Professional Documents
Culture Documents
Osidation
Osidation
Oxidation of Alloys in
Steam Environments
- A Review
ABSTRACT
Traditionally materials designed for use at high temperature have been developed
primarily for their mechanical properties but there is now a growing realisation that
oxidation may limit lifetime. The purpose of this report is to review existing
information regarding oxidation of alloys, which are intended for use at high
temperatures, under atmospheres containing steam or high water vapour content and to
identify areas where more understanding is required.
ii
ISSN 1473-2734
Extracts from this report may be reproduced provided the source is acknowledged and
the extract is not taken out of context.
iii
iv
CONTENTS
1
2
3
4
5
6
7
Introduction........................................................................................................................1
The Influence of Chromium Content.................................................................................1
Low Alloy Ferritic Steels...................................................................................................6
9-12% Cr Martensitic Steels ............................................................................................12
Austenitic Steels...............................................................................................................17
Ni- and Co-based alloys...................................................................................................24
Mechanisms of Oxidation Under Environments Containing Water Vapour ...................28
7.1
Dissociation Mechanism..........................................................................................28
7.2
Oxidant-gas Penetration Mechanism .......................................................................30
7.3
Formation and Volatilisation of Fe(OH)2 ................................................................30
7.4
Formation and Volatilisation of CrO2(OH)2 ............................................................31
7.5
Changes in the Oxide Defect Structure via Proton Dissolution...............................31
8
Discussion and Conclusions ............................................................................................33
9
References........................................................................................................................35
Introduction
During service in high temperature plant, tubing, piping and headers in boilers are exposed to
steam at high pressure flowing at high rates and often in the presence of heat flux. Similarly
components in steam turbines are exposed to high-pressure steam atmospheres.
Service lifetimes of these components may be limited by creep, fatigue or oxidation.
Traditionally materials designed for use at high temperature have been developed primarily
for their mechanical properties but there is now a growing realisation that oxidation may limit
lifetime, either directly through metal wastage or indirectly through raising local temperatures
(and consequently reducing creep-controlled lifetimes) due to the lower thermal conductivity
of the oxide scale.
The purpose of this report is to review existing information regarding oxidation of alloys,
which are intended for use at high temperatures, under atmospheres containing steam or high
water vapour content and to identify areas where greater understanding is required. One of
the difficulties encountered in comparing test results within the literature is ensuring that the
conditions of the exposures are similar. In many cases the conditions of the exposure and/or
the water chemistry were not stated, this information has been included in this review where
available.
2
The most significant improvements in oxidation resistance are achieved by changing the
inherent nature of the protective oxide scale. One of the most dramatic demonstrations of this
mechanism is the influence of chromium additions to iron-base alloys. During oxidation,
carbon steels form a scale consisting of one or more layers of wustite, haematite and
magnetite depending upon the specific oxidation conditions. The addition of chromium to
the alloys firstly changes the scale such that the inner layer (magnetite in plain carbon steels)
is an iron-chromium spinel in low alloy ferritic (up to 3%Cr) or martensitic (9-12%Cr) steels.
Further increases in chromium will eventually lead to the formation of chromia, e.g. in
austenitic stainless steels.
The effect of increasing the chromium content in ferritic steels has been demonstrated in both
laboratory and service exposures. Solberg et al 1 demonstrated a systematic decrease in metal
loss with increasing chromium content in short-term laboratory tests at 593 C. Their data
are presented in Figure 1.
Similarly Eberle et al 2 demonstrated (see Table I) a similar trend in material that had been
withdrawn from service after 36000 hours at 554 C / 160 atm.
Page 1 of 39
Figure 1 Corrosion of steel bars in contact with steam at 593 C for 200, 500, 1000 and 2000 h (after
Solberg et al 1)
Table I Metallographically measured scale thickness on exposure to steam at 554 C/160 atm for 36000 h
(after Eberle et al 2)
Material
Inner scale,
m*
SA-210 C-steel
90
Mo
80
Cr Mo
70
1 Cr Mo
75
2 Cr Mo
75
5 Cr Mo
70
7 Cr Mo
65
9 Cr 1Mo
65
* Measured at one location
#
Range of measurements
Outer scale,
m*
70
70
55
60
60
50
55
55
Total thickness,
m*
160
150
125
135
135
120
120
120
Total thickness
range, m#
125 160
125 150
125 165
135 165
125 145
110 150
120 125
120 145
Heiermann et al 3 have also shown that there is a trend for the higher Cr-containing steels to
produce thinner scales: 160 and 260 m scales at 560 C for 12 and 9% Cr and 290 and
510 m scales at 600 C for 12 and 9% Cr ferritic steels respectively after 200000 hours
exposure. Thus the more Cr an alloy contains the better it can protect itself from oxidation,
and therefore alloys with high Cr are usually used in corrosive environments. The
improvement in steam oxidation resistance of the 9-12Cr materials over the low alloy ferritic
steels is demonstrated in Figure 2 4. At 560 C the rate of scale growth on 9%Cr steels is half
that on 2%Cr material. Increasing the chromium content to 12% produces another factor of
two improvement.
