- Email: [email protected]
Oxidation of alumina formers at 1173 K: effect of yttrium ion implantation and yttrium alloying addition
Corrosion Science 45 (2003) 1815–1831 www.elsevier.com/locate/corsci
Oxidation of alumina formers at 1173 K: effect of yttrium ion implantation and yttrium alloying addition R. Cueff *, H. Buscail, E. Caudron, C. Issartel, F. Riffard Laboratoire Vellave sur l’Elaboration et l’Etude des Mat eriaux, Equipe locale Universit e Blaise Pascal Clermont-Ferrand II, 8 rue J.B. Fabre, B.P. 219, 43006 Le Puy-en-Velay Cedex, France Received 16 October 2002; accepted 11 December 2002
Abstract The oxidation behaviour of three alumina forming FeCrAl alloys has been investigated during isothermal exposures in air at 1173 K. Two of them were Kanthal A1, differing by the presence or not of implanted yttrium. The third one, Kanthal AF contains alloying additions of yttrium. Kinetic results indicate that only yttrium implantation significantly reduces the growth rate of the oxide scale during the early oxidation stage. For longer oxidation times, the reactive element markedly influences the oxidation rate and the composition of the oxide scale, whatever its introduction mode in the alloy. In situ X-ray diffraction shows that yttrium suppresses the formation of transition alumina and promotes the growth of a-Al2 O3 , thereby leading to the earlier formation of a protective oxide scale. Ó 2003 Published by Elsevier Science Ltd. Keywords: A. Rare earth elements; B. Ion implantation; B. X-ray diffraction; C. High temperature corrosion; C. Oxidation
1. Introduction FeCrAl alloys offer excellent properties for high temperatures industrial applications, especially for fabrication of electric resistances for furnaces. The oxidation
*
Corresponding author. Address: Laboratoire Vellave sur lÕElaboration et lÕEtude des Materiaux, Departement de Chimie Sciences des Materiaux, Institut Universitaire de Technologie, 8 rue J.B. Fabre, B.P. 219, 43006 Le Puy-en-Velay Cedex, France. Fax: +33-04-71099049. E-mail address: cueff@iut.u-clermont1.fr (R. Cueff). 0010-938X/03/$ - see front matter Ó 2003 Published by Elsevier Science Ltd. doi:10.1016/S0010-938X(02)00254-8
1816
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
resistance is attributed to the formation of an adherent and slow growing a-alumina scale. Nevertheless, the anticorrosive properties can be drastically reduced, particularly under thermal cycling when the scale spallation is excessive. The corrosion resistance of these alumina forming alloys is thus usually improved by doping alloys with small concentrations of reactive elements (such as yttrium or cerium) [1–5]. The modes of introduction of the reactive element include alloying additions of the reactive element or its oxide, ion implantation of the element and superficial application of the reactive element oxide as a thin coating (sol–gel, electrophoresis, MOCVD methods). The growth and development of protective Al2 O3 scales during oxidation at temperatures from 1073 to 1273 K, is complicated by the presence of transition aluminas which are less protective and faster growing than a-alumina [6– 9]. Numerous studies have reported that the influence of yttrium on the oxidation behaviour of alumina forming alloys is related to the effect on the alumina phase transformations and/or to the effect on the nucleation of a-alumina [6,10–14]. Many oxidation mechanisms proposed have been based on oxide scale characterization after cooling, only a few works have been dedicated to the in situ investigation of the reactive element effect [15–17]. In the present study, we used in situ high temperature X-ray diffraction to investigate the influence of yttrium ion implantation and yttrium alloying addition on the oxidation of alumina forming alloys at 1173 K. From the comparison between doped (yttrium-implanted Kanthal A1 and Kanthal AF, containing alloying addition of yttrium) and undoped (Kanthal A1) FeCrAl specimens, the effect of the reactive element is discussed in terms of kinetics related to the scale structure evolution at high temperature.
2. Experimental procedures The chemical compositions (obtained by glow discharge optical spectroscopy) of the FeCrAl alloys (produced by melting), are reported in Table 1. One mm thick specimens of rectangular shape with a total area of about 4 cm2 were abraded with SiC paper up to 800 grade and ultrasonically cleaned in ethanol prior to oxidation. Uniform yttrium implantation of the two main coupon faces, at a nominal dose of 1017 ions cm2 , was undertaken at the Institute for Health and Consumer Protection (Ispra, Italy), using an ion beam energy of 180 keV. The implantation chamber oxygen partial pressure was lower than 102 Pa and a rise in sample temperature to 200 °C was observed during yttrium implantation process. The sample was not cleaned by sputtering before implantation. Oxidation kinetics were followed at 1173 K, in laboratory air, during 120 h, using a Setaram TGDTA 92-1600 thermobalance. In situ scale structure evolution was analysed in a high temperature Anton PAAR HTK 1200 chamber with an integrated sample spinner in a Philips XÕpert MPD diffractometer. The XRD goniometer is equipped with a curved Cu monochromator to cut off the diffracted CuKa wavelength from all other wavelengths such as iron fluorescence radiation. In situ X-ray diffraction analyses were performed using CuKa1 (0.15406 nm) radiation. Series of X-ray diffractograms were recorded every hour during the 30 h oxidation test. The most representative in situ X-ray diffraction
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1817
Table 1 Composition of Kanthal A1 and Kanthal AF (weight percentage) Elements Fe Cr Al Si Mn Zr C Ti Y Ce Mg S Hf
Alloys Kanthal A1
Kanthal AF
71.3 22.2 5.66 0.26 0.13 0.083 0.022 0.016 <0.005 <0.005 <0.005 <0.005 <0.005
71.4 22.8 5.23 0.21 0.17 0.054 0.036 0.071 0.028 <0.010 <0.005 <0.005 <0.001
