Materials Science and Engineering, A 153 ( 19 9 2) 5 7 3- 5 7 7 5 73

Behavior of iron aluminides in oxidizing and oxidizing/sulfidizing

environments

P. F. T o r t o r e l l i a n d J. H. D e V a n
Oak Ridge National Laboratory, Oak Ridge, TN 37831 (USA)

Abstract
Fe3AI alloys have potential as structural materials and coatings for use in hostile environments because they form protec-
tive scales at very low oxygen partial pressures at elevated temperatures. Weight gains and parabolic rate constants were
small for exposures to dry air at 800 and 900 °C. Chromium additions to Fe3AI resulted in higher initial rates of oxidation
in dry air at 800 and 900 °C and substantially increased overall reaction rates in a very aggressive oxidizing/sulfidizing
mixed gas when greater than 2 at.%. There was some evidence that an a-alumina layer formed during extended air
exposure at 900 °C. The ability to rapidly form an oxide layer appeared to limit damage caused by any loss of scale during
thermal cycling. Preoxidation reduced the corrosion of Fe3AI-5%Cr in the oxidizing/sulfidizinggas, but aluminizing this
alloy to form an FeAI coating appeared to be the best way to assure long-term protection in sulfidizing environments.

1. Introduction under argon, drop casting into copper molds, and hot-
rolling to plate and sheet. For comparison, an FeCrAI
Iron aluminides based on Fe3A1 or FeAI have alloy consisting of 18Cr and 12A1 (this and all con-
potential as structural materials and coatings because, centrations hereafter are expressed in at.%) was
in many cases, they form protective scales at relatively obtained from Harwell Laboratory. Rectangular
low oxygen partial pressures at elevated temperatures coupons, approximately 16 m m × 76 mm × 1 mm, were
[1]. While most high temperature corrosion studies of used in the corrosion studies. All major surfaces were
iron aluminum alloys have been concerned with their mechanically ground with 600-grit abrasive before
oxidation properties in air or pure oxygen, some recent exposure. For Fe-28A1-5Cr-0.5Nb-0.2C, the surface
work on the corrosion of specific iron aluminides has aluminum concentration was increased to 50% by
also included examination of the effects of oxidizing/ aluminizing. Coated specimens were prepared at Ohio
sulfidizing and molten salt environments [2]. The State University using a developmental pack cementa-
present paper describes some recent results on the tion process [4] to produce FeA1 surface layers 50-150
growth and nature of the scales that form on iron /~m in thickness.
aluminides, and observations on the apparent ways The corrosion behavior was first evaluated by
they impart corrosion resistance under different en- thermogravimetric analysis (qGA). Controlled flows of
vironmental and alloying conditions. either dry air or H2S/H2/H20/Ar were used. For the
It should be noted that the exposure temperatures at mixed gas, the chemical activity of oxygen was continu-
which these corrosion experiments are conducted are ously monitored by contact with a Y203-stabilized
normally above those that are practical for structural ZrO2 electrochemical cell at the exposure temperature
applications of either Fe3AI or FeAI. However, because and was set by the controlled addition of water vapor.
the iron aluminides often show superior resistance to The partial pressure of sulfur at the reaction tempera-
aggressive environments at higher temperatures, the
ture was calculated from the known initial H 2 :H2S ratio
possibility exists for using such compositions as coat- assuming chemical equilibrium at the exposure tem-
ings or claddings on higher strength, but less corrosion
perature. In the present study, the oxidizing/sulfidizing
resistant, materials. Initial results on the high tempera- environment consisted of a gas mixture yielding O r and
ture corrosion of steels coated with iron aluminide $2 pressures of 10 -22 atm and 10 -6 arm respectively.
have recently been reported [3]. Such a gas mixture is a rather severe simulation of
service conditions associated with a heat recovery
2. Experimental details steam generator designed to cool product gas from a
coal gasifier. Details of the TGA exposure systems and
Fe3AI alloys (all consisting of 28 at.% AI) were procedures can be found elsewhere [5].
prepared by arc melting high purity component metals Following exposure in a microbalance system, the