Page 2 of 39
800
1% & 2% Cr @ 600oC
700
600
9% Cr @ 600oC
500
400
1% & 2% Cr @ 560oC
300
12% Cr @ 600oC
200
9% Cr @ 560oC
100
12% Cr @ 560oC
0
0
20
40
60
80
100 120 140
EXPOSURE TIME (103 hours)
160
180
200
Figure 2 Bore oxide scale thickness as a function of operating time and temperature for
ferritic (1-2%Cr) and martensitic (9-12%Cr) steels (after Ref. 4)
This improved protection can also be seen in Figure 3, which shows the oxidation of T22,
T23, TP347, T91 and T92 in flowing pure water vapour at 650C 5. In this instance the
samples were hung on a kanthalmade support fixed on a corundum basket, which was then
placed into the mullite tubes inside a tubular furnace. The steam was produced by ebullition
of deionised water. The corrosion rates of the steels were determined by mass change
measurements prior to and after exposure and descaling by etching in sodium hydroxide
melted at 450 C 4.
800
700
600
T22
500
T23
400
TP347
T91
300
T92
200
100
0
0
1000
2000
3000
4000
5000
6000
Time, h
Manning and Metcalfe 6 have also demonstrated this improvement when they compared the
parabolic rate constants of 2Cr1Mo with 9Cr1Mo and 12Cr ferritic steels over the
Page 3 of 39
16
14
12
482C Ground
482C Annealed
10
482C Pickled
538C Ground
538C Annealed
538C Pickled
9Cr 482C
4
9Cr 538C
2
0
0
5000
10000
15000
20000
25000
30000
Time, h
Figure 4 2Cr1Mo and 9Cr steels exposed to superheated plant steam (after Ref. 7)
It appears that the level of benefit arising from increased chromium content is not constant
but is dependant on the moisture content of the oxidising atmosphere. To illustrate this point
when comparing the oxidation in steam of the low Cr containing alloys T22 and T23 to the
higher Cr containing ferritic steels it has been reported by Gambier 8 that the corrosion rates
of the 9% Cr steels were only about 3 times lower than those of the 2% Cr steels. These
results clearly demonstrate the deleterious effect of water vapour since the corrosion rate in
air is usually 60 70 times lower for 9Cr steels than for 2 Cr steels. This reduction of the
favourable effect of Cr in steam compared to in air has been attributed to changes in the
morphology of the oxide scales developed in steam environments. Lepingle 5 showed that
Page 4 of 39
In the case of alloys with Cr content over 12% benefits in steam oxidation resistance have
again been observed from increases in the Cr content. Tests have shown that increasing the
Cr content from 12 to 18% increased the duration of the protective stages of some alloys from
1 hour to more than 25 hours. In studies by Tavast 9 and Otsuka and Fujikawa 10 it was
concluded that if the Cr content was over 25% (Otsuka quotes 22%), then a continuous
chromia layer quickly forms at the metal-metal oxide interface, slowing down the diffusion of
iron to form magnetite at the gas-metal interface. This is a general rule and there will be
exceptions. For example, in a comparative study the austenitic steel TP347 was found to
have mediocre corrosion resistance, despite its high level of Cr content 5. The relatively poor
behaviour was attributed to corrosion product scaling phenomena and the non-uniform
healing of the chromia layer during the test (this has also been observed by Armanet 11,
Montgomery 12 and Otsuka 10). This may suggest that there is a limit to the maximum
amount of Cr that is beneficial in terms of scale adherence. Figure 5 shows the metal loss
rate for alloys with differing amounts of Cr. The data show how the amount of metal loss
decreases with increasing Cr content and approaches a plateau level, thus illustrating an upper
limit to the benefits of Cr additions.
Page 5 of 39
Low alloy ferritic steels are most widely used in pressurised applications in boilers fired by
fossil fuels and also in Heat Recovery Steam Generators (HRSGs). Specifically these are
tubing alloys that are used particularly in the cooler sections of superheaters and reheaters
and also for the waterwalls in the temperature range where mild steel becomes too susceptible
to creep. These alloys are also used in thick-section components such as headers and steam
pipes.
Commonly the key requirements that this class of material must exhibit are
For the purpose of this review the most pertinent of the above requirements is the resistance
to steam oxidation. These materials do have limited resistance to steam oxidation, and
improving this resistance would result in enhanced heat transfer and increased operating
temperature, thereby contributing to increased plant efficiency with little increase in cost.
Table II shows the composition of some common low alloy 1-3% Cr steels, together with
commonly used alternative nomenclature of these alloys.
Table II Composition of Low Alloy 1-3% Cr Steels
Alternative
Nomenclature
Alloy
Grade
11
Grade
22
Grade
23
Grade
24
1CrMoV
P11/T11/13CrMo
44
P22/T22/
10CrMo 9 10
P23/T23/HCM2S
T24/7CrMoVTiB
10-10
Min
Max
Min
Max
Min
Max
Min
Max
Mn
0.05
0.15
0.05
0.15
0.04
0.1
0.05
0.10
0.25
0.3
0.6
0.3
0.6
0.1
0.6
0.3
0.7
0.8
Si
0.025
0.025
0.5
1.0
0.025
0.025
0.5
0.03
0.01
0.02
0.01
0.01
0.02
0.5
0.15
0.45
0.2
Composition, wt%
Cr Mo V
1.0
1.5
1.9
2.6
1.9
2.6
2.2
2.6
1.0
0.44
0.65
0.05
1.13
0.05
0.3
0.9
1.10
1.0
0.2
0.3
0.2
0.3
0.3
Nb
B
ppm
0.03
0.02
0.08
1.45
1.75
5
6
15
70
0.012
0.004
The scale formed during steam oxidation of low alloy steels has three distinct regions when
cross-sectioned. These comprise a thick compact outer layer, a subscale region consisting of
extensive internal oxide precipitate formation mostly along the grain boundaries, and a thin
layer between the outer layer and the subscale. It is claimed that a critical balance of the
selectively oxidised metal, the oxidant species and the growth rate of the external scale
govern the extent of the internal oxide precipitation and subscale formation 13.