spectra will only be presented in our study. Diffraction peaks will be considered as significant if their relative intensity is greater than 10% of the intensity reported on their corresponding Joint Committee on Powder Diffraction Standards (JCPDS) file. Glancing angle X-ray diffraction (GAXRD), performed on the same diffractometer at various incident angles ranging from 0.5° to 1.5° (maximum penetration depths of approximately 50–140 nm), was carried out to study any possible effect of ion implantation on the structure of the metallic substrate. The oxide scale surface and cross-section morphologies were observed by scanning electron microscopy (SEM) coupled with energy dispersive X-ray spectrometry (EDXS). The EDXS point analyses were performed with an electron probe focused to a 1 lm spot. For crosssection analyses, the specimens were mounted in resin before polishing (with 1 lm diamond suspension) in order to preserve the integrity of the brittle oxide scales on the FeCrAl substrates. An evaporated carbon coating of about 10 nm thickness was also deposited on these samples to improve conducting properties. 3. Results 3.1. Oxidation kinetics Fig. 1 presents the oxidation curves Dm=S ¼ f ðt1=2 Þ (Dm=S is the mass gain per unit area and t the exposure time) of Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF. It can be observed that the mass gains after 120 h oxidation are clearly higher for the yttrium-free specimen than for the yttrium-containing ones. These mass gains are respectively, 0.26 mg cm2 , 0.08 mg cm2 and 0.12 mg cm2 for Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF. The kinetic curves of the yttrium-free and the yttrium-alloyed specimens exhibit an initial transient stage during the first 6 h, followed by a parabolic regime. The kinetic curve of the yttrium-implanted specimen is different since two subsequent parabolic regimes are observed. The first one, acting
1818
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
0.3 Kanthal A1 Kanthal AF
-2 ∆m/S (mg.cm )
0.25
Kanthal A1
0.2 0.15 0.1 0.05 0 0
2
4
6 1/2
(t)
8
10
12
1/2
(h)
Fig. 1. Mass-gain curves versus square root time for Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF at 1173 K in air.
during the first 6–8 h, is characterized by a much higher rate constant (about two orders of magnitude). However, the mass gain observed for the yttrium-implanted specimen after the early oxidation stage (first 6 oxidation hours) is clearly lower (0.025 mg cm2 ) than the mass gains obtained for the yttrium-alloyed (0.048 mg cm2 ) and the yttrium-free (0.057 mg cm2 ) specimens. This result indicates that only yttrium implantation significantly influences the growth rate of the oxide scale during the initial stage of the oxidation process. For longer oxidation times, the oxidation rate of Kanthal A1 was significantly higher than the ones exhibited by yttrium-implanted Kanthal A1 and Kanthal AF. The parabolic rate constants kp , calculated after the early oxidation stage, are respectively 1:3 107 mg2 cm4 s1 for Kanthal A1, 1:2 108 mg2 cm4 s1 for yttrium-implanted Kanthal A1 and 1:4 108 mg2 cm4 s1 for Kanthal AF. These kinetic constants clearly reveal the beneficial effect of yttrium implantation and yttrium alloying addition on the oxidation rate. 3.2. Initial specimens characterizations by XRD and GAXRD XRD analyses performed at 278 K on the initial specimens reveal diffractograms corresponding to the a-Fe structure (JCPDS 06-696) (Fig. 2). For both alloys, the relative intensities of the three main characteristic diffraction peaks are equal to those reported on the a-Fe JCPDS file (JCPDS file relative intensities are respectively 100%, 20% and 30% for the (1 1 0), (2 0 0) and (2 1 1) peaks). On the opposite, the GAXRD patterns (Fig. 3) of the yttrium-implanted Kanthal A1 exhibit significant differences in the relative intensities of these peaks compared to those expected. This result is indicative of some preferred crystallographic orientation of the metal surface. The broad low intensity peak located near 2h ¼ 30° reveals the presence of a
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1819
Fig. 2. XRD patterns performed on Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF at 278 K in air.
nanocrystalline phase. The reduction of the incident angle from 1.5° to 0.5° leads to a progressive disappearance of the broad peak together with the metal peaks, thus indicating that the phase responsible for the broad peak is located throughout the implanted region. 3.3. In situ high temperature X-ray diffraction study 3.3.1. Oxidation of Kanthal A1 In situ high temperature X-ray diffraction patterns obtained on Kanthal A1 specimens are given in Fig. 4. The relatively high intensity of the main characteristic diffraction peaks of the FeCrAl alloys during the whole experiment suggests that the
1820
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 3. GAXRD patterns performed at various glancing angles on yttrium-implanted Kanthal A1 at 278 K in air.
sample is slowly oxidized. The diffractograms clearly show that a-Al2 O3 (JCPDS 461212) is formed during the first hours. The typical diffraction peaks of a-Al2 O3 are detected after 2 h and their intensity increases continuously up to 30 h. The presence of transition aluminas, d-Al2 O3 (JCPDS 46-1131) and h-Al2 O3 (JCPDS 35-0121), is also detected during the oxidation test after 7 h. The high intensity of the peak near 2h ¼ 37:7° leads to the assumption that this peak does not contain only the aAl2 O3 (1 1 0) peak. It probably corresponds to the overlapping of the d-Al2 O3 (0 2 5) (JCPDS file relative intensity: 43%), d-Al2 O3 (1 2 3) (JCPDS file relative intensity: 38%), d-Al2 O3 (1 0 9) (JCPDS file relative intensity: 34%) and the a-Al2 O3 (1 1 0) (JCPDS file relative intensity: 21%) diffraction peaks. All the experimental peaks which allow identification of h-Al2 O3 correspond to the overlapping of h-Al2 O3 and
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1821
Fig. 4. Selected in situ XRD patterns performed on Kanthal A1 at 1173 K in air (0–30 h).
