- Email: [email protected]
Gaseous corrosion resistance of Fe–Al based alloys containing Cr additions
Materials Science and Engineering A 405 (2005) 102–110
Gaseous corrosion resistance of Fe–Al based alloys containing Cr additions Part II. Scale morphology J.R. Regina ∗ , J.N. DuPont, A.R. Marder Lehigh University, Materials Science and Engineering, 5 E. Packer Ave., Bethlehem, PA 18015, USA Received in revised form 12 May 2005; accepted 19 May 2005
Abstract Iron–aluminum based weld overlay claddings are currently being considered as corrosion resistant coatings for corrosion protection at 500 ◦ C in aggressive sulfidizing and oxidizing atmospheres. These alloys rely on the formation of a thin passive Al2 O3 layer for their corrosion protection. Once the alloy can no longer maintain the protective oxide scale, passive layer breakdown can occur, where non-protective nodules form at random locations along the surface of the sample. These nodules can continue to grow together and eventually overgrow the protective oxide rendering the sample no longer corrosion resistant. In this current study, 10 iron–aluminum based alloys containing chromium and titanium additions were exposed to three high-temperature corrosive environments at 500 ◦ C for 100 h. The passive layer breakdown phenomenon was observed in all three environments on several non-protective alloys. The extent of nodule growth was measured and used to identify corrosion resistant alloy compositions. It was found that aluminum and chromium additions help to form and maintain a protective oxide layer and critical alloying contents were required to prevent passive layer breakdown during 100 h of exposure. Critical alloying values necessary to prevent nodule formation were compared to critical alloying levels required to prevent rapid corrosion kinetics. While corrosion kinetics are helpful in determining protective alloy compositions, the corrosion scale morphology must be characterized to ensure that the alloy is completely protective during the exposure time. © 2005 Elsevier B.V. All rights reserved. Keywords: Corrosion; Iron–aluminum; Chromium; Oxidation; Sulfidation; Scale morphology
1. Introduction When describing high-temperature corrosion behavior, corrosion kinetics are most commonly used to express the corrosion rate and extent of corrosion during an exposure time [1–5]. Although corrosion kinetics are a good indicator as to whether or not an alloy is protective in a specified environment, these types of results may not be sensitive enough to accurately describe the corrosion resistance of an alloy. An even more accurate way to describe the corrosion behavior of an alloy is to observe the corrosion scale morphology after exposure. For iron–aluminum based alloys, resistance to high-temperature corrosion comes from the ability to from ∗
Corresponding author. Tel.: +1 610 758 5000; fax: +1 610 758 6407. E-mail address: [email protected] (J.R. Regina).
0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2005.05.085
and maintain a passive Al2 O3 layer, which acts as a barrier between the corrosive environment and the underlying alloy [6–11]. When the passive layer can no longer protect the alloy, faster growing corrosion products, such as sulfides and oxides, can form initially as small nodules and eventually completely overgrow the Al2 O3 layer [12–15]. The presence of nodules on the sample surface therefore indicates that the passive layer is not completely protective and if exposure continues, may lead to breakaway corrosion. 1.1. Passive layer breakdown The corrosion resistance of an alloy depends primarily on the ability to form and subsequently maintain a passive oxide layer. Therefore, an important phenomenon that can occur during high-temperature corrosion is breakdown of the
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
103
Fig. 1. Schematic of nodule formation and growth [16].
passive layer. This can be defined as the moment when the passive layer is no longer able to suppress the formation of unwanted corrosion nodules. Although the formation of nodules through a protective layer has been observed in previous studies [8,12,16–18], the mechanism behind nodule formation is not very well understood. The formation of nodules on iron–aluminum alloys during initial stages of oxidation was described in detail by Tomaszewicz and Wallwork and can be seen schematically in Fig. 1 [16]. This explanation holds only for oxidation and suggests that there has to be enough aluminum to initially form a continuous Al2 O3 scale. The process begins with the initial stages of oxidation, when alumina (Al2 O3 ) and iron oxides (FeO) can simultaneously nucleate at the surface of the sample (Fig. 1a). The alumina continues to grow until a uniform Al2 O3 layer almost completely covers the sample, while the iron oxide nodules remain at random sites and continue to grow into the alloy (Fig. 1b). The FeO nodules continue to grow outward from the alloy and form various iron-oxide species (Fig. 1c). At the nodule–alloy interface, the alumina reacts with the FeO to form the spinel FeAl2 O4 , while the iron continues to oxidize into Fe3 O4 and Fe2 O3 at the nodule–gas interface. Oxidation continues through the nodules and alumina becomes more stable at the nodule–alloy interface where the oxygen partial pressure is very low (Fig. 1d and e). Finally, the oxygen partial pressure can become too low at the nodule–alloy interface to form any iron-oxide compounds and a continuous layer of Al2 O3 can form at the base of the nodule to prevent any further nodule
growth (Fig. 1f). It is important to note that the formation of an Al2 O3 layer below the nodule can act as a diffusion barrier, thus preventing any further corrosion. Another proposed explanation for the formation of nodules was given by Banovic et al. [12]. This alternative explanation, shown in Fig. 2, considers nodules that form after a protective alumina layer has already completely formed on the alloy. The process begins with the continuous passive alumina layer covering the alloy, keeping reactive oxygen
Fig. 2. Schematic of passive layer breakdown and subsequent nodule growth [12].
104
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
or sulfur gases from reaching the metal (Fig. 2a). During further exposure, the passive Al2 O3 layer breaks down either by cracks forming through the protective layer due to scale stresses (Fig. 2b) or through fast diffusion pathways such as grain boundaries. Once the passive layer begins to break down, the reactive gases can travel through the cracks and come in direct contact with the alloy. If the alloy cannot reform the passive layer fast enough, iron-rich oxide or sulfides can rapidly form and grow both inward and outward form the metal substrate (Fig. 2c). Eventually, these nodules can meet up and overgrow the protective Al2 O3 layer forming a thick non-protective corrosion scale (Fig. 2d). In the current study, 10 iron–aluminum based alloys containing chromium and titanium additions were exposed to three high-temperature corrosion environments. Careful consideration was given to the sample surface where the scale morphology was observed. The type of corrosion products that formed were considered and the extent of corrosion was classified based on the amount of surface area covered by unwanted corrosion products. The purpose of this study was to find alloys that were protective in various corrosion environments and to characterize the passive layer and corrosion scale morphology.