0921-5093/92/$5.00 © 1992 - Elsevier Sequoia. All rights reserved

574 P. F. Tortorelli, J. H. De Van / Behavior of iron aluminides

coupons were visually checked for evidence of loss of of F e - 2 8 A I - 6 C r and - 1 0 C r - 0 . 1 Z r - 0 . 0 5 B - 0 . 5 N b -

corrosion products by spallation. Selected specimens 0.5Mo-0.02Y. An influence of the minor alloying
were then analyzed by optical imaging of polished elements or variations between 4 and 10Cr on weight
cross sections and/or by scanning electron microscopy gains was not readily apparent. Assuming parabolic
(SEM) with associated energy dispersive X-ray analysis oxidation kinetics, this difference between (0-2)Cr and
(EDX), F%AI alloys with higher chromium concentrations was
also evident from the measured parabolic rates con-
stants, kp:
3. Results (Ao )21t = % (1)
3.1. A i r oxidation where A w is the change in weight per unit surface area
Weight change data from exposures of iron after exposure for time t (see Table 1). The kp values
aluminides to 800 °C air are shown in Fig. 1. All the for the iron aluminides with 4Cr or more were approxi-
measured weight gains were quite small. However, the mately a factor of 10 greater than those for
eight weight change curves of Fig. 1 tended to fall into Fe-28AI-(0-2Cr). For comparison, the weight gains
two groupings. The lower grouping consisted of the and parabolic rate constant for the FeCrAI alloy were
three Fe-28A1 coupons and one composed of also measured and found to be comparable with those
Fe-28AI-2Cr, while the higher cluster included of the Fe3AI alloys with lower concentrations of
two specimens of Fe28AI-4Cr-0.1B and one each chromium (see Table 1).
At 900 °C, the effect of chromium on the oxidation
behavior of Fe3AI in air was less pronounced than at
0.25 Air Oxidation, 800°C ' ' I 800 °C. As shown in Fig. 2, increasing chromium con-
Fe-28A1 [ centration of ternary Fe-28A1-Cr alloys led to a more
'~" 0.20 imnmIl U g• < 10CT+ 1
rapid weight gain during initial exposure. However,
there was no significant influence on the oxidation rate
,," . ~ 4Cr / after about 100 h. The parabolic rate constants (calcu-
iiii ~ n
o.15 , ,• , . . . ' . : o""Ill ~ 4Cr+ J] lated from curve fits over the entire exposure range)
0011, O000 / fell within a range of (1-3)× 10 -7 g2 m-4 s-1 (4-10 x
• ='.=II', °° o" I 10-5 mg2 cm-4 h - l ) and did not vary monotonically
= • .....o°°':..° J
0.10
~ AmI~A'~.Oooo" 1I with chromium concentration. The FeCrA1 alloy
/ =I#° °°° x ~ OCr showed similar oxidation behavior at 900 °C.
I d* .,8o00 x x * - ~ I As expected from the small weight gains, the scales
that formed on the oxidized Fe3AI alloys were quite

0.00 -~ - " ~ 0Cr

0 50 100 150 200

I I I i I I
Time (h) Air Oxidation, 900°C
Fig. 1. Weight change versus exposure time for Fe-28Al alloys
exposed to dry air at 800 °C. Chromium concentrations are in
at.%. Plus sign indicates presence of additional elements in small
concentrations.

TABLE 1. Parabolic rate constants measured for iron

aluminides and FeCrAI exposed to dry air at 800 °C for greater ,%
than 160 h 0.10

Alloya ko

(g2 m-4 S-1) (mg2cm -4 h-1)

28A1 (6.4 + 3.1) x 10 -7 (2.3 + 1.1)x 10 -5

28AI-2Cr 3.6 × 10 -7 1.3 x 10 -5 0.00
0 50 100 150 200 250 300
28AI-4Cr (5.0 + 2.0) x 10 -a (1.8 + 0.7) x l0 -4
28AI-10Cr 5.6 x 10 -6 2.0 x 10 -4 Time (h)
20Cr-12AI 4.2 X 10 -7 1.5 X 10 -5
Fig. 2. Weight change versus exposure time for binary Fe-28Al
aConcentrations in at.%, balance is iron and minor alloying and ternary Fe-28Al-Cr alloys exposed to dry air at 900 °C.
additions. Chromium concentrations are in at.%.
 P. F, Tortorelli, J. H. De Van / Behavior of iron aluminides 575

Fig. 3. Polished cross section of Fe-28A1-4Cr exposed to dry air

for 168 h at 900 °C.