d-Al2 O3 characteristic diffraction peaks. Among them, the peak located near 2h ¼ 32:8° exhibits the greatest intensity. It can be attributed to the overlapping of the h-Al2 O3 (2 0 2) (JCPDS file relative intensity: 100%), h-Al2 O3 (2 0 0) (JCPDS file relative intensity: 100%), d-Al2 O3 (0 2 2) (JCPDS file relative intensity: 71%) diffraction peaks. This peak intensity increases significantly from the 18 to 30 h oxidation whereas the intensity of the d-Al2 O3 (2 2 0) peak located at 2h ¼ 45:6° remains roughly constant during the same period. This relative evolution of the transition alumina characteristic diffraction peaks underlines the more pronounced evidence of the h-Al2 O3 presence during the last 10 h oxidation. 3.3.2. Oxidation of yttrium-implanted Kanthal A1 The diffractograms of Fig. 5 clearly indicate that the composition of the oxide layer formed on yttrium-implanted Kanthal A1 is completely different from the one determined for the yttrium-free alloy. Peaks derived from a-Al2 O3 , (Fe,Cr)2 O3 (JCPDS 34-0412) and YAlO3 (JCPDS 33-0041) were detected. The relative intensities of the characteristic diffraction peaks of YAlO3 and (Fe,Cr)2 O3 , which do not show any significant evolution during the first 8 oxidation hours of the test, tend to decrease slightly during the last 20 h exposure. This indicates that the formation of these two mixed oxides is probably achieved after the very first oxidation hours. However, the yttrium–aluminium mixed oxide YAlO3 seems to be the prevalent oxidation compound whereas the iron–chromium mixed oxide (Fe,Cr)2 O3 is present in very low amount. The stable phase of alumina (a-Al2 O3 ) is clearly identified in the oxide layer after 1 h exposure and exhibits a gradually growth during the whole test.
1822
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 5. Selected in situ XRD patterns performed on yttrium-implanted Kanthal A1 at 1173 K in air (0–30 h).
The relative intensities of the characteristic diffraction peaks of a-alumina indicate some orientation effects during the initial growth of the oxide. It is interesting to note the absence of transition aluminas, which were coexisting with a-Al2 O3 in the oxide scale formed on Kanthal A1. 3.3.3. Oxidation of Kanthal AF Fig. 6 reveals that the growth of a-alumina starts from the beginning of the test, at least after the first hour exposure. The diffraction peaks of a-Al2 O3 are clearly detected all along the 30 h oxidation test. As observed in the case of yttrium-implanted Kanthal A1, transition aluminas are not detected. Yttrium seems to suppress the formation of transition alumina and favours early a-Al2 O3 formation and growth. However, the in situ X-ray diffraction analysis has not allowed the identification of any yttrium or yttrium–aluminium oxides. Nevertheless, we cannot reliably conclude on the absence of the reactive element in the oxide scale since its concentration can be below the detection limit of XRD. It would not be surprising since the yttrium concentration in Kanthal AF is very low compared to that in yttrium-implanted Kanthal A1. 3.4. Oxide scales morphology The scale surface morphology is strongly affected by the scale composition. On the one hand, Fig. 7(a) shows the blade-like whiskers morphology of the outer surfaces of the oxide scale developed on the yttrium-free alloy. This typical morphology appears to be indicative of h-Al2 O3 formation [12,18,19]. On the other hand, Fig. 7(c)
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1823
Fig. 6. Selected in situ XRD patterns performed on Kanthal AF at 1173 K in air (0–30 h).
Fig. 7. Surface scale morphology (SEM secondary electron images) of Kanthal A1 (a), yttrium-implanted Kanthal A1 (b), Kanthal AF (c), oxidized during 120 h in air at 1173 K.
reveals that equiaxed oxide grains (with diameter ranging from 1 to 2 lm) entirely cover the surface of the Kanthal AF. This microstructure is typical of a-Al2 O3 [13,20]. The surface morphology of the oxidized yttrium-implanted alloy (Fig. 7(b)) is slightly different. The metal is covered by a thin oxide layer (polishing stripes have not disappeared) composed of very small grains (diameter less than 0.5 lm), without any whisker. These SEM observations are consistent with X-ray diffraction patterns and confirm that the fast oxide growth is related to the presence of transition alumina and the low kinetic rate corresponds to a-alumina. 3.5. SEM cross-sections and EDXS analysis The SEM cross-sections of the oxidized specimens (120 h isothermal exposure) reveals that the oxide layer formed on the yttrium-free alloy (Fig. 8(a)) is thicker than the one developed on the yttrium-containing alloys (Fig. 8(b) and (c)). The
1824
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 8. SEM backscattered electron cross-section images of the oxide scales formed on Kanthal A1 (a), yttrium-implanted Kanthal A1 (b), Kanthal AF (c), in air at 1173 K (120 h exposure).
