Вы находитесь на странице: 1из 20

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

net/publication/222653587

Continuous and cyclic oxidation of T91 ferritic steel under


steam

Article  in  Corrosion Science · March 2004


DOI: 10.1016/S0010-938X(03)00173-2

CITATIONS READS

105 613

3 authors, including:

Francisco Castro
CEIT Centro de Estudios e Investigaciones Técnicas de Gipuzkoa
114 PUBLICATIONS   1,282 CITATIONS   

SEE PROFILE

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

Low cost novel high speed steels via continuous sintering in nitrogen atmospheres (Brite-EuRam II: CRAFT-BER2-CT94-1425) View
project

All content following this page was uploaded by Francisco Castro on 04 January 2018.

The user has requested enhancement of the downloaded file.


Corrosion Science 46 (2004) 613–631
www.elsevier.com/locate/corsci

Continuous and cyclic oxidation of


T91 ferritic steel under steam
Dionisio Laverde a,1, Tom omez-Acebo b,
as G
Francisco Castro b,*
a
Universidad Industrial de Santander, Escuela de Ing. Quımica, AA 678,
Bucaramanga, Colombia
b
CEIT and Universidad de Navarra, P de Manuel Lardizabal 15, 20018,
San Sebasti
an, Spain
Received 14 November 2002; accepted 2 July 2003

Abstract

The oxidation behaviour of T91 ferritic steel in steam has been studied under isothermal
and non-isothermal conditions within a temperature range between 575 and 700 C. Iso-
thermal treatments resulted in parabolic oxidation kinetics. Three clearly defined oxide layers
constituted the oxide scales. The innermost layer was a (Fe,Cr)3 O4 . The intermediate layer was
porous magnetite (Fe3 O4 ) followed by a compact thinner layer of hematite (Fe2 O3 ). Under
non-isothermal conditions the oxide scales were irregular and evidently cracked. An increase
of the oxidation temperature produces an acceleration of the oxidation process, causing an
increase of the oxide scale thickness that depends on the temperature increase and the ex-
posure time.
 2003 Elsevier Ltd. All rights reserved.

Keywords: A. Chromium steel; B. XRD; B. SEM; B. Thermal cycling; C. Oxidation

1. Introduction

Ferritic steels, containing chromium and molybdenum are well known for their
excellent mechanical properties combining high temperature strength and creep

*
Corresponding author. Fax: +34-943-213076.
E-mail addresses: dlaverde@uis.edu.co (D. Laverde), tgacebo@ceit.es (T. G
omez-Acebo), fcastro@
ceit.es (F. Castro).
1
Temporary Visiting Professor at CEIT. Fax: +57-7-6352540.

0010-938X/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0010-938X(03)00173-2
614 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

resistance with high thermal fatigue life, as well as with good thermal conductivity,
weldability, and resistance to corrosion and graphitisation. Because of these char-
acteristics this type of steels have attracted special interest for application in in-
dustrial processes related to carbochemistry, oil refining, carbon gasification and
energy generation in thermal power plants, where components like, heat exchangers,
boilers and pipes operate at high temperatures and pressures for long periods of time
[1,2]. Amongst these the modified T91 steel, containing 9Cr–1Mo with small addi-
tions of V and Nb, favourably compares to the austenitic grades, for instance the
AISI 316 or 304 types, because of its better mechanical properties that allows it to
support higher stresses at operating temperatures up to 600 C [3]. Also, because of
its higher rupture stress, as compared for instance to the 2 14Cr–1Mo ferritic steel, a
reduced wall thickness may be used resulting in important weight reductions and
savings in the welding process [2–6].
Additionally, during high temperature exposure the interaction between a metal
or an alloy and the surrounding gases and combustion products leads to corrosion,
which is one of the main causes of failure for materials and structures [7–9]. One
corrosion mechanism of this type is the oxidation of materials operating in oxygen
rich environments, which causes chemical attack by reaction with the oxygen con-
tained in the surrounding media. The kinetics of the oxidation process, as well as the
characteristics of the oxidation products are critically dependent on the oxygen
partial pressure of the aggressive environment [9–11,18]. A particular case is found in
thermal power generating plants where metallic materials in the heater and re-
heater zones are subjected to temperatures between 500 and 600 C in contact with
steam.
Given the importance of oxidation on the design, optimisation, safety and the
performance of power generating plants during their lifetime, extensive research
has been devoted to study the oxidation behaviour of chromium containing steels
in contact with steam [12–17,20]. In these works, it is commonly reported that as
a result of the oxidation process under isothermal conditions a protective
Cr-containing oxide is developed on the surface of the steel causing a decrease of the
oxidation rate with time. There also seems to be wide agreement that the oxidation
resistance of these alloys is appreciably reduced in steam as compared to that in
air. Additionally, several reports confirm that the oxide scale is constituted by a
layered structure with compositional and microstructural variations from the
substrate to the outer interface [12–16,20]. On the other hand, depending on
the oxidation temperature and the chemical composition of the steel, both,
the mechanisms of formation and the microstructural characteristics of the
oxide scale, along with the degree of protection it provides, are reportedly
different.
The present report is intended as a contribution to the knowledge of the oxida-
tion behaviour of the T91 ferritic steel in a steam saturated atmosphere under cy-
clic and isothermal conditions. Especially designed experiments along with the
microstructural characterisation of the oxide scales formed, were carried out
in an attempt to understand more fully the mechanisms of formation of the oxide
scale.
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 615