2. Experimental procedure Cast alloys were used for corrosion testing because it was previously shown that the high temperature corrosion behavior of weld overlays could be explained by using cast alloys of equivalent composition [19]. Alloys contained 14.5 or 19 at.% Al, chromium levels ranged from 0 to 5 at.%, and two quaternary alloys contained both chromium and titanium additions. All alloy compositions can be seen in Table 1. Preparation of corrosion samples and the corrosion testing experimental procedures were outlined in detail in Section 1 of this paper. The three gas compositions used for this study can be seen in Table 2. The sulfur and oxygen partial pressures were calculated for the gases using the HSC chemistry computer program [20]. Selected exposed samples were cut Table 1 Alloy compositions used for corrosion testing Alloy designation
Fe
Al
Cr
Ti
Fe–14.5Al Fe–14.5Al–1Cr Fe–14.5Al–2Cr Fe–14.5Al–5Cr Fe–14.5Al–2Cr–1.5Ti
Bal. Bal. Bal. Bal. Bal.
14.2 14.3 14.5 14.8 14.9
– 1.0 2.1 5.0 2.2
– – – – 1.8
Fe–19Al Fe–19Al–1Cr Fe–19Al–2Cr Fe–19Al–5Cr Fe–19Al–2Cr–1.5Ti
Bal. Bal. Bal. Bal. Bal.
18.8 18.8 19.0 19.9 19.2
– 1.0 2.1 5.0 2.1
– – – – 1.7
All values are in atomic percent.
Table 2 Gas compositions used for corrosion testing (vol.%) Gas component
Sulfidizing gas
Mixed oxidizing/ sulfidizing gas
Oxidizing gas
O2 CO CO2 H2 H2 O H2 S SO2 N2
– 15 – 3 – 0.12 – Bal.
– 10 5 – 2 0.12 – Bal.
2 – 15 – 6 – 0.12 Bal.
Log Po2 Log Ps2
−28 −6
−19 −8
−2 −46
approximately 80% through and submersed into liquid N2 . The samples were then cracked and dropped into ethanol to obtain fractured surface images [12]. Fracturing the samples in this manner was done to observe the passive layer in cross-section and corrosion scale morphology along the sample surface. This technique allowed for scale morphology observations to be made in three dimensions, and therefore, gave more information than could be obtained by a polished cross-section. A JEOL 6300 Scanning Electron Microscope (SEM) was used to obtain surface images as well as fractured cross-sections. Samples were observed with a 17 mm working distance and accelerating voltages of 5 and 10 keV. Energy-Dispersive Spectroscopy (EDS) was used to identify corrosion products and was taken with an accelerating voltage of 15 keV. SEM surface images were used with an imaging program to obtain area fractions.
3. Results and discussion The formation of nodules and the breakdown of the passive layer were observed on the Fe–Al–Cr alloys tested in the current study. The Fe–19Al–2Cr sample exposed to the sulfidizing gas was fractured in liquid N2 and the surface can be seen in Fig. 3. The majority of the surface was covered by a thin (<1 m) uniform layer (Fig. 3a), but there were areas where initial stages of external corrosion were observed (Fig. 3b). These small external corrosion products could not be seen when observing the surface on the SEM, but were obvious when looking at the cross sections. Similar results were found for Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas, which can be seen in Fig. 4. It can be seen from these images that a thin passive layer was present over the entire sample surface, but large nodules could be seen protruding from the surface at several areas. When observing a non-protective sample (Fe–14.5Al– 2Cr) exposed to the oxidizing environment, the passive layer breakdown phenomenon can be seen. For this sample, a thin passive layer could be detected in areas where nodule growth did not completely cover the surface (Fig. 5a and b). At areas where the passive layer could no longer re-heal itself, spher-
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
105
Fig. 3. Fractured cross-section of Fe–19Al–2Cr exposed to the sulfidizing gas. The thin passive layer covered the entire sample (a), but external nodules (arrows) were sporadically observed across the surface (b).
Fig. 4. Fractured cross-section of Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas. The passive layer was observed over the entire crosssection (a), but large nodules were seen protruding from the surface as well (b).
ical iron-oxides began to form as individual nodules (Fig. 5b and c). Areas that were no longer protective consisted of clusters of nodules (Fig. 5d), which could eventually completely overgrow the sample. Fractured cross-sections revealed that a passive layer, less than 1 m thick, formed on the alloys exposed to the three different environments. Non-protective samples, such as the one shown in Fig. 5, still formed the passive layer but large areas of nodules were observed to overgrow this layer. Some alloys seemed to be protective according to the kinetic data [1], but after examination of the scales through electron microscopy, nodules were observed on the samples. This was evident from Fe–19Al–2Cr exposed to the sulfidizing environment and Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas. Each of these seemed to be protective after 100 h of exposure in their respective environments according to the kinetic data, but small areas where the passive layer broke down resulted in external nodule growth. From these images it was concluded
that kinetic data alone was not enough to determine the corrosion resistance of an alloy. The integrity of the passive layer needed to be observed as well in order to determine if nodule formation can occur and overgrow the protective scale after 100 h, as seen in Fig. 5d. Combining the kinetic results presented in Section 1 of this paper and observations of nodule growth on the corrosion sample surfaces, it was determined that the formation of corrosion nodules could have formed by either proposed mechanism depending on the alloy composition and exposure environment. For example, when considering the Fe–19Al–1Cr alloy exposed to the mixed oxidizing/sulfidizing gas, small nodules were observed on the alloy surface even though the kinetic results showed that no significant weight gain was observed during exposure. This could indicate that the nodules formed at the onset of exposure and a passive layer formed beneath the corrosion nodule, as proposed by Tomaszewicz and Wallwork [16].
106
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Fig. 5. Fe–14.5Al–2Cr exposed to the oxidizing environment. Note that the passive layer was observed to be less than 1 m thick (a and b) and the nodules that formed were several microns in diameter (c). The nodules were also observed to form at random sites and then grow together to cover large areas of the sample (d).