thickness of the oxide layer formed on Kanthal A1 is equal to 2 lm, which can be compared to the approximately 0.5 and 1 lm thick oxide scales formed respectively on yttrium-implanted Kanthal A1 and Kanthal AF. These SEM observations can be related to the kinetic results, i.e. to the higher mass gains obtained after 120 h oxidation for the yttrium-free specimen than for the yttrium-containing alloys. The very low thickness of the oxide layer formed on yttrium-implanted Kanthal A1 is also consistent with its surface aspect. The EDXS analyses of the oxide scales formed on Kanthal A1 and Kanthal AF (Figs. 9 and 10) show that mainly aluminium and oxygen are detected in the intermediate and the outer parts of the layer. In the inner part, low intensity peaks of chromium and iron are detected, which is consistent with a contribution of the metal to the X-ray emission in the spotted region. Within the detection limit of EDXS (about 0.5 at.%), yttrium was not detected in the oxide scale formed on Kanthal AF. This corroborates the XRD results and suggests that yttrium compounds, if exist in Kanthal AF, are present in very low amounts. The EDXS analyses performed on the yttrium-implanted alloy (Fig. 11) clearly reveal the presence of the reactive element in the oxide scale. The most intense YLa and YLb peaks are relative to X-ray emission from the inner part of the oxide layer. This is consistent with a more likely presence of yttrium at the metal–oxide interface. Peaks attributed to iron and chromium are detected whatever the spotted region of the scale. Although the presence of such peaks derived from EDXS analyses in the inner and the intermediate parts of the oxide may partly be explained by the significant contribution of the metal to the X-ray emission (related to the low oxide scale thickness), their identification in the outermost part of the scale cannot primarily be accounted by the influence of the underlying matrix. This iron and chromium detection at the oxide– gas interface, together with a poor alumina concentration (the AlKa peak intensity is strongly reduced in this outermost oxide region) is consistent with an outer part of the oxide layer mostly composed of iron–chromium mixed oxides.
4. Discussion GAXRD analyses performed on the yttrium-implanted Kanthal A1 indicate, besides some preferred orientation of the metal, the presence of a nanocrystalline
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1825
Fig. 9. EDXS spectra of the oxide scale formed on Kanthal A1: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
phase. This phase, which is present throughout the implanted region, is not due to a surface oxide. A previous study on yttrium-implanted Ni–20Cr reported the presence of a nanocrystalline phase upon initial nucleation [21]. Its author identified, by XRD and EXAFS analyses, this phase to be Y2 O3 . Other studies relating to the yttriumimplantation effects using XRD, GAXRD and RBS also reported the presence of Y2 O3 as the main compound promoted by yttrium implantation on pure iron and various steels [22,23]. These conclusions are somehow different from those reported in a study performed on yttrium-implanted chromium [24]. Their authors stated that a large proportion of yttrium is not contained in the Y2 O3 phase and is thus in ‘‘random’’ positions in the implanted region. With our experimental implantation conditions, the possible formation of yttria in the implanted-specimen would be
1826
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 10. EDXS spectra of the oxide scale formed on Kanthal AF: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
related to the presence of oxygen which can appear in the sample by the combination of two phenomena: the sputtering of the naturally oxidized sample surfaces (the sample was not cleaned by sputtering before implantation) during the implantation process and the low oxygen partial pressure existing in the implantation chamber [25,26]. Kinetic results suggest that the oxidation behaviour during the initial stage of the oxidation process is strongly dependent on the introduction mode of yttrium in the alloy: the oxidation rate is only significantly reduced when yttrium is implanted in the alloy, probable consequence of the formation of an oxide scale immediately acting as an effective diffusion barrier. In situ X-ray diffraction analyses performed during this oxidation period revealed drastic changes in the nature of the main oxide
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1827
Fig. 11. EDXS spectra of the oxide scale formed on yttrium-implanted Kanthal A1: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
compounds formed, consecutive to yttrium implantation in the alloy: the oxide scales developed in the first 6 h on Kanthal A1 and Kanthal AF are mainly composed of aAl2 O3 whereas oxidation of yttrium-implanted Kanthal A1 leads to the formation of YAlO3 , (Fe,Cr)2 O3 and a-Al2 O3 , with a much greater amount of the yttrium– aluminium mixed oxide. These results must be approached by taking into accounts general considerations regarding the transient stage of oxide development on FeCrAl alloys. During the early stage of oxidation, transient oxides, rich in iron and chromium are usually formed before the development of the a-Al2 O3 layer at the inner interface. The formation of the a-Al2 O3 layer beneath the transient oxides arises from the selective oxidation of aluminium which occurs with the resulting reduction in oxygen potential at the oxide–alloy interface [27]. In the present study, the (Fe,Cr)2 O3 phase probably forms during the very early stage of oxidation and
1828
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
remains on the surface of the growing oxide layer. The formation of these (Fe,Cr)2 O3 oxides during the initial transient oxidation stage of yttrium-bearing aluminaforming alloys has been previously reported in other studies [28,29]. As the iron– chromium mixed oxide is only detected in the yttrium-implanted specimen, we can suppose that the (Fe,Cr)2 O3 enrichment and the formation of the yttrium–aluminium mixed oxides (YAlO3 ) are closely linked. Indeed, the yttrium–aluminate formation may causes the thin alloy surface layer to be aluminium depleted in the early stage of oxidation [2]. However, there is no reliable indication suggesting any possible influence of this (Fe,Cr)2 O3 formation on the oxidation kinetic in the early stage of oxidation. We only observed that the scale growth kinetic of the yttriumimplanted alloy, in the initial oxidation stage, is dominated by diffusion of reactants through a growing compact scale. In the presence of oxygen, yttrium has a very high affinity to aluminium, leading to the formation of yttrium aluminates (YAlO3 and/or Y3 Al5 O12 ), which have a high thermodynamic stability. In the present study, the implanted yttrium was found to incorporate in the scale in the perovskite structure (YAlO3 ), in the alumina-rich part of the scale (at the metal–oxide interface). This yttrium–aluminium mixed oxide, which forms during the very first oxidation hours, provides from aluminium reaction with oxygen in the presence of yttria, according to the reaction [30]: Y2 O3 þ Al2 O3 ! 2YAlO3
