M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8

Evolution of secondary phases in austenitic stainless steels

during long-term exposures at 600, 650 and 800 °C

Marián Vach a,⁎, Terézia Kuníková b , Mária Dománková a , Peter Ševc c , Ľubomír Čaplovič a ,
Peter Gogola a , Jozef Janovec a
a
Faculty of Materials Science and Technology, Slovak University of Technology, Bottova 23, 917 24 Trnava, Slovak Republic
b
PSL, a.s., Robotnícka ul., 017 01 Považská Bystrica, Slovak Republic
c
Institute of Materials Research, Slovak Academy of Sciences, Watsonova 47, 040 01 Košice, Slovak Republic

AR TIC LE D ATA ABSTR ACT

Article history: Three austenitic steels (18Cr–8Ni, 18Cr–10Ni, 21Cr–30Ni), used for long-term applications at
Received 14 April 2008 temperatures between 600 and 800 °C were investigated. In the investigation,
Accepted 30 April 2008 metallography, transmission electron microscopy, selected area electron diffraction,
energy dispersive X-ray spectroscopy, and scanning electron microscopy were used. In
Keywords: additional to the experimental measurements, thermodynamic predictions were done
Austenitic steels using the ThermoCalc software and the non-commercial database STEEL16F. Various
ThermoCalc combinations of M23C6, sigma, and MC phases were identified in the austenite matrix of
TEM these steels. It was confirmed experimentally that extra large particles (up to 10 μm)
Secondary phases observed in the 21Cr–30Ni steel are M23C6, even though this carbide was not predicted as the
equilibrium carbide at service temperature (800 °C). The analytical-experimental approach,
combining thermodynamic predictions and experimental measurements, was found to be
reliable for the characterization of austenitic steels.
© 2008 Elsevier Inc. All rights reserved.

1. Introduction
The austenitic stainless steels (referred to subsequently as
austenitic steels) are typical representatives of the high alloy
corrosion resistant steels. They are widely used in various
applications from cutlery, trough rocket engine parts to steam
generators [1–4]. Annealing at temperatures around 600 °C can
lead to the formation of secondary phases that usually evoke
changes in the steel properties. Thus, the evolution of
secondary phases has to be well known and understood. The
main precipitate in austenitic steels is M23C6. Its formation has
been widely observed upon long-term exposures at tempera-
tures around 550 °C [5–7]. If M23C6 particles are present at the
grain boundaries they can lead to susceptibility to intergra-
nular corrosion. To protect the steel from such corrosion,
stabilization is necessary. The stabilizing elements (e.g. Ti, Nb,
V) have as a rule higher affinity to carbon than chromium, and
they form carbides of the MC type [1,5]. These carbides bind
carbon within the grains and reduce in this way the suscept-
ibility to intergranular corrosion. The precipitation of MC has
been observed after short-term annealing at temperatures
around 750 °C.
The brittle σ-phase (furthermore referred to as Sigma) is the
most common intermetallic phase precipitating in austenitic
steels. It has a detrimental effect on creep properties when
precipitated on grain boundaries. The χ-phase is a relatively
minor phase and has been mainly found in austenitic steels of
the type 316 at temperatures above 750 °C. A Laves phase has
also been observed in various grades of austenitic steels. It is
often a minor phase precipitating intragranularly in the form
of equiaxed particles, with occasional occurrence at the grain
boundaries. Another phase present in austenitic steels is
polymorphous δ-ferrite, which promotes fatigue crack initia-
tion on the steel surface and affects crack propagation [8].

⁎ Corresponding author. Tel.: +421 905 872401; fax: +421 33 5521007.

E-mail address: marian.vach@stuba.sk (M. Vach).

1044-5803/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.matchar.2008.04.009
 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8 1793

Table 1 – Chemical compositions of austenitic stainless

steels investigated [wt.%]
Steel C Mn Si Cr Ni Mo Ti Nb S P

A 0.06 1.67 0.62 18.48 8.37 0.30 – – 0.025 0.030

B 0.10 1.59 0.74 17.71 9.32 0.52 – 0.56 0.026 0.030
C 0.12 0.6 1.00 21.02 30.07 0.25 0.45 – 0.005 0.030

Marshall [7] examined several steels of the AISI 316 type after
long-term in-plant service and found M23C6 as the dominant
phase. On the other hand, Terada [9] did not find any M23C6 after
long-term exposure of the AISI 316 L(N) steel. M23C6 and sigma
were found by Tanaka [10] in an AISI 304 steel after annealing at
550–750 °C for a maximum of 180,000 h. Similarly, Källqvist [11]
reported the occurrence of NbC and sigma in an AISI 347 steel
exposed between 500 and 700 °C for up to 70,000 h.
The characterization of secondary phases in austenitic
steels after long-term service at temperatures between 600
and 800 °C is the main aim of this work. An approach
combining experimental techniques and thermodynamic
predictions was used in the investigation.

