Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Oxid Met (2013) 79:201–224

DOI 10.1007/s11085-012-9318-2

ORIGINAL PAPER

Transition Between Different Oxidation Modes


of Binary Fe–Si Alloys at 600–800 °C in Pure O2

L. L. Liu • Q. Q. Guo • Y. Niu

Received: 10 July 2012 / Revised: 29 August 2012 / Published online: 25 October 2012
Ó Springer Science+Business Media New York 2012

Abstract The oxidation behavior of three Fe–Si alloys containing approximately


5, 9 and 13 at.% Si has been studied at 600–800 °C under 1 atm O2. Fe–5Si and
Fe–9Si followed multi-stage parabolic kinetics, while Fe–13Si showed more com-
plex kinetics at all temperatures. The increase in Si content resulted in a transition
from the internal oxidation of Si beneath an external scale of iron oxides to the
exclusive external formation of silica. The critical contents required for the tran-
sitions between the various possible oxidation modes of the binary Fe–Si alloys
were calculated and compared with the experimental results. A thermodynamic
mechanism for the formation of Fe-rich oxide nodules observed in the oxidation of
Fe–5Si and Fe–9Si has been proposed.

Keywords Fe–Si alloys  Oxidation modes  Transitions  Nodules

Introduction

The formation of a continuous external SiO2 layer can sharply decrease the high temperature
oxidation rate of metallic materials such as Fe [1, 2], Ni [3] and Fe–Cr steels [4, 5] due to the
very slow rate of growth and to the highly stoichiometric structure of silica, which acts as a
barrier for the diffusion of metal ions. For example, the diffusion of Fe3? through vitreous
SiO2 below 1,000 °C is so slow to be negligible according to Atkinson [6]. The oxidation of
Fe–Si alloys in different atmospheres has been subject of many researches: for instance, this
has been studied in air [7], in CO2 ? 1 % CO mixtures [8–10] and in air ? H2O mixtures
[11]. Conversely, the oxidation of Fe–Si alloys by pure oxygen at high temperatures has
seldom been reported previously, except for the oxidation of Fe–Si thermal spray coatings

L. L. Liu  Q. Q. Guo  Y. Niu (&)


State Key Laboratory for Corrosion and Protection, Institute of Metal Research, Chinese Academy
of Sciences, Wencui Road 62, Shenyang 110016, China
e-mail: yniu@imr.ac.cn

123
202 Oxid Met (2013) 79:201–224

[12] and for the heat treatment of Fe94Si6 [13]. The present work, concerning the oxidation of
three Fe–Si alloys containing approximately 5, 9 and 13 at.% in 1 atm of pure O2 at 600, 700
and 800 °C, is part of a more general study of the corrosion of these alloys by H2–CO2 and
H2–H2S–CO2 mixtures in the same temperature range. The aim of the whole work is to find
materials presenting a good corrosion resistance to sulfidation as well as to sulfidation–
oxidation in atmospheres containing high sulfur pressures and low oxygen pressures similar
to those typical of coal gasification processes [14–18].
Experimental Procedures

Three alloys used for this study contain approximately 5, 9, and 13 at.% Si, named as
Fe–5Si, Fe–9Si and Fe–13Si respectively. Their actual composition, measured by ICP
spectrometry analysis, is shown in Table 1. The alloys were prepared by vacuum
induction melting of mixtures of appropriate amounts of the two pure elements (99.8
wt% Fe and monocrystalline Si) and then cast in cylinders of 5 cm diameter, 10 cm
high. The alloy ingots were subsequently annealed in 1 atm argon at 1,000 °C for 36 h
to remove the residual mechanical stresses and achieve a better alloy equilibration. In
agreement with the Fe–Si phase diagram (Fig. 1), all of the alloys are single-phase a or
a1. The average grain sizes of the three alloys (mm), calculated according to the
measurement in GB/T 6394-2002, were 1.59 (Fe–5Si), 1.12 (Fe–9Si) and 0.506
(Fe–13Si). Specimens with a size of 10 9 8 9 1.5 mm were cut from the ingots using
a line saw and a 1 mm hole was drilled near one edge. All the specimens were
mechanically abraded on successively finer abrasive papers down to 2,000 grit and
finally cleaned with water, acetone and ethanol and dried immediately before each test.
The oxidation experiments were carried out in a microbalance Cahn Thermax 700.
The corroded samples were examined by means of scanning electron microscopy
(SEM), generally using the back-scattered electron image (BEI) mode, energy-
dispersive X-ray microanalysis (EDX), electron probe microanalysis (EPMA) and
XRD to study the structure of the scales, to identify the nature of the phases and
their distribution and for elemental analysis.

Results

Corrosion Kinetics

The kinetic curves for the oxidation of the three Fe–Si alloys in 1 atm O2 at
600–800 °C for 24 h are shown separately in Figs. 2a, 3a and 4a as linear plots and

Table 1 Nominal and actual composition (at.%/wt%) of the three Fe–Si alloys measured by ICP
spectrometry analysis
Nominal (at.%) Actual (at.%/wt%)

Fe–5Si Fe-5.35/2.75Si
Fe–9Si Fe-9.47/4.97Si
Fe–13Si Fe-13.00/6.96Si

123
Oxid Met (2013) 79:201–224 203

Temperature oC

at.% Si

Fig. 1 Binary phase diagram of Fe–Si alloy system (Fe-rich corner)

