See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225257686

Scaling of Carbon Steel in Simulated Reheat Furnace

Atmospheres

Article  in  Oxidation of Metals · February 2005

DOI: 10.1007/s11085-005-1949-0

CITATIONS READS

25 1,090

3 authors, including:

Brian M Gleeson David Young

University of Pittsburgh UNSW Sydney
186 PUBLICATIONS   4,311 CITATIONS    324 PUBLICATIONS   7,389 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Study of the solubility, permeability and diffusivity of oxygen in the Fe-Ni system at high temperatures under carbon free
and carbon bearing gases View project

corrosion and Materials science & engineering research View project

All content following this page was uploaded by Brian M Gleeson on 31 March 2015.

The user has requested enhancement of the downloaded file.

Oxidation of Metals, Vol. 63, Nos. 1/2, February 2005 (© 2005)
DOI: 10.1007/s11085-005-1949-0

Scaling of Carbon Steel in Simulated Reheat Furnace

Atmospheres
V. H. J. Lee∗ , B. Gleeson,∗ † and D. J. Young∗ ‡

Received May 21, 2004; revised October 13, 2004

A low carbon steel was exposed to a series of laboratory-simulated combus-

tion gases at temperatures of 700–1200 ◦ C. The gases corresponded to the
products of burning natural gas with amounts of air varying from 95% to
112% of air–fuel stoichiometric equivalence. At all temperatures, fast, para-
bolic scaling kinetics were observed in the 112% hyperstoichiometric environ-
ment, whereas linear kinetics were always found in a 99% air–fuel gas. At
1100 ◦ C, linear kinetics were found in 95%, 99% and 101% air–fuel gases,
initial linear kinetics followed by a parabolic regime in 108% air–fuel gas,
and simple parabolic kinetics in the 112% gas. At equilibrium, Fe3 O4 would
be stable in the 95%, 99% and 101% air–fuel gases; however, only FeO
was formed during reaction. Also at equilibrium, Fe2 O3 would be stable
in the 112% and 108% gases, and this phase was found at the surface of
scales grown in these gases. Mass transfer calculations show that molecular
oxygen was not the principal reactant in any of these gases; instead CO2
and/or H2 O were the important species. It is concluded that surface reactions
control the linear scaling rates.

KEY WORDS: linear kinetics; surface reactions; mixed gases.

INTRODUCTION
The high temperature oxidation of carbon and low alloy steels is of prac-
tical importance because their secondary processing requires reheating in
∗ School of Materials Science and Engineering, University of New South Wales, Sydney 2052,
Australia.
† Current address: Department of Materials Science and Engineering, Iowa State University,
Ames, IA 50011, USA.
‡ To whom correspondence should be sent. E-mail: d.young@unsw.edu.au
15
0030-770X/05/0200–0015/0 © 2005 Springer Science+Business Media, Inc.
16 Lee, Gleeson, and Young

direct-fired furnaces to temperatures between 1100 and 1250 ◦ C. Scaling

during reheating increases production costs as a result of steel yield losses,
which are typically 1.5–2%.1,2 The extent of scaling is expected to depend
on furnace temperature and atmosphere, residence time and steel compo-
sition.
Considerable information is available for iron oxidation at high tem-
peratures in oxygen,3,4 CO–CO2 mixtures,5−8 and H2 –H2 O mixtures.9−12
In general, if the scale is thin, solid-state diffusion is rapid, and the reac-
tion rate is controlled by gas-phase mass transport or chemical processes
at the scale–gas interfaces. Either of these leads to linear scaling kinetics.
As the scale grows, it reaches a thickness at which diffusion within the
oxide becomes rate controlling and the kinetics become parabolic. Because
diffusion in iron oxides at high temperature is so fast, the scale thickness
at which parabolic kinetics commence can be large.
This information is not useful in predicting scaling rates for steels in
reheat furnaces. The addition of even very small concentrations of alloy-
ing elements to iron significantly affects oxidation rates. Furthermore, the
combinations of oxidants present in a furnace combustion gas (O2 , CO2
and H2 O) give rise to complex interactions with the growing scale. Thus,
for example, the addition of water vapor to oxygen affects scale–metal
adhesion during iron oxide scale growth11,13 and changes mass transport
within porous scales.12 These latter effects are particularly important, as
a potential route to the reduction of scaling losses is the use of less oxi-
dizing reheat furnace atmospheres,1,2,14 which contain predominantly CO2
and H2 O together with some residual oxygen. Very few studies of steel
oxidation in such gases have been reported1,13,15 and no systematic inves-
tigation appears to have been undertaken.
The aim of this study was to examine the scaling behavior of a com-
mercial low carbon steel in simulated reheat furnace environments corre-
sponding to the combustion of natural gas with air. The stoichiometric
ratios of the simulated air–natural gas combustion mixtures varied from
95% to 112%, and temperatures ranged from 700 to 1200◦ C.

