Materials Science and Engineering A 447 (2007) 266–276

Oxidation of AISI 304 and AISI 439 stainless steels

A.M. Huntz a,∗ , A. Reckmann a , C. Haut b , C. Sévérac b , M. Herbst a ,
F.C.T. Resende c , A.C.S. Sabioni c
a Laboratoire d’Etude des Matériaux Hors Equilibre, CNRS ICMMO UMR 8182, Université Paris XI, F-91405 Orsay, France
b CNRS ICMMO UMR 8182, Université Paris XI, F-91405 Orsay, France
c Laboratório de Difusăo em Materiais, Departamento de Fisica, UFOP/REDEMAT, Ouro-Preto, M.G., Brazil

Received 26 June 2005; received in revised form 22 September 2006; accepted 6 October 2006

Abstract
The oxidation behaviour of AISI 304 and AISI 439 stainless steels was studied at high temperatures, under various oxygen pressures and
in the presence or not of water vapour. Thermogravimetric analyses were conducted in isothermal conditions from 850 to 950 ◦ C for 50 h and
microstructural and chemical analyses of the oxide films grown by oxidation were performed by SEM and EDX. The oxide films were also analysed
by grazing X-ray diffraction and by X photoelectron spectroscopy (XPS).
The AISI 439 steel has higher oxidation resistance than AISI 304, above 850 ◦ C, under high oxygen pressures. On the other hand, the AISI 304
steel has higher oxidation resistance under low oxygen pressures in the whole temperature range.
In order to check whether the growth kinetics of Cr2 O3 formed by the oxidation of stainless steels was controlled by oxygen or/and chromium
diffusion through the oxide film, the oxidation constants were theoretically calculated on the basis of diffusion data using Wagner’s Theory. The
calculated values of the oxidation constants agree reasonably well with experimental values, therefore confirming the role of diffusion processes
on the growth of chromia on stainless steels.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Oxidation; AISI 304 stainless steel; AISI 439 stainless steel; Oxygen; Water vapour

1. Introduction of a 17% Cr ferritic stainless steel (AISI 439) containing tita-

nium and niobium (see Table 1). These elements are generally
Chromia protective films grown on stainless steels efficiently added in stainless steels to prevent the risks of intergranular cor-
reduce high temperature corrosion [1,2]. Among various appli- rosion of weldments, due to the formation of stable titanium and
cations, stainless steels are nowadays widely used in exhaust niobium nitrides and carbides instead of non-stable chromium
pipes to improve the service life of their components, especially carbides [3,4]. Moreover, due to the additions of nitride and car-
the upstream part of the exhaust line (manifold, down-pipe, bide forming elements, the structure of the stabilised 17% Cr
converter shell), where temperatures can reach 1100 ◦ C. steels is “fully” ferritic at all temperatures. This austenite has a
Austenitic stainless steels like AISI 304 are traditionally used, beneficial effect on the oxidation resistance of those steels [5],
but, more recently, ferritic stainless steels have been developed as it prevents damage to the protective oxide film.
for those applications. Besides offering the advantage of lower In most applications, steels can be affected by water vapour,
costs than austenitic grades (as the former does not include therefore, oxidation resistance was studied from 850 to 950 ◦ C
nickel), they also present a lower expansion coefficient than in various atmospheres, namely pure oxygen, synthetic air, air
austenitic steels, which is clearly an advantage when temperature with H2 O and argon–hydrogen–H2 O mixtures.
cycling resistance is needed.
In this work, the high temperature oxidation behaviour of an
austenitic stainless steel (AISI 304) is compared to the behaviour 2. Materials and experiments

The samples of AISI 304 austenitic and AISI 439 ferritic

∗ Corresponding author. Tel.: +33 1 69156318; fax: +33 1 69154819. stainless steels used in this work were supplied by Acesita S.A.
E-mail address: [email protected] (A.M. Huntz). The chemical analyses of these steels are given in Table 1. For

0921-5093/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2006.10.022
 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276 267

Table 1
Chemical composition of the 304 and 439 stainless steels
Steel wt% N (ppm)

C Mn Si P S Cr Ni Nb Ti

304 austenitic 0.0466 1.37 0.46 0.028 0.0006 18.07 8.11 – – 322
439 ferritic 0.0060 0.18 0.42 0.033 0.0010 17.01 0.23 0.17 0.15 122

