Vous êtes sur la page 1sur 4

JOURNAL OF MATERIALS SCIENCE LETTERS 12 (1993) 1144-1147

Sigma-phase formation and its prevention in duplex stainless steels


S. ATAMERT, J.E. KING

Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2
3QZ, UK

Duplex stainless steels (DSS) are gaining popularity


for use in corrosive environments because of their
excellent stress corrosion cracking resistance in
CO2/C1- environments and their high yield strength.
The high strength of these steels relies on the
presence of a two-phase mixture of austenite and
ferrite. Resistance to corrosion is mainly derived
from high levels of Cr, Mo and N, but in many cases
galvanic coupling is also believed to be beneficial.
There is a growing concern in welding of DSS that
slow cooling conditions can result in the formation of
various solid-state transformation products in addition to austenite and ferrite. The presence of
transformation products such as sigma, nitrides,
laves, etc., is detrimental to both the corrosion
performance and the mechanical properties. Among
these precipitates, sigma-phase is of particular interest. It is a thermodynamically stable phase below
900 C and due to its high hardness it reduces
toughness significantly, resulting in brittle weldments. This problem is generally overcome in the
production of plates by quenching from high solution heat-treatment temperatures (e.g. 1150C).
However, in welding, continuous cooling is often
unavoidable and hence the cooling rate is of primary
importance in determining whether sigma-phase
forms. Fortunately, the formation of sigma-phase is
relatively sluggish because of the large tetragonal
unit cell of 32 atoms.
Despite the sluggishness of the transformation,
the formation of sigma-phase can be encouraged if
cooling conditions and microstructural variations
favour nucleation. For example, alloying additions,
such as Cr, Mo and W, accelerate the formation of
sigma-phase by enlarging the sigma-phase field on
the time-temperature-transformation (TTT) diagram (Fig. 1). As recently developed super duplex
stainless steels (SDSS) are particularly enriched with
these elements, sigma-phase formation can be promoted. In addition, the presence of the ferrite in
austenitic stainless steels has been shown to speed
sigma-phase formation, indicating that DSS are, in
general, more susceptible to sigma-phase formation
than austenitic stainless steels.
This work is a part of a systematic investigation
which aims to establish the relationship between the
weldment microstructures and properties of DSS
[1, 2]. In this study we aimed to establish optimum
welding conditions to prevent the formation of
sigma-phase in SDSS weldments.
SDSS (Zeron 100) with a composition of (wt %)
Fe-24.8Cr-7.3Ni-3.6Mo-0.65W-0.66Cu-0.19Si1144

Si~~a

Effec! of increasing
ICr,Mo, W
"

Time

Figure 1 Schematic diagram showing the effect of alloying additions on the position of the C-curve for the sigma-phase
transformation.

0.7Mn-0.21N was supplied by Weir Material Services, Manchester. The 12.5 ram-thick plate had
been rolled and solution-treated at 1150 C followed
by water quenching. Some small plates were furnace-cooled to stimulate the formation of the sigmaphase.
Hounsfield-type tensile test specimens were prepared from both quenched and furnace-cooled
plates and pulled using a 50 kN capacity screwdriven Schenk testing machine at a strain rate of
1 mm min-1. Fracture surfaces were examined using
a Camsean $4 operated at 30 kV. The other experimental techniques used in this work were described
elsewhere [2, 3].
The microstructure of the as-received quenched
plate was found to be a mixture of banded austenite
particles, aligned parallel to the rolling direction, in
a ferrite matrix (Fig. 2a). It is apparent that the
quenching rate is sufficient to suppress the formation
of other phases such as carbides, nitrides or intermetallics, as confirmed by transmission electron microscopy in [3]. On the other hand, a plate that was
cooled in a furnace showed extensive precipitation
(etched black) in the ferrite and at ferrite-austenite
interfaces (Fig. 2b). Transmission electron microscopy (TEM) and diffraction analysis of these
particles confirmed that they were sigma-phase
which nucleates preferentially at austenite-ferrite
and ferrite-ferrite interfaces (Fig. 2c). It has been
proposed that the nucleation of sigma-phase at
interfaces is favoured because of a crystallographic
orientation relationship between the austenite and
sigma phases [4]. However, Fig. 2c illustrates that
sigma-phase behaves like a grain boundary allotriomorph covering both 7/7 and 7/6 grain boundaries.
0261-8028 1993 Chapman & Hall