On the other hand, a sample, such as Fe–14.5Al–2Cr exposed to the oxidizing environment resulted in a significant amount of corrosion nodules on the sample surface coupled with high corrosion kinetics. Careful examination of the corrosion kinetic results showed that very low corrosion rates were observed for approximately 5 h before the corrosion kinetics significantly increased and remained relatively high for the remainder of the exposure time (Fig. 6). These results
may indicate that a passive layer originally formed during the onset of exposure and then subsequently broke down according to the model proposed by Banovic et al. [12]. Although the formation of nodules on these FeAlCr based alloys could not be solely attributed to one nodule formation model, it was inferred that at a given exposure time, the amount of surface area covered by nodules was a direct indication of the ability of the passive layer to remain protective. 3.1. Extent of nodule growth
Fig. 6. Enlarged kinetic results for Fe–14.5Al–2Cr alloy exposed to the oxidizing environment showing that there appeared to be a period of time where significant corrosion was suppressed before high corrosion rates dominated the corrosion behavior of the alloy.
The integrity of the scales was explored by observing whether the protective oxide scale was completely uniform or if breakdown of the passive layer took place during 100 h of exposure. The amount of surface area covered by nodules was measured by taking pictures of the sample surface using the SEM and then using an image analysis program to threshold the images and measure the area fraction covered by nodules. Typical corrosion products that formed on non-protective samples exposed to the sulfidizing environment can be seen in Fig. 7a. It can be seen from this figure that block-like corrosion products (arrow) were present and completely covered some non-protective samples, but scratch marks (arrow) from sample preparation could still be seen on protective samples (Fig. 7b). The block-like nodules were identified as ironsulfide compounds using EDS. Similar block-like scales have been observed on iron–aluminum alloys when exposed to sul-
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
107
Fig. 7. Typical scale morphology of non-protective samples (a) and protective samples (b) exposed to the sulfidizing gas for 100 h. Note the arrows indicating the block-like corrosion products (a) and the direction of the scratch marks still present on protective samples (b). Table 3 Alloying content factor (φ) for each binary and ternary alloy used in the study
Fig. 8. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the sulfidizing environment.
fidizing environments and were also identified as FeS scales [21–23]. Nodule area fraction measurements were plotted against the total aluminum and chromium content of each alloy, known as the alloying content (φ) [1], and can be seen in Fig. 8. The alloying content factor (φ) was found to be: φ = (at.% Al) + (at.% Cr)
(1)
Alloy designation
Al
Cr
φ = (at.% Al) + (at.% Cr)
Fe–14.5Al Fe–14.5Al–1Cr Fe–14.5Al–2Cr Fe–14.5Al–5Cr Fe–19Al Fe–19Al–1Cr Fe–19Al–2Cr Fe–19Al–5Cr
14.2 14.3 14.5 14.8 18.8 18.8 19.0 19.9
– 1.0 2.1 5.0 – 1.0 2.1 5.0
14.2 15.3 16.6 19.8 18.8 19.8 21.1 24.9
Values of φ for each alloy used in this study can be seen in Table 3. The alloying content (φ) factor was created to numerically represent each alloying element based on both the aluminum and chromium concentration. It can be seen from Fig. 8 that an alloying content factor of 16.5 was required to decrease nodule growth during 100 h of exposure to the sulfidizing gas, but approximately 25 was required to completely prevent nodules from forming in this environment. Typical scales that formed on non-protective samples exposed to the mixed oxidizing/sulfidizing environment can be seen in Fig. 9. It was observed that non-protective samples were completely covered by a block-like scale that cracked
Fig. 9. Typical scale morphology of non-protective samples (a) and protective samples (b) exposed to the mixed oxidizing/sulfidizing gas for 100 h.
108
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
and spalled when removed from the furnace. This blocklike scale was similar in appearance to the nodules observed on non-protective samples exposed to the sulfidizing environment and was again very similar to scales formed in previous studies [21,22,24]. This was confirmed by EDS, which showed that the block-like scales shown in Fig. 9a were also identified as iron-sulfide compounds. On the other hand, scratch marks from sample preparation could still be seen on protective samples after exposure (Fig. 9b). When considering the extent of surface area covered by the block-like scale (Fig. 10) it could be seen that a critical alloying content of approximately 21 was required to significantly reduce the amount of nodules that formed during 100 h of exposure to the mixed oxidizing/sulfidizing environment. Although this alloying content would reduce the amount of nodules that formed, an alloying content of 25 was required to completely prevent nodules from forming during 100 h of exposure. Two types of nodule morphologies were observed on samples exposed to the oxidizing environment. The first type of nodules to form was sphere-like in appearance and was observed to initially form as individual nodules and then continued to grow together to cover large areas (Fig. 5c and d). These nodules have been previously observed on oxidized iron–aluminum alloys and the outermost layer of these nodules were identified as Fe2 O3 [8,14,17]. The second type of corrosion morphology was plate-like in appearance and was observed primarily at locations where the sphere-like nod-
Fig. 10. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the mixed oxidizing/sulfidizing environment.
ules completely covered an area of the sample. Both types of corrosion products can be seen in Fig. 11. Because the plate-like corrosion products were only present at areas where the sphere-like nodules were located, it was inferred that the sphere-like nodules were the first type of corrosion products to form once the passive layer began to break down. The platelike corrosion product formed only after the passive layer was no longer protective and unwanted sphere-like corrosion products already covered the surface. Despite the type of cor-
Fig. 11. Different types of scale morphologies that were observed on samples exposed to the oxidizing environment. Non-protective samples initially formed sphere-like nodules (a), which grew together to cover the sample surface. After the sphere-like nodules covered the surface, the plate-like nodules formed (b and c). Again, scratch marks from grinding could still be seen on protective samples after 100 h of exposure (d).