in Figs. 2b, 3b and 4b as parabolic plots. The corrosion rates of Fe–5Si and Fe–9Si
at these temperatures all followed a multi-step parabolic rate law after an initial
incubation period of small rate and variable duration. The initial and final times for
each stage as well as the corresponding kp values for the oxidation of Fe–5Si and
Fe–9Si at different temperatures are shown in Table 2, while Table 3 gives the kp
values for the oxidation of pure Fe [19–21] and Si [8, 21]. The oxidation kinetics of
Fe–13Si, not clearly evident from these figures because of the quite small mass
gains, are examined separately later.
At 600 °C the corrosion kinetics of Fe–5Si (Fig. 2a, b), after an initially period of
slow rate (stage 1), denoted as an incubation period, showed a stage (stage 2) where
the instantaneous slope of the parabolic plot, denoted as instantaneous parabolic rate
parameter (iprp), increased quickly with time. This was followed by a truly
parabolic period (stage 3), and then by a period presenting an iprp decreasing with
time (stage 4) and finally by a second parabolic stage (stage 5) with a rate constant,
kp, smaller than during stage 3. The oxidation behavior of the same alloy at 700 °C
was similar to that shown at 600 °C, but the drop of kp between the first and the
second parabolic stage was larger than at 600 °C. The kinetics of oxidation of this
alloy at 800 °C were again similar to those already observed at 600 and 700 °C, but
kp for the last parabolic stage was smaller than those measured at 600 and 700 °C,
similar to each other. The kinetic curve measured at 600 °C was lowest, while that
at 800 °C was intermediate between the results observed at 600 and 700 °C.
In the oxidation of Fe–9Si at 600 °C (Fig. 3a, b) an initial incubation period,
longer than for Fe–5Si, was followed by two parabolic stages, connected by a short

123
204 Oxid Met (2013) 79:201–224

(a) 2.0
1.8 600
700
1.6 800
Mass gain, mg/cm2
1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20 25
Time, h
(b) 2.0
600
1.8
700
800
1.6
Mass gain, mg/cm 2

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0
0 1 2 3 4 5
(Time, h)1/2

Fig. 2 Kinetic curves for the oxidation of Fe–5Si in 1 atm O2 at 600, 700 and 800 °C for 24 h: a normal
plots; b parabolic plots

transition period, while kp of the last parabolic stage was significantly larger than
that for the first parabolic stage. Oxidation of this alloy at 700 °C was similar to that
observed at 600 °C: an initial incubation period, longer than at 600 °C, was
followed again by two parabolic stages connected by an intermediate stage of
variable iprp. However, contrary to what observed at 600 °C, kp for the last
parabolic stage was smaller than that for the first parabolic stage. Finally, the
oxidation kinetics of Fe–9Si at 800 °C were similar to those observed at 700 °C, but
with a larger decrease of kp of the last parabolic stage with respect to the first
parabolic stage. For this alloy an increase of temperature produced a decrease of the

123
Oxid Met (2013) 79:201–224 205

(a)
0.8 600
700
800
Mass gain, mg/cm2

0.6

0.4

0.2

0.0
0 5 10 15 20 25
Time, h

(b)
0.8 600
700
800
Mass gain, mg/cm2

0.6

0.4

0.2

0.0
0 1 2 3 4 5
1/2
(Time, h)

Fig. 3 Kinetic curves for the oxidation of Fe–9Si in 1 atm O2 at 600, 700 and 800 °C for 24 h: a normal
plots; b parabolic plots

final mass gains after 24 h, which was particularly large in going from 700 to
800 °C. Also the final parabolic rate constant at 800 °C was considerably smaller
than those measured at 600 and 700 °C.
The curves for the oxidation of Fe–13Si are shown in Fig. 4a as linear plots and in
Fig. 4b as parabolic plots. At 600 and 700 °C the reaction rate increased more rapidly
than according to the linear rate law. The instantaneous slope of the linear plot
increased regularly with time at 600 °C, while at 700 °C it was more irregular,
presenting especially an acceleration after about 22 h. The kinetic curve at 800 °C was
even more irregular, presenting an initial stage of incubation followed by a short

123
206 Oxid Met (2013) 79:201–224

(a) 0.15
600
700
Mass gain, mg/cm2 800

0.10

0.05

0.00
0 5 10 15 20 25
Time, h

(b) 0.15
600
700
800
Mass gain, mg/cm2

0.10

0.05

0.00
0 1 2 3 4 5
(Time, h)1/2

Fig. 4 Kinetic curves for the oxidation of Fe–13Si in 1 atm O2 at 600, 700 and 800 °C for 24 h:
a normal plots; b parabolic plots

period with an iprp increasing with time and then by a short parabolic stage, with a
kp = 1.23 9 10-14 g2 cm-4 s-1. In turn this was followed by a longer period with an
iprp increasing with time and finally by a second parabolic stage with a kp value of
6.42 9 10-16 g2 cm-4 s-1. Altogether, the final mass gain, as well as the final
oxidation rate, was smaller at 800 °C than at 600 and 700 °C. Moreover, the mass
gains at 700 °C were smaller than those measured at 600 °C for the same reaction time,
except after about 22 h, due to a significant rate increase observed at 700 °C in the final
reaction stage. In a previous study [7], the oxidation of four Fe–Si alloys with 5–20
wt% Si at 900–1,100 °C was reported to follow the linear rate law with very small rates

123
Oxid Met (2013) 79:201–224

Table 2 Parabolic rate constants (g2 cm-4 s-1) and durations of the various stages (h) for the oxidation in 1 atm O2 for 24 h at 600–800 °C of Fe–5Si and Fe–9Si
T (°C) Stage1 Stage 2 Stage 3 Stage 4 Stage 5

5Si 9Si 5Si 9Si 5Si 9Si 5Si 9Si 5Si 9Si
(h) (h) (h I) (h I) (h D) (h D)

600 0–0.5 0–1 0.5–1.5 1–5 1.5–6 h 5.96 9 10-11 5–16 h 2.35 9 10-11 6–12 16–24 12–24 h 2.0 9 10-11 –
-10 -11
700 0–0.2 0–1 0.2–0.5 1–5 0.5–2.5 h 1.79 9 10 5–11 h 1.3 9 10 2.5–14 11–24 14–24 h 1.34 9 10-11 –
800 0–0.1 0–0.6 0.1–0.5 0.6–1.2 0.5–3 h 1.5 9 10-10 1.6–2.2 h 8.01 9 10-12 3–9 2.2–7 9–24 h 4.59 9 10-12 7–24 h 1.77 9 10-13
207

123
208 Oxid Met (2013) 79:201–224

Table 3 Parabolic rate constants (g2 cm-4 s-1) for the oxidation of pure Fe [19–21] and Si [8, 21] in
1 atm O2 for 24 h at 600–800 °C
T (°C) Fe Si

600 1.6 9 10-10 3.12 9 10-15


-8
700 1.06 9 10 2.34 9 10-15
-8
800 6.52 9 10 9.24 9 10-15

which were mostly attributed to the growth of an external layer of Fe2O3 produced by
diffusion of Fe through the innermost silica layer.
The kinetic curves for the oxidation of the three alloys at 600, 700 and 800 °C are
finally compared to each other in Fig. 5a–c. At all temperatures, an increase of the
Si content of the alloy produced a large decrease of the mass gains. In particular,
this effect was quite significant when the Si content increased from 5 to 9 at.%,
especially at high temperatures. In all cases, the overall mass gains measured for
Fe–13Si were much smaller than those observed for the other two alloys.