the thermogravimetric analyses, the samples were cut to dimen- a stainless steel, the chromium to iron ratio at the outer sur-
sion (10 mm × 10 mm × 0.6 mm), with a small hole of 0.8 mm in face can vary according to the surface preparation and this will
diameter near an edge, to hang them in the thermobalance. They have an influence on the nucleation of the possible oxides [6].
were all polished with diamond paste, and submitted to oxidation It was also shown [7] on nickel that the polishing procedure
treatments in a SETARAM TGDTA 92 thermobalance (sensitiv- could modify the growth kinetics of NiO. Moreover, the first
ity of ±1 ␮g). The oxidation treatments were performed at 850, oxidation stage has probably an important effect on the adher-
900 and 950 ◦ C for 50 h, under 1 atmosphere pressure, either ence of the oxide film after long oxidation treatments. Finally,
made of pure oxygen or synthetic air, and, in order to observe the concerning the stresses at the surface of the sample, the initial
possible influence of water vapour, in Ar–H2 –H2 O or air–H2 O surface condition is particularly important as significant differ-
mixtures. The Ar–H2 –H2 O mixture was obtained from Ar with ences are obtained according to the nature of the polishing or to
100 vpm H2 , the gas passing through a catalyst which allows a shot-peening [8], etc. That is why all our studied samples were
the elimination of oxygen by promoting the 2H2 + O2 → 2H2 O polished in the same way. The polishing procedure was chosen
reaction, then through a cryostat at −60 ◦ C which maintains a in order to obtain a surface roughness in adequation with the
pH2 O = 5.26 × 10−6 atm. In the furnace, the oxygen pressure, further surface analyses.
pO2 , was imposed by the equilibrium H2 + (1/2)O2 ↔ H2 O, The microstructure and composition of the oxidised surface
for which G◦ (J) = −249700 + 57.07T = −RT ln K with K = were examined by scanning electronic microscopy (SEM) and
pH2 O /(pH2 p1/2
O2 ). Concerning the mixture air + H2 O, it was energy dispersive spectroscopy (EDX). Grazing X-ray diffrac-
obtained by passing air through water maintained at 15 ◦ C, pro- tion experiments and photoelectron spectroscopy (XPS) gave
ducing a pH2 O = 1.68 × 10−2 atm. The theoretical oxygen and supplementary information. The chemical analyses of oxide lay-
water vapour pressures are gathered in Table 2. The growth ers by XRD have been carried out on a Panalytical X’Pert Pro
kinetics of the scales formed on the steels were established by MRD apparatus in grazing incidence fixed at 2◦ . The copper
measuring the mass gain per unit area versus oxidation time. It anticathode of the diffractometer presents a λK␣ (Cu) wavelength
is well known that the sample initial surface condition can have equal to 0.154184 nm. The 2θ scanning range was fixed between
an important effect on the oxidation kinetics and on the scale 20◦ and 80◦ by step of 0.02◦ with a counting time of 4.5 s per
adherence. Indeed, the surface preparation will act on several step. Finally, XPS analyses were performed on a Leybold LHS10
parameters which characterise the surfaces: analytical, struc- apparatus, operating under a 10−7 to 10−8 Pa vacuum. It used
tural, mechanical parameters. In the case of an alloy, especially a 1253.6 eV radiation of an Mg anticathode X-ray source and
worked in constant analyser transmission mode (CAT). Argon
ions sputtering was carried out for 5 min in order to remove the
Table 2 contamination layer (especially carbon contamination). Most of
Gas pressure in the various tested atmospheres (total pressure = 1 atm)
the analyses were performed with a beam at 90◦ relative to the
T (◦ C) p (atm) sample.
pO2 pH2 pH2 O pH2 O /pO2

850 3. Results
O2 1 10−6 10−6 10−6
Air 0.21 10−6 10−6 4.76 × 10−6
Air + H2 O 0.1932 10−6 0.0168 8.7 × 10−2
3.1. Thermogravimetry (TG)
Ar–H2 –H2 O 1.44 × 10−20 10−4 5.26 × 10−6 3.65 × 1014
All the results concerning the kinetic curves (mass gain curves
900
O2 1 10−6 10−6 10−6 as a function of time) are gathered in Fig. 1 for all the atmo-
Air 0.21 10-6 10−6 4.76 × 10−6 spheres tested as well as the two steels. It clearly appears that
Air + H2 O 0.1932 10−6 0.0168 8.7 × 10−2 for AISI 439, the atmosphere nature has little effect on oxida-
Ar–H2 –H2 O 1.40 × 10−19 10−4 5.26 × 10−6 3.76 × 1013 tion which is not the case for the AISI 304, particularly when the
950 temperature reaches 900 ◦ C and beyond. In all cases, for AISI
O2 1 10-6 10−6 10−6 439, the kinetics follows a parabolic law, as shown in Fig. 2a. In
Air 0.21 10−6 10−6 4.76 × 10−6 contrast, for AISI 304, when plotting the kinetic curve at 900 and
10−6 8.7 × 10−2
950 ◦ C in quadratic coordinates, two successive stages appear,
Air + H2 O 0.1932 0.0168
Ar–H2 –H2 O 1.34 × 10−18 10−4 5.26 × 10−6 3.92 × 1012
as shown in Fig. 2b, each of them following a parabolic law.
268 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276