There is a distinct difference in the structure and morphology of the oxide scales formed
under dry air oxidation and steam oxidation. In terms of the morphology of the oxide scales
a recent study conducted by Raman 14, in which specimens were oxidized at 600 C in a
Page 6 of 39
Al
0.03
0.02
0.01
Exposures time
Temperature
(days)
550 C
600 C
625 C
7
M+H
M/W
M
14
M+H
M
M+(W)
28
M+H
M
M+W
56
M
M
M+W
M magnetite, H haematite, W wustite.
650 C
W/M+W
/
M+W
/
In the experiments performed by Arztegui et al 16 a constant flux of high purity argon was
used to carry the steam through three conduits. The first fed into a reference tube in the
furnace, in which the sample was exposed to high temperatures in the absence of steam. The
remaining two identical conduits flowed through a water flask before reaching the furnace
thereby producing an Argon / steam mixture.
Under isothermal conditions the oxidation kinetics not surprisingly followed a parabolic rate
law, with calculated activation energy of oxidation calculated from the scale thickness of
324 kJ mol-1 (308 kJ mol-1 from weight gain data), this value is reported to agree well with
Page 7 of 39
Time spent at
Slow Cooling
550 C + 700 C
7 days + 4 hours
M/M+(W)
7 days + 8 hours
/
7 days + 24 hours
M/M+W
7 days + 4 hours + 7 days 550 C
M+H/M
M magnetite, H haematite, W wustite.
Page 8 of 39
Rapid Cooling
M/M+(W)
M/M+W
M+W/W+M
/
Figure 7 Mass change of T22 and T23 steels in water vapour at 650 C (after Ref. 5)
Page 9 of 39
Steel
T22
T23
500 C
0.46
0.41
z exponent at
550 C
600 C
0.49
0.62
0.41
0.71
650 C
0.83
0.71
This suggests that at temperatures below 600 C, T22 and T23 obey a parabolic law, which
indicates that at these temperatures the materials are self-protecting. However at high
temperatures the scale becomes non-protective and the kinetics approach linear behaviour.
Although interesting this material class is rarely used at such elevated temperatures.
Although many of the reported relationships claim to be following parabolic rate laws, others
rate laws have been cited i.e. logarithmic, combined functions etc. In a report by Bueno and
Marino 20 data relating to steam oxidation from three different sources were compared with
their own dry oxidation data. Figure 8 shows three graphs comparing the four sets of data.
In each plot the dry oxidation data of Bueno and Marino 20 has been compared to different
oxidation rate laws suggested by other authors 21-22, namely parabolic (plot a) 21, logarithmic
(plot b) 22 and as a combined function (plot c) 22. It is apparent that there is much
disagreement between them; and that the presence of steam greatly affects the observed
reaction kinetics. Interestingly, the data indicate that the oxidation rate changes around 700
C, such that at 600 C kp dry > kp steam, whilst at 700 C kp dry kp steam and at 800 C
kp dry < kp steam.
Data in the literature are generated using three main experimental methods to produce high
temperature steam environments, static steam, pure flowing steam and a mixture of steam and
a carrier gas. The oxide scales formed and the oxidation rate are dependant on which method
is used and the temperature of the test. To illustrate this consider Figure 9, which shows the
oxidation rate at 600C for T22 and T23 in air 20, pure water vapour 5 and a mixture of steam
and argon 16.
It is clear from Figure 9 that:
The difficulty comes in determining which exposure method best replicates the in-service
conditions, in terms of both the oxidation rate and the morphology of the resultant oxide
scale. From the literature examined there is no consensus on these points and also little
mention of the potential effects of steam chemistry.
Page 10 of 39
(a)
(c)
(b)
Figure 8 Comparison of data obtained from four separate works, illustrating how different rate laws can be
used. The symbols show dry oxidation data compared rate laws under steam oxidation conditions. Plot (a)
shows a parabolic rate, (b) logarthimic and (c) is a combined function. All data relates to 2 Cr-1Mo steel
(After Ref. 20)
200
180
160
140
120
100
80
60
40
20
0
0
1000
2000
3000
4000
5000
Time (h)
Page 11 of 39
6000
The 9-l2% Cr steels are used in both boilers and in steam turbines where the alloys used are
exposed to high pressure steam environments at temperatures up to 620 C and pressures in
excess of 300 bar.
In boilers that use fossil fuel these steels are used for tubing in superheaters and reheaters,
Page 12 of 39
9Cr1Mo
T91
T92
HCM12
X20
Mn
Si
Cr
Mo
0.12
0.45
0.6
1.0
0.1
0.45
0.4
0.07
0.45
0.06
0.1
0.55
0.3
0.20
1.0
0.5
1.0
0.2
0.049
0.5
0.2
0.06
1.8
0.004
12
1.0
0.25
12
1.0
0.3
0.03
1.0
Ni
0.8
0.6
(a)
(b)
(c)
(d)
(e)
(f)
Figure 10 Arrhenius plots of steam oxidation constants for (a) 10CrMo910 (T22), (b) X10CrMoVNb91 (T91), (c)
X20CrMo121, (d) HCM12, (e) NF616 and (f) comparison of all the previous materials (after Ref. 12)
Page 13 of 39
log kp, m h
-1
-1
9Cr1Mo
X20
-2
HCM12
-3
0.0010
0.0011
0.0012
0.0013
-1
Reciprical Temperature, K
Comparison of the relative oxidation rate of the steels containing 9%Cr shows that 9Cr1Mo is
substantially better than T91 and T92, with the improvement increasing with increasing
temperature. This observation implies that one or more of the additional alloying elements in
T91 and T92 i.e. V, W or Ni, is detrimental to steam oxidation resistance.