2. Experimental procedure

The oxidation experiments performed during this work were carried out using
samples of a T91 steel whose chemical composition in wt.% is: 8.51Cr–0.90Mo–
0.5Mn–0.31Si–0.1C–0.222V–0.08Nb–410(ppm)N and Fe-balance. The experimental
samples were cut to specified dimensions from tubular sections as illustrated in Fig.
1. The specimens were subsequently polished on SiC emery paper down to the 1200
grade corresponding to about 6–10 lm finish. Polishing was carried out on five faces
leaving the concave bottom surface in the initial as-passivated condition. The
specimens were dried, cleaned in ethanol and accurately weighed and measured to
determine the total surface area exposed to the oxidative environment.
The oxidation treatments were performed using an experimental set up especially
designed for this purpose, so that, three independent lines introduced into the fur-
nace can be used without interfering with each other. The layout of the experimental
device illustrating one of these lines is schematically shown in Fig. 2.

20
10

Fig. 1. Geometry and dimensions in mm of the experimental samples obtained from T91-steel tubular
sections.

Fig. 2. Layout of the experimental equipment [17].


616 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

As a steam carrier, a constant flux of high purity Argon gas (99.999% pure) was
established and fed through a water flask, containing high purity distilled water at 90
C, before reaching the lines into the furnace. At this temperature, a vapour pressure
of 70.1 kPa (70% relative humidity) is produced, being the optimum to generate an
oxidative atmosphere through the three experimental lines. Once saturated in steam
the gas stream is conducted towards the entrance of the furnace using 8 mm diameter
copper tubes, heated to 100 C with heating wires, in order to prevent any water
condensation before reaching the furnace. Inside the furnace three U-shaped AISI
304 stainless steel tubes, independently supported by alumina bricks, are coupled to
the copper tubes, which have a sufficient length, to allow any U-tube to be removed
from the furnace at any time, without disturbing the others. The samples, concave
surface down on small alumina supports, and previously located at a specified po-
sition inside the U-tubes, are thus exposed to the flowing gaseous atmosphere (at a
pressure of 1.25 bar) and their temperature was obtained from a previous calibration
with the aid of a thermocouple introduced up to the same position. The excess steam
after the oxidation reaction is condensed and recovered in a cold vessel while the
argon gas bubbles through the water before reaching the environment outside the
system. Before starting any oxidation treatment the system was purged for 2 h using
dry argon gas in order to flush out any traces of air. Purging was also used at the end
of each experiment while waiting for the corresponding U-tube and the specimens
inside to cool down to room temperature.
The oxidation tests were performed under isothermal and cyclic conditions within
a temperature range comprising the current operating temperatures of this alloy
under industrial conditions. The conditions used for the isothermal treatments are
summarised in Table 1, while the profile for the non-isothermal oxidation treatments
is illustrated in Fig. 3. In the figure, points labelled as a, b, c and d indicate the
conditions for obtaining specimens for subsequent characterisation.
With the purpose of obtaining data for the study of the oxidation kinetics, the
oxidised specimens were accurately weighed, with a precision up to 0.1 mg, in order
to determine the corresponding weight gain. Additional observation of the samples
under the SEM also allowed to measure the mean thickness of the oxide scales
generated after every experiment.

Table 1
Combinations of time and temperature used for isothermal oxidation experiments
Temperature (C) 575 600 650
Oxidation time (h) 14 23 20
22 29 44
48 48 68
73 68 90
77 95 138
93 120 164
136 161 210
166 186 242
240 209
329 233
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 617

6h 6h
700 ºC

72 h 45 h 45 h
(a) 650 ºC
a b c d

6h 6h
700 ºC

89 h 65 h 65 h
(b) 600 ºC
a b c d

6h 6h
650 ºC

89 h 65 h 65 h
(c) 600 ºC

Fig. 3. Temperature and time profiles used for cyclic oxidation experiments: (a) Profile 1, (b) Profile 2, and
(c) Profile 3.