Morphology and Structure of the Scales

At 600 °C, the scale formed on Fe–5Si after 24 h (Fig. 6a–c) contained two main
layers: an external porous Fe2O3 layer, as shown by XRD (Fig. 6d), including a
little Fe3O4 at the bottom, as indicated by the arrow in Fig. 6c, and an inner-scale
layer composed of a mixture of FeO ? Fe2SiO4, and similarly to what had been
already reported in the past [11]. Both layers had an irregular thickness and were
followed by a zone of internal oxidation of Si. The particles of SiO2 in the internal
oxidation zone (ioz) were quite small and distributed uniformly. At 700 °C, most of
the scale formed on Fe–5Si (Fig. 7) was similar to that observed at 600 °C, but with
smaller contents of Fe3O4 and FeO ? Fe2SiO4, while the SiO2 particles of the ioz
were much larger than at 600 °C. In addition, a fraction of about 10 % of the sample
surface was covered by a very thin scale composed of two layers of irregular
thickness (Fig. 7c): the outer lighter layer was of iron oxide, while the inner dark
layer was much richer in silicon. No internal oxidation of Si was present beneath
this kind of scale. The scale formed on Fe–5Si at 800 °C (Fig. 8) was of two kinds,
each occupying about 50 % of the surface: the thin and dark scale (Fig. 8a) was rich
in Si, while the large nodules (Fig. 8b) presented two main layers: the most external
region was composed of Fe2O3 with little Fe3O4, while the inner region contained a
mixture of FeO and Fe2SiO4 (Figs. 8c–d). Again, the internal oxidation of Si was
only present beneath the nodular areas.
Oxidation of Fe–9Si at 600 °C (Fig. 9) produced a double type of scale: a
fraction of the surface (about 50 %) was covered by oxide nodules (Fig. 9a), while
the remaining fraction presented a very thin scale rich of Si (Fig. 9b). The nodules
had the usual structure (Fig. 9c), containing Fe oxides in the outer region (Fe2O3,
light gray, outside and Fe3O4, middle gray, inside) plus a mixture of Fe and Si
oxides (darker than Fe3O4) close to the alloy. They were followed by a region
of internal oxidation of Si, which was lacking beneath the thin kind of scale.

123
Oxid Met (2013) 79:201–224 209

(a) 1.6
5Si
1.4 9Si
13Si
1.2

Mass gain, mg/cm2 1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20 25
Time, h
(b) 2.0
5Si
1.8 9Si
13Si
1.6
Mass gain, mg/cm2

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20 25
Time, h
(c)
1.6 Fe-5Si
Fe-9Si
1.4 Fe-13Si

1.2
Mass gain, mg/cm2

1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20 25
Time, h

Fig. 5 Kinetic curves for the oxidation of three Fe–xSi (x = 5, 9 and 13 at.%) alloys in 1 atm O2 for
24 h at 600 °C (a), 700 °C (b) and 800 °C (c)

123
210 Oxid Met (2013) 79:201–224

(a)

20µm

(b)
Fe2O3

Fe/SiOx

5µm i.o.z.

(c)

FeOx

10µm

(d) 1 Fe 2O3
1 1

1 1
1 1 1 1 1
1 1 1

20 30 40 50 60 70 80 90
2θ(degree)

Fig. 6 Micrographs of cross sections of Fe–5Si oxidized in 1 atm O2 at 600 °C for 24 h. a (SEM–BEI):
general view; b enlarged view of the selected area of a; c general view (OM); d XRD spectrum of the
external scale

123
Oxid Met (2013) 79:201–224 211

(a)

i.o.z.

20µm

(b)

20µm

(c)

5µm

Fig. 7 Micrographs of cross sections of Fe–5Si oxidized in 1 atm O2 at 700 °C for 24 h. a, b General
views (SEM–BEI); c enlarged view of the selected area of b

By oxidation of Fe–9Si at 700 °C (Fig. 10) the fraction of the alloy surface covered
with a thin scale (about 800 nm) rich in Si (Fig. 10a) increased up to about 80 %
(the white outermost layer in these figures is due to Ni plating), while the remaining
fraction presented large nodules (Fig. 10b). The external region of these nodules
was composed of an outer lighter layer of Fe2O3 and an inner darker layer of Fe3O4,
while the internal region contained the mixture of Fe and Si oxides (darker than
Fe3O4) (Fig. 10b, c). Beneath the nodules there was a thick region of internal
oxidation of Si, where the SiO2 particles were rather large and approximately
spherical.
The scale formed on Fe–9Si by oxidation at 800 °C (Fig. 11) was mostly made of
a SiO2-rich layer about 3 lm thick (Fig. 11a), present over about 95 % of the
sample surface. In addition, there were still some large nodules (Fig. 11b)
presenting three main layers: an outer layer of Fe2O3, an intermediate layer of Fe3O4

123
212 Oxid Met (2013) 79:201–224

(a)

5µm

(b)

50µm

(c)
Fe2O3

Fe/SiOx

i.o.z.
10µm

(d)

20µm

Fig. 8 Micrographs of cross sections of Fe–5Si oxidized in 1 atm O2 at 800 °C for 24 h. a (SEM–BEI):
general view of the Si-rich scale; b (SEM–BEI): general view of the scale containing nodules; c enlarged
view of the inner scale region of b; d (OM): enlarged view of a nodule

