Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
The influence of submerged-arc flux composition on the inclusion morphology and weld
metal microstructures of low carbon steels is investigated. Systematic weld oxygen varia-
tions and changes in inclusion shape and size are obtained by changing the welding flux
composition. The influence of inclusion content, morphology, and distribution on achiev-
ing specific types of weld metal ferrite is described.
with the inclusion volume fraction for all the flux systems
Table II. Welding Process Parameters studied in this investigation. Since the oxygen in the weld
metal is dependent on the quantity and stability of the
Voltage 33 volts 31 volts 28 volts oxide constituents of the flux, the welds made with the
Current 600 amps 330 amps 550 amps SiO2-MnO-FeO fluxes showed larger quantities of oxygen
Travel speed 12.2 in/min 12.2 in/min 11.88 in/min
Wire speed 75 in/rain 40 in/min 70.0 in/rain with larger additions of FeO because iron oxide is more
Heat input 97.4 kJ/in 50.3 kJ/in 77.8 kJ/in unstable than manganese oxide. In the case of the
3.83 kJ/mm 1.98 kJ/mm 3.0 kJ/in SiO2-MnO-CaF2 system, the behavior is reversed due to
the fact that manganese oxide is being replaced by a non-
oxygen bearing component, calcium fluoride. Figure 5
INCLUSION CHARACTERIZATION shows the variation of oxygen in the weld metal in terms of
flux composition.
As-polished weld samples were examined for inclusion Along with changes in the morphology and the volume
shape, size distribution, and volume fraction. It is observed fraction of the inclusions, a refinement is observed as the
that the composition of the flux affects the shape and the size weld metal oxygen content decreases. Figure 6 is a plot of
of the inclusions. Welds produced with the 40 wt pct inclusion size as a function of flux composition for two of
SIO2-60 wt pct MnO flux have predominantly spherical in- the flux systems examined.
clusions that become primarily irregular and complex in
shape as systematic additions of FeO were made at the
expense of MnO, with the silica held constant. On the other FLUX BASICITY
hand, the weld inclusions remained fairly spherical, with Tuliani et al. 22 proposed a basicity index, BI, formula which
just a decrease in size, when CaF2 additions were made to is the most popular:
the SiO2-MnO flux. Figure 1 denotes the changes observed It was found that, in general, the higher the basicity the
in the inclusion morphology for these two flux systems. The cleaner the weld in regard to nonmetallic inclusions. It has
inclusions in the welds processed with the SiOz-CaO-CaF2 been reported ~2'16that high-flux basicity reduces weld metal
flux systems vary in shape from spherical to irregular and oxygen levels and improves toughness. This correlation be-
other geometric forms such as triangular and square, as tween flux basicity and weld oxygen content for the welds
shown in Figure 2. produced in this investigation is shown in Figure 7. In the
Inclusion volume fraction and size measurements were case of SiO2-CaO-CaF2 flux system, the weld metal oxy-
performed according to ASTM Standard E45 with readings gen content is reduced as the SiO2 level of the flux is de-
taken at six locations using a magnification of 1600x. creased from 40 wt pct to 20 wt pct which corresponds to
Welds from the SiO2-MnO-FeO flux system had the highest increases in the basicity index from 1.6 to 4.0. For each
inclusion levels which increased with larger additions of SiO2-CaO-CaF2 flux system, at a fixed silica level, no
FeO to the flux as seen in Figure 3. The welds processed changes in the basicity index are possible since CaO and
with SiO2-MnO-CaF2 fluxes showed a decrease in the inclu- CaF2 have the same coefficient. This would imply a constant
sion level with greater substitutions of CaF2 for MnO, as weld oxygen level for each of the fluxes of the same silica
observed in Figure 3. This behavior is expected, since level. Figures 8 and 9 show weld oxygen contents and inclu-
oxygen is present in welds primarily in combined form as sion volume fractions which remain fairly constant. Further,
non-metallic oxide inclusions due to the low solubility of the inclusion size remains uniform in each SiO2-CaO-CaFz
oxygen in iron. Figure 4 correlates the weld oxygen content flux system as the CaO and CaF2 amounts are varied. Size
changes occur only when the silica level is altered, as shown the weld microstructure in terms of the acicular ferrite. The
in Figure 10. acicular ferrite volume fraction remained constant at 30 pct
with greater additions of FeO replacing MnO in the
SiO2-MnO-FeO flux system, as seen in Figure 12. An in-
WELD METAL MICROSTRUCTURE
crease of proeutectoid grain boundary and polygonal ferrites
Changes in the weld metal microstructure, particularly in the was also noticed. Such changes in ferrite morphologies,
ferrite morphology, were found to be very apparent with coupled with the increase in inclusion density and size,
variations in flux composition. The primary constituents imply that ferrite nucleates at these inclusions in addition to
identified in the as-deposited weld microstructure of this the grain boundaries.
investigation are proeutectoid grain boundary ferrite which Quantitative metallography was performed on weld sam-
may be allotriomorphic or blocky in appearance, primary or ples of intermediate to high oxygen levels (400 ppm to
secondary side plate ferrite, intragranular plate-like ferrite 1400 ppm) produced with the SiO2-MnO-CaF2 flux system.