The microstructure of the oxide scales was characterised using a SEM fitted with
EDS in order to determine their chemical composition under the area of observation.
The specimens were observed in cross-section after being mounted in conductive
resin and polished down to 1 lm finish. XRD using Cu-K (k ¼ 1:54 A ) radiation was
performed to characterise the phases present in the oxide products of every sample.

3. Results and discussion

The microstructure of the modified T91 steel samples, after polishing and etching
with a 1:1 solution of Nital and Vilellas, according to the ASTM E-407 standard, is
shown in Fig. 4. The micrograph reveals a martensitic structure with a dispersion of
fine M23 C6 and V4 C3 carbides typical of this type of steels.

3.1. Isothermal oxidation tests

As mentioned before a set of isothermal oxidation experiments were carried out in


the temperature range between 575 and 650 C for up to 329 h. Under these con-
ditions measurement of, both, the weight gained by the specimens and the thickness
of the oxide scales were used to follow the oxidation kinetics. In agreement with
previous observations [16,19–21] parabolic oxidation kinetics were observed at all
618 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

Fig. 4. SEM micrograph for a polished and etched sample of the modified T91 ferritic steel in the as
received condition.

140
575 oC
120 600 oC
650 oC
100
Thickness (µm)

80

60

40

20

0
0 50 100 150 200 250 300 350 400
Time (h)

Fig. 5. Oxide scale thickness as a function of oxidation time at three different temperatures.

temperatures. As an illustration, Fig. 5 shows a plot of the data gathered from


measurements of the oxide scale thickness after several oxidation times. As appre-
ciated in the figure the experimental data points, corresponding to every tempera-
ture, can be reasonably fitted by a parabolic curve through the origin of the
coordinate system, represented by the equation
d 2 ¼ kp t ð1Þ
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 619

Table 2
Oxidation rate constants as determined from the measurements of, both, the oxide scale thickness and the
weight gain at different temperatures
Temperature (C) With thickness kp With weight gain kp
(mm2 /min) (mg2 /mm4 /min)
575 8.52 · 108 3.00 · 107
600 2.06 · 107 5.32 · 107
650 1.13 · 106 3.14 · 106
Q (Activation energy, kJ/mol) 224.8 207.9

where d is the total thickness of the oxide scale and t the oxidation time. The values
of kp obtained from curve fitting at every temperature are summarised in Table 2.
Expressing the relationship between the parabolic rate constant and temperature
in an Arrhenius type of equation, kp is given by
kp ¼ k0 expðQ=RT Þ ð2Þ

where Q is the activation energy of the process, R, the gas constant and T , the ab-
solute temperature. A logarithmic plot of Eq. (2), using the values in Table 2, allows
the determination of the activation energy Q for the oxidation process. As observed
in Fig. 6, the data is reasonably fitted by two straight lines yielding the values of
224.8 and 207.9 kJ/mol (Table 2) for the slopes corresponding to the measurements
of the oxide thickness and weight gain, respectively.
The difference between the values for the activation energy is interpreted in terms
of the larger dispersion of the data obtained from the weight gain measurements.
This dispersion of the data is thought to be associated to experimental factors

Fig. 6. Arrhenius plot for the parabolic rate constants as a function of the oxidation temperature.
620 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

leading to difficulties in the precise determination of the area exposed to oxidation


due to the concave surface, as well as, slight damage of the oxide scales during
weighing. Because of these reasons, the value obtained from thickness measurements
is considered to be more precise, being also in reasonable agreement with previously
reported values of the activation energy [16,20] for this kind of steels. It must also be
emphasised that the activation energy corresponds to the overall process and should
therefore be interpreted as an apparent activation energy involving a combination of
mechanisms like the outwards migration of metallic cations as well as the inwards
diffusion of oxygen ions.
To characterise the microstructure of the oxide scale and the distribution of ele-
ments from the outer surface to the oxide/substrate interface a set of polished
specimens were observed in cross section under the SEM. In Fig. 7 for instance, it
may be observed that the microstructure of the oxide scale exhibits a layered
structure with morphological changes from the gas/oxide (bottom of the micro-
graph) to the oxide/substrate interface.
The first layer, in contact with the substrate, is larger than the other two, has a
compact appearance and the oxide/substrate interface is noticeably fractured, most

