite was deter- Fullam EFFA evaporator and then etched
mined by the linear intercept method ap- again (with Nital) to obtain extraction From the flux compositions shown in
plied to ten fields taken at a magnifica- replicas from the same areas which had Table 3, the basicity indices (BI) of each
tion of 500X with ten random lines been examined optically and by SEM; flux were calculated by putting the
analyzed per field. The acicular ferrite these were placed on copper grids and weight percentages of each flux con-
stituent into the following formula (Ref.
64):
CaO + C a F 2 +MgO + N a 2 0 +
BI= K20+Li20+O'5(MnO+FeO)
Si02 +0"5( AI203 +Ti02 +Zr02) (1)
It should be noted that F293 contained
very little MnO compared with the other
fluxes and, as discussed by Palm (Ref. 5),
low flux MnO can lead to Mn losses in
steel weld metal. This effect does not
seem to depend on flux basicity. To avoid
such losses in Navy HY-100 steel about
1.0 wt-% MnO is needed in the flux for-
mulation. However, it should be noted
io that excessive additions of MnO to the
TIME ( SECONOS) flux can lead to increased levels of man-
ganese in HY-100 weld metal which may
Fig. 10 - - CCT diagram for AX90 and AX110 filler metals (Ref. 67). be undesirable (Ref. 68).
The overall chemical analyses of the
O Ti
plotted in Fig. 3A and B. Since oxygen sion size is impor-
activities are very high in SAW weld tant and, accord- C 30- [] O • O Zr
metal, manganese also takes part in the ing to Fleck, et al. • • O A
• Q • Si
deoxidation process. However, the over- (Ref. 32), the in- ~ 20-
all Mn content of the weld metal seems clusion size in .~ • Mn
to be little affected by oxygen content as weld metals "~
n lO-
shown in Fig. 3B, although sample F293 should decrease ..~ 0 0
O0
has a low Mn content due to the low as the oxygen " 0 0 0 Oo
'~ 0
MnO content of the flux. The Zr, AI, Ti content increases ~ As 375
and Si contents all appear to decrease in submerged arc
with increasing oxygen content (with the weld metal. The Weld metal oxygen (ppm)
exception of F296, which appears to inclusion size dis-
have AI, Ti and Zr contents somewhat tributions for all Fig. 12 - - Inclusion Mn, Si, AI, Ti and Zr vs. weld metal oxygen.
higher than its oxygen content of 320 the samples are
ppm would suggest) and this occurs since shown in Fig. 5
increasing the oxygen content leads to and the average
increased consumption of deoxidants in inclusion size is
the molten state to form oxides. Some of plotted vs. oxygen
these deoxidation products float up into content in Fig. 6.
the slag while others remain as non- Between 270 and
metallic inclusions. 300 ppm oxygen
The nature of the size, distribution the average inclu-
and chemical analysis of the nonmetallic sion size drops
inclusions in the weld metal proved very from 0.38 pm to
complex. For example, the inclusion vol- about 0.32 lam
ume fraction should be expected to in- and then stays J _ - ~ - ° ' ~ T" " " . \ w,,.J~,*~t,
70- 70-
mm
[]
w
N
60-
o,m
50-
5o-
40-
40-
30-
[]
30-
I-1
20 ! !
Fig. 14 - - A c i c u l a r ferrite content vs. w e l d metal oxy,~en. Fig. 15 - - A c i c u l a r ferrite content vs. n o r m a l i z e d w e l d m e t a l oxygen.
80 80
70
60
o/ 70-
6o-
r-i
40- 40-
I
.2
0-
[]
30 X
ra
°
20 I I I I 20 , , ,
0.3 0.325 0.35 0.375 0.4 0.425 2.6 2.7 2.8 2.9 3
Fig,. 16 - - A c i c u l a r ferrite content vs. inclusion size. Fig. 17 - - A c i c u l a r f~rrite content vs. basicity index.
suming F296 to have an inclusion vol- weld metal oxygen content by dividing it work the value of A for the weld metal as-
ume fraction which is an average of the by inclusion volume fraction. The rea- sociated with F296 is very different than
other four samples (0.00187) predicts the soning for this comes from a simple ex- F289 through F295. However, we can
oxygen content of F296 to be 290 ppm. pression relating oxygen content (wt-% normalize with respect to A by ignoring
This suggests that if the float-out rate in O) and oxide inclusion volume fraction, the dissolved oxygen, (wt-% O) dis-
F296 had been similar to the other sam- Vf, in the weldments which can be of the solved, (this is usally small compared
ples then its oxygen content would have form with the amount present as inclusions)
been of this order. This is consistent with wt-% O = AVf + (wt-% O)dissolve d (2) and divide wt-% O by Vf. This is impor-
the weld metal total AI, Zr, Ti and Si con- where A should be a constant if the weld- tant since acicular ferrite is nucleated by
tents for F296 discussed earlier. A simple ing conditions for all the weldments stud- inclusions and the weld metal acicular
way of dealing with the effect of these er- ied are the same and the sulfur in the in- ferrite content will increase with increas-
ratic float-out rates is to normalize the clusions is ignored (Ref. 1). In the present ing inclusion volume fraction and de-
338-s I O C T O B E R 1996
crease with decreasing average inclLision
size (and therefore increasing oxygen
content according to Fig. 6). As a result,
• DBTT
a plot of inclusion size vs. this normal- • YS
ized oxygen content (Fig. 7) shows ex-
cellent correlation between these two
-40 850
parameters. This scrutiny of inclusion
size is important because Fig. 5 clearly
shows that reductions in average inclu-
sion size arise as a result of increasing G" -so "'", •
numbers of inclusions with sizes less
than 0.2 tim. As discussed by Fleck, e t a l . e~_ %%%
(Ref. 32), and Liu and Olson (Ref. 34), it
is these small inclusions that are appar-
ently responsible for pinning prior
-60 750 ~
austenite grain boundaries leading to
• / ~'%%.
smaller austenite grain size and thus re-
duced incidence of intragranular acicu-
-7o " """'/ -1oo
lar ferrite.