(aligned M-A-C, martensite-austenite-carbide), acicular fer- It was observed that as the amount of oxygen in the welds
rite and bainite. These microconstitutents are presented in decreased, the volume fraction of acicular ferrite increased,
Figure 11. The changes in the microstructure of the weld as shown in Figure 13. This suggests that since the inclu-
metal were primarily correlated with the weld metal oxygen sions decrease in size, as a result of the lower contents of
content. The manganese level was also found to affect the oxygen in the welds, these will become activated at very
microstructure indirectly through hardenability. large undercoolings. With such large undercoolings, the
The weld samples produced with the SiO2-MnO-FeO flux austenite will transform to ferrite at the low temperatures of
system had the highest oxygen contents, between 1200 and the acicular ferrite range. There are two data points (welds
1400 ppm, as seen in Figure 5. At these levels of oxygen, containing 490 and 1440 ppm of oxygen) in Figure 13, how-
however, variations of about 200 ppm did not appear to alter ever, that seem to contradict this observation, but the man-
I.I
SlO ! 9 4 0 % (r 9
1.0
_~o.9
a-
0 9
0.8
0
0.7
o
~0.6
_1
0.5 FIO or CoF 2 IN THE FLUX (wt %)
I l I I
0.4 60 55 50 4,5 4O
Fig. 3--Inclusion proportion in welds as function of FeO and CaF2 for SiO2-MnO based fluxes.
0.9
9 9 I0
0.8
/ / TO 0 . 6 IJm
Z
9 07
I.-
0
<
/../ 6(1
E
w
0.6 / / 5C
J
0 4(1
> 0.~
/
//.7
Z
0
3C p ~ , /t,,- J
.J
0
Z
0.4
~ 9 4 0 % SiO2 - MnO - FeO
2r
//~/~A/
- \
I. 3 IJm
/7
9 40 % SiO2" M n O - C o F 2
02 9 4 0 % SiO2 CoO CoFz
t~ 3 0 % S i O z C o O CoF2 IC
Oi
i ~--
0
|
--4
I
5
9
q
I0
I
q
15
9
I~I
20 25
I.g
l
lJm
Fig. 4 - - I n c l u s i o n volume fraction as function of oxygen in the weld metal MnO IN THE FLUX (it %)
for all the fluxes in this work.
--OAO I|O|-MmO-C=F! ---ill $iO|-illO-FlO
~1300
o
1400 I 0 9 38 % SiO2- MnO - OaF2 FLUX
== oll ~' 9 4 0 % SiOz- CoO- CaF2 FLUX
I- 1200
Z ~t L o 9 30 % S102- Co O- C aFz ' FLUX
W
I-- .= 1200 ~ o 9 20 %StOz-CoO-CoF 2 FLUX
Z t
~ II1~ J.~t .... o &Do LH
~ -- 9 9 9 9 HHI
I000 ql
Z ,.=,
w 0
X o
800 e~~
O
J 900 600 o ~
=E
W
800 t~J
a SlOz 9 40 % (constant) 3
.J "'"D-. Ib
bJ
9 Eutsctlc Flux 200 , /'D-.._~
70C
I 2 3 4
BASICITY INDEX
I I I I I
0 5 I0 15 20 Fig. 7--Correlation between basicity of the experimental fluxes and the
weld os content in this work. LHI and HHI stand for low and high
FoO or CoF l IN THE FLUX (wt %) heat input.
I I I ] 1
60 55 50 45 40
ganese levels of these welds provide a possible explanation.