100 i
800°C Fig. 5. Preoxidized Fe-28A1-5Cr (with small amounts of
~IPFeCrAI zirconium and boron) exposed to an oxidizing/sulfidizing
environment at 800°C for 170 h. The whiskers and other
~a 10
products are principally chromium sulfides.
Fe-28AI-5Cr

c," OOOOOQOOOOOq'
****~*****~*~****~Pre0xidized Fig. 4, the higher chromium concentration of typical
• • • • • • • • •• Fe-28A1-5Cr FeCrA1 alloys leads to an even greater susceptibility to
Aluminized corrosion in the present mixed gas atmospheres than
.z • ~ Fe-28AI-5Cr
any of the iron aluminides.
~: .1 • ~ , . ~,.,.~w~" m'~'~'~F~-28AI An earlier study had shown that the deleterious
effects of these chromium levels on the corrosion
•D
.0! -I~D i n 8 i I i
resistance of the iron aluminide in high temperature
0 50 100 150 200 H E / H 2 S / H 2 0 mixtures could be offset by raising the
concentration of aluminum [6]. In the present study, the
Time (h) corrosion of aluminized F%A1-5Cr in the oxidizing/
Fig. 4. Weight change vs. exposure time for iron aluminides and sulfidizing environment was examined as an alternative
FeCrAI exposed to an oxidizing/sulfidizing mixed gas at 800 °C
(Po~ = 10 -22 a t • , Ps2 = 10-6 at•). Concentrations are in at.%.
to raising the bulk aluminum concentration of the
(The Fe-28AI-5Cr contained small amounts of zirconium and aluminide. It was found that the presence of an
boron.) Note logarithmic ordinate. Fe-50AI surface layer on the F%AI-5Cr reduced the
weight gains to the low level measured for the pure
binary Fe-28AI (Fig. 4). As als0 shown in Fig. 4, pre-
oxidation of the uncoated 5Cr alloy also improved the
thin and difficult to image and characterize. Examina- resistance to sulfidation. However, examination of an
tion of polished cross sections of selected iron alumi- exposed specimen surface showed some breakdown of
nides oxidized at 900 °C and air-cooled revealed thal the oxide layer and the formation of chromium sulfides
only a fraction of an exposed surface retained pieces ot at these locations (see Fig. 5).
a scale. An example is shown in Fig. 3. EDX analysis of
the thin, detached strip revealed only aluminum, which
is consistent with the formation of an AI203 ,scale 4. Discussion
formed during high temperature exposure. (The X-ray
system was not capable of detecting the oxygen in the The protection of metallic materials exposed to high
scale.) temperature oxidizing environments critically depends
on the formation of a slowly growing, stable, oxide
3.2. Oxidation/sulfidation scale. Alumina scales often provide such protection,
Earlier work has demonstrated that Fe3A1 alloys and the observed low weight gains and oxidation rates
containing more than 2Cr gained considerably more of the iron aluminides are consistent with the growth
weight upon exposure to aggressive oxidizing/sulfidiz- kinetics of a surface A I 2 0 3 product. The present data
ing environments than alloys with lower concentrations agree very well with weight gains and rate constants
of chromium [6]. Weight gains increased with increas- measured for Fe-22AI [7] and Fe-(0-10)Ni-(> 19)A1
ing chromium concentration above 2%. As shown in alloys [8, 9]. The observation of pure alumina scales (as
576 P. F. Tortorelli, J. H. De Van / Behavior of iron aluminides

