Research & Development

August 2012

Effect of cooling rate on solidification

structure and linear contraction of a
duplex stainless steel
Zhong Honggang, Chen Xiangru, Ao Lu and *Zhai Qijie
(Shanghai Key Laboratory of Modern Metallurgy & Materials Processing, Shanghai University, Shanghai 200072, China)

Abstract: Cooling rate is a key factor that can drastically affect the phase transformation and thermal stress of
duplex stainless steels. Therefore, in this research, different sand moulds were used to explore the influence of
cooling rate on the solidification of the 2304 duplex stainless steel (DSS). The macro and micro structures of the
2304 DSS were investigated. Small equiaxed grains are obtained in chromite sand mould sample with a lower
pouring temperature and a higher cooling rate, whereas coarse columnar and equiaxed grains are found in silica
sand and refractory powder mould samples. The size of austenite phase is significantly increased with decreasing
cooling rate, while the ferrite phase content ranging from 51.6% to 53.9% does not change obviously. In addition,
the linear contraction of the 2304 DSS decreases from 2.34% to 1.09% when the mean cooling rate above 1,173
K increases from 0.99 Ks-1 to 3.66 Ks-1.

Key words: duplex stainless steel; solidification structure; linear contraction; cooling rate
CLC numbers: TG142.7+1 Document code: A Article ID: 1672-6421(2012)03-239-05

D uplex stainless steel (DSS), which exhibits favorable

comprehensive properties, consists of both ferrite and
austenite phases with approximately equal amount. The
mechanical properties and the corrosion resistance of DSS are
better than those of austenitic stainless steels. Moreover, the
fabrication of DSS is easier than that of ferritic stainless steels
[1]
. Most of DSSs contain small additions of Mo, Mn, N, and
Cu, which are beneficial to their mechanical and corrosion
properties. Owing to their excellent corrosion resistance
(Cl-) and mechanical properties, DSSs have been extensively
applied to many fields recent years, such as the chemical and
petrochemical industries, marine construction, nuclear industry
and so on [2-4].
In casting process, the cooling rate is a key factor seriously
influencing the ratio of phases and thermal stress of the DSS [5].
The high temperature ferrite phase transforms into austenite
after solidification. Various analytical models, which are
primarily based on the assumption that the diffusion of nitrogen
*Zhai Qijie
Male, born in 1959, is now the Assistant President of Shanghai
University and Director of the Center for Advanced Solidification
Technology of Shanghai University. He obtained his bachelor degree
from the Northeastern University in 1982 and Ph.D degree from the
University of Science and Technology Beijing in 1991. His research
interests mainly focus on metal solidification and microstructure
refinement. His academic research has led to the publication of more
than 300 papers and 5 books. So far he holds 26 invention patents of
China, of which 11 have been put into production.
E-mail: qjzhai@shu.edu.cn
Received: 2011-11-22; Accepted: 2012-03-20
is the controlling mechanism, have been developed to explain
this transformation [4, 6-8]. As the ferrite-to-austenite phase
transformation during a regular solidification is controlled by
diffusion, which is influenced by temperature, the morphology
and content of austenite should be affected by the cooling rate.
However, there is little research on estimation of this influence.
In addition, the overall dimension and porosity are important
indicators for ingots. The solidification shrinkage can induce
shrinkage cavities and porosities, whereas the linear contraction
is a main factor resulting in thermal stresses, cracks and
dimension errors [9]. As the contraction coefficient varies with
different phases and temperature ranges in steels, solid phase
transformation and the degree of supercooling will influence
the linear contraction. Up to now, many works have been
done on the thermal expansion coefficient of alloys [10-13];
however, little attention was paid on the contraction during
continuous cooling process, such as mould casting [5]. Since
DSS represents a complicate metallurgical transformation
which can take place in the ferrite or austenite phases [14], the
solidification contraction may exhibit a different behavior as
compared with that in carbon steels [15].
In the present study, the influence of cooling rate on
solidification structure and linear contraction of the 2304 DSS
has been investigated in mould casting. The correlation of the
pouring temperature, cooling rate, solidification structure and
the linear contraction has been discussed.
1 Experimental procedure
The main chemical composition in weight percentage of