Fig. 7. SEM micrographs showing a cross section of two specimens after isothermal oxidation at 650 C
for (a) 20 h and (b) 210 h.
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 621

Fig. 8. SEM image for the cross section of a specimen after isothermal oxidation at 575 C for 77 h.

probably due to differences in thermal expansion coefficients between the substrate


and the oxide scale, particularly after long oxidation times. In contrast, the layer next
to it, seems to be well adhered but exhibits a high degree of porosity. The last layer,
in contact with the environment, is thinner than the internal ones, is highly compact
and also shows a fractured interface with the neighbouring layer. As appreciated, by
comparing the two micrographs in Fig. 7, and also that in Fig. 8, the characteristics
mentioned above do not suffer alteration regardless the time and temperature of the
oxidation treatment. The only noticeable difference is total thickness of the oxide
scale, being larger for higher temperatures or longer times, as shown earlier in Fig. 5.
It is interesting to notice that the relative thickness of these oxide layers remains
relatively constant for every oxide scale independently of the time and temperature
combination used during oxidation.
The chemical analyses obtained by EDS on cross sectional views revealed im-
portant compositional variations from layer to layer. Additionally, the XRD data
showed that the oxide layers were constituted by different phases. As an example,
this is illustrated in Fig. 9 for a set of specimens oxidised under different conditions.
The profiles shown, account for the content of Cr, Fe and oxygen as measured from
the substrate/oxide interface to the outer surface of the oxide scales. Other elements,
like molybdenum for instance, although present in small amounts in, both, the Cr-
containing spinel oxide layer and the substrate, do not seem to exert any influence on
the oxidation process as reported before [15].
As observed in the figures, the concentration profiles for these three elements
exhibit a similar tendency at all temperatures and times. Basically, it is apparent that
just after a few microns from the substrate the concentration of iron strongly de-
creases (from about 90% in the substrate as shown on the vertical axis of each dia-
gram), while oxygen increases reaching a value of about 60 at%. In contrast,
chromium concentration seems to remain constant up to a point, roughly coinciding
in all cases with half the total thickness of the oxide scale, where it drops to a
negligible concentration.
622 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

100 100
(a) Fe (b) Fe
Cr Cr
80 O 80 O

60

At-%
60
At-%

40 40

20 20

0 0
0 10 20 30 40 50 0 20 40 60
Distance from substrate (µm) Distance from substrate (µm)

100
(c) Fe
Cr
80 O

60
At-%

40

20

0
0 30 60 90
Distance from substrate (µm)

Fig. 9. Chemical composition profiles obtained by EDS on cross sectional views for specimens oxidised at:
(a) 575 C for 329 h, (b) 600 C for 161 h and (c) 650 C for 138 h.

In order to match the concentration profiles with the corresponding microstruc-


tural images observed under the SEM, a set of X-ray mappings were obtained. The
results are shown in Fig. 10.
As indicated by these images the concentration of chromium drops at a distance
exactly coincident with the interface between the two inner layers in the oxide scales.
Simultaneously, as noticed from the X-ray map for iron, the most inner layer must
correspond to a mixed oxide containing chromium and iron, while the most external
ones only contain iron and oxygen, thus corresponding to iron oxides. The sequence
of the oxide layers formed also confirms previous observations [12–15,17,20,22] and
it may therefore be considered as a common feature of the oxidation behaviour of
chromium containing steels.
In order to identify the phases present in the oxidation products, formed after
each oxidation experiment, the oxide layers were characterised by X-ray diffrac-
tometry, through a series of XRD runs followed by careful polishing operations for
removing the layers one by one. It was observed that, both, the sequence of the oxide
layers formed and the phases constituting those, were the same in all cases under the
experimental range of time and temperature conditions used in the present work. As
appreciated in Fig. 11a, the first XRD run carried out on the surface of the outer-
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 623

Fig. 10. EDS X-ray mappings on cross sectional views for specimens oxidised at: (a) 575 C for 329 h,
(b) 600 C for 161 h and (c) 650 C for 138 h.