Fig. 1. Mass gain curves of AISI 304 and 439 under various atmospheres at three temperatures (850–900–950 ◦ C).
 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276 269

Table 3
Global composition (%) of the oxide film grown on AISI 439 at 900 ◦ C
Air Air + H2 O

Cr 39.15 36.91
Mn 13.67 14.02
Ti 4.56 6.42
O 42.63 42.65

3.2. Microstructure and composition: SEM observations

and EDX analyses

In the case of AISI 439 steel, the oxidised outer surface

presents regular equiaxed grains (Fig. 3a) with sometimes sand
roses (Fig. 3b). Whatever the oxidation temperature and atmo-
sphere, global EDX analyses indicate that the scale is mainly
consisting of Cr and oxygen, i.e. Cr2 O3 . Nevertheless, some
Fe, Mn and Ti are detected, and the sand roses are particularly
enriched in Cr, Mn and Ti. After oxidation at 950 ◦ C, whatever
the atmosphere, some bucklings are observed (Fig. 4). They are
already present at 900 ◦ C when water vapour is present in the oxi-
dation atmosphere. Observations of transverse sections by SEM
show that the oxide scale is flat, not convoluted and internal oxi-
dation occurs (Fig. 5a), whatever the atmosphere. The internal
oxide particles are made of silica. Finally, when oxidised with a
high H2 O/O2 ratio, i.e. in Ar/H2 /H2 O mixture, some whiskers
are observed (Fig. 5b) and some few spallings appear at least
when oxidised at 950 ◦ C. At the bottom of the spalled zone,
oxide crystallites made of Si and Ti are detected. Nevertheless,
the global composition of the oxide scales is not significantly
modified by the presence of water vapour in the oxidising gas
(Table 3).
Concerning the AISI 304 steel, the outer surface of the oxide
scale grown at 850 and 900 ◦ C have a similar aspect. The oxide
film is made up of little regular grains (Fig. 6a) with some
Fig. 2. Quadratic plot of the mass variation vs. time for: (a) AISI 439 oxidised thickening along grain boundaries of the underlying substrate
at 900 ◦ C in Ar–H2 –H2 O and (b) AISI 304 oxidised in O2 at 900 ◦ C. (Fig. 6b). EDX analyses indicate that the scale is enriched in
chromium and oxygen with a significant amount of manganese
(10–15%) and some silicon. After oxidation at 950 ◦ C, two types

Fig. 3. Outer surface of the oxide film formed on AISI 439 oxidised at 850 ◦ C: (a) in air and (b) in air + H2 O.
270 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276

Fig. 4. Outer surface of the oxide film formed on AISI 439 oxidised at 950 ◦ C: (a) in air and (b) in air + H2 O.

Fig. 5. AISI 439: (a) transverse section after oxidation in air + H2 O at 900 ◦ C and (b) outer surface of the scale after oxidation at 850 ◦ C in Ar–H2 –H2 O.

of oxide morphology is observed on the outer surface: some After many observations, it seems that the areas with small chro-
zones are made up of little regular grains (Fig. 7a) while, particu- mia grains correspond to spalled areas, this spalling probably
larly on the edges and the periphery of the sample, rough oxide is occurring during cooling. When oxidised with a high H2 O/O2
present (Fig. 7b). The areas with regular oxide grains are made of ratio, i.e. in Ar/H2 /H2 O mixture, whiskers are numerous, espe-
chromia while the convoluted zones are made up of iron oxides. cially on the blisters formed along grain boundaries (Fig. 8a),

Fig. 6. Outer surface of AISI 304 oxidised at 850 ◦ C in air + H2 O.

 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276 271

Fig. 7. Surfaces of AISI 304 oxidised at 950 ◦ C in air: (a) observation of a spalled area showing little chromia grains and (b) unspalled area with iron oxide.