2. Experimental

Chemical compositions of the three austenitic stainless steels

investigated are given in Table 1. All the steels were used in
industry at elevated temperatures for long times; steel A at
600 °C for 3.5 years, steel B at 650 °C for 10 years and steel C at
800 °C for 3 years.
The metallographic samples were prepared by standard
procedures. After grinding, they were polished using diamond
pastes with particle sizes of 6, 3 and 1 μm. Then, the samples
were etched either in a 10% aqueous solution of oxalic acid or
in Groesbeck (4 g KMnO4 + 4 g NaOH + 100 ml H2O) that is an
agent used for the selective etching of carbides. Metallo-
graphic observations were made with a light microscope
Neophot 32 equipped with a CCD camera KC 381 CG. Vickers
hardness was measured using a Buehler IntendaMet 1105.
The identification of secondary phase particles was done
by selected area electron diffraction (SAED) using transmis-
sion electron microscopy (TEM) of thin foils and carbon Fig. 1 – Microstructures of the steel A (a), B (b) and C (c) after
extraction replicas. TEM observations were performed using etching in 10% aqueous solution of oxalic acid, light
a JEOL 200 CX microscope operated at 200 kV. It was equipped microscopy.
with an energy dispersive X-ray spectrometer (EDX). Thin foils
were prepared by jet-electropolishing in a TENUPOL 4 at 13 °C
and 80 V/0.1 A in a solution consisting of 95% acetic acid and coating of the surfaces with a thin layer of carbon, and (c)
5% perchloric acid. Extraction replicas were obtained in three removal of the carbon films from the samples in a 10%
steps: (a) electrolytic etching of the mechanically polished aqueous solution of HCl. Both electron diffraction and EDX
metal surface in a 10% chromic acid aqueous solution, (b) were used to identify the particles.
Scanning electron microscopy was used to observe the
morphology of secondary phase particles and to analyze their
Table 2 – Phases considered in the thermodynamic
chemical compositions. The investigation was done using a
calculations
JEOL JSM-7000F microscope equipped with EDX (Inca Oxford).
Steel Considered phases

A Liquid, Austenite, Ferrite, Laves, Sigma, M23C6, Chi, M3C,

M6C, and M7C3 3. Thermodynamic Calculations
B Liquid, Austenite, Ferrite, Laves, Sigma, M23C6, Chi, M3C,
M6C, M7C3,
The Hillert–Staffansson sublattice model was used for the
C and MC
modeling of secondary phase evolution [12,13]. The
1794 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8

Table 3 – Selected parameters characterizing the steels

Steel Temperature Duration of Average grain Secondary phases identified (TEM, EDX)
of exposure exposure (year) size in μm
(°C) Intergranular area Intragranular area

A 600 3.5 65 ± 27 M23C6 + Sigma M23C6

B 650 10 20 ± 5 M23C6 + Sigma NbC
C 800 3 95 ± 20 M23C6, TiC TiC

calculations have been made with the ThermoCalc software observed to precipitate inside the grains in steel A. In steel B,
using the STEEL16F database formulated by Kroupa et al. [14]. niobium-rich MC (Fig. 5) was identified in the grain interior
The aim of the calculations was the prediction of phase next to the intergranular sigma and M23C6. .In the steel C, MC
equilibrium for various temperatures. All elements present in was found to precipitate inside the grains and to form M23C6/
the investigated steels were considered except for sulphur and MC clusters at the grain boundaries (Table 3). For the same
phosphorus (they are not included in the database). Phases steel, the results of the EDX analysis performed across the
considered in the calculations were chosen with respect to the intergranular particle are illustrated in Fig. 6. In comparison to
literature and are summarized in Table 2. the matrix, the particle contains much more chromium and a
In the calculation procedure the total Gibbs energy of the bit more carbon, and less iron and nickel.
system, consisting of contributions of individual phases, is
minimized at constant pressure. Each of the phases consid-
ered is modeled as a sum of the reference level of Gibbs
energy, entropy term, excess Gibbs energy, and magnetic term
(if plausible the magnetic ordering). Besides phase equilibria,
the chemical compositions and molar fractions of equilibrium
phases were predicted.