239
 CHINA FOUNDRY
the 2304 DSS is Fe (balance), 22.93 Cr, 3.96 Ni, 1.47 Mn,
0.44 Si, 0.28 Mo, 0.02 C and 0.14 N. Its liquidus and solidus
were calculated to be 1,748 K and 1,663 K, respectively, by
software of Thermal-Calc. The raw material was melted in a
medium-frequency induction furnace, and then directly poured
into the sand moulds. In order to acquire different cooling
rates, the moulds were separately made using chromite sand,
silica sand and refractory powders. The thermal conductivities
of these three materials reduce orderly.
The linear contraction as a function of the temperature was
measured by a linear contraction setup [16]. A sketch of the
linear contraction setup is shown in Fig. 1, being composed
mainly of a sand mould, a displacement sensor and a
thermocouple. The thermocouple protected by a F 5 mm SiO2
tube was placed at the 1/4 length of the sample, and inserted
2.5 mm into the sample. The outer dimension of the linear
contraction mould was 100 mm 100 mm 425 mm, while
the sample size was F 20 mm 375 mm. After pouring, the
melt attached to a screwed steel rod that was in connection to
a displacement sensor. The temperature and linear contraction
of samples were recorded with a computer data acquisition
system after pouring the melt into the mould, at a frequency of
10 data points per second.
Back plate Sand mould Thermocouple
Graphite ring Sample Steel rod Displacement sensor
Fig. 1: Scheme of linear contraction measurement
and track record system
(a) 1# (b) 2#
Fig. 2: Macro-structures of as-cast samples: (a) 1,933 K, chromite sand mould; (b) 1,903 K, chromite sand mould;
(c)1,873 K, chromite sand mould; (d) 1,873 K, silica sand mould; (e) 1,873 K, refractory powder mould
240
Vol.9 No.3
To further investigate the microstructure, several slices were
cut from the solidified sample along the longitudinal and cross-
section at 1/4 length of the sample. After being polished, the
slices are etched in a mixture of bluestone (15 g), hydrochloric
acid (40 mL) and deionized-distilled water (60 mL). The
microstructures of the samples were observed using a Leica
DM6000M optical microscope (OM). The volume fraction of
the ferrite phase was measured using a FERITSCOPE FMP30
unit based on the principle of electromagnetic induction.
2 Results and discussion
2.1 As-cast structure
Figure 2 shows the macro-structures of the as-cast samples and
Table 1 lists the experimental results of the samples fabricated
with three different moulds. Different pouring temperatures
or moulding materials can result in different grain structures
(see Fig. 2). Entire equiaxed grains are observed in chromite
sand mould sample, and the mean equiaxed grain size
decreases from 1.15 mm to 0.24 mm with decreasing pouring
temperature, as shown in Figs. 2(a) to (c). As compared with
that of the refractory material mould, the thermal conductivity
of the chromite sand mould is much greater (see Table 1).
As such, when the melt flows across the running channel in
the chromite sand mould, its temperature would be decreased
significantly, so a large number of crystals nuclei form and
survive in the metal liquid. These survived nuclei grow up and
lead to a fully equiaxed grain structure in the chromite sand
mould sample despite different pouring temperatures [17]. On the
other hand, the high pouring temperature, i.e., the increased
superheat, leads to the reduction in quantity of the crystal
nucleus, hence coarse equiaxed grains are acquired.
The coarse columnar and equiaxed grains were found in the
samples of silica sand and the refractory powder moulds [see
Cross section
Vertical section
(c) 3# (d) 4# (e) 5#
 Research & Development
August 2012

Table 1: Results of the experiments with different cooling rates and pouring temperatures
Pouring Mean cooling rate Linear Percentage of Mean equiaxed
No. Moulding material
Temperature (K) (> 1,173 K) (Ks-1) contraction rate (%) equiaxed grains (%) grain size (mm)
1# Chromite sand 1,933 \ 1.65 100 1.15
2# Chromite sand 1,903 3.87 1.50 100 0.58
3# Chromite sand 1,873 3.66 1.09 100 0.24
4# Silica sand 1,873 2.51 1.51 32.60 1.19
5# Refractory powders 1,873 0.99 2.34 66.37 0.87