Abe et al 23 studied the influence of minor alloying additions on the steam oxidation
behaviour of a 9%Cr steel at 600 and 650 C. Si additions improve the steam oxidation
resistance but have the deleterious effect of encouraging carbide agglomeration. The addition
of trace amounts of Ti or Y(0.05%) further improve the steam oxidation resistance of alloys
containing 0.5% Si (Figure 12) 23.
Figure 12 Mass change of the steels during oxidation at 600 and 650 C (after Ref. 23)
Page 14 of 39
The beneficial effect of additions of silicon to the alloy composition has also been reported
by Hurst and Cowen 24. To further investigate this Griess and Maxwell 7 exposed a selection
of alloys containing a range of silicon concentrations to superheated steam at temperatures of
482 C and 538 C. The results are presented in Table VII. Although the exposure time was
short, it is interesting to note that the mass gains were generally lower for higher silicon
contents, an example of this is shown in Figure 13 for 9Cr-1 Mo steel.
Table VII Average mass gains of duplicate Cr-Mo steel specimens in superheated steam (After Ref. 7)
0.01 Si
0.02 Si
0.04 Si
0.36 Si
120
-22 )
Weight
g/m
Mass
gaingain,
(g.m
100
538 C
482 C
80
60
40
20
0
0
1000
2000
3000
4000
5000
6000
7000
Time, hours
Figure 13 Affect of increasing the Si content of 9 Cr-1 Mo Steel, similar affects were exhibited by
lower Cr content alloys as shown in Table VI (After Ref. 7)
Page 15 of 39
Igarashi et al 25 have shown that Pd additions to 9Cr 3.3W steels improve creep resistance at
over 650 C and steam oxidation rates at up to 750 C decreased by factor of 10.
Figure 14 Oxidation curves of stainless steels with various (a) chromium and (b) silicon (for 15% Cr)
contents, oxidised at 900 C in Ar+15%H2O (after Ref. 26)
Henry et al 26 exposed a reference steel sample to dry air, and established that under these
conditions the oxidation followed a pure parabolic rate law for durations of over 400 hours
(kp=8.76x10-13 g2 cm-4 s-1). In the presence of water vapour the same material showed
breakaway corrosion after only 8 hours. Before the breakaway occurred the kinetics followed
a parabolic law (kp =2.42x10-12 g2 cm-4 s-1) and then accelerated to a linear rate after
breakaway. A comparison was then conducted to illustrate the effect of increasing the Cr
content from 12 to 18% and increasing Si from 0 to 1%. In conjunction with this, a study on
the effect of the water vapour partial pressure was conducted, varying the pressure from
150 mbar to 10 mbar. The results are presented in Table VIII and in Figure 14.
Table VIII Time to breakaway at 900 C as a function of the composition of the steel and the water
partial pressure (after Ref. 26)
0% Si
0.5% Si
1% Si
12% Cr
10 mbar
150 mbar
Not studied
t = 1h
t = 1h
Not studied
15% Cr
10 mbar
150 mbar
t = 5-8h
t = 2-3h
t = 15h
t = 9h
t > 50h
t > 50h
18% Cr
10 mbar
150 mbar
Not studied
t > 25h
t > 25h
Not studied
The results showed that increasing the Si content from 0 to 1% in a Fe-15Cr alloy, increased
time to breakaway from 2 hours to more than 50 hours. Decreasing the water vapour partial
pressure from 150 mbar to 10 mbar delayed the initiation of breakaway. This behaviour is
consistent with observations previously reported.
Page 16 of 39
Austenitic Steels
Austenitic steels are more expensive than corresponding ferritic steels and have greater
coefficients of thermal expansion and relatively poor thermal conductivity. Consequently
applications for austenitic steels are restricted to the higher temperature boiler tubes i.e.
superheaters and reheaters and to specific situations where severe corrosion conditions occur,
e.g. in black liquor units. Austenitic materials are often used as weld overlay on ferritic
materials to repair corroded areas or to provide protection in areas where corrosion could be a
problem.
Key requirements are:
Low costs
Good fabricability
Resistance to alkali salt corrosion
Weldability
The philosophy behind the alloy development was to increase the volume fraction of
strengthening precipitates by replacing chromium carbides with other more stable carbides,
while at the same time freeing chromium back to the matrix to give improved corrosion
resistance. The composition and mechanical properties of several common alloys are given
in Table IX.