and they are rich in Cr and Mn. Looking at the oxide–steel inter- 3.4. X photoelectron spectroscopy
face along transverse sections, it is observed that the interface
is more convoluted than for the 439 steel, as shown by Fig. 8b. It is important to note that the thickness of the sample anal-
Silica particles are observed along the inner interface of the scale ysed by XPS is very small, depending on the mean distance
and as internal penetrations. covered by the emitted electrons, this last being a function of
its energy. The order of magnitude of the analysed thickness
3.3. Structure characterisation by grazing XRD when the X beam is perpendicular to the sample surface is
included between 0.5 and 10 nm [9]. Thus, the obtained infor-
XRD analyses allowed to verify that the oxide scale mainly mation is coming from the very outer surface of the oxide
consisted of chromia, except for the 304 steel oxidised at 950 ◦ C film. When decreasing the beam angle by tilting the sample,
for which the amount of iron oxide is important. Nevertheless, the information becomes more and more characteristics of the
many diffraction peaks are observed on the X patterns as many surface.
elements constituting the steel are more or less oxidised, espe- In nearly all the cases, two components must be considered to
cially chromium, but also iron, silicon, manganese and titanium rebuild the experimental chromium peak (Fig. 9a). The energy
in case of 439 steel. Thus, mixed oxides and solid solution of of the more intense peak corresponds to Cr3+ in Cr2 O3 (or
chromium–iron oxide (Cr,Fe)2 O3 are present. With the 304 steel (Cr,Fe)2 O3 ) while the energy of the small peak indicates the
oxidised at 950 ◦ C, Fe2 O3 is detected at the outer surface. With presence of Cr in mixed oxides (probably Mn1.5 Cr1.5 O4 ) at the
the 439 steel, Mn1.5 Cr1.5 O4 and MnTiO3 oxides are detected. very outer surface. Moreover, at 950 ◦ C, the chromium peak is
These oxides are the constituent of the observed sand roses on the more intense in the 439 steel than in the 304 steel (Fig. 9b), due
outer surface (see Fig. 3b). No important differences are linked to the presence of iron oxides on 304. Three components are
to the presence of water vapour in the oxidising atmosphere, necessary for Mn (Fig. 9c). Due to their energy, it can be said
which conforts the result shown in Table 3. that the first peak corresponds to Mn2+ (i.e. MnO compound),

Fig. 8. AISI 304: (a) outer surface after oxidation at 850 ◦ C in Ar–H2 –H2 O and (b) cross section after oxidation at 900 ◦ C in air + H2 O.
272 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276

Fig. 9. XPS peaks and their reconstruction: (a) Cr peak in 439 after oxidation at 950 ◦ C in air, (b) comparison of Cr peak in 439 and 304 after oxidation at 950 ◦ C in
air, (c) Mn peak on 304 oxidised in air at 950 ◦ C, (d) Ti peak on 439 after oxidation at 950 ◦ C in air, and (e) comparison of Fe peaks on 304 according to oxidation
temperature and atmosphere.
 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276 273

Table 4
Intensity ratios of XPS peaks determined on the oxide film formed on AISI 439
at 950 ◦ C in air according to the beam angle
Beam angle Intensity ratios

Ti459 /Cr576 Ti459 /Mn641 Cr576 /Mn641

90◦ 0.58 0.92 1.59

40◦ 0.72 0.97 1.35
(40◦ − 90◦ ) +0.14 +0.05 −0.24

the second to Mn bound in a mixed oxide (Mn1.5 Cr1.5 O4 and/or

MnTiO3 ) and the last one corresponds to a satellite [9]. Ti is not
detected on the oxidised surface of the 304 steel, but a significant
amount is present on the outer surface of 439 steel (Fig. 9d). One
peak, corresponding to Ti4+ , is sufficient to fit the experimental
peak. In agreement with the previous results, iron is particularly
present on the outer surface of the 304 steel when oxidised at
950 ◦ C (Fig. 9e), and its fraction increases with the presence of
water vapour.
For the 439 steel oxidised at 950 ◦ C in air, analyses were also
made with the beam at 40◦ from the sample. It means that a
Fig. 10. Arrhenius plot of oxidation parabolic constants for 439 and for the first
smaller depth is concerned by the analysis when compared to stage of 304 steel.
the previous results. Thus, as shown by the results in Table 4, it
indicates that the titanium–manganese rich oxide (MnTiO3 ) is
located on the outer surface of the oxide scale, while chromia a protective chromia film [10–15]. Indeed according to Hindam
and Mn–Cr rich oxide lie underneath. and Whittle [15], the growth of chromia films is characterised
by kp values in the range 4 × 10−13 to 1 × 10−11 g2 cm−4 s−1 at
850 ◦ C, and 2 × 10−12 to 7 × 10−11 g2 cm−4 s−1 at 950 ◦ C. Our
4. Discussion
values are close to the smaller values. This work indicates that
the growth mechanism of chromia films depends neither on the
4.1. Kinetic aspect
oxygen pressure, nor on the presence of water vapour.
The situation is somewhat more complex with the AISI
In the case of AISI 439 steel, it appeared from the results
304 steel. Indeed, it can be seen in Fig. 1 that, except under
(Figs. 1 and 2) that, whatever the temperature and the atmo-
Ar–H2 –H2 O atmosphere, there is a gap between the mass evolu-
sphere, the mass gain curves follow a parabolic law given
tion at 850 and that at 900 or 950 ◦ C according to the atmosphere.
by (M/S)2 = a + kp t, i.e. a law which is particularly interest-
At 950 ◦ C and for any of the three atmospheres O2 , air and
ing for the applications as the oxidation rate decreases with
air + H2 O, the mass gain curve clearly shows two stages, a first
time. The so-determined kp constants are presented in Table 5
short one characterised by a small mass gain, followed by a
and their evolution with the temperature has been plotted in
fast mass gain. The analysis of quadratic plots indicates that
an Arrhenius graph leading to the determination of the oxi-
the first stage leads to parabolic oxidation constants kp of the
dation activation energy (Fig. 10 and Table 6). Whatever the
same order of magnitude than in the case of AISI 439 steel,
oxidising atmosphere, the curves in the Arrhenius plot are quasi-
see Table 5, i.e. corresponding to the values found in the liter-
superimposed and the activation energy is included between 230
ature for chromia growth [10–15]. In contrast, the second stage
and 250 kJ mol−1 . All these results agree with the formation of
leads to important parabolic constants close to those found for
iron oxides: literature data on the oxidation constants of iron (or
Table 5 iron base alloys) [10,16,17] lead to scattered values, between
Oxidation parabolic constants (kp ) values in g2 cm−4 s−1 for AISI 439 and for 10−11 and 10−9 g2 cm−4 s−1 at 950 ◦ C, according to the for-
the first stage of AISI 304 (i.e. for Cr2 O3 growth)
mation of haematite or magnetite, respectively, while the value
Air Air + H2 O Ar/H2 + H2 O O2