EXPERIMENTAL
The steel used was a low carbon grade produced as strip to be used
for galvanizing. Its measured composition is shown in Table I. Reaction
samples were coupons about 10 × 10 × 2 mm, ground to a 1200 grit SiC
finish and ultrasonically cleaned in ethanol immediately before use. These
were exposed to flowing gas mixtures of CO–CO2 –H2 –H2 O–N2 to sim-
ulate environments generated by combustion of CH4 at various air–fuel
ratios and a total pressure of 1 atm. The linear flow rate was 24 cm min−1 .
Scaling of Carbon Steel 17

Table I. Steel Compositions (wt%)

C P Mn Si S Ni Cr Cu Al
0.04 0.015 0.26 <0.004 0.008 0.02 0.01 < 0.01 0.04

Combustion of CH4 with the stoichiometrically equivalent amount of

dry air, i.e. an air–fuel ratio of 100%, can be written
CH4 + 2(O2 + 3.776 N2 ) = CO2 + 2 H2 O + 7.552 N2
In practice, complete combustion is commonly ensured by using excess air,
i.e. more than 100% air, resulting in remnant free oxygen in the product
gas. Conversely, combustion with sub-stoichiometric, i.e. less than 100%,
air yields less CO2 and H2 O, but more CO and H2 .
The constituent partial pressures for such combustion ratios were cal-
culated using free energy minimization software, Chemix.16 Table II shows
results for equilibrium at 1100 ◦ C. Dry gas flows were regulated with mass
flow controllers, and the mixtures saturated with deionized water held at
appropriate temperatures. The saturated mixtures were catalysed using a
section of platinised ceramic honeycomb situated near the hot zone of the
reaction furnace.
Reaction rates were measured by exposing steel samples to reaction
for various times and observing the resultant weight change. In some
cases, continuous weight change measurements were obtained using a
microbalance. Reacted samples were characterized using X-ray diffraction
(XRD) of scale surfaces, standard metallographic techniques and laser
Raman microscopy.

RESULTS
Oxidation in 112% air combustion gas led to parabolic weight
gain kinetics at all temperatures. High temperature results are shown in
Fig. 1. Regression of the data on the equation
(
W/A)2 = 2kp t + C (1)

Table II. Equilibrium Partial Pressures (atm) of Reaction Gases

Air–fuel CO2 CO N2 O2 H2 O H2
112 8.57 × 10−2 3.5 × 10−7 0.72 2.1 × 10−2 0.171 3.5 × 10−7
108 8.86 × 10−2 4.4 × 10−7 0.72 1.4 × 10−2 0.177 4.4 × 10−7
101 9.42 × 10−2 1.3 × 10−6 0.72 1.9 × 10−3 0.188 1.3 × 10−6
99 9.38 × 10−2 1.9 × 10−3 0.71 8 × 10−10 0.190 1.9 × 10−3
95 8.90 × 10−2 9.6 × 10−3 0.70 3 × 10−11 0.187 1.0 × 10−2
18 Lee, Gleeson, and Young

12000

700
10000 800
1000
1100
1200
(∆W/A)2, mg2/cm4

8000

6000

4000

2000

0
0 20 40 60 80 100 120 140 160
t / min

Fig. 1. Oxidation kinetics in 112% air–fuel gas.

with
W/A the weight change per unit area formed in time, t, and C
a constant, led to estimates of the rate constant, kp , which are shown
in the Arrhenius plot of Fig. 2. The apparent activation energy was
125 ± 7 kJ/mole.