30 30
Intensity1/2 (arbit. unit)
Intensity1/2 (arbit. unit)

H M Magnetite
(a) H Hematite (b) W Wüstite
M Magnetite
25 F Ferrite 25 M F Ferrite
20 H
H H HH 20
H M
15 H 15 M
M M
10 F H HH 10 F M M
H M W W MM
MH F
5 5
0 0
20 40 60 80 100 20 40 60 80 100
Diffraction angle, 2θ (deg) Diffraction angle, 2θ (deg)

Fig. 11. XRD traces from a sample oxidised at 600 C for 186 h as obtained on: (a) outer surface; (b) same
as (a) after careful polishing for removing the upper layer.

most layer indicated the presence of hematite (Fe2 O3 ) as the dominant iron oxide
formed. The faint reflections corresponding to magnetite (Fe3 O4 ) were interpreted as
belonging to the second oxide layer as verified in Fig. 11b which corresponds to the
XRD trace of such a layer after removal of the first one. As a somewhat unexpected
feature, according to the iron–oxygen phase diagram [23], a small amount of ferrite
was detected in both cases by the reflection appearing at an angle of 2h ¼ 44:5.
However, this phase has also been previously reported, by Saeki et al. [12] who found
small amounts of ferrite in the iron oxide layers formed after oxidation of an AISI
304 stainless steel.
Taking into account the results shown in Fig. 11 it is clear, however, that the main
constituents of the two outer oxide layers are hematite and magnetite respectively.
624 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

After removal of the porous magnetite layer the XRD trace obtained was identical to
that of Fig. 11b. In this case, however, combining this information with the EDS
analyses in Fig. 9 the results presented in Fig. 10, the inner chromium-containing
oxide layer was identified as (Fe,Cr)3 O4 whose XRD spectrum is entirely coincident
with that of magnetite in Fig. 11b according to the XRD database (PC PDF WIN)
used.
As a summary, it may consequently be stated that within the temperature range
used in this work, the oxide scales formed after oxidation in steam of this T91 steel
consist of three distinct oxide layers mainly constituted by (Fe,Cr)3 O4 , Fe3 O4 , and
Fe2 O3 respectively.

3.2. Re-oxidation experiments

In an attempt to gain a more fully understanding of the mechanisms of formation


of the oxide scale and the role played by the oxide layers in the passivation process of
the steel, especially designed re-oxidation experiments followed by microstructural
characterisation of the oxide scales formed were carried out on partially polished
specimens. The starting situation was chosen from a set of specimens isothermally
oxidised at 650 C for 190 h. As expected, the characteristics of the oxide scales
developed and illustrated in Fig. 12a, closely followed the pattern described above.

Fig. 12. SEM micrographs of specimens obtained (a) after isothermal oxidation at 650 C for 190 h; (b) re-
oxidised for 70 h after removing the external layer, and (c) re-oxidised for 70 h after removal of both the
external and the intermediate layers.
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 625

Fig. 13. X-ray mapping showing distribution of constituting elements for sample in Fig. 12c.

The mean thickness of the oxide scales formed was about 109 lm and were con-
stituted by an innermost Cr-containing layer of about 51 lm, followed by a 38 lm
porous magnetite intermediate layer and an outer 20 lm hematite layer. One sample,
designated as SCM (Substrate + Cr-spinel + Magnetite), out of this set of specimens,
was carefully polished to remove the outermost hematite layer, while polishing of a
second specimen, designated as SC (Substrate + Cr-spinel), was continued until the
magnetite layer was totally removed. These two specimens were therefore subjected
to a re-oxidation treatment under identical conditions for a further period of 70 h.
After re-oxidation the results illustrated by Figs. 12 and 13, revealed two different
final situations. On the one hand, re-oxidation of specimen SCM lead to the resto-
ration of the hematite layer accompanied by growth of the whole oxide scale (Fig.
13). At the end of the 70 h exposure time the Cr-containing layer had slightly grown
from 51 lm to about 59 lm, the magnetite layer from 38 to 59 lm and the hematite
re-generated layer was around 16 lm. As a result the total thickness of the oxide
scale formed had noticeably increased and was comparable to that developed after
isothermal oxidation for an equivalent total oxidation time corresponding to the sum
of oxidation and re-oxidation (190 + 70 h).
In contrast, re-oxidation of specimen SC only lead to restoration of a thin (7 lm)
non-adherent (notice in Figs. 12c and 13, that this layer was detached from the
specimen by the metallographic mounting resin) hematite layer on top of the Cr-
containing layer without re-formation of magnetite.
A schematic summary of these observations, from the standpoint of the oxide
products formed, could be written as
626 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