Page 17 of 39
Alternative
nomenclature
SUS321J1HTB
SUS309J4HTB
SUS304J1HTB
SUS310J1TB
HR6W
NF709
Esshete 1250
SUS310J2TB
P
max
0.045
0.045
0.045
0.045
0.045
0.045
0.045
Composition, wt%
S
Si
Cr Mo
max max
0.03
1
18
0.03
1
18
0.03
1
17
0.03
1
17
0.03
1
17 2.5
0.03
1
23
0.03
1.5 24
0.15
0.08
0.08
0.08
0.08
0.2
0.25
Mn
Max
2
2
2
2
2
2
2
0.08
1.6
0.6
18
0.12
0.05
0.1
0.06
1.6
1.5
0.8
1.2
0.6
0.4
0.2
0.4
18
22
18
25
0.10
2.0
1.0
23
0.15
0.5
20
1.5
0.09
0.6
15
0.03
0.03
0.03
0.003
Other
N
Nb
Ni
9
8
11
11
12
14
19
0.15
0.1
0.2
0.04
0.8
10
0.1
0.7
0.4
0.45
0.4
max
0.2
10
15
9
20
40
25
10
0.15 Ti
0.8 Nb+Ta
0.08 Ti
0.002 B
3.0 Cu
6.0 W, 0.2 max
Ti
0.1 Ti
0.004 B, 0.02 Al,
0.9 Nb, 0.25 V,
0.14 Cu
(a)
(b)
(c)
(d)
Figure 15 Arrhenius plots of steam oxidation constants for (a) 1714CuMo, (b) AISI347 (coarsed
grained), (c) AISI347 (fine grained), (d) comparison of all the previous materials. Compostion
presented in Table X. (after Ref. 12)
Page 18 of 39
Alloy
A
B
C
D
C
SUS 347HTB 0.07
17-14CuMo
0.11
20-Cr-25Ni type 0.07
22Cr-35Ni type 0.08
Ni
11.3
14.5
25.2
33.7
Composition (wt%)
Cr Mo
Ti
Nb
B
N Cu
17.8
0.72
0.01 15.9 2.50 0.24 0.43
0.01 3.1
20.1 1.56 0.06 0.27 0.004 0.18 22.1 1.46 0.12 0.19 0.005 0.03 -
Calculations based on the assumption of formation of Cr2O3, Fe3O4 and FeO were made and
the experimental data compared to predicted values. It was reported 29 that alloys C and D
fall very close to the theoretical line, which corresponded to light spalling, A and B deviate
far from them indicating heavier spallation, as shown in Figure 16. This figure shows
theoretical predictions between mass gain form oxidation and mass loss from descaling.
Otoguro et al 29 noted that the resistance to steam oxidation was influenced by four factors;
steam pressure, temperature, Cr content and Ni content. Montogomery et al 12 suggest that
by increasing the pressure you can increase the oxidation rate by decreasing the integrity of
the scale. This has the secondary affect of increasing the oxidation rate by increasing the
temperature of the underlying metal substrate. The effect of pressure is illustrated in
Figure 17. Under high steam pressure, voids and cavities were found to have formed in the
outer scale layer. It is claimed that this is attributed to the local cessation of the scale growth,
the difference in the diffusion rates of the metallic ions in the outer and inner scale, and/or a
weakening of the oxide grain boundaries 29.
Page 19 of 39
Figure 16 Weight loss and gain for the steels tested by Otoguro et al (after Ref. 29).
Steel compositions are shown in Table X.
Figure 17 Relationship between steam pressure and maximum scale thickness of steels A, B, C and
D from work conducted by Otoguro et al 29. Steel compositions are shown in Table X.
Page 20 of 39
The gas velocity has been shown to strongly affect the oxidation of austenitic and ferritic
chromium steels. In work conducted by Asteman et al 30 it was concluded that the
vaporization of chromium from the oxide in the form of chromium (VI) oxyhydroxide
increased the oxidation rate of X20, 304L and 310 steels in O2/H2O environments. They
found that at low gas velocities the X20 and 310 showed completely protective behaviour. In
contrast the 304L material did not form a completely protective oxide at low gas velocities.
For higher gas velocities (>2.5 cm/s), X20 exhibited breakaway corrosion while the oxide on
the 304L and 310 was observed to fail locally. The effect of evaporation is schematically
illustrated in Figure 18.
30-100 nm
30-100nm
Cr-rich (Cr, Fe)2O3
Oxidation in dry O2
CrO2(OH)2(g)
CrO2(OH)2(g)
O2/H2O
Low evaporation rate
Protective oxide
304/310
Fe-rich (Cr, Fe)2O3
O2/H2O
Medium evaporation rate
Partly protective oxide
X20
Fe2O3
Fe2O3 or Fe-rich (Cr, Fe)2O3
304/310
Fe-rich (Cr, Fe)2O3
O2/H2O
High evaporation rate
Non-protective oxide
304/310
X20
Figure 18 Schematic illustration of the oxidation of austenitic and ferritic chromium steels in O2/H2O
environments. The illustration represents the situation after 168 hours as reported by Ref. 30
Whilst it is generally accepted that steels containing high levels of Cr have better corrosion
resistance, it was observed by Otoguro et al 29 that the weight loss in alloy B is less than that
of alloy A in spite of lower Cr content, as shown in Figure 19. The authors suggest that this
improvement in steam oxidation is caused by the higher Ni content. The benefit of Ni
additions was also reported by Croll and Wallwork 31 who demonstrated the benefit on the
oxidation resistance in Fe-Cr-Ni alloys. Caplan and Cohen 32 also showed that Ni increased
oxidation resistance of stainless steels in wet air. It is suggested that this is due to the
formation of a Ni enriched layer. As with martensitic alloys it has also been demonstrated
that Si plays an important role in enhancing the oxidation resistance of austenitic steels by
forming an inner layer of silica between the oxide scale and the alloys by promoting the rapid
formation of Cr2O3. This layer of silica, which grows much more slowly than chromia, acts
as a diffusion barrier thus controlling the transport of Cr ions 33. Chromium consumption
decreases as a result and the depleted zone forms more slowly. Silica, it is claimed, also
decreases the activity of Fe at the interface 33.