0 0 0 0 0
1200 C 1100 C 1000 C 800 C 700 C
2.5
A1006
Pure Iron
2.0
Log (kp / mg2 cm-4 min-1)

1.5

1.0

0.5

0.0

-0.5
6 7 8 9 10 11
104 K / T

Fig. 2. Arrhenius plot for 112% air–fuel gas. Data for pure iron from Refs. [18, 19].
.
Scaling of Carbon Steel 19

80

700
70
800
1000
60
1100
1200
∆W/A, mg/cm2

50

40

30

20

10

0
0 30 60 90 120 150 180 210
t / min

Fig. 3. Oxidation kinetics in 99% air–fuel gas.

Reaction in simulated 99% air combustion gas produced linear oxida-

tion kinetics, as shown in Fig. 3. Regression on the linear rate equation

(
W/A) = kl t + C (2)

with C as constant, led to estimates of the rate constant, kl , which are

shown in the Arrhenius plot of Fig. 4. The activation energy was esti-
mated as 135 ± 4 kJ/mole.
The effect of varying air–fuel ratio was explored more widely at a
reaction temperature of 1100 ◦ C. Weight gain kinetics are compared in
Fig. 5, where linear kinetics are seen to prevail (within the times exam-
ined) up to 108% air, where a transition to parabolic behavior is evident.
Linear and parabolic rate constants are summarized in Table III. Con-
tinuous weight gain kinetic measurements (Fig. 6) yielded the same rate
laws.
A typical scale grown in 112% air to fuel gas at 1200 ◦ C is shown
in Fig. 7. It was found to consist of an inner layer of FeO containing
Fe3 O4 precipitates in its outer region, an intermediate layer of Fe3 O4 and
a thin outermost layer of Fe2 O3 . Separation and cracking were frequently
observed in cooled scales. Scales grown in this gas were invariably uniform
in thickness, with an almost flat outer surface.
Reaction in 99% air to fuel gas led to the growth of scales which were
non-uniform in thickness, frequently developing highly irregular surfaces.
The example in Fig. 8(a) consists of FeO containing globular precipitates
20 Lee, Gleeson, and Young

12000C 11000C 10000C 8000C 7000C

0

-1
Ln (kL / mg cm-2 min-1)

-2

-3

-4

-5

-6

-7
6 7 8 9 10 11
104 K / T

Fig. 4. Arrhenius plot for 99% air–fuel gas.

Fig. 5. Effect of air–fuel ratio on oxidation weight uptake at 1100◦ C.

of Fe3 O4 ; that in Fig. 8(b) is FeO with a thin outer layer of Fe3 O4 . Sim-
ilar scales developed in 95% and 101% air–fuel gases at 1100◦ C.
Scaling of Carbon Steel 21

Table III. Parabolic (mg2 /cm4 min) and Linear (mg/cm2 min) Rate
Constants at 1100 ◦ C

Air–fuel ratio/(%)
95 99 101 108 112
kl 0.14 0.22 0.23 0.48
kp 24 40

60
112%

50

40
∆W/A, mg/cm2

30

99%
20

10 95%

0
0 10 20 30 40 50 60 70
t / min

Fig. 6. Continuous weight gain kinetics at 1100◦ C.

DISCUSSION
The phase constitutions of the oxide scales reflect what was produced
at reaction temperatures together with any transformations occurring dur-
ing subsequent cooling to ambient conditions. In view of the regular kinet-
ics observed, it is believed that the metallographically observed cracking
and scale separation occurred after reaction. Precipitation of Fe3 O4 within
FeO (and widening of the Fe3 O4 layer) have been shown17 in quench-
hold-quench experiments to result from the eutectoid (and pro-eutectoid)
partial decomposition of FeO during cooling. The present results therefore
reveal two forms of behavior: growth at high air–fuel ratios of a three-lay-
ered FeO/Fe3 O4 /Fe2 O3 scale according to parabolic kinetics, and forma-
tion at low air–fuel ratios of a single-phase FeO scale according to linear
kinetics.
22 Lee, Gleeson, and Young

Fig. 7. Cross-section of scale grown in 112% air–fuel gas at 1200◦ C.