• For SCM: re-oxidation of Substrate + Cr-spinel + Magnetite fi S + C + M + H


with continued growth of the oxide scale.
• For SC: re-oxidation of Substrate + Cr-spinel fi S + C + H without re-formation
of magnetite and without growth of the Cr-spinel layer,

where Cr-spinel means (Fe,Cr)3 O4 , S stands for substrate and H for hematite.
From these results it can be concluded that, under the experimental conditions
used in the present work, the formation of hematite as an outermost layer is a
common feature under all oxidation conditions. However, the re-oxidation experi-
ments suggest that it could be formed by at least two different mechanisms. The first
one and more evident, may be based on the thermodynamic instability of magnetite
in contact with an atmosphere of high oxygen potential such as steam, thus leading
to the chemical reaction:
2Fe3 O4 þ H2 O ! 3Fe2 O3 þ H2 ð3Þ
This mechanism could reasonably account for the experimental observation after re-
oxidation of specimen SCM. Further support to this argument is gained by in-
spection of the isothermal section of the Fe–Cr–O phase diagram in Fig. 14, which
was obtained by calculation at 650 C using the Thermo-Calc computing programme
and assessed thermodynamic data [24]. In the figure it may be expectedly noticed
that Fe2 O3 is the stable iron oxide at high oxygen potentials.
On the other hand, the result obtained after re-oxidation of specimen SC seems to
be based on a different mechanism, since the formation of hematite by selective
oxidation of Fe in a Cr-containing material at the present concentrations does not

Fig. 14. Calculated pO2 ––composition diagram of the Cr–Fe–O ternary system at 650 C using data from
Ref. [24]. Black dot indicates the composition of the T91 steel. Broken line is the tie-line of the Cr2 O3 oxide
in equilibrium with the steel.
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 627

seem thermodynamically likely. In order for hematite to be formed it is clear,


however, that a flux of iron cations must be established. But, the outwards migration
of Fe cations could be promoted by the steep Fe gradient created from the metal/
oxide to the oxide/gas interface and the inverse oxygen potential. This would lead to
the formation of an iron oxide on the surface of the Cr-containing layer, where the
gaseous atmosphere imposes high oxygen potential. The reason for the direct for-
mation of hematite, without any previous magnetite is not well understood, however,
comparing the results obtained from the re-oxidation of specimens SCM and SC it
seems that the growth of the Cr-containing layer is catalysed by the presence of
magnetite.
According to these observations and based on the isothermal section in Fig. 14,
the oxidation process of this T91 steel could be reasonably described by the following
steps.
Taking into account Fe and Cr only, this simplified composition of the steel is
illustrated in Fig. 14. Upon oxidation a horizontal tie-line from this point indicates
that equilibrium is established with Cr2 O3 . This is also consistent with a recent report
[14] where the development of this oxide was observed as the first oxidation product.
This would be followed by the outwards migration of Fe cations under the gradients
mentioned above that could even be reinforced by the possible generation of a Cr-
depleted, or alternatively Fe-rich, zone on the side of the steel at the oxide–metal
interface. Since the oxygen potential (pO2 for the test atmosphere of 1.8 · 108 bar) is
higher at the oxide/gas interface, according to Fig. 14, the continued flux of Fe ca-
tions could take the system into any of the bi-phasic regions eventually forming the
observed (Fe,Cr)3 O4 spinel and magnetite. Since the oxygen potential gradually in-
creases from the substrate to the oxide/gas interface the isothermal section also
possibly predicts the most external hematite layer. The formation of hematite by
reaction (3) would probably cause the porosity observed in the magnetite layer by
trapping or ejection of hydrogen or steam.
As the oxidation time increases the iron flux, and hence the growth of the oxide
scale continues. This produces the consumption of the metal with parabolic kinetics
under the main passivating effect provided by the (Fe,Cr)3 O4 oxide layer. The inward
diffusion of oxygen anions is obviously another part of the process in order to
maintain the magnetite-catalysed growth of the Cr-containing layer towards the
interior of the steel substrate. Incidentally, a secondary reaction between the released
hydrogen and the surrounding iron oxides, could lead under specific local reduction
conditions to the formation of small amount of ferrite as detected by XRD. Fig. 15
represents a schematic illustration of the proposed mechanism.