123
Oxid Met (2013) 79:201–224 213

(a)

200 µm

(b)

2µm

(c)
Fe2O3

FeOx
Fe/SiOx

i.o.z
10µm

Fig. 9 Micrographs of cross sections of Fe–9Si oxidized in 1 atm O2 at 600 °C for 24 h. a (SEM–SEI):
scale surface; b (SEM–BEI): enlarged view of the Si-rich scale; c (OM): enlarged view of a nodule

with some silicon and an innermost dark layer containing a mixture of FeO ?
Fe2SiO4. Finally, these nodules were followed again by a region of internal
oxidation of Si. The special case of Fig. 11b shows an accumulation of SiO2
particles at the interface between the ioz and the alloy, suggesting that the nodule
may be subsequently healed with time.
The scale grown on Fe–13Si by oxidation at 600 °C (Fig. 12) was mostly made
of a dark layer about 800 nm thick of vitreous SiO2 (Fig. 12a). Only over about 5 %
of the sample there were small nodules, followed a thin region of internal oxidation
of Si (Fig. 12b, c) (the white outermost layer in these figures is due to Ni plating).
Oxidation of Fe–13Si at 700 °C (Fig. 13) formed a continuous, protective single
SiO2 layer, about 3 lm thick, which improved clearly its oxidation resistance. The
oxide nodules were substantially absent at this temperature. Finally, oxidation of
Fe–13Si at 800 °C (Fig. 14) developed a continuous layer of silica about 5 lm thick

123
214 Oxid Met (2013) 79:201–224

(a)
Ni-plated

2µm

(b)
Ni-plated

20µm

(c)
Ni-plated
Fe2O3
FeOx

Fe/SiOx
20µm
i.o.z.

Fig. 10 Micrographs (SEM–BEI) of cross sections of Fe–9Si oxidized in 1 atm O2 at 700 °C for 24 h.
a General view of the Si-rich scale; b, c enlarged views of nodules

with a very thin external Fe2O3 layer, while neither oxide nodules nor zones of
internal oxidation of Si were present.
Altogether, an increase of the Si content and of the reaction temperature
produced an increase of the area on the alloy surface covered by the layer rich in
SiO2, until it became continuous with the alloy richest in Si, especially at the highest
temperature tested.

Discussion

During the incubation stage, the alloys developed a protective kind of scale, as
already observed for Fe–1.5Si [11] and for Fe–Al [22] alloys, associated with very

123
Oxid Met (2013) 79:201–224 215

(a)

3µm

(b)
Fe2O3

Fe/SiOx

20µm
i.o.z.

Fig. 11 Micrographs (SEM–BEI) of cross sections of Fe–9Si oxidized in 1 atm O2 at 800 °C for 24 h.
a General view of the Si-rich scale; b general view of a nodule

slow mass gains. With further oxidation, Fe-rich oxide nodules formed at some
locations over the initial scale grown on Fe–5Si and Fe–9Si, producing the kinetic
breakaway corresponding to stage 2. The subsequent parabolic stages of slower
oxidation rates corresponded very likely to a partial healing of the nodules by a
layer of SiO2 in contact with the scale. However, this may also be partly due to the
enrichment of Si-rich oxides within the nodules, which grow less rapidly than pure
iron oxides.
The formation of Fe-rich oxide nodules has often been reported in the oxidation
of binary and ternary Fe-based alloys such as Fe–Si [7, 11], Fe–Cr [23, 24], Fe–Ni–
Cr [25] and Fe–Cr–Al [26–28]. According to previous authors [23–28] they can
form either from the beginning of oxidation or after an initial incubation stage of
slow oxidation in the presence of a thin layer of a protective oxide such as chromia
or silica. In the former case the nodules can simply grow due to the continued
expansion of iron oxide nuclei already present during an initial transient stage when
all the alloy components form their respective oxides. If, on the contrary, the
nodules grow after the establishment of a protective layer, a situation essentially
applying to the present results, the mechanism proposed involves a local breakdown
of the protective scale, very likely as a consequence of a localized penetration of the
oxidant from the gas phase through some physical defects of the scale [27, 29]. In
this case the less stable oxide can form even more easily than during the initial stage
because the reactive element may be severely depleted in the alloy next to the
interface with the scale. The continued growth of the nodules is strongly favored if

123
216 Oxid Met (2013) 79:201–224

(a)
Ni-plated

1µm

(b)

Ni-plated
c

10µm

(c) Ni-plated

i.o.z.

1µm

Fig. 12 Micrographs (SEM–BEI) of cross sections of Fe–13Si oxidized in 1 atm O2 at 600 °C for 24 h.
a General view of the Si-rich scale; b general view showing two small nodules; c enlarged view of
selected area of b showing internal oxidation

the reactive component is oxidized internally beneath the nodule itself, as observed
in the present system. Conversely, the internal oxidation may be suppressed and the
oxidation rate strongly decreased if after some time a healing layer of the oxide of
the reactive component forms at the base of the nodule.
Before discussing a possible alternative mechanism for the breakaway of the
initial silica-rich scales after an incubation period, it is instructive to develop a
theoretical prediction of the critical Si contents needed for the transitions between
the different oxidation modes expected for binary A–B alloys, where A is the most
noble component, exposed to high temperatures and high oxygen pressures
sufficient to form the oxides of both metal components, denoted as AO and BOm,
(high oxygen pressures). This reaction may in principle produce three different
scaling modes, i.e.: 1) exclusive growth of external AO scales, generally coupled to

123
Oxid Met (2013) 79:201–224 217

2µm

Fig. 13 Micrograph (SEM–BEI) of a cross section of Fe–13Si oxidized in 1 atm O2 at 700 °C for 24 h