Parabolic Kinetics
The high air–fuel ratio combustion gases contained significant levels
of free oxygen at equilibrium: 2% in the 112% gas and 1.4% in the 108%
gas. At these oxygen potentials, Fe2 O3 is stable, and the scale constitution
observed reflects a close approach to local scale–gas equilibrium. The fur-
ther observation of parabolic kinetics indicates that the rate was controlled
by steady-state diffusion within the scale. In this case, the rate would be
expected to be similar to that at which pure iron is oxidized to the same
products. Rates at 112% air are compared in Fig. 2 with pure iron oxida-
tion data18,19 and agreement is seen to be reasonable.
The consistently slower rate of carbon steel oxidation compared to
that of iron has long been known.20,21 It is commonly attributed to slight
decreases in scale–metal adhesion and scale plasticity, leading to brief peri-
ods of localized scale separation and temporary cessation of reaction.
Recent work by Abuluweful et al.22 on the oxidation of a very similar
carbon steel in dilute N2 /O2 mixtures led to a finding of initial linear rates
Scaling of Carbon Steel 23

Fig. 8. Cross-section of scale grown in 99% air–fuel gas (a)

at 1100◦ C and (b) at 1200◦ C.
24 Lee, Gleeson, and Young

followed by parabolic kinetics at high temperatures, and kinetics described

as logarithmic at lower temperatures. The linear kinetics were ascribed to
rate control by O2(g) diffusion and the parabolic kinetics to solid-state
diffusion control. As discussed in the next section, the gas compositions
used in the present work were such as to avoid a regime of gas-phase
diffusion control.

Linear Kinetics
Linear scaling kinetics were observed in 99% air–fuel ratio gas at all
temperatures, and at 1100 ◦ C in gases corresponding to air–fuel ratios of
95–101%. A gas corresponding to 108% air oxidized the steel at 1100◦ C
initially according to linear kinetics and then, after a weight uptake of
10–20 mg cm−2 , according to parabolic kinetics. The oxygen potentials of
these gases varied considerably. Comparison with standard thermochemi-
cal data23 shows that the 108% air gas was oxidizing enough to equilibrate
with Fe2 O3 , whereas the lower air–fuel ratio gases could equilibrate with
Fe3 O4 throughout the temperature range examined. However, the observa-
tion of linear kinetics demonstrates conclusively that scale–gas local equi-
librium was not achieved.
If Fe3 O4 had formed at the surface, then the FeO layer beneath it
would be expected to grow at the same diffusion-controlled (parabolic)
rate as observed in the 112% air experiments. This would result from
the existence of an internal FeO/Fe3 O4 boundary which would control
the oxygen activity at the same value in the two experiments, producing
the same diffusion flux (for the same conditions at the FeO/metal inter-
face). Since the FeO layer makes up the bulk of the scale, the same over-
all scaling rate would therefore result. Since this was not in fact the case, it
is concluded that no continuous FeO/Fe3 O4 boundary existed at reaction
temperatures in the 95%, 99% or 101% gases. No continuous Fe3 O4 layer
formed, because the solid–gas interface was not at equilibrium. It is con-
cluded, therefore, that the non-equilibrium process occurring at the scale
surface was slow and controlled the overall reaction rate.
As already concluded, the thin layers of Fe3 O4 observed at the scale
surface after reaction with these low air–fuel ratio gases formed during
cooling, via solid-state phase transformation.
The scale structure which grew according to linear kinetics differed
from those grown in 112% air. In the latter case, the diffusion-controlled
mechanism eliminated irregularities in scale thickness, producing a rela-
tively smooth, flat surface. In contrast, the scales grown in low air–fuel
ratio gases had very uneven surfaces. These were able to develop because
the scale growth rate was independent of thickness and varied locally. Such
Scaling of Carbon Steel 25

variations in rate could reflect either surface reaction occurring at differ-

ent rates on different FeO crystallographic orientations, or rate control by
gas-phase mass transfer, leading to an unstable oxide–gas interface.
Gaseous mass transfer rates can be calculated from gas transport the-
ory in the viscous flow regime.24 The molar flux of species i, Ji , is given by
km(i)
(o)