3.3. Cyclic oxidation tests

During operation of the power generating plants the components subjected to


high temperature are not always under isothermal conditions due for instance to
thermal transients or the formation of hot spots. It was therefore considered at-
tractive to carry out a set of experiments to study the oxidation behaviour of the T91
steel under thermal cycling. As mentioned in the experimental procedure the thermal
628 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

Fig. 15. Sketch of the proposed oxidation mechanism of T91 steel under steam.

cycles consisted in an initial isothermal treatment between 72 and 89 h followed by


two holding periods of 6 h each at an oxidation temperature, either, 50 or 100 C
higher than the initial one, according to the illustration of the profiles in Fig. 3. Four
specimens were oxidised per cycle to be withdrawn at different stages according to
points a, b, c and d in the same figure. The results obtained were compared to those
corresponding to the isothermal oxidation treatments.
After non-isothermal treatments the oxide scales formed were very irregular and
evidently cracked due to differences in thermal expansion coefficients between the
constituting layers and the substrate. As an example, this is illustrated in the mi-
crograph of Fig. 16 obtained from a specimen after oxidation at 600 C followed by
a period at 700 C and another at 600 C again. The specimen corresponds to point b
in Fig. 3b.
In terms of the oxidation kinetics the general trends were very similar for the three
profiles used and could be followed by reference to Fig. 17, which shows the mea-
surements obtained after each oxidation treatment. As observed in the figure mea-
surements taken after the first temperature increase (point a) always evidenced the
accelerated growth of the oxide scale due to exposure at the higher temperature. Fig.
17a, for instance, shows that exposure of the specimen to 700 C for 6 h causes an
increase in thickness equivalent to more than 60 h under the initial isothermal
conditions at 600 C. Also, considering Fig. 17c for instance, it is clear that the final
thickness of the oxide scale only increased up to an intermediate point between those
corresponding to the isothermal treatments at either temperature. Continued ex-
posure of the specimens at the initial isothermal treatment temperature after the first
temperature increase, thus reaching point b in Fig. 17, resulted in an oxidation ki-
netics which could be very closely expressed by a parallel curve to that of the initial
isothermal treatment (dotted line in the figure).
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 629

Fig. 16. SEM micrograph of the oxide scale formed after non-isothermal oxidation of a specimen under
the conditions corresponding to point B in Fig. 3b.

Fig. 17. Kinetic data obtained from non-isothermal oxidation treatments as compared to the curves of
previous isothermal results, corresponding to temperature combinations (a) 650–700 C, (b) 600–700 C
and (c) 600–650 C.

Unfortunately, specimens subjected to a second temperature increase presented


extreme difficulties for accurately measuring, either, weight gains or thickness of the
oxidation products. This was due to the irregularity, brittleness and loss of adherence
of the fractured oxide scales, even leading to the detachment, during handling, of
630 D. Laverde et al. / Corrosion Science 46 (2004) 613–631

small pieces from the oxide scales. Points c and d in Fig. 17 reflect this fact mis-
leadingly indicating a decrease in thickness of the oxide scales after longer oxidation
periods. Given this situation it was not possible to draw a clear conclusion after the
second thermal cycling. However, it may be worth mentioning that after detailed
examination of the specimens under the electron microscope, it is believed that the
second thermal cycle did not produce a similar increase in thickness as that produced
after the first cycle. In any case, considering only the first thermal cycling, it seems
that any temporary temperature increase or presence of hot spots occurring in
practice will produce an acceleration of the oxidation process, causing a decrease of
lifetime, which is dependent on the magnitude of the temperature difference and the
length of the exposure time at the higher temperature.

4. Conclusions

The isothermal oxidation in steam of this T91 ferritic steel follows parabolic ki-
netics in the range of temperatures between 575 and 650 C. The passivation effect is
related to the development of an oxide scale formed by three clearly distinct oxide
layers mainly constituted by (Fe,Cr)3 O4 , Fe3 O4 , and Fe2 O3 from the substrate to the
oxide/gas interface respectively. The continuous growth of the oxide scale during the
exposure time requires the outwards migration of Fe cations and the inwards dif-
fusion of the oxygen ions. A mechanism of formation of the oxide layers is suggested
based on experimental observations obtained from re-oxidation treatments and the
calculation of an isothermal section of the Fe–Cr–O phase diagram. During non-
isothermal oxidation treatments it was observed that a temporary temperature in-
crease produces an acceleration of the oxidation process that is dependent on the
temperature difference and the exposure time at the higher temperature.

Acknowledgements

The authors gratefully acknowledge ‘‘Direcci on de Polıtica Cientıfica’’ of the


Basque Government for the grant provided to one of the authors, Dr. D. Laverde,
and for the financial support for the realisation of this work.