3µm

Fig. 14 Micrograph (SEM–BEI) of a cross section of Fe–13Si oxidized in 1 atm O2 at 800 °C for 24 h

an internal oxidation of B (OM1), 2) formation of external scales composed of


mixtures of the oxides of both A and B in the absence of any internal oxidation of B
(OM2), and finally 3) exclusive growth of an external scale of BOm (OM3) [30–32].
The first mode (OM1) is typical of A–B alloys dilute in B, while the third mode
(OM3) is characteristic of alloys sufficiently rich in B to make AO unstable in
contact with the alloy. The second mode (OM2) may appear at B contents
intermediate between those typical of the other two modes. However, according to a
previous analysis [32], the binary alloys may actually present all the three or only
two oxidation modes, depending on the relative magnitude of the critical B contents
needed for the transitions between OM1 and OM2 and between OM2 and OM3,
denoted as NB°*(1) and NB°*(2), respectively. More precisely, if NB°*(1) \ NB°*(2)
(condition 1), the alloys will present all the three oxidation modes in sequence with
an increase of the B content, involving two different transitions. In fact, an increase
of the bulk B content in the alloy, NB°, will produce first a transition from OM1 to
OM2 (transition1) as soon as NB° [ NB°*(1). The mode OM2 will remain stable as
long as NB°*(1) \NB° \NB°*(2), while a second transition from OM2 to OM3 will
occur when NB° [ NB°*(2). Conversely, if NB°*(1) [ NB°*(2) (condition 2) only
two oxidation modes are possible, i.e. OM1 and OM3, while mode OM2 is n NB°
exceeds NB°*(1) BOm forms directly as an external scale, so that only one transition,
i.e. that between OM1 and OM3, is possible.

123
218 Oxid Met (2013) 79:201–224

According to a criterion proposed by Wagner [31], the transition from internal to


external oxidation of B for binary A–B alloy occurs when the volume fraction of BOm,
fv, in the ioz exceeds a critical value, fv*, usually set equal to 0.3 after a study of the
oxidation of Ag–In alloys [33, 34]. In the presence of an external AO scale, the critical
content of B needed for this transition, NB°*(1), is calculated by solving a system of two
equations. The first equation, concerning the kinetics of internal oxidation, is [35, 36]
NO s ¼ mNB  ½GðcÞ=FðhÞ½erf ðcÞ  erf ðuÞ=erf ðcÞ ð1Þ
where NOs is the solubility of oxygen (mole fraction) in the alloy under the oxygen
pressure for the equilibrium of A/AO, and m is the O/B ratio in BOm, which here is equal
to two. Moreover, c is a dimensionless parameter related to the parabolic rate constant
of internal oxidation according to the equation n2 = 4 c2 DO t = kn t, while n is the
distance of the front of internal oxidation zone from the original alloy surface and DO
the diffusion coefficient of oxygen in the base metal A. In addition, u is equal to ‘
[kc(AO)/DO]1/2, where kc(AO) is the parabolic rate constant of the growth of AO scale
in terms of thickness of metal consumed, X, according to the equation X2 = kc(AO) t.
Moreover, h = c/1/2, with / = DO/DB, where DB is the diffusion coefficient of B in
the base alloy. Finally, G(r) and F(r) are two auxiliary functions defined as [37]
 
GðrÞ ¼ p1=2 r exp r2 erf ðrÞ and FðrÞ ¼ p1=2 r exp r2 ½1  erf ðrÞ
The second equation is [31, 32]

NB  ð1Þ ¼ f v FðhÞ=q ð2Þ
where q is the ratio between the molar volumes of BOm and the alloy.
In turn, the calculation of NB°*(2) involves a second criterion, still defined by
Wagner [30], which requires that the B content at the alloy-scale interface during
the oxidation does not fall below the value for the equilibrium between the alloy and
the oxides of both A and B, NBeq. On this basis, NB°*(2) is given by [30, 32]

NB  ð2Þ ¼ Neq eq
B þ ð1  NB ÞFðuB Þ ð3aÞ
where uB is equal to ‘[kc(BOm)/DB]1/2 and kc(BOm) is the parabolic rate constant for
the growth of BOm in terms of thickness of metal consumed, Y, according to the
equation Y2 = kc(BOm) t. If the difference between the thermodynamic stability of
AO and BOm is sufficiently large, as in the present case, NBeq in Eq. (3a) is negli-
gible, so that one has simply

N B  ð 2 Þ ffi Fð u B Þ ð3bÞ
The mole fraction of oxygen dissolved in a-Fe under the oxygen pressure for the
Fe/FeO equilibrium is given by [38]
NsO ¼ 1:5 expð128:33 kJ=mol=RTÞ
while the diffusion coefficient of oxygen in a-Fe is given by [39]
DO ¼ 1:79  103 expð85:7kJ=mol=RTÞcm2 s1

123
Oxid Met (2013) 79:201–224 219

Moreover, the diffusion coefficient of Si in a-Fe, DSi, is given by Borg and Lai in
the form [40]
DSi ¼ 0:735  ð1 þ 12:4NSi Þexpð219:88kJ=mol=RTÞcm2 s1
where NSi is the mole fraction of Si in the alloy.
Before proceeding, it is noted that the parabolic rate constant kc(AO) appearing
in the definition of the kinetic parameter u in Eq. (1) is actually a measure of the rate
of inward displacement of the alloy/scale interface. This parameter is important
because it affects the critical B content required for transition 1: in fact, it reduces
the degree of enrichment of B inside the ioz and thus produces an increase of
NB°*(1) with respect to the case of absence of an external scale, an effect which is
larger for larger values of kc(AO). The rate constant involved in Eq. (1) is generally
denoted in the form reported above, i.e. as kc(AO), because it is commonly assumed
that the external scales growing on A–B alloys undergoing an internal oxidation of
B are substantially composed of the oxide of A. The situation of the present Fe–Si
alloys is considerably different from this ideal model because the corresponding
external scales contain also significant amounts of silicon compounds. In addition,
these scale contain mostly Fe2O3, while the lower oxides, Fe3O4 and FeO, which in
the oxidation of pure iron are quantitatively prevailing, especially FeO, and grow
more rapidly than Fe2O3, are only present in minor amounts. Conversely, the Si-rich
oxides present in the external scales give also a contribution to the inward
displacement of the alloy/scale interface, which should be included in an
appropriate evaluation of kc(AO). Therefore, reliable values of this rate constant
should be smaller than the corresponding rate constants measured in the oxidation of
pure iron, kc(FeOx), but larger than those corresponding to the growth of the only
oxide Fe2O3, which in any case is almost impossible to evaluate correctly in the
resent complex situation.
For the reasons outlined above, approximate NB°*(1) values are initially calculated
using the parabolic rate constants for the oxidation of pure iron, kp(FeOx) and
kc(FeOx). On the basis of what recalled above, this procedure gives upper limiting
values of this parameter. More reliable values are subsequently calculated below using
experimental data about the kinetics of oxidation of these alloys. The parabolic rate
constant for the oxidation of pure iron in oxygen in terms of mass gain, kp(FeOx), is
[20, 41]
kp ðFeOx Þ ¼ 3:05 expð157:539kJ=mol=RTÞg2 cm4 s1
which, after transformation into the value expressed in terms of thickness of metal
consumed, kc(FeOx) becomes equal to
kc ðFeOx Þ ¼ 0:60 expð157:539kJ=mol=RTÞcm2 s1
Similarly, the parabolic rate constant for the oxidation of Si in terms of the thickness
of silica scales formed, ky(SiO2), is [8, 21]
ky ðSiO2 Þ ¼ 1:1  109 expð119:4kJ=mol=RTÞcm2 s1
which, when expressed in terms of thickness of metal consumed, becomes