Although the basicity of the welding
flux is a very subjective measure of the 65O
likely oxygen activity in the weld metal, -80 3; 4; 5; 6; 7; 80
the weld metal oxygen should, in princi-
ple, decrease as the basicity increases. % Acicular Ferrite
However, examination of the work of Tu-
liani, et al. (Ref. 64), and Dallam, et al.
(Ref. 63), suggests that such a variation is Fi,~. 18 Yield strength and ductile/brittle transition temperature vs. acicu-
small and difficult to determine for ba- lar t~rrite content.
sicity indices greater than about 1. This is
no doubt because of the scatter in the
weld metal oxygen contents and inclu-
sion volume fractions discussed previ-
with appropriate size distribution and steels since the intragranularly nucle-
ously. This is illustrated in Fig. 8, which
is a plot of basicity index (BI) vs. weld chemical composition, the possible for- ated ferrite in HY-IO0 and similar steels
metal oxygen for the welds studied in the mation of acicular ferrite depends on the has a much smaller aspect ratio and a
present work, and it is clear that there is nature of the CCT curve for the welding finer morphology than that observed in
very little correlation between these vari- wire and the cooling rate through the lower strength varieties as shown in the
ables, with sample F296, which has the austenite transformation region. Simple optical and transmission electron micro-
abnormally high inclusion volume frac- calculations based on the approximate graphs of Figs. 2 and 11. The TEM with
tion, really going against the trend of de- equation developed by Christensen and EDX chemical analyses of the weld
creasing BI with increasing weld metal Simonsen (Ref. 65) metal inclusions in the present work in-
oxygen. However, a plot of BI vs. the nor- dicate that these contain MnO, Al203,
malized oxygen content (Fig. 9) shows a At8 5 = 5tiE (3) SiO2, TixOy (x and y unknown) and a
high degree of correlation; this finding small amount of ZrO 2 with occasionally
suggests that it is important to take into a trace of MnS. The average inclusion
account in some way the erratic float-out indicate that the time for the weld metal composition for each sample (excluding
rate of the nonmetallic inclusions in to cool from 800°C to 500°C, At 8 5, for oxygen) is given in Table 8. Careful
order to develop a quantitative relation- the weldments studied in the present scrutiny of the light element (atomic
ship between welding parameters, flux work is around 1] s. Liu and Indacochea number, Z < 11 ) region of the energy dis-
basicity, inclusion oxygen content, total (Ref. 66) have indicated that 15 s is ideal persive x-ray spectra of the inclusions
oxygen content and, ultimately, acicular for nucleating optimum amounts of aci- failed to detect the presence of any ni-
ferrite content. Figure 9 also suggests that cular ferrite in HSLA steels. In Equation trogen. A plot of Mn, Si, AI, Ti and Zr in
the oxygen content of the inclusions is in- 3, 11 is the welding efficiency (N0.95 for the inclusions is shown in Fig. 12, and it
creasing with increasing basicity index, SAW) and E the gross heat input in kJ can be seen that the amount of strong de-
which would imply that oxides such as m m J. Figure 10 (Ref. 67), which shows oxidants (AI, Zr and Ti) decreases with
TiO 2 and AI203 are increasing at the ex- the CCT behavior of AX110 weld metal increasing oxygen content while the
pense of oxides such as MnO as the oxy- and which is very similar to that used in amount of the weaker deoxidants (Si and
gen content decreases. This is difficult to the present work, indicates that a cool- Mn) increases. This is consistent with the
prove because of the problem of quanti- ing rate of 11s between 800°C and fact that, as the oxygen activity rises,
fying the oxygen content of the inclu- 500°C would lead to the formation of weaker deoxidants such as Mn and Si are
sions by x-ray analysis in the TEM, as bainite in the absence of appropriate in- more likely to take part in the deoxida-
pointed out previously. However, as dis- clusions for acicular ferrite formation. tion process. Once again, F296 appears
cussed in the next paragraph, there is Consequently, it would seem that the to have more AI, Ti and Zr in the inclu-
clear evidence for an increase in inclu- weld thermal cyle for the fusion zones of sions than would be expected for an oxy-
sion MnO content with increasing weld the weldments studied in the present gen content of 320 ppm. This suggestion
metal oxygen. work is close to ideal for acicular ferrite reinforces the assertion that the oxygen
As indicated in the introduction, in formation. The term acicular ferrite may, content of F296 weld metal would have
addition to the presence of inclusions in fact, be a misnomer for high-strength been somewhat less than 300 ppm if the
342-s I O C T O B E R 1996