Page 21 of 39
Figure 19 Effect of steam pressure on loss in weight for steels tested by Otoguro et al 29
There has been little investigative work into the comparative effect of the grain sizes within
the base metal and the effect this has on steam oxidation. Although it has been reported by
Teranishi 35-36 that there is a marked improvement in the corrosion resistance of fine grained
(ASTM 10), AISI 347, compared to a coarse grained (ASTM 3) version of the alloy. It is
Page 22 of 39
Composition (wt%)
C
Cr
Ni
Mo
Fe
Mn
Si
Other
Alloy 800
0.08 20.78 33.90
43.46
0.44 Ti, 0.46 Al
Alloy 617
0.07 20.30 57.35 8.58 1.01 0.05 0.16 11.72 Co, 0.76 Al
304 stainless steel 0.06 18.04 8.90
Bal. 1.84 0.55
Alloy
Surface preparation is a further factor that will influence the oxidation rate of a material. In
the case of the Alloy 800 samples half were solution annealed at 1177 C for 30 minutes in
argon and air cooled and then pickled in HNO3-HF, the other half were tested as ground
without being heat-treated. It is immediately apparent that the ground sample oxidised at a
very low rate. From further study of the pickled samples it was noted that they had
apparently been sensitised during the cooling process and subsequently corroded
intergranularly during the pickling process. The depth of the grain boundary penetration
produced by pickling was about 200 m, it is reported that this depth did not increase as a
result of steam exposure. It is clear that care should be taken when performing surface
treatments as it is established that ground surfaces of alloy 800 corrode less in steam than
annealed ones, other alloys may adhere to this observation.
All of the 304 stainless steel samples were annealed and pickled, and then ground. There was
no evidence of the sensitising effect as seen with alloy 800. The penetrations and corrosion
rates were small when compared with the ferritic steels that were also tested. The growth of
oxides scales on Cr-Mo steels and the austenitic stainless alloys is generally considered to be
from both outward migration of iron ions to the surface, where they react with water
molecules, and the inward diffusion of oxygen ions to the metal-oxide interface, where they
react with the metal 37. No explanation for the change from parabolic to a linear (Figure 21)
rate was offered by the authors for the previous work.
Page 23 of 39
8
7
6
5
4
3
2
1
0
0
5000
10000
15000
20000
25000
30000
Time (h)
Alloy 800 Ground 538C
Alloy 800 Sensitized and pickled 482C
304 SS 538C
Figure 21 Mass gains for alloy 800 and 304 Stainless Steel in superheated steam (after Ref. 7)
From the literature two main points are clear regarding this class of material, namely that the
oxidation rate increases with high pressure atmospheres and that fine grain sizes improve the
oxidation resistance by supplying Cr to the surface.
6
Ni- and Co-based alloys are used for hot-end parts in gas turbines and although there are
some other minor applications, it is the requirements defined for gas turbines that drive
developments. In the absence of improved materials the only option open to the design
engineer faced with developing engines with greater performance is to design improved
cooling technologies. The key property requirements are indicated below.
Page 24 of 39
Discs:
Combustors:
Care should be exercised when testing this class of materials as the environmental conditions,
primarily the presence of oxygen and/or moisture in the test environments, have been
observed to have significant effects 38.
Commonly coatings are used to confer corrosion resistance. Nonetheless corrosion resistant
alloys have been developed where generally a higher Cr/Al ratio is required than alloys
developed purely for creep resistance. Mo which was present as a solid solution
strengthening element has poor hot corrosion performance and this is normally replaced by
refractory element such as W, Ta and Nb 39.
Although quite a widely used class of material, very little information is available in the open
literature regarding the steam oxidation behaviour. Exposures in different forms of steam,
that is flowing plant, static etc, have been reported in work previously discussed in the earlier
sections. Thus Griess and Maxwell 7 exposed samples of Alloy 617 to plant steam. Very
little corrosion occurred during the test, although for Ni-based alloys the temperature was low
(482 and 538 C). The oxide layer formed on the samples was so thin that it was claimed to
be undetectable at a magnification of 500x. The composition of the alloy is presented in
Table XII.
Page 25 of 39
Alloy
Alloy 617
Incoloy
800
Inconel
600
Inconel
625
Inconel
X750
Rene 41
Composition (wt%)
C
Cr
Ni
Co
Mo Fe Mn Si
0.07 20.30 57.35 11.72 8.58 1.01 0.05 0.16
Other
0.76 Al
0.04
20.0
33.0
0.007 S, 0.41 Cu
0.04
16.3
75.9
0.007 S
22.0
62.3
In much older work conducted by Ruther and Greenberg 40 a number of alloys were tested at
650 C in static autoclaves for short durations. The results of these tests are presented in
Figure 22. Results from earlier corrosion tests on type 304 and 406 steel have been included
to establish reference levels. Of the Ni-based alloys tested Inconel 600 and Inconel X750
(Inconel X in Figure 22) offered the weakest corrosion resistance. The three remaining alloys
all showed metal losses less than type 406 (reference line shown in Figure 22) for exposures
up to 56 days (1344 hours).
It has been demonstrated in earlier sections that the surface preparation can influence the
oxidation kinetics in steam exposures. This class of alloy has also been observed to display
this effect. Incoloy 800 is reported to have exhibited a strong surface preparation effect. The
metal loss increased from 0.13 to 4.67 mg.cm-2 when the surface was electro-polished as
opposed to being just wet ground. Jansson et al 41 have conducted further investigations
concerning the effect of surface preparation, comparing ground surfaces with pickled ones.