123
220 Oxid Met (2013) 79:201–224

kc ðSiO2 Þ ¼ 3:11  1010 expð119:4kJ=mol=RTÞcm2 s1 :


Finally, use of the density of silica glass (2.2 g cm-3) yields q = 3.846.
Introduction of these parameters into Eqs. (1) and (2) yields NB°*(1) = 7.8 9 10-2
at both 700 and 800 °C. Conversely, the critical Si level for the transition from OM2
to OM3 is calculated by introducing DSi and kc(Si) into Eq. (3b), yielding
NB°*(2) = 8.7 9 10-3 and 5.0 9 10-3 at 700 and 800 °C, respectively.
Extrapolation of the parabolic rate constants for the oxidation of Fe reported by Chen
and Yuen [20] down to 600 °C yields kp(FeOx) = 2.3 9 10-10 g2 cm-4 s-1. However,
an experimental measurement of the oxidation of iron in a dry mixture 95 % N2 ? 5 %
O2 at 600 °C yields kp(FeOx) = 1.6 9 10-10 g2 cm-4 s-1 [19]: the same value applies
also to oxidation of Fe under 1 atm O2 because this rate constant is expected to be
practically independent of the external oxygen pressure when this exceeds the value
required for the stability of Fe2O3 [42]. The corresponding rate constant in terms of
thickness of metal consumed, kc(FeOx), becomes equal to 3.2 9 10-11 cm2 s-1.
Extrapolation of the parabolic rate constant for the oxidation of Si reported by Deal and
Grove [21] down to 600 °C yields ky(SiO2) = 2.3 9 10-17 cm2 s-1. However, since
this temperature is rather far from the range investigated by those Authors, the oxidation
of pure Si at 600 °C has been studied experimentally. The kinetic curve was rather
irregular, but presented two quasi-parabolic stages: the kp value of second stage, which
lasted for 15 h up to the end of the experiment, was smaller than that of the first stage and
was equal to 3.1 9 10-15 g2 cm-4 s-1. When expressed in terms of thickness of metal
consumed this kp yields kc(Si) = 4.4 9 10-16 cm2 s-1. Introducing the appropriate
parameters at 600 °C into Eqs. (1) and (2) yields NB°*(1) = 7.7 9 10-2, while use of
Eq. (3b) yields NB°*(2) = 6.1 9 10-2.
The combination of the two critical values of Fe–Si alloys calculated above
shows that at all temperatures NB°*(1) [NB°*(2), corresponding to condition 2
mentioned earlier. On this basis, the oxidation of Fe–Si alloys is expected to involve
only one transition, i.e. that from OM1 to OM3. In practice, of the three Fe–Si alloys
used in the present study only Fe–9Si and Fe–13Si have an Si content larger than
NB°*(1), i.e. sufficient to form external SiO2 scales, while the Si content of Fe–5Si is
smaller than this limit. On this basis Fe–5Si is expected to form external scales
based on iron oxides, coupled to the internal oxidation of silicon, while Fe–9Si and
Fe–13Si should form exclusively external silica scales. This prediction is only in
partial agreement with the present experiment results because a real protective
behavior, corresponding to the oxidation mode OM3, was only observed for
Fe–13Si, especially at 800 °C, while Fe–9Si tended initially to form a kind of
protective scale, as shown by the initial stages of slow reaction rate, but later
developed also numerous Fe-rich oxide nodules coupled to the internal oxidation of
Si. For this alloy the fraction of surface covered by thin scales ranged from 50 % at
600 °C to 80 % at 700 °C and to 95 % at 800 °C (approximate values). A similar
behavior was also shown by Fe–5Si, for which, however, the fraction of surface
covered by a thin Si-rich scale was smaller than for Fe–9Si and increased again with
temperature, rising from nearly zero at 600 °C to about 10 % at 700 °C and to about
50 % at 800 °C. Thus, the actual behavior of the two more dilute alloys is
intermediate between the oxidation modes OM1 and OM3, with a tendency of the