4. Results

The heterogeneous austenitic microstructure (from the grain

size point of view), with relatively homogeneous distribution
of precipitates, was observed in all samples except for steel C.
Steels A (Fig. 1a), B (Fig. 1b), and C (Fig. 1c) exhibited respective
grain sizes of 65 ± 27 μm, 20 ± 5 μm, and 95 ± 20 μm (Table 3). The
large precipitates at the grain boundaries were observed in
steel C only. They are documented in Fig. 2 for the micro-
structure etched in Groesbeck.
Particles of M23C6 and sigma (Figs. 3 and 4) were found at
the grain boundaries of steels A and B. Moreover, M23C6 was

Fig. 2 – Morphology and distribution of large carbides in Fig. 3 – Particle of M23C6 identified in steel A (a) and the
microstructure of the steel C etched in Groesbeck (selective corresponding electron diffraction pattern (b), TEM/SAED,
etchant for carbides), light microscopy. carbon extraction replica.
 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8 1795

advantages of the above concept, near-equilibrium experi-

mental results have to be available to validate the predictions
and also to complete the thermodynamic databases. This
work contributes excellently to this concept, because the
samples of three steels used for long times (between 3 and
10 years) were studied.
The characteristic microstructure of all the steels consists
of austenitic grains of different sizes containing particles of
secondary phases localized both in the grain interior and at
the grain boundaries. Steel A shows greater variance in grain
sizes than the other two steels (Fig. 1a, Table 3), probably due
to the pre-service heat treatment. It is probable that small
particles of MC identified in steels B and C precipitated during
the pre-service heat treatment performed at about 1000 °C
[16]. These treatments resulted in the absence of extremely
large grains in the microstructure, mainly for steel B.
In the investigation, three types of secondary phases
(sigma, MC and M23C6) were identified. With respect to the
suggested heat treatment conditions [16], sigma and M23C6

Fig. 4 – Particle of Sigma phase identified in steel A (a) and the

corresponding electron diffraction pattern (b), TEM/SAED,
carbon extraction replica.

The results of thermodynamic calculations performed for

the temperature range of 550–900 °C are shown in Table 4.
According to the predictions, sigma and M 23 C6 are in
equilibrium with austenite at 600 °C (service temperature)
for steel A. For the steel B, M23C6 and MC were predicted to be
in equilibrium with austenite at 650 °C. MC and austenite are
in equilibrium in steel C at the service temperature of 800 °C.
For the same steel, changes in the composition of the
equilibrium M23C6 are shown as a function of temperature in
Fig. 7.

5. Discussion

In modern concepts of materials characterization, both

experimental and theoretical approaches are used in interac-
tion. When thermodynamic predictions are combined with Fig. 5 – Particle of MC (NbC) identified in steel B (a) and the
experimental measurements, reliable results are obtained corresponding electron diffraction pattern (b), TEM/SAED,
usually after a short time [15]. To exploit effectively the carbon extraction replica.
1796 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8

Fig. 7 – Mass fractions of Cr, C, Fe, and Ni in M23C6 as a

function of temperature. The data were obtained for a system
corresponding to steel C using ThermoCalc.