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Fig. 12. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the oxidizing environment.
rosion scale morphology, both the sphere-like and plate-like scales were identified using EDS as iron-oxide compounds. When considering the amount of surface area covered by unwanted corrosion products, Fig. 12 shows that there again was a critical alloying content required to significantly reduce the amount of nodules present at a given exposure time. As was observed with alloys exposed to the mixed oxidizing/sulfidizing environment, an alloying content value of 25 was required to completely suppress nodule growth from occurring during 100 h of exposure. Comparing the corrosion kinetic results to microstructural observations, it can be seen that there is a discrepancy as to the amount of chromium required to obtain a protective coating. Although alloys containing 19 at.% Al showed relatively no weight gains after 100 h in the sulfidizing gas [1], the microstructure revealed that a small portion of the surface contained nodules. The formation of nodules can eventually lead to increased corrosion kinetics at longer exposure times. In the mixed oxidizing/sulfidizing gas, Fe–19at.% Al alloys containing at least 1% Cr seemed protective after 100 h according to the kinetic data [1], but nodules were found on the surface of alloys containing less than 5 at.% Cr. Although only a small number of nodules were present on these samples, at longer times further breakdown of the passive layer may occur, leading to rapid corrosion rates. In the oxidizing environment, alloys containing 19 at.% Al seemed to be protective despite the chromium concentration in these alloys according to the kinetic data [1], but microstructural observations revealed that Fe–19Al had small areas where external nodules could be seen. It was observed that 5 at.% Cr was needed to totally prevent external nodules from forming during 100 h of exposure. From these results it can be seen that additions of chromium are beneficial to the corrosion resistance of iron–aluminum alloys at 500 ◦ C. Critical alloying contents were found for alloys that prevented the formation of nodules during exposures (Fig. 13). Although the critical alloying content required to significantly reduce the amount of nodules
109
Fig. 13. Critical alloying contents required to prevent nodule growth during 100 h exposures in all three test environments.
varied between the three gases, an alloying content of approximately 25 was required to completely suppress the formation of nodules during 100 h of exposure. From this study it has been observed that kinetic data alone cannot be used to accurately describe the high-temperature corrosion behavior of these alloys. Microstructural analysis must also be performed to determine the integrity of the passive scale. Observance of only a few nodules even after 100 h of exposure could lead to unacceptable corrosion rates at longer times. The kinetic results are valuable as they can be used to find the possible candidates for protective coatings, but careful microscopy must also be employed to determine if the alloy will continue to be protective at times greater than the exposure time.
4. Conclusions Several alloys were exposed to three gaseous corrosive environments: a sulfidizing atmosphere, a mixed oxidizing/sulfidizing environment and an oxidizing environment. It was determined that: • Kinetic data can help to determine likely candidates for protective coatings, but these candidates must be scrutinized using microscopic analysis to determine if nodule formation is prevented as well. For a given exposure time, it was found that the amount of surface area covered by unwanted corrosion products could give a direct indication of the corrosion resistance of an alloy. Electron microscopy observations showed that samples that appeared passive from kinetic results could still be problematic at long exposure times due to the formation of nodules. • Aluminum and chromium additions were shown to help enhance formation of a passive layer, and also inhibit nodule growth in all three environments. An alloying content of approximately 25 (corresponding to the Fe–19Al–5Cr alloy) was required to completely suppress nodule growth during 100 h of exposure in all three environments.
110
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Acknowledgements This research was sponsored by the Fossil Energy Advanced Research and Technology Development (AR&TD) Materials Program, U.S. Department of Energy, under contract DE-AC05-96OR22464 with U.T. Battelle. The authors would like to thank V.K. Sikka, P.F. Tortorelli and B.A. Pint from ORNL for the cast alloys used in corrosion testing.
References [1] [2] [3] [4]
J.R. Regina, J.N. DuPont, A.R. Marder, Mater. Sci. Eng. A, in press. J.H. DeVan, P.F. Tortorelli, Mater. High Temp. 11 (1993) 30–35. J.H. DeVan, P.F. Tortorelli, Corros. Sci. 35 (1993) 1065–1071. P.F. Tortorelli, J.H. DeVan, in: J.H. Schneibel, M.A. Crimp (Eds.), Proc. Symp. Annu. Meet. Miner., Met. Mater. Soc., 1994, pp. 257–270. [5] S.W. Banovic, J.N. DuPont, A.R. Marder, Scr. Mater. 38 (1998) 1763–1767. [6] C.G. McKamey, J.H. DeVan, P.F. Tortorelli, V.K. Sikka, J. Mater. Res. 6 (1991) 1779–1805. [7] S.W. Banovic, J.N. DuPont, A.R. Marder, Metall. Mater. Trans. A 31A (2000) 1805–1817.
[8] W.E. Boggs, J. Electrochem. Soc. 118 (1971) 906–913. [9] P. Tomaszewicz, G.R. Wallwork, Rev. High Temp. Mater. 4 (1978) 75–105. [10] R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 233–254. [11] R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 73–87. [12] S.W. Banovic, J.N. DuPont, A.R. Marder, Oxid. Met. 54 (2000) 339–369. [13] C.H. Kao, C.M. Wan, J. Mater. Sci. 23 (1988) 1943–1947. [14] M. Sakiyama, P. Tomaszewicz, G.R. Wallwork, Oxid. Met. 13 (1979) 311–330. [15] F. Saegusa, L. Lee, Corrosion 22 (1966) 168–177. [16] P. Tomaszewicz, G.R. Wallwork, Oxid. Met. 19 (1983) 165–185. [17] P.J. Smith, R.M. Beauprie, W.W. Smeltzer, D.V. Stevanovic, D.A. Thompson, Oxid. Met. 28 (1987) 259–276. [18] P.C. Patnaik, W.W. Smeltzer, Oxid. Met. 23 (1985) 53–75. [19] P.F. Tortorelli, G.M. Goodwin, M. Howell, J.H. DeVan, in: K. Natesan, G. Krishnamurti, G.Y. Lai (Eds.), Proc. Int. Conf. Heat-Resistant Mater. II, 1995, pp. 585–590. [20] HSC Chemistry for Windows, Version 4.0. Oy, Finland, Outokupmu Research, 1997. [21] S.W. Banovic, J.N. DuPont, A.R. Marder, Proc. Corrosion 99, San Antonio, TX, Pub. NACE, 1999, p. 12. [22] P.J. Smith, P.R.S. Jackson, W.W. Smeltzer, J. Electrochem. Soc. 134 (1987) 1424–1431. [23] W. Kai, J.P. Chu, R.T. Huang, P.Y. Lee, Mater. Sci. Eng. A A239–A240 (1997) 859–870. [24] K. Przybylski, T. Narita, W.W. Smeltzer, Oxid. Met. 38 (1992) 1–32.