They found that ground surfaces exhibited lower mass gain than pickled ones for all the
alloys they tested up to 650 C. The beneficial effect of grinding was found to decrease with
increasing temperature up to 800 C, the retardation caused by grinding being reduced to a
factor of 1-3. The results for an alloy with nominal composition of 16% Cr and 75% Ni are
shown in Figure 23.
Page 26 of 39
Figure 22 Corrosion of nickel alloys in flowing steam; 650 C, 42 kg.cm-2, 61 m/s velocity, 30 ppm oxygen
(after Ref. 40)
Page 27 of 39
The improvement bestowed by grinding can be explained in the following way. The grinding
process, it is suggested, causes a high density of dislocations or sub-grain boundaries, these in
turn provide chromium enhanced diffusion paths, which leads to a higher level of chromium
in the metal layer beneath the oxide, thus imparting greater protection to the base metal. The
absence of this retarding effect at the higher temperatures is a consequence of recovery and
recrystallisation of the surface structure. It was also observed that increasing the oxygen
content of the steam caused the retarding effect of grinding to diminish.
One of the main difficulties in reviewing the steam oxidation characteristics of this class of
material has been the lack of information openly available with the literature, although some
interesting points have been highlighted above.
7
It is well known that the oxidation behaviour of alloys in environments that contain water
vapour differs strongly from that in dry atmospheres e.g. oxygen or air. However there is
still not a complete understanding of the mechanisms behind these observations.
There are five main mechanisms commonly used to account for the effect of water vapour on
oxidation kinetics. They are 46:
7.1
Dissociation mechanism
Oxidant-gas penetration mechanism
Formation and volatilisation of Fe(OH)2
Formation and volatilisation of CrO2(OH)2
Changes in the oxide defect structure via proton dissolution
Dissociation Mechanism
The dissociation mechanism, proposed by Fujii and Meussner 47-48, is illustrated in Figure 24.
Metal ions that are generated from reactions at interfaces (2) and (4), migrate to the outer
interface of the oxide scale and the atmosphere (1), where water vapour is adsorbed. The
subsequent reaction forms wustite, defects in the oxide and adsorbed hydrogen. Most of the
hydrogen is believed to desorb, but some will dissolve in the oxide.
Page 28 of 39
(1)
FeO
H2(g)
Fe+2
2e
(2)
Void
H2O(g)
(3)
FeO
+
Spinel
+
Void
(4)
Fe+2
2e
H2(g)
Subscale
Alloy
Figure 24 Dissociation mechanism proposed by Fujii and Meussner (after Ref. 46)
where Fevac indicates a vacant iron ion site and is an electron defect, H* represents
hydrogen dissolved within the oxide.
The iron oxide at interface (2) is believed to dissociate and produce an oxide ion as an
adsorbed species. The permeating hydrogen permits these adsorbed species to react to form
the carrier gas for oxygen transport to the void, as follows,
FeO + Fevac + 2 Fe 2+ + O 2
2 H ( ox ) H 2( ads )
H 2 ( ads ) H 2 ( gas )
H 2 ( ads ) + O 2 ( ads ) H 2 O( g ) + 2e
Wustite layers of the outer scale are formed in this resultant atmosphere by the reaction as
defined for the interfaces (1) and (2). In this way reactions similar to those found at (1) also
occur at interface (3). The composition of the spinel phase is believed to be altered, such
that,
Page 29 of 39
In oxidation studies conducted by Ikeda and Nii 49-50 on Fe-Cr alloys (5-30%Cr) in dry
oxygen and in O2-10% H2O at 750 C and 900 C, it was observed that alloys with chromium
content greater than 15% had a mass gain an order of magnitude greater in wet oxygen than
in dry oxygen. Two possible reaction mechanism were compared, i.e. the dissociation
mechanism (Section 7.1 above) and oxidant-gas penetration. The authors showed by simple
gas kinetics based calculations that the dissociation mechanism was not an appropriate
explanation, and thus proposed the oxidant-gas penetration mechanism. This basically states
that via the development of microcracks and pores in the scale, oxygen and water vapour can
pass directly to the metal surface, assuming the size and frequency of the cracking is large.
Further calculations by Honda et al 51 have confirmed that the supply of oxygen via water
vapour was faster than in dry atmospheres, although it was still insufficient to account for all
the observed weight gains thus causing further doubt regarding the dissociation mechanism.
Oxidation would proceed in the following manner. Initially a chromia layer would cover the
alloy surface. Gas penetrating through a crack or pore into a void could increase the partial
pressure of oxygen within the void in such a way that the iron is oxidized. The resultant
mixed-oxide scale would not be able to prevent outward diffusion of iron ions, and so an
iron-rich outer scale is likely to form. The presence of these microcracks or pores was
confirmed by sulphur decoration.
Ikeda and Nii 50 observed that under dry oxygen conditions the gas penetration was
infrequent, while in wet oxygen conditions it was always observed. It was suggested that the
water vapour caused more frequent scale cracking and also prevented the scale from healing,
thereby accelerating the oxidation process.
7.3
Surman and Castle 52-53 proposed a mechanism based on formation and volatilisation of
Fe(OH)2 to account for the accelerated oxidation of chromium steels in the presence of water
vapour. They suggested that the rate-determining step could be one of three things
The model that they used assumed the following phase boundary reactions for the formation
and dissociation of the volatile hydroxide.