References

[1] J.C. van Wortel, C.F. Etienne, F. Arav, Application of modified 9chromium steels in power
generation components, in: VDEh ECSC Information Day, The Manufacture and Properties of Steel
91 for the Power Plant and Process Industries, Dusseldorf, 5th November, 1992, paper 4.2.
[2] T. Fujita, Current progress in advanced high Cr steel for high temperature applications, ISIJ Int. 32
(2) (1992) 175.
[3] J. Orr, A. Di Gianfrancesco, The effect of compositional variations on the properties of steel 91, in:
VDEh ECSC Information Day, The Manufacture and Properties of Steel 91 for the Power Plant and
Process Industries, Dusseldorf, 5th November, 1992, paper 2.4.
D. Laverde et al. / Corrosion Science 46 (2004) 613–631 631

[4] ASTM, Standard specification for seamless ferritic and austenitic alloy-steel boiler, superheater and
heat-exchanger tubes, A 213/A 213M-90a, 1990.
[5] ASTM, Standard specification for seamless ferritic alloy-steel pipe for high-temperature service, A
335/A 335M-95a, 1996.
[6] R. Panton-Kent, Phase balance in 9%Cr1%Mo steel welds. TWI report, Bulletin 1, January/February
1991, p. 15.
[7] G.E. Birchenall, A brief history of the study of oxidation of metals and alloys, in: High Temperature
Corrosion, Proceedings, NACE, San Diego, CA, 1981, p. 3.
[8] D.A. Jones, Principles and Prevention of Corrosion, second ed., Prentice Hall, USA, 1996.
[9] P. Kofstad, High-temperature Corrosion, Elsevier Applied Science, London, 1988, pp. 382–385
(Chapter 11).
[10] J.A. Gonz alez, Corrosion metalica a temperaturas elevadas: Aspectos teoricos de la Oxidaci on y
Sulfuracion, Rev. Metal. CENIM 7 (4) (1971) 307.
[11] R.A. Rapp, The high temperature oxidation of metals forming cation-diffusing scales, Metall. Trans.
B 15B (2) (1984) 195.
[12] I. Saeki, T. Saito, R. Furuichi, M. Itoh, Growth process of protective oxides formed on type 304 and
430 stainless steels at 1273 K, Corros. Sci. 40 (8) (1998) 1295.
[13] S. Jianian, Z. Longjiang, L. Tiefan, High temperature oxidation of Fe–Cr alloys in wet oxygen, Oxid.
Met. 48 (3, 4) (1997) 347.
[14] Z. T€okei, H. Viefhaus, H.J. Grabke, Initial stages of oxidation of a 9CrMoV steel: role of segregation
and martensite laths, Appl. Surf. Sci. 165 (1) (2000) 23.
[15] A.P. Greeff, C.W. Louw, H.C. Swart, The oxidation of industrial FeCrMo steel, Corros. Sci. 42 (10)
(2000) 1725.
[16] A. Arıztegui, T. G omez-Acebo, F. Castro, Steam oxidation of ferritic steels: kinetics and
microstructure, Bol. Soc. Esp. Ceram. Vidr. 39 (3) (2000) 305.
[17] M.R. Rodrıguez, Predicci on de vida bajo condiciones de fatiga, oxidaci
on y fluencia en componentes
de centrales termicas, Tesis doctoral, Escuela Superior de Ingenieros, Univ. de Navarra, 2002.
[18] H. Taimatsu, Kinetic analysis of high-temperature oxidation of metals accompanied by scale
volatilization, J. Electrochem. Soc. 146 (10) (1999) 3686.
[19] B.J. Downey, J.C. Bermel, P.J. Zimmer, Kinetics of the nickel–chlorine reaction at temperatures
between 350 and 600 C, Corrosion 25 (12) (1969) 502.
[20] A.S. Khanna, P. Rodriguez, J.B. Gananamoorthy, Oxidation kinetics, breakaway oxidation, and
inversion phenomenon in 9Cr–1Mo steels, Oxid. Met. 26 (3, 4) (1986) 171.
[21] J.P.T. Vossen, P. Gawenda, K. Rahts, M. Schorr, M. Schutze, Limits of the oxidation resistance
of several heat-resistant steels under isothermal and cyclic oxidation as well as under creep in air at
650 C, Mater. High Temp. 14 (4) (1997) 387.
[22] H. Asteman, J.-E. Svensson, L.-G. Johansson, Oxidation of 310 steel in H2 O/O2 mixtures at 600 C:
the effect of water-vapour-enhanced chromium evaporation, Corros. Sci. 44 (11) (2002) 2635.
[23] T.B. Massalski, in: Binary Alloy Phase Diagrams, vol. 1, ASM, 1986.
[24] J.R. Taylor, A.T. Dinsdale, A thermodynamic assessment of the Cr–Fe–O system, Z. Metallkd. 84 (5)
(1993) 335.

View publication stats

Вам также может понравиться