Gaseous corrosion resistance of Fe–Al based alloys containing Cr additions Part II. Scale morphology J.R. Regina ∗ , J.N. DuPont, A.R. Marder Lehigh University, Materials Science and Engineering, 5 E. Packer Ave., Bethlehem, PA 18015, USA Received in revised form 12 May 2005; accepted 19 May 2005
Abstract Iron–aluminum based weld overlay claddings are currently being considered as corrosion resistant coatings for corrosion protection at 500 ◦ C in aggressive sulfidizing and oxidizing atmospheres. These alloys rely on the formation of a thin passive Al2 O3 layer for their corrosion protection. Once the alloy can no longer maintain the protective oxide scale, passive layer breakdown can occur, where non-protective nodules form at random locations along the surface of the sample. These nodules can continue to grow together and eventually overgrow the protective oxide rendering the sample no longer corrosion resistant. In this current study, 10 iron–aluminum based alloys containing chromium and titanium additions were exposed to three high-temperature corrosive environments at 500 ◦ C for 100 h. The passive layer breakdown phenomenon was observed in all three environments on several non-protective alloys. The extent of nodule growth was measured and used to identify corrosion resistant alloy compositions. It was found that aluminum and chromium additions help to form and maintain a protective oxide layer and critical alloying contents were required to prevent passive layer breakdown during 100 h of exposure. Critical alloying values necessary to prevent nodule formation were compared to critical alloying levels required to prevent rapid corrosion kinetics. While corrosion kinetics are helpful in determining protective alloy compositions, the corrosion scale morphology must be characterized to ensure that the alloy is completely protective during the exposure time. © 2005 Elsevier B.V. All rights reserved. Keywords: Corrosion; Iron–aluminum; Chromium; Oxidation; Sulfidation; Scale morphology
1. Introduction When describing high-temperature corrosion behavior, corrosion kinetics are most commonly used to express the corrosion rate and extent of corrosion during an exposure time [1–5]. Although corrosion kinetics are a good indicator as to whether or not an alloy is protective in a specified environment, these types of results may not be sensitive enough to accurately describe the corrosion resistance of an alloy. An even more accurate way to describe the corrosion behavior of an alloy is to observe the corrosion scale morphology after exposure. For iron–aluminum based alloys, resistance to high-temperature corrosion comes from the ability to from ∗
Corresponding author. Tel.: +1 610 758 5000; fax: +1 610 758 6407. E-mail address: [email protected] (J.R. Regina).
0921-5093/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2005.05.085
and maintain a passive Al2 O3 layer, which acts as a barrier between the corrosive environment and the underlying alloy [6–11]. When the passive layer can no longer protect the alloy, faster growing corrosion products, such as sulfides and oxides, can form initially as small nodules and eventually completely overgrow the Al2 O3 layer [12–15]. The presence of nodules on the sample surface therefore indicates that the passive layer is not completely protective and if exposure continues, may lead to breakaway corrosion. 1.1. Passive layer breakdown The corrosion resistance of an alloy depends primarily on the ability to form and subsequently maintain a passive oxide layer. Therefore, an important phenomenon that can occur during high-temperature corrosion is breakdown of the
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
103
Fig. 1. Schematic of nodule formation and growth [16].
passive layer. This can be defined as the moment when the passive layer is no longer able to suppress the formation of unwanted corrosion nodules. Although the formation of nodules through a protective layer has been observed in previous studies [8,12,16–18], the mechanism behind nodule formation is not very well understood. The formation of nodules on iron–aluminum alloys during initial stages of oxidation was described in detail by Tomaszewicz and Wallwork and can be seen schematically in Fig. 1 [16]. This explanation holds only for oxidation and suggests that there has to be enough aluminum to initially form a continuous Al2 O3 scale. The process begins with the initial stages of oxidation, when alumina (Al2 O3 ) and iron oxides (FeO) can simultaneously nucleate at the surface of the sample (Fig. 1a). The alumina continues to grow until a uniform Al2 O3 layer almost completely covers the sample, while the iron oxide nodules remain at random sites and continue to grow into the alloy (Fig. 1b). The FeO nodules continue to grow outward from the alloy and form various iron-oxide species (Fig. 1c). At the nodule–alloy interface, the alumina reacts with the FeO to form the spinel FeAl2 O4 , while the iron continues to oxidize into Fe3 O4 and Fe2 O3 at the nodule–gas interface. Oxidation continues through the nodules and alumina becomes more stable at the nodule–alloy interface where the oxygen partial pressure is very low (Fig. 1d and e). Finally, the oxygen partial pressure can become too low at the nodule–alloy interface to form any iron-oxide compounds and a continuous layer of Al2 O3 can form at the base of the nodule to prevent any further nodule
growth (Fig. 1f). It is important to note that the formation of an Al2 O3 layer below the nodule can act as a diffusion barrier, thus preventing any further corrosion. Another proposed explanation for the formation of nodules was given by Banovic et al. [12]. This alternative explanation, shown in Fig. 2, considers nodules that form after a protective alumina layer has already completely formed on the alloy. The process begins with the continuous passive alumina layer covering the alloy, keeping reactive oxygen
Fig. 2. Schematic of passive layer breakdown and subsequent nodule growth [12].
104
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
or sulfur gases from reaching the metal (Fig. 2a). During further exposure, the passive Al2 O3 layer breaks down either by cracks forming through the protective layer due to scale stresses (Fig. 2b) or through fast diffusion pathways such as grain boundaries. Once the passive layer begins to break down, the reactive gases can travel through the cracks and come in direct contact with the alloy. If the alloy cannot reform the passive layer fast enough, iron-rich oxide or sulfides can rapidly form and grow both inward and outward form the metal substrate (Fig. 2c). Eventually, these nodules can meet up and overgrow the protective Al2 O3 layer forming a thick non-protective corrosion scale (Fig. 2d). In the current study, 10 iron–aluminum based alloys containing chromium and titanium additions were exposed to three high-temperature corrosion environments. Careful consideration was given to the sample surface where the scale morphology was observed. The type of corrosion products that formed were considered and the extent of corrosion was classified based on the amount of surface area covered by unwanted corrosion products. The purpose of this study was to find alloys that were protective in various corrosion environments and to characterize the passive layer and corrosion scale morphology.