Ji = pi − pi (3)
RT
where km(i) is the mass transfer coefficient and pi the partial pressure of
species i, the superscript (o) indicates the value at the scale surface and R,
T have the usual meanings. The value of km(i) is given by
 1/6
Di4
v 1/2
km(i) = 0.664 (4)
υi l

with Di the binary gas diffusion coefficient, υi the kinematic viscosity, v

(o)
the bulk gas velocity and l the sample length. Approximating pi as zero
and Di by the value for a binary N2 -i mixture and evaluating it and υi
by standard methods,24 values for km(i) and hence Ji were calculated for
the relevant conditions. Results at 1100◦ C are summarized in Table IV,
where they are compared with oxygen fluxes corresponding to the mea-
sured weight uptake rates. In the case of parabolic kinetics in the 112%
air gas, the weight uptake rate was taken at t = 30 min.
It is clear that, at least in the short term, molecular oxygen was not
the reactant species in any gas at 1100◦ C. Even in the 112% gas, where
the free oxygen content was about 2%, its gas-phase mass transfer rate
was too slow to keep up with the solid-state diffusion controlled scaling
process. It may be calculated that only after 24 hr reaction (and a scale
thickness of 1.9 mm) would diffusion through the scale be slow enough for
delivery of O2 (g) to keep pace. On the other hand, gas-phase transport
of H2 O is more than fast enough to sustain observed oxidation rates in

Table IV. Oxygen Fluxes (mole O/cm2 s) at 1100 ◦ C

Air–fuel ratios
95 99 101 108 112
As O2(g) 1.7 × 10−16 2.9 × 10−15 1.2 × 10−8 9 × 10−8 1 × 10−7
As CO2(g) 6 × 10−7 6 × 10−7 5 × 10−7 5 × 10−7 5 × 10−7
As H2 O(g) 1 × 10−6 1 × 10−6 1 × 10−6 1 × 10−6 1 × 10−6
Scale formation 1 × 10−7 2 × 10−7 2 × 10−7 5 × 10−7 6 × 10−7 (30 min)
ratea
a Linear rate except for 112% (parabolic) where rate = d(
W )/dt = kp /2
W .
26 Lee, Gleeson, and Young

all gas mixtures studied, and the rate of CO2 (g) transport is slightly faster
than or, in the case of more oxidizing conditions, similar to oxygen uptake
rates. On this basis it would be concluded that gas-phase transport would
not be rate controlling if H2 O were the reactant. However, possible inac-
curacies in the calculation of JCO2 make uncertain any inferences as to the
reactivity of CO2 .
The uncertainty is resolved by the temperature effect on kl in 99% air
gas (Fig. 4). The activation energy of 135 kJ/mole is much larger than the
temperature effect on km(i) .23 Furthermore, measurements22 of linear steel
oxidation rates in dilute O2 –N2 gases, where the rate is controlled by gas-
eous mass transfer, yielded an apparent activation energy of 17 kJ/mole. It
is therefore concluded that the rate controlling step in linear oxidation in
these simulated combustion gases is a surface reaction.
This conclusion is in agreement with that of earlier workers using
simple CO2 /CO gases5−8 or H2 O/H2 gases.9 Activation energies for the
surface reaction controlled oxidation of pure iron were measured9 as
80 kJ/mole in H2 O/H2 and 217 kJ/mole in CO2 /CO. The presently observed
value in 99% air gas of 135 kJ/mole is closer to the former, but indicates
contributions by both H2 O and CO2 to the reaction. This is consistent
with the finding by Turkdogan et al.9 that the linear rate constants for
iron oxidation for the two gases were similar to each other in the temper-
ature range 1050–1200◦ C.
The absence of any significant contribution to rate control by
gas-phase mass transfer in these combustion gases is an important obser-
vation. Unlike dilute O2 mixtures, combustion gases contain large concen-
trations of CO2 and H2 O, which provide substantial additional oxygen
mass transfer capability.
The dependence of the observed linear rate constant on oxidant par-
tial pressures is shown in Figs. 9 and 10, where two regimes of behaviors
are evident. Whether CO2 or H2 O is the reactant, the same difficulties are
apparent. Firstly, in the more reducing gases, a small fractional change in
oxidant partial pressure leads to a large change in rate. Secondly, the 108%
air gas has, through dilution, lower pCO2 and pH2 O values than the 99%
air gas, but produces a significantly faster linear oxidation rate. The for-
mer observation can be understood from a simple kinetic analysis.
A possible sequence of CO2 surface reactions can be written as

S + CO2(g) = CO2 | S (5)

k1
CO2 |S
O = |S + 2h· + CO(g) (6)
k−1
k2
O X |S −→ OO
X
+ VFe + S (7)
Scaling of Carbon Steel 27

0.6

0.5

0.4
kl

0.3

0.2

0.1

0
8.5 8.7 8.9 9.1 9.3 9.5
102 PCO2 / atm

Fig. 9. Dependence of linear scaling rates at 1100◦ C on pCO2 .