123
Oxid Met (2013) 79:201–224 221

last mode to become more important with an increase of the Si content of the alloy
and of temperature. Conversely, the oxidation mode OM2 is not shown by any alloy.
Before attempting to propose an interpretation of the peculiar oxidation behavior
shown by the present alloys, it is noted that both Fe–5Si and Fe–9Si show
incubation stages of variable duration with very small oxidation rates, indicating an
initial formation of rather protective scales. Thus, a second and more reliable
estimate of NB°*(1) is carried out by using the approximate kp values obtained from
the initial slopes of the parabolic plots of Fe–9Si, which amount to 7.2 9 10-14,
1.9 9 10-15 and 5.0 9 10-14 g2 cm-4 s-1 at 600, 700 and 800 °C, respectively.
Introduction of the corresponding kc values (equal to 1.4 9 10-14, 1.9 9 10-15 and
9.9 9 10-15 cm2 s-1, respectively) into Eqs. (1) and (2) yields NB°*(1) values of
5.1 9 10-2 at 600 °C and 5.2 9 10-2 at 700 and 800 °C.
These last NB°*(1) values, reported in Table 4 together with the corresponding
values of NB°*(2), are significantly smaller than those calculated earlier, but are also
smaller than the actual Si content of Fe–5Si, so that, according to these data, all the
three alloys should in principle be able to form external silica scales. This
conclusion is in agreement with the ability of both alloys to produce thin scales over
a fraction of their surface, especially for Fe–9Si and at 700 and 800 °C.
The appearance of Fe-rich nodules is very interesting. A possible interpretation
of this behavior is based on the following thermodynamic considerations. The three
compounds which may grow in contact with Fe–Si alloys sufficiently dilute in Si are
FeO, SiO2 and the double oxide Fe2SiO4, each of which is only stable in a limited
range of alloy composition. Using the standard free energy of formation of the two
pure oxides and the standard free energy of formation of Fe2SiO4, which at 500 °C
is evaluated as -1,195 kJ/mol [8], the standard free energy change associated with
the reaction
2FeO þ SiO2 ¼ Fe2 SiO4
at 500 °C is equal to -17.66 kJ/mol. Assuming that it remains constant even at
higher temperatures, it is possible to calculate the ranges of NSi values where each
compound is stable. Considering for brevity only the case of 600 °C, it is concluded
that FeO is stable in contact with the alloy only if NSi \ 3.78 9 10-21. Conversely,
SiO2 is only stable if NSi [ 4.90 9 10-19. Within these two limits the compound
stable in contact with the alloy is Fe2SiO4. Similar conclusions are also reached at
700 and 800 °C, even if both critical Si contents increase with temperature.
The previous considerations allow to formulate the following tentative interpre-
tation of the sequence of events occurring during the initial stages of the oxidation

Table 4 Critical NB° values for the transitions between the various oxidation modes of the Fe–Si alloys
in 1 atm O2 at 600–800 °C
T (°C) Nb°* (1) Nb°* (2)

600 5.1 9 10-2 6.1 9 10-2


-2
700 5.2 9 10 8.7 9 10-3
-2
800 5.2 9 10 5.0 9 10-3

123
222 Oxid Met (2013) 79:201–224

of the two most dilute Fe–Si alloys. At the very beginning of the reaction all the
alloys form the oxides of both Fe and Si, roughly in proportion of the concentration
of the respective elements in the bulk alloy, due to the absence of a situation of
equilibrium at the alloy/scale interface. As reaction proceeds, a condition of
equilibrium, gradually established at this site, favors the formation of silica over
most of the alloy surface, including also the areas initially covered by iron oxide,
which will then be cut out of contact with the alloy. However, at places where the
silica layer is free from iron oxide it may continue to grow due to a continuous
sufficient supply of silicon from the bulk alloy. Conversely, where the silica layer is
surmounted by iron oxide the two compounds tend to react by forming the spinel
Fe2SiO4. Since the rate of growth of this compound is certainly considerably faster
than that of silica, the Si content of these two alloys may be insufficient to support
its continued growth. As a result, the alloy in contact with the silicate layer will be
strongly depleted of Si until Fe2SiO4 becomes unstable. After this, the alloy will
start to form iron oxide, resulting in the development of the nodules associated with
the internal oxidation of silicon. In turn, their appearance will involve a significant
increase of the overall reaction rate, which represents a kind of weighted average
between the contribution by the two kinds of scale. Even if the coexistence of two
different kinds of scale over the same sample is surprising, it is justified by the fact
that silica is stabilized by the continuous supply of Si from the bulk alloy, due to the
absence of its local internal oxidation, while the nodules are stabilized by the lack of
supply of Si which is blocked inside the alloy by internal oxidation. Thus, the two
kinds of scale may continue to grow side by side, even if with time they may partly
convert into each other around their zones of contact. Conversely, the Si content of
Fe–13Si is apparently sufficient to support the formation of SiO2 over the whole
alloy surface.
As noted above, the fraction of sample surface covered by the Si-rich scale for
the same alloy increases with temperature, while for a given temperature it increases
with the bulk Si content. The reduction of the parabolic rate constants between the
first and second parabolic stage observed for both Fe–5Si and Fe–9Si seem to be
related to a change of the fraction of alloy surface covered by the protective scale,
which apparently increases with time. It is also possible that this decrease is partly
due to the process of healing of the nodules by formation of a Si-rich scale in
contact with the alloy. Conversely, the bulk Si content of Fe–13Si is sufficient to
make silica stable in contact with the alloy directly from the start of oxidation, even
if some exceptions are noted at 600 and 700 °C.
The previous analysis suggests that the main mechanism of formation of the
oxide nodules in the Fe–Si system is of chemical nature and is associated with a
destabilization of silica at places where it is surmounted by a layer of iron oxide due
to a local depletion of Si at the alloy/scale interface produced by the transformation
of silica into the Fe–Si double oxide. The alternative mechanism of development of
oxide nodules recalled above based on the penetration of oxygen down to the alloy
through physical discontinuities of the scale seems less likely. In fact, it is not clear
why its effectiveness could change so strongly with temperature and Si content,
until it disappears completely for Fe–13Si at 800 °C. Furthermore, use of acoustic
emission during the oxidation four Fe–Si alloys at 900–1,100 °C showed that there

123
Oxid Met (2013) 79:201–224 223

was no measurable cracking of the silica scales during the tests [7]. Thus, it is
believed that the chemical alternative proposed here is the most likely and important
mechanism for the formation of double types of scales observed in the present
system.