Figs. 2(d) and 2(e)], very different from the complete equiaxed
grains in the chromite sand mould [see Figs. 2(a) to (c)]. Due
to the smaller thermal capacity and heat conductivity of the
refractory powders, a higher superheat during the pouring
and filling processes of the metal liquid could be maintained,
resulting in the decrease of the nucleation rate and the coarse
grains.
It should be noted that in the four solidified samples, a
similar phenomenon of shrinkage porosity is observed, as
marked in Figs. 2(a) to (d). Due to the rapid cooling rate,
the feeding channel is blocked immediately after the filling
process, there is little liquid flows from the pouring cup to the
samples to feed the solidification shrinkage, thus a shrinkage
cavity or porosity should be observed in samples. However,
the interval between the liquidus and solidus of 2304 DSS is
85 K, a rather large solidification range. Samples are solidified
in the mushy state, and the melt is partitioned by dendritic
framework, and consequently, dispersive porosities are formed.
This problem can be avoided by using of the refractory powder
mould [Fig. 2(e)], because the feeding channel from the
pouring cup to sample maintains a longer time due to its low
cooling rate.
2.2 Austenite and ferrite phases
According to the calculated phase diagram, the nucleation
and growth of austenite would occur in the temperature range
from 1,573 K to 1,173 K. To consider the solid state phase
transformation, the mean cooling rates in different moulds are
calculated from the peak temperature to 1,173 K and listed in
Table 1. The DSS mainly consists of two phases, ferrite (d /a )
and austenite (g ); and sometimes it may precipitate a number
of intermetallic phases, such as s phase, phase and carbide,
etc. [18]. The micro and macro structures of the samples solidified
in different moulds were observed by an OM as shown in Fig.
3. All five samples contain both isolated and Widmanstatten
austenite structure, whereas no evidence of s phase is
observed. Furthermore, the ferrite phase volume fractions
were measured at 30 points of each sample by a FMP30 ferrite
content measuring instrument, and the average content of
ferrite, ranging between 51.6% and 53.9%, does not change
significantly with pouring temperature and cooling rate [Fig.
4(a)]. However, the decrease in cooling rate appears to increase
the size of the austenite phase. The mean width of island-
shaped austenite increases from less than 6.23 mm to 32.86 mm

(a) (b) (c)

a
g
g
a a
g

Fig. 3: Island-shaped austenite of 2304 DSS solidified at different cooling rates: (a) 3#, 3.66 Ks-1;
(b) 4#, 2.51 Ks-1; (c) 5#, 0.99 Ks-1

(a) (b)

Fig. 4: Mean volume fraction of ferrite phase and austenite width at different mean cooling rate above 1,173 K