ð1Þ
The presence of YAlO3 particles within the scale of oxidized yttrium-doped aluminaforming alloys is in agreement with other studies [31–34]. It is often suggested [31,35] that yttrium diffuses rapidly along the metal–oxide interface and segregates to the alumina grain boundaries. Other authors have detected the presence of this perovskite, together with yttria, in the outermost region of an oxidized yttrium-doped alumina-forming alloy [33]. This was attributed to internal oxidation which occurred below the alumina–alloy interface, particularly in the alloy grain boundaries. Other studies concluded on the internal oxidation of yttrium-doped NiAl [36] or yttriumimplanted pure iron [37], leading to the formation of yttrium-containing oxide particles (respectively Al2 Y4 O9 and Fe2 YO4 ) located inside the metal. In the present work, such an internal oxidation of yttrium in the alloy during the oxidation process is possible, owing to the very high intensity observed for the YAlO3 (1 0 0) peak (JCPDS file relative intensity: 100%) during in situ X-ray diffraction. The steady-state stage of oxidation corresponds, for both alloys, to an oxidation process limited by diffusion through the growing oxide scale. The mass gain curves clearly show a marked influence of yttrium on the scale growth rate during this oxidation stage. The addition of the reactive element leads to a reduction of the parabolic rate constant by a factor of 10. Furthermore, the in situ X-ray diffraction analysis revealed that transition alumina is not detected in the oxide scales developed on the yttrium-doped alloys. These structural analyses correlated with kinetic studies suggest that the effect of the reactive element, by hampering the formation of transition aluminas or by leading to an acceleration of the phase transformation of transition aluminas into stable alumina, promotes the formation of a protective aAl2 O3 scale. This influence of the reactive element on the rate of transformation of
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1829
metastable alumina into a-alumina at temperatures below 1273 K is still a matter of discussion. Some authors have reported that yttrium addition delays the phase transformation [12,38] whereas other investigations led to opposite conclusions, i.e. an acceleration of the rate of transformation [15,39]. A possible explanation of these various existing controversies is proposed by Jedlinski [40]. This author suggests that the influence of the reactive element can be accounted for in terms of several mechanisms, which lead to an acceleration or a retardation of the phase transformation. The relative contribution of each of these mechanisms depends on the amount and form of the reactive element [40]. Our results support the assumption that the involved mechanisms favour the formation of a-Al2 O3 . This leads to a reduced mass gain since transition aluminas are less protective and faster growing than a-alumina [6–9]. Yttrium may affect the properties of unstable aluminas and their transformation into a-alumina by virtue of microstructural effects, involving additional sites for heterogeneous nucleation of the oxide as well as hampering the grain growth of unstable aluminas [40]. Earlier nucleation of the oxide layer results in accelerated evolution of the scale, and consequently, in earlier formation of the stable alumina [40]. Chemical effects involving injection of vacancies into unstable aluminas, due to the formation of mixed yttrium–aluminium oxides in the scale, can also account for the influence of yttrium in promoting the growth of stable aalumina [40]. Some authors also suggested that yttrium promotes the selective oxidation of aluminium and facilitates the development of a-Al2 O3 scales, probably by acting as an additional getter for oxygen in the alloy [1]. Other studies have reported that the presence of titanium in alumina forming alloys could play a role in the phase transformation of transition alumina to a-alumina [6,12,13]. Small additions of this element to the alloy would cause an acceleration of the phase transformation resulting in an early formation of stable a-Al2 O3 . Since our study results are in agreement with the above assumption (Kanthal AF alloy contains the greatest amount of Ti and does not reveal the presence of transition alumina in the oxide scale), a possible effect of Ti on the phase transformation of alumina cannot be neglected. It is well established that the growth mechanism of a-alumina involves both inward oxygen diffusion and outward cation diffusion via short-circuit paths such as grain boundaries [41]. Some authors proposed that yttrium modifies the transport properties along the grain boundaries in a-alumina scales [42,43]. These studies suggest that the lower parabolic rate constant observed with addition of yttrium is related to the reduction of the aluminium outward transport through the oxide scale. The formation on the yttrium-free specimen surface of whiskers protruding towards the atmosphere may likely be due to a relatively fast diffusion of aluminium ions through the preformed layer towards the surface. It has been also reported that reactive elements have an important influence on the morphology of the scale. Several works have indicated that these elements can modify the grain size in the oxide scale. In many cases, it is mentioned that yttrium induces a decrease of the grain size of the alumina scale [20,44,45]. This indicates a change in the oxide growth process, which can affect the oxide growth rate and the mechanical properties (spallation resistance) of the oxide scale. Recently, the ‘‘dynamic segregation model’’
1830
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
[46] established that the reactive element particles are not stagnant dopants in the grain boundaries but diffuse outward during scale growth, thus suggesting that the effect of the reactive element on the growth mechanism and on the scale microstructure is attenuated as the temperature increases. In the present study, the above theory of physical blocking of cation diffusion may rather apply to the effect of implanted-yttrium since the presence of the reactive element in the oxide scale formed on the yttrium-alloyed specimen is probably very limited.