noted by EDX analysis, see above) is in agreement with the beneficial effect of chromium additions on room
the findings of other investigators [7, 9], who showed temperature ductility of iron aluminides [11] may relate
that AI203 is the predominant surface oxide (with little to an oxidation effect during high temperature treat-
of no oxides of iron) on alloys with aluminum con- ment such that the scale that forms then serves as a
centrations greater than about 15 at.%. While the better barrier to hydrogen uptake by the alloy at
present scales were too thin to analyze by conventional ambient conditions [12].
X-ray diffraction, the work of Sakiyama et al. [8] Because FeCrA1 is also an alumina-former under the
suggests that, at least at 800 °C, they were y-alumina. oxidation conditions of this study, it is not unexpected
At higher temperatures, a-Al203 should be expected, that it shows comparable oxidation rates to that of
but it is unclear whether such has formed on the Fe3AI Fe3A1 (see Table 1). However, the ability of FeCrAI
under the exposure conditions of this study. Hagel alloys (containing about 7-12 at.% AI) to form AI203
reported both 7- and a-alumina formed on oxidized critically depends on the presence of substantial
Fe-22AI at 950°C, but observed only y-AI203 at amounts of chromium, which promotes the growth of
900°C [7]. While the present kinetic curves for a this surface product [7, 13]. In the case of Fe3AIalloys,
particular temperature show no clear evidence of a the aluminum levels of the aluminide are well in excess
transition from one alumina form to another [10], the of the critical concentration needed for external A120 a
fact that the long-term parabolic rate constants for formation, even in the absence of chromium.
Fe-28AI are essentially the same at both exposure For high temperature corrosion protection, it is not
temperatures may indicate that the slower growing only necessary that a stable AI203 surface product
a-A1203 forms on the iron aluminide at 900°C. form, but also that the scale remains attached to the
Furthermore, rate constants on the order of those metallic substrate. As can be seen from Fig. 3, as well
measured here, 10 -7 g2 m-4 s (10 -5 mg 2 cm -4 h), as the prior work of DeVan [6], the scales on Fe3A1and
appear to be closer to those measured for the growth its alloys tend to spall during cooling from the exposure
of a-Al203 than for less stable forms of alumina temperatures, which would seem to bode ill for the
[10]. Since a-Al203 is the most chemically stable oxide corrosion behavior of these iron aluminides under
form (neglecting surface energy effects), continued thermal cycling conditions. However, as also shown by
growth of another alumina may lead to its transforma- DeVan [6, 14], the repeated loss and regrowth of scale
tion to the a-phase. Thus, the relatively long exposure when an iron aluminide was thermally cycled in an
times of the present work, compared with previous oxidizing/sulfidizing environment did not lead to sig-
studies (10-25 h) [7, 8], could promote the formation nificant changes in the chemistry of the preoxidized
of a-alumina, particularly at 900°C. Diffraction surface nor to significant increases in the overall corro-
analysis of these thin scales, though difficult because of sion loss relative to exposure under isothermal con-
the small volumes of oxide, is obviously needed to ditions. It appears that the ability of Fe3A1 to rapidly
clearly determine the structural form of the surface form surface alumina in this temperature range (900 °C
product. However, regardless of which alumina is or lower), particularly after initial oxides have spalled
present at the respective temperatures, the scale [14], is sufficient to limit damage caused by any loss of
affords protection of the underlying metallic material. scale during thermal cycling. Furthermore, the high
The above results show that while chromium in Fe3AI aluminum concentration of the Fe3AI assures that there
(up to 10 at.%) does not exert a large influence on the is a sufficient surface activity of this element to form
oxidation of these aluminides, there is a definite effect new alumina despite preferential removal by repeated
at concentrations of 4% and above. At 800 °C, this was selective oxidation. The rapid growth of alumina has
manifested by faster oxygen uptake during the initial also been observed by Sakiyama et al. [8] for
stages of exposure and overall higher parabolic rate Fe-Ni-( > 14)A1 alloys exposed at 800 °C. As discussed
constants (Fig. 1). At 900 °C, there was no apparent above, the influence of chromium in promoting faster
difference in rate constants for the Fe3AI with various initial oxygen uptake by the aluminide may further
levels of chromium, but the amount of mass gain during enhance this phenomenon.
the initial stages of oxidation was directly proportional The excellent corrosion resistance of Fe3AI in the
to the chromium concentration (Fig. 2). Therefore, it very aggressive oxidizing/sulfidizing atmosphere used
appears that the presence of this element in significant in this and other iron aluminide studies [6, 14] derives
amounts leads to more rapid formation of an AI203 directly from the rapid growth of, and protection
surface product. Under certain circumstances, this can imparted by, the alumina scale despite the very low
actually be helpful in more quickly establishing an oxygen partial pressure in the mixed gas (10 -22 atm).
external protective oxide layer prior to exposure to While chromium may be beneficial to oxidation in
more aggressive or deleterious (hydrogen-containing) certain cases (see above), its tendency to form sulfides
environments. Indeed, it has been hypothesized that in the mixed gas environment degrades the corrosion
 P. F. Tortorelli, J. H. De Van / Behavior of iron aluminides 577