241
 CHINA FOUNDRY
as cooling rate above 1,173 K decreases from 3.66 Ks-1 to 0.99
Ks-1 [Fig. 4(b)]. The low cooling rate can also bring out coarse
austenite Widmanstatten structure, which grows up from the
grain boundaries.
It is suggested that the ferrite/austenite ratio is mainly
determined by composition[1, 6]; therefore the appropriate ratio
of d /g phase could be easily achieved during DSS casting.
However, the size of austenite could be evidently affected by
cooling rate. If not taking thermal stress into account, a high
cooling rate is allowed to reduce the size of austenite during
hot rolling or heat treating.
Generally, coarse austenite phase structure may reduce the
mechanical properties and the corrosion resistance of DSS [1]. A
part of black spots in Fig. 3 are corrosive pits. Most of them
are observed at the interfaces of ferrite and austenite phases
(marked by arrows in Fig. 3), and some pits can also be
identified in ferrite phase (marked by ellipses in Fig. 3).
2.3 Linear contraction
The measured linear contraction rates at different pouring
temperatures and different cooling rates (i.e., in different sand
moulds) are listed in Table 1. The results suggest that the
linear contraction rates increase with the decrease of cooling
(a)
Fig. 5: Effect of cooling rate on linear contraction of 2304 DSS: (a) Linear contraction rate and
cooling curves; (b) Linear contraction rate versus temperature at different cooling rates
Being limited by the measuring method, the linear contraction
could not be detected before the tensile strength of the sample
becomes stronger than the frictional resistance of the sand mould.
However, the sample solidifies with a remarkable temperature
gradient in axis direction. This large temperature gradient can
result in that the metal under the pouring cup still remains in
liquid state while both ends of the sample may have solidified
and contracted. As a result, the values of these linear contraction
rates (LCR) could be smaller than those of some other steels,
which have been measured in a uniform thermal field [5, 15].
Furthermore, the linear contraction rate below 973 K is
plotted as a function of the temperature in Fig. 5(b). The mean
thermal linear contraction coefficient below 973 K, calculated
by the linear relationship between contraction and temperature,
increases from 16.510 -6 to 23.210 -6 K -1 with deceasing
cooling rate. These thermal linear contraction coefficient
values are very close to that of austenitic stainless steel D9
242
Vol.9 No.3
rate or superheat.
Figure 5(a) shows the typical cooling curves in chromite
sand and refractory powder moulds (with the highest and
the lowest cooling rate, respectively). It is revealed that the
refractory powder mould exhibits a better thermal retarding
capacity than the chromite sand mould. Their mean cooling
rates are 0.99 Ks-1 and 3.66 Ks-1, respectively. To avoid being
destroyed by running melt, a quartz glass tube was used to
protect the B type thermocouple. Because of this, the highest
recorded temperature was lower than 1,700 K. The linear
contraction curves of different cooling rates are also shown in
Fig. 5(a). Due to the mould friction, creeping phenomenon was
observed during the solidification and contraction, showing as
steps on linear contraction curves. The linear contraction rate of
2304 DSS increases from 1.09% to 2.34% with the cooling rate
decreasing from 3.66 Ks-1 to 0.99 Ks-1, as shown in Fig. 5(a)
and Table 1; while it decreases with the decreasing pouring
temperature. As the thermal contraction of austenite phase is
greater than that of ferrite phase in steels [10], the progressive
phase transformation of ferrite to austenite could only
accelerate the linear contraction of the samples, so the linear
contraction curves of all three types of moulds are all smooth,
and no obvious inflection point arises [Fig. 5(b)].
(b)
[10]
, but higher than that acquired from product manuals
(approximately 12 10-6 to 14 10-6 K-1). This result indicates
that the linear contraction coefficient is affected by cooling
rate during the continuous cooling process.
The LCR of an ingot is not only controlled by thermal
contraction coefficient, but also influenced by thermal field and
mould resistance. In these experiments, it is obvious that the
chromite sand mould has the greatest temperature gradient and
mould resistance, so its LCR is smaller than those of the other
two. On the other hand, the superheat of melt can change the
temperature gradient of the sample, and lead to the increase of
LCR with increasing superheat. Therefore, to obtain accurate
size of ingots, the mould size should be calculated according to
the moulding material, pouring temperature, solid state phase
transformation and thermal field, rather than considering only
the thermal contraction coefficient. These linear contraction
curves could be considered as a reference to mould design.