5. Conclusions This study suggests that the oxidation behaviour of the yttrium-doped FeCrAl alloys closely depends on the introduction mode of yttrium in the alloy. On the one hand, the kinetic curves of the yttrium-free and the yttrium-alloyed specimens exhibit an initial transient stage during the first 6 h, followed by a parabolic regime. On the other hand, the scale growth kinetic of the yttrium-implanted alloy obeys a parabolic rate law with two subsequent stages with a much lower rate constant during the second stage. Thus, during the early stage of oxidation, the oxide scale formed on the yttrium-implanted alloy immediately acts as a diffusion barrier and the oxidation rate is significantly reduced. In situ X-ray diffraction analyses performed during this oxidation period revealed drastic changes in the nature of the main oxide compounds formed, consecutive to yttrium implantation in the alloy: besides the formation of a-Al2 O3 , which is also effective on Kanthal A1 and Kanthal AF, oxidation of yttrium-implanted Kanthal A1 leads to the formation of YAlO3 and (Fe,Cr)2 O3 . The steady-state stage of oxidation corresponds, for both alloys, to an oxidation process limited by diffusion through the growing oxide scale. The mass gain curves clearly show a marked influence of yttrium on the scale growth rate during this oxidation stage. The addition of the reactive element by ion implantation or alloying addition leads to a reduction of the parabolic rate constant by a factor of 10. The in situ X-ray diffraction study revealed that this reduction of the oxidation rate is related to the absence of transition alumina in the oxide layer formed on the yttrium-containing alloys. This phenomenon, which has been established as part of the reactive element effect on the oxidation of alumina forming alloys, directly reflects the influence of yttrium in promoting the nucleation and growth of a-alumina. As a consequence, the formation of a protective oxide scale occurs earlier on the yttrium-containing alloys than on the yttrium-free alloy.
References [1] F.A. Golightly, F.H. Stott, G.C. Wood, Oxid. Met. 10 (1976) 163. [2] W.J. Quadakkers, J. Jedlinski, K. Schmidt, M. Krasovec, G. Borchardt, H. Nickel, Appl. Surf. Sci. 47 (1991) 261. [3] H.J. Schmutzler, H. Viefhaus, H.J. Grabke, Surf. Interf. Anal. 18 (1992) 581. [4] P.Y. Hou, J. Stringer, Oxid. Met. 38 (1992) 323. [5] C. Mennicke, E. Schumann, C. Ulrich, M. Ruhle, Mater. Sci. Forum 251–254 (1997) 389.
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831 [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46]
1831
B.A. Pint, M. Treska, L.W. Hobbs, Oxid. Met. 47 (1997) 1. K.D.D. Lagerlof, B.J. Pletka, T.E. Mitchell, A.H. Heuer, Radiat. Defects 74 (1983) 87. R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 73. K. Messaoudi, A.M. Huntz, B. Lesage, Mater. Sci. Eng. A 247 (1998) 248. R. Prescott, D.F. Mitchell, J.W. Fraser, M.J. Graham, J. Phys. IV 3 (1993) 301. E. Schumann, Oxid. Met. 43 (1995) 157. B.A. Pint, J.R. Martin, L.W. Hobbs, Solid State Ionics 78 (1995) 99. K.M.N. Prasanna, A.S. Khanna, R. Chandra, W.J. Quadakkers, Oxid. Met. 46 (1996) 465. H.J. Grabke, Intermetallics 7 (1999) 1153. I. Rommerskirchen, V. Kolarik, Mater. Corros. 47 (1996) 625. J.M. Herbelin, M. Mantel, Mater. Sci. Forum 251–254 (1997) 349. V. Kolarik, H. Fietzek, W. Engel, B. Eltester, H.J. Grabke, Mater. Sci. Forum 251–254 (1997) 251. C. Mennicke, E. Schumann, M. Ruhle, R.J. Hussey, G.I. Sproule, M.J. Graham, Oxid. Met. 49 (1998) 455. B.A. Pint, L.W. Hobbs, J. Electrochem. Soc. 141 (1994) 2443. M.W. Brumm, H.J. Grabke, Corros. Sci. 33 (1992) 1677. T.A. Crabb, Ph.D. Thesis, University of Strathclyde, 1993. E. Caudron, H. Buscail, R. Cueff, Y.P. Jacob, M.F. Stroosnijder, Thin Solid Films 350 (1999) 168. E. Caudron, H. Buscail, V.A.C. Haanapel, Y.P. Jacob, C. Josse-Courty, M.F. Stroosnijder, Appl. Surf. Sci. 153 (2000) 156. M.J. Cristobal, P.N. Gibson, M.F. Stroosnijder, Corros. Sci. 38 (1996) 805. E. Caudron, H. Buscail, Corros. Sci. 43 (2001) 1477. E. Caudron, F. Riffard, H. Buscail, R. Cueff, C. Issartel, S. Perrier, Eur. Phys. J. AP 17 (2002) 87. F.A. Golightly, G.C. Wood, F.H. Stott, Oxid. Met. 14 (1980) 217. S. Weinbruch, A. Anastassiadis, H.M. Ortner, H.P. Martinz, P. Wilhartitz, Oxid. Met. 51 (1999) 111. P. Perez, V.A.C. Haanapel, M.F. Stroosnijder, Surf. Coat. Technol. 139 (2001) 207. H.G. Jung, K.Y. Kim, Oxid. Met. 49 (1998) 403. T.A. Ramanarayanan, M. Raghavan, R.P. Luton, J. Electrochem. Soc. 131 (1984) 923. F. Clemendot, J.M. Gras, J.C. van Duysen, J. Phys. IV 3 (1993) 291. P. Choquet, R. Mevrel, Mater. Sci. Eng. A 120 (1989) 153. N. Czech, M. Juez-Lorenzo, V. Kolarik, W. Stamm, Surf. Coat. Technol. 108–109 (1998) 36. M.J. Graham, J.I. Eldrige, D.F. Mitchell, R.J. Hussey, Mater. Sci. Forum 43 (1989) 207. E. Schumann, J.C. Yang, M.J. Graham, Script. Mater. 9 (1996) 1365. C. Josse-Courty, H. Buscail, M.F. Stroosnijder, P. Dufour, J.P. Larpin, Oxid. Met. 52 (1999) 321. Y. Saito, B. Onay, T. Maruyama, J. Phys. IV 3 (1993) 217. H.J. Grabke, Mater. Sci. Forum 251–254 (1997) 149. J. Jedlinski, Oxid. Met. 39 (1993) 55. F.H. Stott, Mater. Sci. Forum 251–254 (1997) 19. B.A. Pint, J.R. Martin, L.W. Hobbs, Oxid. Met. 39 (1993) 167. B.A. Pint, A.J. Garratt-Reed, L.W. Hobbs, Mater. High Temp. 13 (1995) 3. M.K. Loudjani, C. Haut, J. Eur. Ceram. Soc. 16 (1996) 1099. M.K. Loudjani, C. Haut, S. Parisot, Radiat. Effects Defects Solids 134 (1995) 233. B.A. Pint, Oxid. Met. 45 (1996) 1.