Austenitic AISI 304 Table 6

850 ◦ C 2.1 × 10−13 9.0 × 10−14 1.6 × 10−13 1.3 × 10−13 Activation energy associated to the oxidation constants in Table 5
◦
900 C 6.9 × 10−13 6.5 × 10−13 3.2 × 10−13 4.6 × 10−13
950 ◦ C 1.4 × 10−12 1.5 × 10−12 6.4 × 10−13 1.1 × 10−12 AISI 304 (kJ mol) AISI 439 (kJ mol)

Ferritic AISI 439 Air 231 231

850 ◦ C 4.0 × 10−13 4.1 × 10−13 2.9 × 10−13 3.2 × 10−13 Air + H2 O 323 235
900 ◦ C 1.6 × 10−12 1.8 × 10−12 1.5 × 10−12 1.4 × 10−12 Ar/H2 /H2 O 158 252
950 ◦ C 3.0 × 10−12 3.2 × 10−12 2.6 × 10−12 2.5 × 10−12 O2 239 239
274 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276

determined in this work for the second stage at 950 ◦ C is about due to through scale cracking. This does not appear through the
10−9 g2 cm−4 s−1 . When plotting Arrhenius graphs with the oxi- SEM observations as, when spalling occurs, a compact chromia
dation constants determined in the first stage (Fig. 10), it appears film is observed underneath (Fig. 7a). It cannot be suggested
that the values of the activation energy (Table 6) are much more that water vapour is efficient for scale failure as, in our case,
scattered than those of the previous steel. Thus, for the three a similar oxidation rate occurs in pure oxygen and air + H2 O
atmospheres O2 , air and air + H2 O, a transition occurs from chro- (Fig. 1b and c). Thus, creation of microcracks, or voids at the
mia growth to iron oxides growth and the higher the temperature inner interface, does not seem to be responsible for the increase
the faster the transition. of oxidation rate. A depletion in the chromium concentration
It could be suggested that the data for the AISI 439 in Fig. 10 in the substrate, near the substrate–oxide interface, as observed
are the consequence of the superimposition of two processes, recently by Galerie et al. [18], cannot be invoked for a mech-
one which would correspond to the volatilisation of chromium anism change in presence of water vapour as these authors did
species which would induce a mass decrease, and the other not observe significant difference in the chromium interface con-
which would lead to a mass increase due to the growth of centration according to the atmosphere nature (with or without
iron oxides, as for AISI 304 [18]. This possibility is rejected water vapour). The same authors also showed that stress in the
on account of the commonly accepted negligible pressure oxide film could not be the thermodynamic force for haematite
obtained for the chromium trioxide at temperatures of our tests nucleation.
[19–21]. Indeed, according to Hagel for instance [20], at 950 ◦ C The behaviour difference with AISI 439 is not related to
the vaporisation constant is equal to 1.7 × 10−21 g cm−2 s−1 . the steel composition as both have roughly the same chromium
Moreover, in oxygen-water vapour atmospheres, chromium(VI) amount (it is even more important in 304), but to the steel struc-
oxo-hydroxide CrO2 (OH)2 exhibits much higher partial pres- ture. AISI 439 has a ferritic structure and diffusion coefficients
sure than CrO3 [22], but the present results do not show are much greater (approximately 10 times greater), than in an
significant differences according to the nature of the environ- austenitic matrix. Thus, the amount of chromium in the steel is
ment for both studied steels, though the H2 O ratio is very sufficient to feed the oxide film growth and there is no variation
important in Ar–H2 –H2 O atmosphere (see Table 2). It can be according to the atmosphere composition as, in all cases, the
concluded that scale volatilisation is not a valid damaging pro- oxygen pressure is sufficient to form this oxide. In the case of
cess for the oxide layer. AISI 304, even 18% chromium is not always sufficient to feed
the growth of a chromia film. Indeed, the minimum amount of
4.2. Effect of the atmosphere on the oxidation mechanism chromium in the steel necessary to lead to a chromia protective
and high temperature resistance film can be derived from Wagner equation [23]:
  