Figure2 Optical microstructures of (a) as-quenched base plate and (b) furnace-cooledbase plate; (c) TEM micrograph showing the
precipitation of sigma-phase at grain boundaries (arrows). The diffraction pattern (inset) showsthat the grain boundaryphase is the
sigma-phase; the zone axisof the pattern is [21 1]sigma.

This in turn suggests that nucleation is heterogeneous in nature and does not depend strongly on the
crystallographic orientation relationships between
the phases. In addition, the preferential growth of
sigma-phase into the ferrite cannot be explained on
the basis of crystallographic orientation since, as
sigma-phase adopts a particular orientation relationship with respect to the austenite, it should also be
crystallographically related to the ferrite, within the
same grain, because of the Kurdjumov-Sachs orientation relationship between the austenite and ferrite
phases. It has also been claimed that preferential
growth of the sigma-phase into the ferrite is mainly a
result of the higher Cr and Mo concentrations in the
ferrite matrix [4]. A-fundamental reason why the
sigma-phase preferentially grows into the ferrite is
that the ferrite is thermodynamically metastable at
temperatures where the sigma phase precipitates so
that the ferrite phase should decompose into an
equilibrium state. The evidence for this comes from
recent investigations which have shown that an
equilibrium structure is a mixture of sigma and
austenite phases in the temperature range
650-900 C [5]. It should be noted that the compositions of the ferrite and sigma phases are very similar,
indicating that growth of sigma-phase does not
require long-range diffusion of substitutional alloying elements, such as Cr and Mo. The growth of
sigma-phase is likely to be controlled by small-scale
atomic rearrangement of b c c atoms into a tetragonal lattice in a manner similar to the in situ
transformation of one carbide phase into another
[6]. The absence of any intragranular precipitation
of sigma-phase confirms that nucleation is heterogeneous and a rate-controlling step is the nucleation
stage as suggested previously [7].
Although long-range diffusion of Cr and Mo may
not be essential for the nucleation and growth of
sigma-phase, small compositional variations can
influence the formation of sigma. For example,
sigma-phase grows preferentially near grain boundary austenite allotriomorphs rather than intragranular austenite in the weld metal, as shown in Fig. 3
[5]. This is probably a result of an extensive
partitioning of Cr and Mo into the ferrite phase near

Figure3 An optical micrograph of a weld metal aged 8 days at


800 C. TOe dark etched phase is ferrite, lightly etched phase is
austenite and sigma-phase appears light. (Original source of the
micrograph [7]).

the grain boundary austenite, consistent with recent


observations [3].
The effect of sigma-phase is of considerable
interest since its presence has been associated with
the degradation of mechanical properties. Charpy
room-temperature impact energy for the microstructure in Fig. 2b is found to be as low as 7 J. Tensile
test data agree with impact tests results, showing
that the presence of sigma-phase reduces the total
elongation to failure from 40% in the base plate to
7%. Both yield strength and ultimate tensile
strength showed a small increase in the presence of
sigma-phase. The fracture path in samples containing sigma-phase appears to involve both fracture of
brittle sigma-phase and interfacial failure as opposed
to a typical ductile failure in an as-quenched sample
(Fig. 4).
Three-dimensional Rosenthal heat-flow equations
were used in this study to calculate thermal cycles as
a function of the welding conditions. The original
equations were given elsewhere [2] and are not
presented here. In order to calculate a critical
cooling rate to prevent sigma-phase formation,
knowledge of the T T T diagrams for sigma-phase
formation is essential. This information is already
available in the literature for low-grade, super
duplex and super ferritic stainless steels [4, 8, 9]. The
results are overall in agreement that the kinetics of
1145