Fig. 6 – SEM micrograph (a) and changes in contents of carbon,

chromium, iron, and nickel across the intergranular particle than MC. The differences in particle size are caused by the
in steel C (b), EDX/SEM, secondary electron image. earlier precipitation and higher growth rate of M23C6 com-
pared to MC [19]. Both the intergranular M23C6 and the
intragranular MC particles are larger in steel C than in steel
B (Fig. 1), probably due to the higher service temperature of
probably precipitated during the service rather than MC. The
steel C. The size, quantity and distribution of MC particles are
formation of the Fe–Cr sigma phase [7] in steels A and B was
also dependent on a bulk concentration of the stabilizing
driven by two factors. The first resides in a higher Cr content
element. Unfortunately, it is lower than recommended for
and rather low concentrations of Ni and C (Table 1) retarding
both steels [18,20].
the formation of sigma [17,18]. The second factor applicable
for steel B consists in the formation of carbon depletion zones
around M23C6 particles due to the MC precipitation. The
depletion zones seem to be ideal places for sigma precipita-
tion. The enhanced amounts of Ni and C are probably the
reason for the lack of sigma in steel C. In the non-stabilized
steel A, sigma precipitated preferentially at the grain bound-
aries due to the energetically favourable heterogeneous
nucleation.
MC in steels B and C was classified as intragranular carbide
(Table 3) even though some particles of MC were also found at
the grain boundaries. Evidence about the presence of M23C6/
MC clusters at the grain boundaries is presented in Fig. 8
showing the distribution of elements in the form of elemental
EDX maps. Larger dark particles are the Cr-rich M23C6 and
smaller bright particles correspond to the Ti-rich MC. Janovec
et al. [19] showed that the precipitation of Cr(Fe)-rich M3C and
M7C3 at the grain boundaries preceded the intergranular
precipitation of V-rich MC in a 2.7Cr–0.6Mo–0.3 V steel
annealed at 720 °C. If the mechanism described above is also
applicable for the stabilized austenitic steels, M23C6 as a Cr(Fe)-
rich phase should start to precipitate in steels B and C earlier

Table 4 – Equilibrium phases predicted using ThermoCalc

Secondary phases Temperature range (°C)
predicted to be
in equilibrium Steel A Steel B Steel C
with austenite