It was already mentioned that the composition of the oxidis- NB (m) − NB (i) 1 kc 1/2
=F (1)
ing atmosphere had no effect on the oxidation behaviour of AISI 1 − NB (i) 2 D̃
439, contrary to the case of AISI 304.
For AISI 439, whatever the temperature and the atmosphere where NB (m) is the atomic fraction of the most oxidisable ele-
composition, chromia is always formed as a protective film and ment in alloy AB, NB (i) the atomic fraction at the metal–oxide
its growth kinetics depends neither on the oxygen pressure, nor interface, D̃ the coefficient of chemical diffusion of the alloy
on the presence of water vapour. It is well represented in Fig. 10. and F(u) = π1/2 u(1 − erf u) exp(u2 ).
For AISI 304 steel at 850 ◦ C (Fig. 1), the mass gains are rela- By substituting NB (i) by 0, and applying the equation to
tive to chromia growth and the effect of oxygen or water vapour iron–chromium alloys, it enables the nominal chromium con-
pressure is not significant. When temperature increases, some centration of the alloy required for the development of a compact
differences are observed according to the atmosphere composi- layer of Cr2 O3 to be ascertained [14]:
tion and it appears that the oxidation kinetics speeds up with the  1/2 
oxygen pressure and with the presence of H2 O (Fig. 1c). In this 1 k c (Cr 2 O 3 )
NCr
min
=F (2)
case, chromia is formed as a continuous protective film only at 2 D̃
the smaller temperatures or in atmosphere with small oxygen
pressure. where kc (Cr2 O3 ) is the oxidation constant of chromia and D̃ is
Recently, in the literature, several mechanisms have been the interdiffusion coefficient in the FeCr based substrate.
suggested to justify the oxidation of stainless steels occurring Thus, the smaller D̃, the greater the amount of Cr necessary
in water vapour. Most of them take into account the possible to form the continuous film. In the 304 steel case, 18% Cr is
integrity loss of the chromia scale in such atmospheres [18] not sufficient when temperature (or time) increases due to the
which would justify the behaviour of AISI 304 at temperatures fact that diffusion coefficients are smaller than in the case of
greater than 850 ◦ C. The scale volatilisation has been invoked AISI 439. Thus, at the beginning, even if a thin chromia film
to justify the loss of chromia scale protection. It was already forms, the chromium concentration and diffusivity cannot feed
discussed and concluded that it was negligible in the present the further growth, which leads to the growth of iron oxide, thus
conditions. Moreover, it can be seen in Fig. 1b and c that, even in to an oxidation kinetics significantly accelerated. This does not
dry oxygen, a second stage is obtained with a greater parabolic act under Ar–H2 –H2 O atmosphere as, in this case, the oxygen
constant. The loss of the chromia film integrity could also be pressure is too low to allow the formation of iron oxides.
 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276 275

Table 7
Values of the oxidation constants calculated from diffusion data using Wagner’s theory
Atmosphere kp (g2 cm−4 s−1 )

850 ◦ C 900 ◦ C 950 ◦ C

Polycrystals Films Polycrystals Films Polycrystals Films

O2 1.18 × 10−12 3.17 × 10−12 1.66 × 10−12 6.04 × 10−12 2.34 × 10−12 1.15 × 10−11
Ar–H2 –H2 O 2.04 × 10−13 5.50 × 10−13 2.57 × 10−13 9.31 × 10−13 3.61 × 10−13 1.78 × 10−12
Air 1.14 × 10−12 3.08 × 10−12 1.61 × 10−12 5.85 × 10−12 2.26 × 10−12 1.12 × 10−11
Air + H2 O 1.14 × 10−12 3.08 × 10−12 1.61 × 10−12 5.85 × 10−12 2.26 × 10−12 1.12 × 10−11