0.6

0.5

0.4
kl

0.3

0.2

0.1

0
0.15 0.16 0.17 0.18 0.19 0.2
PH2O / atm

Fig. 10. Dependence of linear scaling rates at 1100◦ C on pH2 O .

where S represents a free surface site and X| S an adsorbed species, the

oxide point defect species h· , VFe and oxide anions OO
X are represented
25
using the Kroger–Vink notation and kn represents a rate constant.
28 Lee, Gleeson, and Young

Assuming a fixed number of surface sites

M = [S] + [O = |S ] + [CO2 |S ] (8)
defining K5 as the equilibrium constant for Eq. (5) and adopting the
steady-state approximation for the surface concentration [O = |S ]:
d [O = |S ]  2
= 0 = k1 [CO2 |S ] − k−1 [O = |S ] h· pCO − k2 [O = |S ] (9)
dt
it is found by substituting K5 pCO2 [S] for [CO2 |S ] in Eqs. (8) and (9),
followed by elimination of [S], that
k1 K5 MpCO2
[O = |S ] =
 (10)
K5 pCO2 k1 + k−1 [h· ]2 p
CO + k · 2
2 + k−1 [h ] pCO + k2

A similar reaction scheme can be proposed for H2 O species:

S + H2 O(g) = H2 O |S (11)
k3
H2 O |S
O = |S + 2h. + H2(g) (12)
k−3
k2
O = |S −→ OO
X 
+ VFe +S (13)
leading to
k3 K12 MpH2 O
[O = |S ] =
 (14)
K11 pH2 O k3 + k−3 [h· ]2 pH2 + k2 + k−3 [h· ]2 pH2 + k2

Here, K12 is the equilibrium constant for Eq. (12). The rate of the
oxygen uptake reaction (7) and (13) is proportional to [O = |S ], given by
Eq. (10) or (14). Considering the latter, it is concluded that k3 is not the
dominant term in the numerator (because the oxidation rate is not directly
proportional topH2 O ) and the reverse of reaction (12) must therefore be
significant. Similarly, it can be concluded that the surface is not saturated
with adsorbed H2 O, as the rate does change with changing gas composi-
tions, and therefore K11 pH2 O cannot be large.
Proceeding on the assumption that, in fact,K11 pH2 O is small, Eq. (14)
is approximated as
k3 K12 MpH2 O
[O = |S ] = (15)
k3 + k−3 [h· ]2 pH2
Assuming further that [h· ] is essentially constant in FeO, and using Eq.
(15) in the rate expression for Eq. (13)
Rate = k2 [O = |S ] (16)
Scaling of Carbon Steel 29

it is predicted that
pH2 O
kl ≈ (17)
a + bpH2
where a and b are constants. This expression is found to fit the data for
95%, 99% and 101% air atmospheres well, with a = 0.75 mg−1 cm2 s atm
and b = 58.7 mg−1 cm2 s estimated from a non-linear regression procedure.
The large change in pH2 had a much greater effect than did the very small
change inpH2 O .
A similar treatment of the CO2 surface reaction leads to
pCO2
kl ≈  (18)
a + b pCO
Again, this describes the data for 95%, 99% and 101% air atmospheres
well, with a  =0.375 mg−1 cm2 s atm and b =27.3 mg−1 cm2 s. Once more,
the large change in pCO had a much greater effect than the small change
in pCO2 .
It is concluded on this basis that the slow linear oxidation rates
observed in substoichiometric air–fuel combustion gases result from the
significant rate at which reducing species remove adsorbed oxygen from
the scale surface.
The faster rate observed in 108% air gas cannot be explained in this
way. However, it is likely that the 1.4% free oxygen in this gas makes some
contribution to the increased rate. Reference to Table IV shows that the
gas-phase flux of molecular oxygen to the scale surface does increase sig-
nificantly from the 101% to the 108% air gas. Nonetheless, it still makes
up only about 20% of the measured oxidation rate, accounting for about
35% of the increase in rate observed in the 108% over the 101% air gas.
Although the precision of the rate measurement in 108% air gas was lim-
ited by the transient nature of the kinetics, the extent of disagreement is
unacceptable, and more extensive measurements are desirable.
Since real combustion gases contain both CO2 and H2 O, interactions
between the two reaction paths via Eqs. (6) and (12), and through sharing
surface sites would need to be considered, along with the possible incorpo-
ration of hydrogen into the scale. The present results set does not contain
a sufficient range of variation in pCO2 or pH2 O to justify such consider-
ation.