Oxidation of alumina formers at 1173 K: effect of yttrium ion implantation and yttrium alloying addition R. Cueff *, H. Buscail, E. Caudron, C. Issartel, F. Riffard Laboratoire Vellave sur l’Elaboration et l’Etude des Mat eriaux, Equipe locale Universit e Blaise Pascal Clermont-Ferrand II, 8 rue J.B. Fabre, B.P. 219, 43006 Le Puy-en-Velay Cedex, France Received 16 October 2002; accepted 11 December 2002
Abstract The oxidation behaviour of three alumina forming FeCrAl alloys has been investigated during isothermal exposures in air at 1173 K. Two of them were Kanthal A1, differing by the presence or not of implanted yttrium. The third one, Kanthal AF contains alloying additions of yttrium. Kinetic results indicate that only yttrium implantation significantly reduces the growth rate of the oxide scale during the early oxidation stage. For longer oxidation times, the reactive element markedly influences the oxidation rate and the composition of the oxide scale, whatever its introduction mode in the alloy. In situ X-ray diffraction shows that yttrium suppresses the formation of transition alumina and promotes the growth of a-Al2 O3 , thereby leading to the earlier formation of a protective oxide scale. Ó 2003 Published by Elsevier Science Ltd. Keywords: A. Rare earth elements; B. Ion implantation; B. X-ray diffraction; C. High temperature corrosion; C. Oxidation
1. Introduction FeCrAl alloys offer excellent properties for high temperatures industrial applications, especially for fabrication of electric resistances for furnaces. The oxidation
*
Corresponding author. Address: Laboratoire Vellave sur lÕElaboration et lÕEtude des Materiaux, Departement de Chimie Sciences des Materiaux, Institut Universitaire de Technologie, 8 rue J.B. Fabre, B.P. 219, 43006 Le Puy-en-Velay Cedex, France. Fax: +33-04-71099049. E-mail address: cueff@iut.u-clermont1.fr (R. Cueff). 0010-938X/03/$ - see front matter Ó 2003 Published by Elsevier Science Ltd. doi:10.1016/S0010-938X(02)00254-8
1816
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
resistance is attributed to the formation of an adherent and slow growing a-alumina scale. Nevertheless, the anticorrosive properties can be drastically reduced, particularly under thermal cycling when the scale spallation is excessive. The corrosion resistance of these alumina forming alloys is thus usually improved by doping alloys with small concentrations of reactive elements (such as yttrium or cerium) [1–5]. The modes of introduction of the reactive element include alloying additions of the reactive element or its oxide, ion implantation of the element and superficial application of the reactive element oxide as a thin coating (sol–gel, electrophoresis, MOCVD methods). The growth and development of protective Al2 O3 scales during oxidation at temperatures from 1073 to 1273 K, is complicated by the presence of transition aluminas which are less protective and faster growing than a-alumina [6– 9]. Numerous studies have reported that the influence of yttrium on the oxidation behaviour of alumina forming alloys is related to the effect on the alumina phase transformations and/or to the effect on the nucleation of a-alumina [6,10–14]. Many oxidation mechanisms proposed have been based on oxide scale characterization after cooling, only a few works have been dedicated to the in situ investigation of the reactive element effect [15–17]. In the present study, we used in situ high temperature X-ray diffraction to investigate the influence of yttrium ion implantation and yttrium alloying addition on the oxidation of alumina forming alloys at 1173 K. From the comparison between doped (yttrium-implanted Kanthal A1 and Kanthal AF, containing alloying addition of yttrium) and undoped (Kanthal A1) FeCrAl specimens, the effect of the reactive element is discussed in terms of kinetics related to the scale structure evolution at high temperature.