It can be noted that the presence of large amount of water the grain boundaries. Bulk and grain boundary diffusion coef-
vapour (Ar–H2 –H2 O mixtures) promotes the formation of ficients of chromium and oxygen in chromia were taken from
whiskers especially at 850 ◦ C (see Figs. 8a and 5b). Such literature data [13,30] and f was considered as being equal to
whiskers are not observed in air + H2 O atmosphere in agreement 3δ/Φ, δ being the grain boundary width and Φ being the grain
with recent works of Hansson and Somers [24]. On the basis of size [31]. Once kc obtained from D values, kp was calculated by
our results, it is not possible to suggest, as they do, a change in the equation:
the oxidation mechanism from controlled by countercurrent dif-  2
fusion of anions and cations in air + H2 O to a control by cation Mox
kc = kp (4)
diffusion only in Ar–H2 –H2 O atmosphere. Moreover, it seems b ρox MO
that silicon oxidises more easily when water vapour is present
(Figs. 5a and 8b) while the reverse occurs for Mn oxidation. The results of our calculations are given in Table 7 and corre-
spond to the formation of Cr2 O3 only. Thus, the best comparison
with our experimental values of the oxidation constant concerns
4.3. Comparison of the two steels
the results obtained in Ar–H2 –H2 O atmosphere as in this gas
mixture, for both steels and the three tested temperatures, the
To conclude on the differences between the two steels on
oxide film is only constituted by chromia (see kinetic curves in
the oxidation resistance, it can be said that the AISI 439 steel
Fig. 1). So, the calculated oxidation constants and the experi-
has a higher oxidation resistance than AISI 304 at temperatures
mental ones obtained in Ar–H2 –H2 O mixture (Fig. 11) present
above 850 ◦ C linked to the growth of a protective chromia film
a reasonable agreement. Moreover, the comparison of the values
in all oxidation conditions. Nevertheless, the oxidation constant
given in Tables 5 and 7, including all atmospheres used in our
is somewhat higher for 439 steel than for 304 steel when con-
work, is sufficiently satisfying to say that Wagner’s theory can
sidering the cases where only Cr2 O3 is formed, i.e. for the first
be applied to a complex alloy. The results given in the literature
stage of the kinetic curves of 304. This is probably an effect
data concerning oxidation of either pure chromium or chromia
of the impurity as Ti is present in 439 steel but not in the 304
forming alloys [32], are also plotted in Fig. 11, showing a rea-
steel, and it oxidises and gives sand roses at the outer surface
sonable agreement even if a noticeable scattering is observed,
of the chromia films (cf. Fig. 3b). Thus, according to literature
due to the great variety of tested alloys and of oxidation.
data [25–27], the ferritic steel exhibits a better high temperature
oxidation resistance than the austenitic steel.

4.4. Oxidation mechanism

In order to confirm the previous suggestion, oxidation con-

stants were calculated by means of Wagner’s theory [11,28] and
compared to the experimental oxidation constants.
The following equation was used to calculate the oxidation
constant kc expressed in cm2 s−1 :
 
zc (pO2 )e
kc = Dcat + DO ln (3)
za (pO2 )i
zc and za being the valence of the cation and anion, respectively.
This equation implies that the defect concentration in the chro-
mia film does not depend on the oxygen pressure as suggested
by previous works [29]. In Eq. (2), Dcat and DO were consid-
Fig. 11. Comparison, in an Arrhenius plot, of the experimental oxidation
ered as effective diffusion coefficients given by Deff = (1 − f) parabolic constants for 304 and 439, and those calculated from diffusion data
Db + fDgb where Db is related to bulk diffusion and Dgb to grain according to Wagner theory, in Ar–H2 –H2 O atmosphere. Data from the literature
boundary diffusion, and f is the fraction of atoms diffusing in concerning growth of chromia film by oxidation are also given.
276 A.M. Huntz et al. / Materials Science and Engineering A 447 (2007) 266–276