Conclusions

The oxidation of three Fe–Si alloys containing from 5 to 13 at.% Si in 1 atm O2 at


600–800 °C produced only one transition between the three possible oxidation
modes of binary alloys, i.e. that from the internal oxidation of Si beneath external
iron oxide scales to the exclusive external growth of SiO2. The critical Si contents
required for this transition, NB°*(1), calculated from the available thermodynamic
and kinetic data using the parabolic rate constants for the oxidation of the two pure
elements are equal to 7.7 9 10-2 at 600 °C and to 7.8 9 10-2 at both 700 and
800 °C. Smaller NB°*(1) values, calculated by using the parabolic rate constants
measured during the initial stage of protective behavior exhibited by Fe–5Si and
particularly by Fe–9Si, predict that all the alloys should be able to form external
silica scales. However, in practice only Fe–13Si, due to its large Si content, was able
to form continuous silica layers preventing almost completely the growth of iron
oxides. Conversely, Fe–5Si and Fe–9Si formed rather protective Si-rich scales only
during a short initial stage, but later developed fast-growing nodules of iron oxide
over a fraction of their surface, resulting in the formation of mixed scales with
complex kinetics. This behavior is attributed to a breakdown of an initial silica layer
over places where silica reacts with external islands of iron oxide forming a layer of
iron silicate growing more rapidly that silica and becoming quickly unstable,
leading to a localized growth of iron oxides. In general, the addition of Si to iron
reduced quite significantly the oxidation rate of the base metal.

Acknowledgments A financial support by the National Natural Scientific Foundation of China (NSFC)
under the research project No. 50971129 is gratefully acknowledged.

References

1. L. S. Darken, Transactions AIME 150, 1942 (157).


2. C. W. Tuck, Corrosion Science 5, 1965 (631).
3. D. L. Douglass, P. Nanni, C. De Asmundis and C. Bottino, Oxidation of Metals 28, 1987 (309).
4. D. Caplan and M. Cohen, Transactions AIME 194, 1952 (1057).
5. G. C. Wood, J. A. Richardson, M. G. Hobby and J. Bottino, Corrosion Science 11, 1971 (659).
6. A. Atkinson and J. W. Gardner, Corrosion Science 21, 1981 (49).
7. T. Adachi and G. H. Meier, Oxidation of Metals 27, 1986 (347).
8. A. Atkinson, Corrosion Science 22, 1982 (87102).
9. P. T. Moseley, G. Tappin and J. C. Riviere, Corrosion Science 22, 1982 (69).
10. P. T. Moseley, G. Tappin, J. A. A. Crossley and J. C. Riviere, Corrosion Science 23, 1983 (901).
11. M. Fukumoto, S. Maeda, S. Hayashi and T. Narita, Oxidation of Metals 55, 2001 (401).
12. J. Porcayo-Calderon, E. Brito-Figueroa and J. G. Gonzalez-Rodriguez, Material Letters 38, 1999
(45).
13. A. R. Lashin and O. Schneeweiss, Czechoslovak Journal of Physics 56, 2006 (23).
14. R. A. Perkins, in High Temperature Corrosion, ed. R. A. Rapp, (NACE, Houston, 1983), p. 345.

123
224 Oxid Met (2013) 79:201–224

15. K. Natesan, Corrosion 41, 1985 (646).


16. F. Gesmundo, D. J. Young and S. K. Roy, High Temperature Material Processes 8, 1989 (149).
17. J. Stringer, in High Temperature Oxidation and Sulfidation Processes ed. J. D. Embury, (Pergamon
Press, New York, 1990), p. 257.
18. F. Gesmundo, in High Temperature Materials for Power Engineering, vol. 1, ed. E. Bachelet (Kluwer
Academic Publishers, Dordrecht, 1990), p. 67.
19. B. Pujilaksono, T. Jonsson, M. Halvarsson, J. Svensson and L. Johansson, Corrosion Science 52,
2010 (1560).
20. R. Y. Chen and W. Y. D. Yuen, Oxidation of Metals 59, 2003 (433).
21. B. E. Deal and A. S. Grove, Journal of Applied Physics 36, 1965 (3770).
22. Z. G. Zhang, F. Gesmundo, P. Y. Hou and Y. Niu, Corrosion Science 48, 2006 (741).
23. J. M. Perrow and W. W. Smeltzer, Journal of the Electrochemical Society 109, 1962 (1023).
24. G. C. Wood and D. P. Whittle, Corrosion Science 7, 1967 (763).
25. G. C. Wood, Oxidation of Metals 2, 1970 (11).
26. P. Tomaszewicz and G. R. Wallwork, Reviews on High Temperature Materials 4, 1978 (75).
27. F. H. Stott, G. C. Wood and M. G. Hobby, Oxidation of Metals 3, 1971 (103).
28. M. Lambertin, A. Stoklosa and W. W. Smeltzer, Oxidation of Metals 15, 1981 (355).
29. R. Prescott and M. J. Graham, Oxidation of Metals 38, 1992 (73).
30. C. Wagner, Journal of the Electrochemical Society 99, 1952 (369).
31. C. Wagner, Zeit Elektrochem 63, 1959 (772).
32. F. Gesmundo and Y. Niu, Oxidation of Metals 50, 1998 (1).
33. R. A. Rapp, Corrosion 21, 1965 (382).
34. R. A. Rapp, Acta Metallurgica 9, 1961 (730).
35. F. Maak, Zeitschrift für Metallkunde 52, 1961 (545).
36. F. Gesmundo and F. Viani, Oxidation of Metals 25, 1986 (269).
37. J. Crank, The Mathematics of Diffusion, (Oxford University Press, New York, 1994).
38. W. Frank, H. J. Engell and A. Seeger, Transactions of the Metallurgical Society 242, 1968 (749).
39. J. Takada, S. Yamamoto, S. Kikuchi and M. Adachi, Oxidation of Metals 25, 1986 (93).
40. R. J. Borg and D. Y. F. Lai, Journal of Applied Physics 41, 1970 (5193).
41. J. Paidassi, Acta Metallurgica 3, 1955 (447).
42. N. Birks, G. H. Meier and F. S. Pettit, Introduction to high temperature oxidation of metals, 2nd edn.
(Cambridge University Press, 2006), p. 56.

123

You might also like