Page 30 of 39
Oxide/gas interface,
7.4
This mechanism would suggest that the evaporation of volatile chromium species primarily
along the grain boundaries accounts for scale cracking and spallation of oxide scales in the
presence of water vapour. Armanet et al 57 exposed a selection of alloys to pure oxygen, air
and wet air. They found that the oxidation of Ni-20Cr in both dry air and wet air resulted in
the formation of three oxides. These were Cr2O3, NiO and the spinel oxide NiCr2O4. Whilst
in the dry air the NiO phase gradually disappeared with time by reaction with chromia to
generate the spinel, in the wet atmosphere, the NiO persisted. Significant cracking and
spallation is reported to have occurred. It is suggested that during volatilisation compounds
rich in chromium would exert an over-pressure, which would cause the oxide layer to burst.
When the chromia-layer was continuous at the alloy/oxide interface, compressive stresses
would be induced causing breakdown of the NiO and NiCr2O4 layers. The chromia layer is
then in contact with the atmosphere and would volatilise as CrO3 and CrO2(OH). The alloy
now depleted in chromium would reoxidise and a new layer of NiO, and eventually spinel,
would develop. This process would continue in such a manner so that the steady state seen
during dry oxidation would not be attained.
Cr2 O3 + 3 2 O2 2CrO3( g )
Cr2 O3 + O2 + H 2 O 2CrO2 (OH ) ( g )
2CrO3 + H 2 O 2CrO2 (OH ) + 1 2 O2( g )
7.5
Water vapour may have an effect on the defect structure and defect-dependant properties of
many oxide systems. It is suggested that water vapour serves as a source of hydrogen, which
Page 31 of 39
where s is an adsorption site on the oxide surface. This would yield a linear rate law and thus
the effect of the water vapour pressure would follow:
r = k .[ ]
where r is the reaction rate, k the rate constant and the fraction of the oxide surface covered
by OH species, this is given by Langmuirs equilibrium equation:
KP
1 + KP
where P and K are the water vapour pressure and equilibrium constants, thus
KP
r = k
1 + KP
In the case of acidic oxides Galerie et al report that the rate of OH surface decomposition was
high enough to supply oxide ions to the scale as rapidly as they were being consumed, and
the reaction rate was not limited by the decomposition. A general rule is proposed: Water
vapour may react more rapidly with metals than oxygen provided that the formed oxide can
Page 32 of 39
Although there is a large body of work concerned with oxidation in steam or high moisture, it
is difficult to extract coherent themes even within a single alloy class. Galerie 58 suggested
the major reasons behind this confusion as being:
Within industrial situations actual pure water vapour, with a maximum oxygen partial
pressure equal to its decomposition pressure, is seldom, if ever encountered.
Atmospheres consisting of pure water vapour are very difficult to produce and
maintain, thus reported observations within the literature are subject to caution.
As a consequence of the two points already mentioned, much of the reported work
deals with atmospheres consisting of O2/H2O, air/H2O or H2/H2O mixtures.
The apparent lack of basic understanding of steam oxidation is understandable given the
complexities encountered. Further work is required to enhance the current level of
understanding, in particular the influence of laboratory test procedures on the resultant data.
The precise in-service conditions are difficult if not impossible to reproduce in the laboratory
and some degree of simplification is necessary. Several exposure procedures are in common
use in the laboratory in ascending order of complexity these are:
However, there is little evidence of comparisons between the different techniques. Such a
comparison between the different methods with in-service exposure would be of great benefit
to the industry as it would give clear guidance as to which method produces oxide scales
which most closely resemble those seen from real exposures.
The review has also shown that subtle changes in the alloy composition can have marked
effects on oxidation resistance in steam atmospheres, for example the addition of Si or rare
earth elements. Further investigation of the effects of alloy composition changes could be
beneficial in terms of future alloy development and plant design.
In addition to these points it is important to remember that oxidation is a dynamic process,
the presence of oxide scales will act to reduce the thermal diffusivity of components. Within
the course of this review very little information regarding the thermal conductivity of
components with oxide scales was apparent. Thus a further avenue of work to aid in the
lifetime evaluation of components must include some activity on thermal conductivity
Page 33 of 39
This work was carried out as part of project LPM2.1 on the Principles of Steam Oxidation,
which is part of the LPM programme on Life Performance of Materials, funded by the UK
Department of Trade and Industry.
The authors are grateful to T. B. Gibbons and to S. R. J. Saunders for their critical review of
the manuscript.
Page 34 of 39
References
10
11
12
13
14
Page 35 of 39
16
17
18
19
20
21
22
23
24
P. Hurst and H. C. Cowen, The Oxidation of 9 Cr-Mo and Other Steels in 6.9
MN/m2 Steam at 748 and 823K, paper 62 in Proc. Int. Conf. On Ferritic Steels for
Fast Reactor Steam Generators, British Nuclear Energy Society, London, 1977.
25
26
Page 36 of 39
28
D.V. Thornton, K.H. Mayer. New Materials for Advanced Steam Turbines.
Proceedings of the 4th International Charles Parsons Turbine Conference.
29
30
31
32
33
34
35
H. Teranishi et al, Fine grained TP347 Steel Tubing with High ElevatedTemperature Strength and Corrosion Resistance for Boiler Applications. The
Sumitomo Search 38, May 1989 pp. 63-74
36
37
38
39
40
Page 37 of 39
42
43
44
45
46
47
48
49
50
Y. Ikeda and K. Nii. Microcrack Generation and its Healing in the Oxide Scale
Formed on Fe-Cr Alloys, Oxidation of Metals, 1978, 12, pp. 487.
51
52
53
54
Page 38 of 39
56
57
58
59
60
61
B.M. Kats, V.V. Kutarov, L.M. Kutovaya, Colloids. Surf. A, 1999, 157, 1-3, 95.
Page 39 of 39