sigma-phase formation from ferrite follows classic


C-curve behaviour. The position of the nose of the
C-curve varies with the alloy composition as shown
schematically in Fig. 1. As emphasized above,
higher Cr and Mo concentrations push the C-curve
towards the left-hand side, resulting in the relatively
fast formation of sigma-phase. Therefore, calculations were carried out for both super ferritic stainless
steel and SDSS. The composition of the super
ferritic stainless steel represents the ferrite phase
which is enriched with Cr and Mo (29Cr and 4Mo) as
a result of an excessive partitioning of alloying
elements during solid-state transformation in highheat-input welding. The positions of the nose of the
C-curves of the super ferritic and duplex stainless
steels are 890 C and 2.5 rain, and 900 C and 70 s
respectively [4, 8, 9]. The results of the calculations
are shown in Fig. 5 for three different preheat
temperatures. As expected, the sigma-phase-free
region depends strongly on the heat input
and the preheat temperature. An approximately
0.3 kJ mm -1 lower heat input is necessary to prevent
sigma-phase formation in ferrite which is enriched
with respect to Cr and Mo additions.

'

'

tE 4

E
Sigma-phase/
possible / /

,"-I~ 3

_=~

(a)

4.1

l/ /

==> Sg
i ma-fre

e region

~2
"r
I

0.2

0.4

, I

0.6

0.8

Plate thickness (cm)


7

4.0

"

'

'

'

"

'

E
E 3.5

"3

~_. 3.o
+a

~2.5

Sigma-phase
poss,ble / / .

.t.a

r~ 2.0
7"
1.5

1.0

~-,

3.5

_~ 2.5

(b)
,

Sigma'free

//

region

0.2

'

//

~
,

0.4

0.6

....

0.8

Plate thickness (cm)


'

'

'

Sigma-phase

'

'

'

'

//

possible
e_ 2.0

'-

/ ~ - Sigma-free

t-

1.0

(c)
0.2

0.4

0.6

0.8

Plate thickness (cm)


Figure 5 Plate thickness versus heat input for three different
interpass temperatures: (a) 25 C, (b) 100C and (c) 200C,
showing sigma-free and sensitized regions for both (II~) SDSS
(25Cr, 3.7Mo) and (A) SFSS (29Cr, 4Mo).

Acknowledgements

Figure 4 SEM micrographs of tensile fracture surfaces of parent


metal: (a) furnace-cooled sample showing both fracture of the
sigma-phase (arrows) and intergranular failure and (b) ductile
failure in as-quenchedsample.

1146

The authors thank British Gas plc, SERC and the


Fellowship of Engineering for funding and the staff
at ERS (British Gas) for helpful discussions. Special
thanks to K. Prosser (ERS) who initiated this
particular study. Thanks are due to Professor C. J.
Humphreys for the provision of laboratory facilities
at the University of Cambridge. The authors are
grateful to Dr N. I. A. Haddad and Dr H. K.~D. H.
Bhadeshia for allowing the publication of Fig. 3 .

References
1.
2.
3.
4.

5.
6.

s. ATAMERT and J. E. KING, Z. Metallkde 82 (1991) 230.


S. ATAMERT, R. C. REED and J. E. KING, Mater. Sci.
Technol. in press.
S. ATAMERT and J. E. KING, Acta Metall. Mater. 39
(1991) 273.
A . J . STRUTT, G. W. LORIMER, C. V. ROSCOE and K.
J. GRADWELL, in Proceedings of Duplex Stainless Steels
86, October 1986, p. 310.
N . I . A . HADDAD, PhD thesis, University of Cambridge
(1989).
R . W . K . HONEYCOMBE and A. K. SEAL, JISI 188
(1958) 9.

7.
8.

9.

J. M. VITEK and S. A. DAVID, Weld. Res. Suppl. 65


(1986) 106-s.
J. CHARLES, in Proceedings of the Duplex Stainless Steels
91, October 1991, edited by J. Charles and S. Bernhardsson,
p. 3.
G. HERBSLEB and P. SCHWABB, in Proceedings of the
Duplex Stainless Steels 82, October 1982, edited by R. A,
Lula, p. 15.

Received 24 A u g u s t
and accepted 23 N o v e m b e r 1992

1147

Vous aimerez peut-être aussi