2. Experimental procedure Cast alloys were used for corrosion testing because it was previously shown that the high temperature corrosion behavior of weld overlays could be explained by using cast alloys of equivalent composition [19]. Alloys contained 14.5 or 19 at.% Al, chromium levels ranged from 0 to 5 at.%, and two quaternary alloys contained both chromium and titanium additions. All alloy compositions can be seen in Table 1. Preparation of corrosion samples and the corrosion testing experimental procedures were outlined in detail in Section 1 of this paper. The three gas compositions used for this study can be seen in Table 2. The sulfur and oxygen partial pressures were calculated for the gases using the HSC chemistry computer program [20]. Selected exposed samples were cut Table 1 Alloy compositions used for corrosion testing Alloy designation
Fe
Al
Cr
Ti
Fe–14.5Al Fe–14.5Al–1Cr Fe–14.5Al–2Cr Fe–14.5Al–5Cr Fe–14.5Al–2Cr–1.5Ti
Bal. Bal. Bal. Bal. Bal.
14.2 14.3 14.5 14.8 14.9
– 1.0 2.1 5.0 2.2
– – – – 1.8
Fe–19Al Fe–19Al–1Cr Fe–19Al–2Cr Fe–19Al–5Cr Fe–19Al–2Cr–1.5Ti
Bal. Bal. Bal. Bal. Bal.
18.8 18.8 19.0 19.9 19.2
– 1.0 2.1 5.0 2.1
– – – – 1.7
All values are in atomic percent.
Table 2 Gas compositions used for corrosion testing (vol.%) Gas component
Sulfidizing gas
Mixed oxidizing/ sulfidizing gas
Oxidizing gas
O2 CO CO2 H2 H2 O H2 S SO2 N2
– 15 – 3 – 0.12 – Bal.
– 10 5 – 2 0.12 – Bal.
2 – 15 – 6 – 0.12 Bal.
Log Po2 Log Ps2
−28 −6
−19 −8
−2 −46
approximately 80% through and submersed into liquid N2 . The samples were then cracked and dropped into ethanol to obtain fractured surface images [12]. Fracturing the samples in this manner was done to observe the passive layer in cross-section and corrosion scale morphology along the sample surface. This technique allowed for scale morphology observations to be made in three dimensions, and therefore, gave more information than could be obtained by a polished cross-section. A JEOL 6300 Scanning Electron Microscope (SEM) was used to obtain surface images as well as fractured cross-sections. Samples were observed with a 17 mm working distance and accelerating voltages of 5 and 10 keV. Energy-Dispersive Spectroscopy (EDS) was used to identify corrosion products and was taken with an accelerating voltage of 15 keV. SEM surface images were used with an imaging program to obtain area fractions.
3. Results and discussion The formation of nodules and the breakdown of the passive layer were observed on the Fe–Al–Cr alloys tested in the current study. The Fe–19Al–2Cr sample exposed to the sulfidizing gas was fractured in liquid N2 and the surface can be seen in Fig. 3. The majority of the surface was covered by a thin (<1 m) uniform layer (Fig. 3a), but there were areas where initial stages of external corrosion were observed (Fig. 3b). These small external corrosion products could not be seen when observing the surface on the SEM, but were obvious when looking at the cross sections. Similar results were found for Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas, which can be seen in Fig. 4. It can be seen from these images that a thin passive layer was present over the entire sample surface, but large nodules could be seen protruding from the surface at several areas. When observing a non-protective sample (Fe–14.5Al– 2Cr) exposed to the oxidizing environment, the passive layer breakdown phenomenon can be seen. For this sample, a thin passive layer could be detected in areas where nodule growth did not completely cover the surface (Fig. 5a and b). At areas where the passive layer could no longer re-heal itself, spher-
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
105
Fig. 3. Fractured cross-section of Fe–19Al–2Cr exposed to the sulfidizing gas. The thin passive layer covered the entire sample (a), but external nodules (arrows) were sporadically observed across the surface (b).
Fig. 4. Fractured cross-section of Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas. The passive layer was observed over the entire crosssection (a), but large nodules were seen protruding from the surface as well (b).
ical iron-oxides began to form as individual nodules (Fig. 5b and c). Areas that were no longer protective consisted of clusters of nodules (Fig. 5d), which could eventually completely overgrow the sample. Fractured cross-sections revealed that a passive layer, less than 1 m thick, formed on the alloys exposed to the three different environments. Non-protective samples, such as the one shown in Fig. 5, still formed the passive layer but large areas of nodules were observed to overgrow this layer. Some alloys seemed to be protective according to the kinetic data [1], but after examination of the scales through electron microscopy, nodules were observed on the samples. This was evident from Fe–19Al–2Cr exposed to the sulfidizing environment and Fe–19Al–1Cr exposed to the mixed oxidizing/sulfidizing gas. Each of these seemed to be protective after 100 h of exposure in their respective environments according to the kinetic data, but small areas where the passive layer broke down resulted in external nodule growth. From these images it was concluded
that kinetic data alone was not enough to determine the corrosion resistance of an alloy. The integrity of the passive layer needed to be observed as well in order to determine if nodule formation can occur and overgrow the protective scale after 100 h, as seen in Fig. 5d. Combining the kinetic results presented in Section 1 of this paper and observations of nodule growth on the corrosion sample surfaces, it was determined that the formation of corrosion nodules could have formed by either proposed mechanism depending on the alloy composition and exposure environment. For example, when considering the Fe–19Al–1Cr alloy exposed to the mixed oxidizing/sulfidizing gas, small nodules were observed on the alloy surface even though the kinetic results showed that no significant weight gain was observed during exposure. This could indicate that the nodules formed at the onset of exposure and a passive layer formed beneath the corrosion nodule, as proposed by Tomaszewicz and Wallwork [16].
106
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Fig. 5. Fe–14.5Al–2Cr exposed to the oxidizing environment. Note that the passive layer was observed to be less than 1 m thick (a and b) and the nodules that formed were several microns in diameter (c). The nodules were also observed to form at random sites and then grow together to cover large areas of the sample (d).