August 2012
3 Conclusions
In the present study, the linear contraction rates have
been measured and the macro- and micro-structures of the
solidification samples have been investigated by varying the
pouring temperature and the cooling rate in different moulds.
The correlations among the cooling rate, linear contracting
rate and the structure have been discussed. Based on the
experimental and analytical results, the following conclusions
can be drawn:
(1) Full equiaxed grain structure can be obtained in chromite
sand mould sample, and the mean equiaxed grain size
decreases from 1.15 mm to 0.24 mm with pouring temperature
decreasing from 1,933 K to 1,873 K. However, coarse
columnar grains can be observed both in silica sand mould and
refractory powder mould samples.
(2) The width of austenite phase changes dramatically from
less than 20 mm to about 50 mm with decreasing cooling rate,
while the ferrite contents of about 52% on average on all
samples, do not seem to change obviously with the changing
conditions.
(3) The linear contraction rate of the 2304 DSS increases
with increasing pouring temperature, and decreases from 2.34%
to 1.09% as the mean cooling rate above 1,173 K increases
from 0.99 Ks-1 to 3.66 Ks-1.
References
[1] Alvarez-Armas I. Duplex stainless steels brief history and some
recent alloys. Recent Patents on Mechanical Engineering
2008, 2008, 1(1): 51-57.
[2] Iza-Mendia A, Pinol-Juez A, Urcola J J, and Gutierrez I.
Microstructural and mechanical behavior of a duplex stainless
steel under hot working conditions. Metall. Mater. Trans. A,
1998, 29(11): 2975-2986.
[3] Qiu D and Zhang W. A TEM study of the crystallography of
austenite precipitates in a duplex stainless steel. Act. Mater.,
2007, 55(20): 6754-6764.
[4] Chen C, Yen H, and Yang J. Sympathetic nucleation of
austenite in a Fe-22Cr-5Ni duplex stainless steel. Script. Mat.,
2007, 56(8): 673-676.
[5] Ridolfi M R, Vito A De, and Ferro L. Effect of alloying elements
The project is financially supported by the National Natural Science Foundation of China (Grant No. 50904044).
Research & Development
on thermal contraction and crack susceptibility during in-mold
solidification. Metall. Mater. Trans. B, 2008, 39(4): 581-592.
[6] Badji R, Bouabdallah M, Bacroix B, Kahloun C, Belkessa B,
and Maza H. Phase transformation and mechanical behavior in
annealed 2205 duplex stainless steel welds. Mater. Charact.,
2008, 59(4): 447-453.
[7] Southwick P D and Honeycombe R W K. Decomposition of
ferrite to austenite in 26%Cr-5%Ni stainless steel. Met. Sci.,
1980, 14(7): 253-261.
[8] Sieurin H and Sandstrom R. Austenite reformation in the heat-
affected zone of duplex stainless steel 2205. Mat. Sci. Eng. A,
2006, 418(1-2): 250-256.
[9] Li Z, Zhong H, Sun Q, Xu Z, and Zhai Q. Effect of cooling rate
on hot-crack susceptibility of duplex stainless steel. Mat. Sci.
Eng. A, 2009: 191-195.
[10] Banerjee A, Raju S, Divakar R, Mohandas E, Panneerselvam
G, and Antony M P. Thermal property characterization of a
titanium modified austenitic stainless steel (alloy D9). J. Nucl.
Mater., 2005, 347(1-2): 20-30.
[11] Tan X, Conway P P, and Sarvar F. Thermo-mechanical
properties and regression models of alloys: AISI 305, CK 60,
CuBe2 and Laiton MS 63. J. Mater. Process. Tech., 2005,
168(1): 152-163.
[12] Daw J E, Rempe J L, Knudson D L, and Crepeau J C. Thermal
expansion coefficient of steels used in LWR vessels. J. Nucl.
Mater., 2008, 376(2): 211-215.
[13] Sun Y, Tu Y, Sun J, and Niu D. Effect of temperature and
composition on thermal properties of carbon steel. In: Proc.
2009 Chinese Control and Decision Conference, CCDC 2009,
IEEE Computer Society, Guilin, China, June 17-19, 2009:
3756-3760.
[14] Johnson E, Kim Y J, Chumbley L Scott, and Gleeson B. Initial
phase transformation diagram determination for the CD3MN
cast duplex stainless steel. Script. Mat., 2004, 50(10): 1351-
1354.
[15] Zhu L G and Kumar R V. Shrinkage of carbon steel by thermal
contraction and phase transformation during solidification.
Ironmak. Steelmak., 2007, 34(1): 71-75.
[16] Chen Guangjun, Liu Xulin, Jin Yunlin, Han Yunpeng, Zhao
Jun and Zhang Lianfang. Design and Investigation of ZQS-
2000 Twin Bar Tester of Hot Tear-Linear Contraction for Alloy.
Foundry, 1988(1): 13-19. (in Chinese)
[17] Chalmers B. The structure of ingots. Journal of the Australian
Institute of Metals, 1963, 8: 255-270.
[18] Huang C and Shih C. Effects of nitrogen and high temperature
aging on phase precipitation of duplex stainless steel. Mat.
Sci. Eng. A, 2005, 402(1-2): 66-75.
243