Sigma + MC + M23C6 550–656

Sigma + M23C6 550–675
MC + M23C6 657–878 550–773
Fig. 8 – SEM micrograph and EDX map showing distribution of
M23C6 676–900
MC 878–900 774–900
elements in selected particles forming intergranular clusters
in steel C.
 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8
M23C6 was found to dominate in the microstructures of all
the steels examined. The average length of the intergranular
particles in steel C was about 4 μm, while the largest particles
reached the length up to 10 μm. Such large dimensions are not
typical for M23C6 precipitates in austenitic steels. This created
some doubt as to whether the observed particles are really
M23C6 and not an intermetallic phase. To obtain evidence
about the structure and composition of the large particles,
additional analyses using electron diffraction were done. The
selective etching by Groesbeck (Fig. 2) revealed that the
particles are truly carbides. Moreover, the EDX line scan
image confirmed enhanced concentrations of chromium and
carbon and reduced concentrations of iron and nickel for the
large intergranular particle, compared to the matrix (Fig. 6).
EDX analysis does not support determination of the exact
content of carbon in analyzed objects, in contrast to metallic
elements. As a consequence, the experimentally determined
carbon content in M23C6 (Fig. 6, 800 °C) is lower than that
thermodynamically predicted (Fig. 7, 773 °C). Thermodynamic
data for 800 °C are not available, because 773 °C was found to
be the upper limit of the M23C6 occurrence in equilibrium.
Thermodynamic predictions of phase equilibria for steels A
and B showed full consistency with the experimental results.
For steel C, the predicted and experimental results differed
from one another. As was mentioned above, M23C6 identified
experimentally was not predicted to be in equilibrium at
800 °C (Tables 3 and 4). This can be due to several factors. The
temperature 773 °C does not differ significantly from the
service temperature (800 °C) that promotes the presence of
non-equilibrium M23C6 in the steel. Also, temperature fluctua-
tions during service cannot be excluded. Thus, it appears that
the M23C6 particles were able to grow at times of temporarily
decreased temperatures. Moreover an added uncertainty can
be ascribed to the fact that the presence and/or amount of
M23C6 particles in the microstructure of the pre-service state
are not known.
The results presented show good general agreement
between the predictions and experiments. Therefore, for the
characterization of austenitic steels, the analytical-experi-
mental approach combining thermodynamic predictions and
experimental measurements seems to be appropriate.
6. Conclusions
Three austenitic steels (A, B, C) used for long-service times at
temperatures between 600 and 800 °C were characterized
using experimental measurements and thermodynamic pre-
dictions. The results obtained can be summarized as follows:
1. Various combinations of M23C6, sigma, and MC were
identified in the austenite matrix of the studied steels.
M23C6 particles were found in the grain interior as well as at
the grain boundaries of all the steels. Sigma was present at
the grain boundaries of steels A and B. In steels B and C,
particles of MC were observed in the grain interior as well
as within intergranular clusters (the latter is plausible for
steel C only).
2. Thermodynamic predictions of phase equilibria for steels A
and B were fully consistent with the experimental results.
1797
3. It was confirmed experimentally that the extra large
particles in steel C (up to 10 μm) are M23C6. However,
M23C6 is not expected to be in equilibrium at a service
temperature of 800 °C according to thermodynamic
predictions.
4. The analytical-experimental approach combining thermo-
dynamic predictions and experimental measurements was
found to be reliable for characterization of austenitic steels.
Acknowledgement
The authors wish to thank the Grant Agency of the Ministry of
Education of the Slovak Republic and the Slovak Academy of
Sciences (VEGA) for the financial support under the grant No.
1/0126/08.
REFERENCES
[1] Erneman J, Schwind M, Liu P, Nilsson JO, Andrén HO, Ågren J.
Precipitation reactions caused by nitrogen uptake during
service at high temperatures of a niobium stabilised
austenitic stainless steel. Acta Mater 2004;52:4337–50.
[2] Padilha AF, Rios PR. Decomposition of austenite in austenitic
stainless steels. ISIJ Int 2002;42:325–37.
[3] Záhumenský P, Tuleja S, Országhová J, Janovec J, Magula V.
Changes in corrosion resistance of 18%Cr–12%Ni-type
stainless steels after sensitization. Corrosion 2001;57:874–83.
[4] Cihal V. Korozivzdorne oceli a slitiny. Praha: Academia; 1999.
[5] Padilha AF, Escriba DM, Materna-Morris E, Rieth M, Klimenkov
M. Precipitation in AISI 316L(N) during creep tests at 550 and
600 °C up to 10 years. J Nucl Mater 2007;362:132–8.
[6] M. Jesenský Structure stability and weldability of stainless
steels, Ph.D. Thesis, MTF STU, Trnava, Slovakia; 2006.
[7] Marshall P. Austenitic stainless steels, microstructure and
mechanical properties. Elsevier; 1984. p. 1–79.
[8] Rho BS, Hong HU, Nam SW. Effect of δ-ferrite on fatigue cracks
in 304 L steels. Int J Fatigue 2000;22:683–90.
[9] M. Terada, D.M. Escriba, I. Costa, E. Materna-Morris, A.F.
Padilha: Investigation on the intergranular corrosion
resistance of the AISI 316 L(N) stainless steel after long time
creep testing at 600 °C. Materials Characterization, in press,
2007.
[10] Tanaka H, Murata M, Abe F, Irie H. Microstructural evolution
and change in hardness in type 304H stainless steel during
long-term creep. Mater Sci Eng 2001;A319⁎–321:788–91.
[11] Källqvist J, Andrén HO. Microanalysis of a stabilised austenitic
stainless steel after long term ageing. Mater Sci Eng 1999;
A270:27–32.
[12] Kroupa A, Výrostková A, Svoboda M, Janovec J. Carbide
reactions and phase equilibria in low-alloy Cr–Mo–V steels
tempered at 773–993 K. Part II: Theoretical calculations. Acta
Mater 1998;46:39–49.
[13] Hillert M, Staffansson LI. Regular solution model for
stoichiometric phases and ionic melts. Acta Chem Scand
1970;24:3618–26.
[14] Homolová V, Janovec J, Kroupa A. Experimental and
thermodynamic studies of phase transformations in Cr–V
low alloy steels. Mater Sci Eng 2002;A335:290–7.
[15] Kroupa A, Havránková J, Coufalová M, Svoboda M, Vřešťál J.
Phase diagram in the iron-rich corner of the Fe–Cr–Mo–V–C
system below 1000 K. J Phase Equilib 2001;22(3):312–23.
[16] ASM Handbook Volume 4: Heat Treating. ASM Int
1991:1682–708.
1798 M A TE RI A L S C H A RAC TE RI ZA T ION 5 9 ( 2 00 8 ) 1 7 9 2–1 7 9 8

[17] Perricone MJ, Anderson TD, Robino CV, Dupont JN, Michael JR.
Effect of composition on the formation of sigma during
single-pass welding of Mo-bearing stainless steels.
Metalurgical Mater Trans A 2007;38:1976–90.
[18] Sourmail T. Precipitation in creep resistant austenitic
stainless steels. Mater Sci Technol 2001;17:1–14.
[19] Janovec J, Magula V, Holý A. Influence of long-term
isothermal exposures upon M7C3 carbide changes in
2.7Cr–0.6Mo–0.3 V steel. Kov Mater 1992;30:44–53.
[20] Bhadeshia HKDH, Sir Honeycombe R. Steels
Microstructure and Properties. Third edition. Oxford;
2006. p. 259–86.