On the other hand, a sample, such as Fe–14.5Al–2Cr exposed to the oxidizing environment resulted in a significant amount of corrosion nodules on the sample surface coupled with high corrosion kinetics. Careful examination of the corrosion kinetic results showed that very low corrosion rates were observed for approximately 5 h before the corrosion kinetics significantly increased and remained relatively high for the remainder of the exposure time (Fig. 6). These results
may indicate that a passive layer originally formed during the onset of exposure and then subsequently broke down according to the model proposed by Banovic et al. [12]. Although the formation of nodules on these FeAlCr based alloys could not be solely attributed to one nodule formation model, it was inferred that at a given exposure time, the amount of surface area covered by nodules was a direct indication of the ability of the passive layer to remain protective. 3.1. Extent of nodule growth
Fig. 6. Enlarged kinetic results for Fe–14.5Al–2Cr alloy exposed to the oxidizing environment showing that there appeared to be a period of time where significant corrosion was suppressed before high corrosion rates dominated the corrosion behavior of the alloy.
The integrity of the scales was explored by observing whether the protective oxide scale was completely uniform or if breakdown of the passive layer took place during 100 h of exposure. The amount of surface area covered by nodules was measured by taking pictures of the sample surface using the SEM and then using an image analysis program to threshold the images and measure the area fraction covered by nodules. Typical corrosion products that formed on non-protective samples exposed to the sulfidizing environment can be seen in Fig. 7a. It can be seen from this figure that block-like corrosion products (arrow) were present and completely covered some non-protective samples, but scratch marks (arrow) from sample preparation could still be seen on protective samples (Fig. 7b). The block-like nodules were identified as ironsulfide compounds using EDS. Similar block-like scales have been observed on iron–aluminum alloys when exposed to sul-
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
107
Fig. 7. Typical scale morphology of non-protective samples (a) and protective samples (b) exposed to the sulfidizing gas for 100 h. Note the arrows indicating the block-like corrosion products (a) and the direction of the scratch marks still present on protective samples (b). Table 3 Alloying content factor (φ) for each binary and ternary alloy used in the study
Fig. 8. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the sulfidizing environment.
fidizing environments and were also identified as FeS scales [21–23]. Nodule area fraction measurements were plotted against the total aluminum and chromium content of each alloy, known as the alloying content (φ) [1], and can be seen in Fig. 8. The alloying content factor (φ) was found to be: φ = (at.% Al) + (at.% Cr)
(1)
Alloy designation
Al
Cr
φ = (at.% Al) + (at.% Cr)
Fe–14.5Al Fe–14.5Al–1Cr Fe–14.5Al–2Cr Fe–14.5Al–5Cr Fe–19Al Fe–19Al–1Cr Fe–19Al–2Cr Fe–19Al–5Cr
14.2 14.3 14.5 14.8 18.8 18.8 19.0 19.9
– 1.0 2.1 5.0 – 1.0 2.1 5.0
14.2 15.3 16.6 19.8 18.8 19.8 21.1 24.9
Values of φ for each alloy used in this study can be seen in Table 3. The alloying content (φ) factor was created to numerically represent each alloying element based on both the aluminum and chromium concentration. It can be seen from Fig. 8 that an alloying content factor of 16.5 was required to decrease nodule growth during 100 h of exposure to the sulfidizing gas, but approximately 25 was required to completely prevent nodules from forming in this environment. Typical scales that formed on non-protective samples exposed to the mixed oxidizing/sulfidizing environment can be seen in Fig. 9. It was observed that non-protective samples were completely covered by a block-like scale that cracked
Fig. 9. Typical scale morphology of non-protective samples (a) and protective samples (b) exposed to the mixed oxidizing/sulfidizing gas for 100 h.
108
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
and spalled when removed from the furnace. This blocklike scale was similar in appearance to the nodules observed on non-protective samples exposed to the sulfidizing environment and was again very similar to scales formed in previous studies [21,22,24]. This was confirmed by EDS, which showed that the block-like scales shown in Fig. 9a were also identified as iron-sulfide compounds. On the other hand, scratch marks from sample preparation could still be seen on protective samples after exposure (Fig. 9b). When considering the extent of surface area covered by the block-like scale (Fig. 10) it could be seen that a critical alloying content of approximately 21 was required to significantly reduce the amount of nodules that formed during 100 h of exposure to the mixed oxidizing/sulfidizing environment. Although this alloying content would reduce the amount of nodules that formed, an alloying content of 25 was required to completely prevent nodules from forming during 100 h of exposure. Two types of nodule morphologies were observed on samples exposed to the oxidizing environment. The first type of nodules to form was sphere-like in appearance and was observed to initially form as individual nodules and then continued to grow together to cover large areas (Fig. 5c and d). These nodules have been previously observed on oxidized iron–aluminum alloys and the outermost layer of these nodules were identified as Fe2 O3 [8,14,17]. The second type of corrosion morphology was plate-like in appearance and was observed primarily at locations where the sphere-like nod-
Fig. 10. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the mixed oxidizing/sulfidizing environment.
ules completely covered an area of the sample. Both types of corrosion products can be seen in Fig. 11. Because the plate-like corrosion products were only present at areas where the sphere-like nodules were located, it was inferred that the sphere-like nodules were the first type of corrosion products to form once the passive layer began to break down. The platelike corrosion product formed only after the passive layer was no longer protective and unwanted sphere-like corrosion products already covered the surface. Despite the type of cor-
Fig. 11. Different types of scale morphologies that were observed on samples exposed to the oxidizing environment. Non-protective samples initially formed sphere-like nodules (a), which grew together to cover the sample surface. After the sphere-like nodules covered the surface, the plate-like nodules formed (b and c). Again, scratch marks from grinding could still be seen on protective samples after 100 h of exposure (d).