MnO IN THE FLUX (wt %) The low hardenability of the weld caused by the high oxy-
9 siOz- - , o - yea o mo=- usa - C=F= gen level seems to be more than compensated by the high
Fig. 5 - - O x y g e n in the weld metal as function of FeD and CaF2 in
manganese content. The weld sample containing only
SiO2-MnO based fluxes. 490 ppm of oxygen experiences a reduction of hardenability
J. MATERIALS FOR ENERGY SYSTEMS VOL. 8, NO. 4, MARCH 1987 395
600 I J)
~50C 0.8
<
i,
0
Z
hi
~-40C 0.~
Z
0
(,.) / A A
/ 0 t,
/ ._.....--
Z
,- 300 0.4
>-
Z
X (2_
0 03
/
2OO 02 o n
tO0 0.( I I I I I I
20 25 30 35 40 45
CoF IN THE FLUX ( wt. % )
I I I I 1 I I I ~ I L I
20 25 3O 35 40 45 40 35 50 25 20 15
CoF t IN THE FLUX ( wt. % ) CoO IN THE FLUX ( wt %), 40 % S;Ot
I I I I l I 1 1 I I I s
40 35 30 25 20 15 50 45 40 55 50 25
CoO IN THE FLUX ( = t . % ), 40 % SI0 t CoO IN THE FLUX" ( wt %), 30 % SiO t
I I I I 1 I I I I I I ,, I
50 45 40 35 :50 25 60 5,5 50 45 40 35
CoO IN THE FLUX ( el. % ), 30 % 5102 CoO IN THE FLUX ( wt. %), 20 % SiOt
I I I I 1 I
60 55 50 45 40 35 0 40 % SiO.
A 30 % $i0~
[3 20 % S=O~
CoO IN THE FLUX ( wt. % ), 20 % SIOt
9 40 % SIO
9 30 % $10'~. Fig. 9 - - Inclusion volume fraction correlated with the CaO and CaF2 in the
9 20 % SiO~ flux SiO2-CaO-CaF~ at three SiO2 levels.
Fig. 8 - - W e l d metal oxygen content for weld specimens produced with
SiO2-CaO-CaF2 fluxes at three SiO2 levels.
I00
+.1 ......... O_. . . . . . . . . L ........ l- ..............
90 9 O O
0 0 Q6 ~m
W t 9 H e --
~ To CoFt IN THE FLUX (wt. % )
W I I I 1 I I
~ so 40 35 5O 25 20 15
0
5O CoO IN THE FLUX ( wt. % ), 40 % SiOt
1 I I 1 I 1
50 45 40 35 30 25
40
),.
CoO IN THE FLUX ( wt+ % ), 30 % SiO•
~ 3o I 1 I I I I
W 9 9 8, 9
60 55 50 45 40 35
~ 2o
F '" ..~._ _ _ 1.3 Mm
CoO IN THE FLUX ( wt % ), 20 % SiOt
I0
.-.~,_--__:__=~___~-._----~=_~~__-~.~ ,~...,. 9 9 9 40 % SlOg
--~ O A O SO % SiO 2
20 25 30 35 40 45 ..... 9 9 gl 20 % $iO 2
Fig. 10--1nclusion size as function of CaF2, CaO and SiO2 levels in the flux.
60
50
W 9
0 . 8 6 M~
40
bJ
1.46 Mn
tl.
30
.J
U
G
2O
~e 4 0
I,iJ microconstitutent present, however, was not bainite but
i-'-
I I
primarily proeutectoid grain boundary ferrite. It is apparent
I i
W that the increase in hardenability provided by a reduction in
the oxygen content is diminished by the low manganese
0::
level. Furthermore, the cooling rate may be too slow to
.J
r activate these small inclusions, which are reduced in size as
<[ the oxygen level decreases (see Figure 10). It is possible,
then, that in failing to become ferrite nuclei, these inclusions
i i i
will pin the grain boundaries (Figure 16) causing a refine-
I0 15 20 ment of the prior austenitic grain size. Figure 17 shows
FeO IN THE FLUX ( wt. % ) the micrographs of two weld samples processed with the
I I I I same flux and heat input; however, weld J7Q was quenched
55 50 45 40 after welding. This resulted in a microstructure consisting
MnO IN THE FLUX ( wt. %)
predominantly of acicular ferrite, proeutectoid veining grain
Fig. 12--Acicular ferrite content in welds produced with the SiOT boundary ferrite, and islands of proeutectoid blocky ferrite.
MnO-FeO fluxes.
Sample J7Q also shows a coarser grain size than the micro-
structure of sample J7.
due to the low manganese level. Figure 14 shows the corre- The chemistry of the inclusions in several welds is being
spgnding micrographs of the welds quantitatively examined examined with the X-ray fluorescence unit (EDS) of the
in Figure 13. scanning electron microscope. Inclusions next to regions of
When the oxygen level in the weld metal falls below acicular and proeutectoid grain boundary ferrites were in-
250 ppm, it has been reported that the amount of acicular spected, noting that Mn, Si, A1, and Ti were the primary
ferrite in the weld is reduced and replaced by a low tem- elements present. Preliminary results indicate that these ele-
perature bainite product. 16'23 Quantitative metallography ments are present in greater proportions in those inclusions
was performed on weld samples of the 20 wt pct SiO2- within the acicular ferrite as seen in Figure 18. The com-
CaO-CaF2 flux system which have oxygen levels between positions of the inclusions did not differ from those of adja-
170 ppm and 210 ppm. The results, shown in Figure 15, cent ferrite areas.
denote this decrease in the proportion of acicular ferrite with The topics discussed in this ongoing work are just a few
reduction in the weld metal oxygen content. The other examples of how basic knowledge of the nature of inclu-
50
w
l-
n.
bJ
b. 40
ix:
<(
.J G64 Mn
30 0.53 Ma
Fig. 17--Photomicrographs of welds produced with the same flux, 40 wt pet SIO2-55 wt pet MnO-5 wt pet CaF2. (a) the weld was allowed to cool at room
temperature, sample J7; (b) the weld was quenched in water immediately after welding, sample J7Q.