CONCLUSIONS
Oxidation of a carbon steel in simulated combustion gases led to
linear scaling kinetics for a range of air–fuel ratios, and parabolic scale
growth only at high air–fuel ratios.
30 Lee, Gleeson, and Young

On the basis of mass transfer calculations and activation energy esti-

mates, it is concluded that the linear rates were controlled by surface reac-
tions rather than gas-phase mass transfer. This was a consequence of the
relatively high partial pressures of CO2 and H2 O. It is further concluded
that these species were the oxidants, as insufficient free oxygen was avail-
able to sustain the observed scaling rates.
As a consequence, it is concluded that dilute oxygen–unreactive gas
mixtures cannot be used to simulate oxidation by real combustion gases.
Linear scaling rates in sub-stoichiometric air–fuel gas mixtures varied
significantly, although the partial pressures of CO2 and H2 O were almost
constant. It is deduced from this finding that reactions between reduc-
ing species, CO and/or H2 , and the oxide surface were important in these
gases.

ACKNOWLEDGMENTS
Support by the Australian Research Council and Bluescope Steel is
gratefully acknowledged.

REFERENCES
1. Y. Suzuki, H. Takashima, and K. Yabuki, CAMP-ISIJ, 1, (397), 1488 (1988).
2. D. M. Obaro, Steel Reheat Furnace Proceedings, (CIM, Iron & Steel Society AIME,
Materialia, 1992), pp. 268–276.
3. A. G. Goursat, and W. W. Smeltzer, Review of High Temperature Materials 1, 351 (1973).
4. P. Kofstad, High Temperature Corrosion, (Elsevier, London, 1988).
5. F. Pettit, R. Yinger, and J. B. Wagner, Acta Materialia 8, 617 (1960).
6. W. W. Smeltzer, Acta Materialia 8, 377 (1960).
7. L. A. Morris, and W. W. Smeltzer, Acta Materialia 15, 1591 (1967).
8. H. J. Grabke, Ber. Bunsenges 69, 48 (1965).
9. E. T. Turkdogan, W. M. McKewan, and L. Zwell, Journal of Physical Chemistry 69, 327
(1965).
10. C. W. Tuck, M. Odgers, and K. Sachs, Corrosion Science 9, 271 (1969).
11. P. Kofstad, and R. Bredesen, Proc. Int. Cong. Metall. Corros., (NRCC, Ottawa, 1984),
p. 12.
12. A. Rahmel, Werkstoffe und Korrosion 16, 662 (1965).
13. J. S. Sheasby, W. E. Boggs, and E. T. Turkdogan, Metal Science 18, 127 (1984).
14. Japanese Patent No. 4-66615 (1992).
15. H. S. Hsu, Oxidation of Metals 26, 315 (1986).
16. CSIRO, ‘Thermochemistry System version 5.1 IBM-PC Program Chemix’ (CSIRO Divi-
sion of Mineral Products, Melbourne, 1988).
17. S. M. M. Hadavi, B. G. Gleeson, and D. J. Young, Materials at High Temperatures 17,
311 (2000).
18. M. H. Davies, M. T. Simnad, and C. E. Birchenall, Transactions AIME, 191, 889 (1951).
19. J. Paidassi, Revue De Metallurgie 54, 569 (1957).
20. K. Sachs, and C. W. Tuck, Reheating for Hot Working, (Iron & Steel Institute, London,
1967).
21. J. Maldy, Revue de Matallurgie Memoires at Etudes Scientifique 57, 379 (1965).
22. X. H. Abuluweful, R. I. L. Guthrie, and F. Ajersch, Oxidation of Metals 46, 423 (1996).
 Scaling of Carbon Steel 31

23. O. Kubaschewski, and C. B. Alcock, Metallurgical Thermochemistry, 5th edn. (Perg-

amon, Oxford, 1983).
24. D. R. Gaskell, Introduction to Transport Phenomena in Materials Engineering, (Macmil-
lan, New York, 1992).
25. F. A. Kroger, and H. J. Vink, Solid State Physics 3, 307 (1956).

View publication stats