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Fig. 12. Aluminum and chromium effect on the amount of surface area covered by unwanted nodule corrosion products for alloys exposed to the oxidizing environment.
rosion scale morphology, both the sphere-like and plate-like scales were identified using EDS as iron-oxide compounds. When considering the amount of surface area covered by unwanted corrosion products, Fig. 12 shows that there again was a critical alloying content required to significantly reduce the amount of nodules present at a given exposure time. As was observed with alloys exposed to the mixed oxidizing/sulfidizing environment, an alloying content value of 25 was required to completely suppress nodule growth from occurring during 100 h of exposure. Comparing the corrosion kinetic results to microstructural observations, it can be seen that there is a discrepancy as to the amount of chromium required to obtain a protective coating. Although alloys containing 19 at.% Al showed relatively no weight gains after 100 h in the sulfidizing gas [1], the microstructure revealed that a small portion of the surface contained nodules. The formation of nodules can eventually lead to increased corrosion kinetics at longer exposure times. In the mixed oxidizing/sulfidizing gas, Fe–19at.% Al alloys containing at least 1% Cr seemed protective after 100 h according to the kinetic data [1], but nodules were found on the surface of alloys containing less than 5 at.% Cr. Although only a small number of nodules were present on these samples, at longer times further breakdown of the passive layer may occur, leading to rapid corrosion rates. In the oxidizing environment, alloys containing 19 at.% Al seemed to be protective despite the chromium concentration in these alloys according to the kinetic data [1], but microstructural observations revealed that Fe–19Al had small areas where external nodules could be seen. It was observed that 5 at.% Cr was needed to totally prevent external nodules from forming during 100 h of exposure. From these results it can be seen that additions of chromium are beneficial to the corrosion resistance of iron–aluminum alloys at 500 ◦ C. Critical alloying contents were found for alloys that prevented the formation of nodules during exposures (Fig. 13). Although the critical alloying content required to significantly reduce the amount of nodules
109
Fig. 13. Critical alloying contents required to prevent nodule growth during 100 h exposures in all three test environments.
varied between the three gases, an alloying content of approximately 25 was required to completely suppress the formation of nodules during 100 h of exposure. From this study it has been observed that kinetic data alone cannot be used to accurately describe the high-temperature corrosion behavior of these alloys. Microstructural analysis must also be performed to determine the integrity of the passive scale. Observance of only a few nodules even after 100 h of exposure could lead to unacceptable corrosion rates at longer times. The kinetic results are valuable as they can be used to find the possible candidates for protective coatings, but careful microscopy must also be employed to determine if the alloy will continue to be protective at times greater than the exposure time.
4. Conclusions Several alloys were exposed to three gaseous corrosive environments: a sulfidizing atmosphere, a mixed oxidizing/sulfidizing environment and an oxidizing environment. It was determined that: • Kinetic data can help to determine likely candidates for protective coatings, but these candidates must be scrutinized using microscopic analysis to determine if nodule formation is prevented as well. For a given exposure time, it was found that the amount of surface area covered by unwanted corrosion products could give a direct indication of the corrosion resistance of an alloy. Electron microscopy observations showed that samples that appeared passive from kinetic results could still be problematic at long exposure times due to the formation of nodules. • Aluminum and chromium additions were shown to help enhance formation of a passive layer, and also inhibit nodule growth in all three environments. An alloying content of approximately 25 (corresponding to the Fe–19Al–5Cr alloy) was required to completely suppress nodule growth during 100 h of exposure in all three environments.
110
J.R. Regina et al. / Materials Science and Engineering A 405 (2005) 102–110
Acknowledgements This research was sponsored by the Fossil Energy Advanced Research and Technology Development (AR&TD) Materials Program, U.S. Department of Energy, under contract DE-AC05-96OR22464 with U.T. Battelle. The authors would like to thank V.K. Sikka, P.F. Tortorelli and B.A. Pint from ORNL for the cast alloys used in corrosion testing.
References [1] [2] [3] [4]
J.R. Regina, J.N. DuPont, A.R. Marder, Mater. Sci. Eng. A, in press. J.H. DeVan, P.F. Tortorelli, Mater. High Temp. 11 (1993) 30–35. J.H. DeVan, P.F. Tortorelli, Corros. Sci. 35 (1993) 1065–1071. P.F. Tortorelli, J.H. DeVan, in: J.H. Schneibel, M.A. Crimp (Eds.), Proc. Symp. Annu. Meet. Miner., Met. Mater. Soc., 1994, pp. 257–270. [5] S.W. Banovic, J.N. DuPont, A.R. Marder, Scr. Mater. 38 (1998) 1763–1767. [6] C.G. McKamey, J.H. DeVan, P.F. Tortorelli, V.K. Sikka, J. Mater. Res. 6 (1991) 1779–1805. [7] S.W. Banovic, J.N. DuPont, A.R. Marder, Metall. Mater. Trans. A 31A (2000) 1805–1817.
[8] W.E. Boggs, J. Electrochem. Soc. 118 (1971) 906–913. [9] P. Tomaszewicz, G.R. Wallwork, Rev. High Temp. Mater. 4 (1978) 75–105. [10] R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 233–254. [11] R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 73–87. [12] S.W. Banovic, J.N. DuPont, A.R. Marder, Oxid. Met. 54 (2000) 339–369. [13] C.H. Kao, C.M. Wan, J. Mater. Sci. 23 (1988) 1943–1947. [14] M. Sakiyama, P. Tomaszewicz, G.R. Wallwork, Oxid. Met. 13 (1979) 311–330. [15] F. Saegusa, L. Lee, Corrosion 22 (1966) 168–177. [16] P. Tomaszewicz, G.R. Wallwork, Oxid. Met. 19 (1983) 165–185. [17] P.J. Smith, R.M. Beauprie, W.W. Smeltzer, D.V. Stevanovic, D.A. Thompson, Oxid. Met. 28 (1987) 259–276. [18] P.C. Patnaik, W.W. Smeltzer, Oxid. Met. 23 (1985) 53–75. [19] P.F. Tortorelli, G.M. Goodwin, M. Howell, J.H. DeVan, in: K. Natesan, G. Krishnamurti, G.Y. Lai (Eds.), Proc. Int. Conf. Heat-Resistant Mater. II, 1995, pp. 585–590. [20] HSC Chemistry for Windows, Version 4.0. Oy, Finland, Outokupmu Research, 1997. [21] S.W. Banovic, J.N. DuPont, A.R. Marder, Proc. Corrosion 99, San Antonio, TX, Pub. NACE, 1999, p. 12. [22] P.J. Smith, P.R.S. Jackson, W.W. Smeltzer, J. Electrochem. Soc. 134 (1987) 1424–1431. [23] W. Kai, J.P. Chu, R.T. Huang, P.Y. Lee, Mater. Sci. Eng. A A239–A240 (1997) 859–870. [24] K. Przybylski, T. Narita, W.W. Smeltzer, Oxid. Met. 38 (1992) 1–32.