CONCLUSIONS
llll
(1) The nucleation of weld metal ferrite has been directly
related to the oxide inclusions, and indirectly to weld metal
f!
oxygen content.
(2) Flux composition affects the morphology and size of
I.I i i P t f IrtV} ILl
the inclusions.
I (3) Inclusion size seems to influence the weld metal mi-
t fJil/- crostructure. Inclusions 0.6 ~m and smaller are located
within the acicular ferrite; inclusions 1 /zm and greater are
found in the proeutectoid grain boundary and polygonal
ferrites. Small inclusions also pin the austenite grain bound-
t"~ll $
aries when failed to be activated, thereby increasing the
grain boundary interface available for nucleation of high
temperature ferrites.
&l ACKNOWLEDGMENTS
'1!
The authors acknowledge and appreciate the support of
l.I
L
i~P[T tlftvj
A II.t
the University of Illinois Research Board, the Engineering
Foundation, and the donation of materials by Inland Steel
Fig. 1 8 - EDS X-ray spectra from inclusions in a weld produced with the
Company.
40 wt pct SiOr20 wt pct MnO-40 wt pct CaFz: (a) at the acicular ferrite,
and (b) at the proeutectoid grain boundary ferrite.
REFERENCES
of the presence of the nonmetallic oxide inclusions. Reduc- 1. W.J. Lewis, G.E. Faulkner, and P.J. Rieppel: Welding Journal,
ing the weld metal oxygen content, and thus the inclusion 1961, vol. 40, pp. 337s-45s.
2. R.A. Kubli and W.B. Sharav: Welding Journal, 1961, vol. 40,
volume fraction, and maintaining the cooling rate constant, pp. 479s-507s.
causes the austenite to remain stable at lower temperatures 3. W.J. Lewis, G. E. Faulkner, D. C. Martin, and P. J. Rieppel: Welding
Journal, 1960, vol. 39, pp. 266-72s.
because of the reduction of the nucleation sites in terms of 4. K. Masubush, R. E. Monroe, and D. C. Martin: WRC Bulletin, 1966,
amount and size. At low temperatures when these oxide p. 1.
5. Y. Ito and M. Nakanish: Sumitomo Search, 1976, vol. 15, p. 42.
inclusions are activated, the austenite is transformed to acic- 6. B.R. Keville and R.C. Cocbrane: 'Integrity of Welding', Conf.
ular ferrite; however, it appears that in those cases when the Proc., New Zealand, Oct. 1981, p. 263.
7. R.C. Cochrane and P.R. Kirkwood: 'Trends in Steel and Con-
supercooling is not large enough to activate these inclusions sumables for Welding", Conf. Proc., London, 1978, p. 103.
and the bulk hardenability of the weld metal is low, the 8. R. Kohno, T. Takami, N. Mori, and K. Nagano: Welding Journal,
inclusions pin the austenite grain boundaries, increasing the 1982, vol. 61, pp. 373s-80s.
9. J.H. Palm: WeldingJournal, 1972, vol. 57, pp. 358s-61s.
volume of grain boundary interface available for nucleation 10. T.W. Eagar: WeldingJournal, 1978, vol. 57, pp. 76s-80s.
of high temperature ferrites. 11. J.E. Indacochea and D.L. Olson: J. Materials for Energy Systems,
1983, vol. 5, pp. 139-48.
Looking at this in the context of the flux behavior in terms 12. D.J. Widgery: 'Trends in Steel and Consumables for Welding,' Conf.
of composition and basicity, acidic fluxes generate high Proc., London, 1978, p. 217.
13. D.D. Schwemmer andD. L. Olson: 'Trends in Steeland Consumables
inclusion contents capable of pinning austenite grain bound- for Welding,' Conf. Proc., London, 1978, p. 265.
aries, thereby giving a high area fraction of grain boundaries 14. A.H. Koukabi, T.H. North, and H.B. Bell: 'Trends in Steel and
Consumables for Welding,' Conf. Proc., London, 1978, p. 281.
available as nucleation sites. In contrast, basic fluxes gener- 15. J.P. Snyder and A.W. Pense: Welding Journal, 1982, vol. 61,
ate inclusion distributions less effective in pinning grain pp. 201s-1 ls.