resistance of Fe3Al when its concentration in the alumi- but the rates were still substantially less than FeCrAl.
nide exceeds 2 at.% [6]. As shown in Fig. 4, preoxida- Preoxidation reduced the corrosion of Fe3AI-5%Cr,
tion of the Fe-28A1-5Cr decreased the weight gain but aluminizing this alloy to form an FeAI coating
upon subsequent exposure to the oxidizing/sulfidizing appeared to be the best way to assure long-term pro-
gas. (This has also been shown for FeCrAI under less tection in sulfidizing environments.
aggressive conditions [15].) However, the protection
starts to locally break down due to the formation of
chromium sulfides (Fig. 5). It therefore appears more
Acknowledgment
prudent to use a thick aluminized coating of Fe-A1 to
protect the underlying Fe-28A1-5Cr rather than to
This research was sponsored by the US Department
rely on the thin AI20 3 scale. As also shown in Fig. 4,
of Energy Fossil Energy A R & T D Materials Program
the coated specimen exhibited weight gain almost as
under contract DE-AC05-84OR21400 with Martin
low as those associated with the F e - 2 8 A I ( - 2 C r )
Marietta Energy Systems, Inc.
alloys.
The successful use of any coating for improved
corrosion behavior depends on some resistance of the
substrate. To this extent, Fe3AI-Cr has sufficiently References
lower reaction rates for oxidizing/sulfidizing environ-
ments; it certainly would not fail catastrophically if a 1 P. Tomaszewicz and G. R. Wallwork. Rev. High Temp.
coating was breached in some way. The weight gains Mater., 4 (1978) 75.
measured for Fe-28A1-5Cr are substantially lower 2 C. G. McKamey, J. H. DeVan, P. F. Tortorelli and V. K.
than those for stainless steel [6] and FeCrAI (Fig. 4), Sikka, J. Mater. Res., 6 ( 1991 ) 1779.
where large concentrations of chromium are available 3 K. Natesan and J.-H. Park, in Fossil Energy Advanced
Research and Technology Development Materials Program
for sulfide formation. Therefore, if chromium is Semiannual Progress Report for Period Ending September 30,
needed as an alloying addition (for ductility [11]), Fe3A1 1990, ORNL/FMP-90/2, US Department of Energy, Oak
can be coated with an iron aluminide that doesn't Ridge, TN, December, 1990, p. 303.
contain this element (or any other that would rapidly 4 F. D. Geib and R. A. Rapp, in V. A. Ravi and T. S. Srivatsan
form sulfides). (eds.), Proc. Symp. Processing and Manufacturing of.
Advanced Materials for High Temperature Applications, The
Metallurgical Society,Warrendale, PA, 1991, to be published.
5 M. Howell, J. H. DeVan, P. F. Tortorelli and J. T. Hutton, A
5. Summary High Sensitivity Thermogravimetric Facility for Elevated
Temperature Gaseous Corrosion Studies, to be published as
an Oak Ridge National Laboratory ORNL/TM report,
Iron aluminides have good oxidation resistance 1992.
because they form external alumina scales that protect 6 J. H. DeVan, in T. Grobstein and J. Doychak (eds.),
the underlying alloy. Spallation of these scales occurs, Oxidation of High-Temperature Intermetallics, The Metallur-
but the effect of such is offset by the ability of these gical Society, Warrendale, PA, 1989, p. 107.
aluminides to quickly form an oxide layer. Weight gains 7 W.C. Hagel, Corrosion, 21 (1965) 316.
8 M. Sakiyama, R Tomaszewicz and G. R. Wallwork, Oxid.
and parabolic rate constants were small. Chromium Met., 13 (1979) 311.
additions to Fe3A1 resulted in higher initial rates of 9 P. Tomaszewicz and G. R. Wallwork, Oxid. Met., 19 (1983)
oxidation in dry air at 800 and 900°C, but only 165.
increased the overall parabolic rate constant at the 10 G.C. Rybicki and J. L. Smialek, Oxid. Met., 31 (1989) 275.
lower temperature. In general, oxidation rates of Fe3AI 11 C.G. McKamey, J. A, Horton and C. T, Liu, J. Mater. Res., 4
(1989) 1156.
alloys were comparable with those of FeCrAI. There 12 C.G. McKamey and C. T. Liu, Scr. Metall., 24 (1990) 2119.
was some evidence from the measured rate constants 13 G. H. Meier, In T. Grobstein and J. Doychak (eds.),
that suggested an a-alumina layer formed during Oxidation of High-Temperature Intermetallics, The Metallur-
extended air exposure at 900 °C. gical Society,Warrendale, PA, 1989, p. 1.
In a highly aggressive oxidizing/sulfidizing environ- 14 J. H. DeVan, in K. Natesan (ed.), Proc. Int. Conf. on Heat-
Resistant Materials, ASM International, Materials Park, OH,
ment, F%A1 alloys with up to 2 at.%Cr showed excel- 1991, p. 235.
lent resistance at 800 °C. Corrosion increased when the 15 P. S. Sidky and M. G. Hocking, Corrosion Sci., 29 (1989)
chromium concentration equaled or exceeded 5 at.%, 735.