5. Conclusions [4] N. Fujita, K. Ohmura, M. Kikuchi, T. Suzuki, S. Funaki, I. Hirishige, Scrip.

Met. 32 (5/6) (1989) 449.
[5] J.P. Baı̈lon, J.M. Dorlot, Des Matériaux, Presses Internationales Polytech-
The oxidation behaviour of two stainless steels, the austenitic niques, 2000.
AISI 304 and the ferritic AISI 439, was studied between [6] G. Amsel, D. David, G. Béranger, P. Poisot, B. De Gélas, P. Lacombe, J.
850 and 950 ◦ C in 1 atm of either oxygen, air, air + H2 O or Nucl. Mater. 29 (1969) 144.
Ar + H2 + H2 O. [7] A.M. Huntz, B. Lefevre, F. Cassino, Mater. Sci. Eng. A 290 (2000)
190–197.
The 439 steel exhibits a very satisfying oxidation resistance
[8] A. Aubry, F. Armanet, G. Béranger, J.L. Lebrun, G. Maeder, Acta Metall.
due to the fact that whatever the temperature and the atmosphere 36 (10) (1988) 2279.
composition tested, its oxidation kinetics follows a parabolic [9] D. Briggs, J.C. Rivière, in: D. Briggs, M.P. Seah (Eds.), Practical Sur-
law, i.e. (M/S)2 = kp t, with kp the parabolic oxidation constant, face Analysis, vol. 1, John Wiley and Sons, Chichester, 1996, pp. 85–141
indicating that the film growth is controlled by diffusion. The (Chapter 3).
[10] A. Atkinson, Rev. Modern Phys. 57 (2) (1985) 437.
oxide scale is mainly made up of Cr2 O3 . Ti rich oxide is present
[11] P. Kofstad, High Temperature Corrosion, Elsevier Applied Science, 1988.
on the outer surface as “sand roses”, and at the lowest tempera- [12] G. Ben Abderrazik, G. Moulin, A.M. Huntz, Oxidat. Met. 33 (3/4) (1990)
ture in the presence of water vapour, whiskers are observed. The 191.
steel–oxide interface is rather straight with silica nodules and [13] S.C. Tsaı̈, A.M. Huntz, C. Dolin, Oxidat. Met. 43 (1995) 581.
internal oxidation of silicon is significant. [14] P. Moulin, A.M. Huntz, P. Lacombe, Acta Metall. 28 (1980) 745.
[15] H. Hindam, D.P. Whittle, Oxidat. Met. 18 (5/6) (1982) 245.
In the case of AISI 304, a parabolic kinetics was observed at
[16] J.H. Park, W.E. King, S.J. Rothman, J. Am. Ceram. Soc. 70 (1987) 880.
low pO2 (Ar/H2 O/H2 ), at all temperatures (850–950 ◦ C), associ- [17] Kvernes, M. Oliveira, P. Kofstad, Corros. Sci. 17 (1977) 237.
ated to Cr2 O3 growth. But, at higher oxygen pressures (O2 , air, [18] A. Galerie, S. Henry, Y. Wouters, M. Mermoux, J.-P. Petit, L. Antoni, Mater.
air + H2 O), a parabolic kinetics was observed only at 850 ◦ C, High Temperatures 22 (1/2) (2005) 105.
and the oxidation rate of the 304 stainless steel shows no sig- [19] C.S. Tedmon, J. Electrochem. Soc. 113 (8) (1966) 766.
[20] W.C. Hagel, Trans. Am. Soc. Met. 56 (1953) 583.
nificant dependence on atmosphere, due to the formation of
[21] J. Paidassi, C. Cadiou, L. Berry, C.R. Acad. Sci. Paris C 226 (1968) 413.
Cr2 O3 . Above 850 ◦ C, a deviation to the parabolic behaviour [22] K. Segerdahl, J.E. Svensson, M. Halvarsson, I. Pans, L.G. Johansson, Mater.
is observed, with two stages, initially formation of Cr2 O3 fol- High Temperatures 22 (1/2) (2005) 69.
lowed by the growth of iron oxides which induces a greater [23] C. Wagner, J. Electrochem. Soc. 99 (1952) 369.
oxidation rate. In presence of H2 O, whiskers rich in Cr and [24] A.N. Hansson, M.A.J. Somers, Mater. High temperatures 22 (3/4) (2005)
223.
Mn can be observed along the substrate grain-boundaries. The
[25] L. Antoni, J.M. Herbelin, Proceedings of EFC Workshop, vol. 27, European
steel–oxide interface is convoluted and internal oxidation of Federation of Corrosion Publications, Frankfurt, 1999, p. 187.
silicon is observed. [26] F. Riffard, H. Buscail, E. Caudron, R. Cueff, C. Issartel, S. Perrier, Appl.
Surf. Sci. 252 (2006) 3697.
[27] S. Henry, Doctor Thesis, INP Grenoble, France, 2000.
References [28] C. Wagner, Z. Phys. Chem. B21 (1933) 25.
[29] A.C.S. Sabioni, B. Lesage, A.M. Huntz, J.C. Pivin, C. Monty, Phil. Mag.
[1] D. Landolt, Traité des Matériaux, Presses Polytechniques et Universitaires A 66 (3) (1992) 333.
Romandes, 1997. [30] S.C. Tsaı̈, Doctor Thesis, University Paris-XI, Orsay, France, 1996;
[2] G. Béranger, J.C. Colson, F. Dabosi (Eds.), Les Editions de Physique, Ecole S.C. Tsaı̈, A.M. Huntz, C. Dolin, Mater. Sci. Eng. A 212 (1996) 6.
d’hiver du CNRS, 1985. [31] E.W. Hart, Acta Metall. 5 (1957) 597.
[3] M. Mantel, B. Baroux, J. Ragot, P. Chemelle, Mémoire et Etudes Scien- [32] A.C.S. Sabioni, A.M. Huntz, J. Philibert, B. Lesage, C. Monty, J. Mater.
tifiques, Revue de Métallurgie, 1990, p. 637. Sci. 27 (1992) 4782.