2. Experimental procedures The chemical compositions (obtained by glow discharge optical spectroscopy) of the FeCrAl alloys (produced by melting), are reported in Table 1. One mm thick specimens of rectangular shape with a total area of about 4 cm2 were abraded with SiC paper up to 800 grade and ultrasonically cleaned in ethanol prior to oxidation. Uniform yttrium implantation of the two main coupon faces, at a nominal dose of 1017 ions cm2 , was undertaken at the Institute for Health and Consumer Protection (Ispra, Italy), using an ion beam energy of 180 keV. The implantation chamber oxygen partial pressure was lower than 102 Pa and a rise in sample temperature to 200 °C was observed during yttrium implantation process. The sample was not cleaned by sputtering before implantation. Oxidation kinetics were followed at 1173 K, in laboratory air, during 120 h, using a Setaram TGDTA 92-1600 thermobalance. In situ scale structure evolution was analysed in a high temperature Anton PAAR HTK 1200 chamber with an integrated sample spinner in a Philips XÕpert MPD diffractometer. The XRD goniometer is equipped with a curved Cu monochromator to cut off the diffracted CuKa wavelength from all other wavelengths such as iron fluorescence radiation. In situ X-ray diffraction analyses were performed using CuKa1 (0.15406 nm) radiation. Series of X-ray diffractograms were recorded every hour during the 30 h oxidation test. The most representative in situ X-ray diffraction
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1817
Table 1 Composition of Kanthal A1 and Kanthal AF (weight percentage) Elements Fe Cr Al Si Mn Zr C Ti Y Ce Mg S Hf
Alloys Kanthal A1
Kanthal AF
71.3 22.2 5.66 0.26 0.13 0.083 0.022 0.016 <0.005 <0.005 <0.005 <0.005 <0.005
71.4 22.8 5.23 0.21 0.17 0.054 0.036 0.071 0.028 <0.010 <0.005 <0.005 <0.001
spectra will only be presented in our study. Diffraction peaks will be considered as significant if their relative intensity is greater than 10% of the intensity reported on their corresponding Joint Committee on Powder Diffraction Standards (JCPDS) file. Glancing angle X-ray diffraction (GAXRD), performed on the same diffractometer at various incident angles ranging from 0.5° to 1.5° (maximum penetration depths of approximately 50–140 nm), was carried out to study any possible effect of ion implantation on the structure of the metallic substrate. The oxide scale surface and cross-section morphologies were observed by scanning electron microscopy (SEM) coupled with energy dispersive X-ray spectrometry (EDXS). The EDXS point analyses were performed with an electron probe focused to a 1 lm spot. For crosssection analyses, the specimens were mounted in resin before polishing (with 1 lm diamond suspension) in order to preserve the integrity of the brittle oxide scales on the FeCrAl substrates. An evaporated carbon coating of about 10 nm thickness was also deposited on these samples to improve conducting properties. 3. Results 3.1. Oxidation kinetics Fig. 1 presents the oxidation curves Dm=S ¼ f ðt1=2 Þ (Dm=S is the mass gain per unit area and t the exposure time) of Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF. It can be observed that the mass gains after 120 h oxidation are clearly higher for the yttrium-free specimen than for the yttrium-containing ones. These mass gains are respectively, 0.26 mg cm2 , 0.08 mg cm2 and 0.12 mg cm2 for Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF. The kinetic curves of the yttrium-free and the yttrium-alloyed specimens exhibit an initial transient stage during the first 6 h, followed by a parabolic regime. The kinetic curve of the yttrium-implanted specimen is different since two subsequent parabolic regimes are observed. The first one, acting
1818
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
0.3 Kanthal A1 Kanthal AF
-2 ∆m/S (mg.cm )
0.25
Kanthal A1
0.2 0.15 0.1 0.05 0 0
2
4
6 1/2
(t)
8
10
12
1/2
(h)
Fig. 1. Mass-gain curves versus square root time for Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF at 1173 K in air.
during the first 6–8 h, is characterized by a much higher rate constant (about two orders of magnitude). However, the mass gain observed for the yttrium-implanted specimen after the early oxidation stage (first 6 oxidation hours) is clearly lower (0.025 mg cm2 ) than the mass gains obtained for the yttrium-alloyed (0.048 mg cm2 ) and the yttrium-free (0.057 mg cm2 ) specimens. This result indicates that only yttrium implantation significantly influences the growth rate of the oxide scale during the initial stage of the oxidation process. For longer oxidation times, the oxidation rate of Kanthal A1 was significantly higher than the ones exhibited by yttrium-implanted Kanthal A1 and Kanthal AF. The parabolic rate constants kp , calculated after the early oxidation stage, are respectively 1:3 107 mg2 cm4 s1 for Kanthal A1, 1:2 108 mg2 cm4 s1 for yttrium-implanted Kanthal A1 and 1:4 108 mg2 cm4 s1 for Kanthal AF. These kinetic constants clearly reveal the beneficial effect of yttrium implantation and yttrium alloying addition on the oxidation rate. 3.2. Initial specimens characterizations by XRD and GAXRD XRD analyses performed at 278 K on the initial specimens reveal diffractograms corresponding to the a-Fe structure (JCPDS 06-696) (Fig. 2). For both alloys, the relative intensities of the three main characteristic diffraction peaks are equal to those reported on the a-Fe JCPDS file (JCPDS file relative intensities are respectively 100%, 20% and 30% for the (1 1 0), (2 0 0) and (2 1 1) peaks). On the opposite, the GAXRD patterns (Fig. 3) of the yttrium-implanted Kanthal A1 exhibit significant differences in the relative intensities of these peaks compared to those expected. This result is indicative of some preferred crystallographic orientation of the metal surface. The broad low intensity peak located near 2h ¼ 30° reveals the presence of a
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1819
Fig. 2. XRD patterns performed on Kanthal A1, yttrium-implanted Kanthal A1 and Kanthal AF at 278 K in air.
nanocrystalline phase. The reduction of the incident angle from 1.5° to 0.5° leads to a progressive disappearance of the broad peak together with the metal peaks, thus indicating that the phase responsible for the broad peak is located throughout the implanted region. 3.3. In situ high temperature X-ray diffraction study 3.3.1. Oxidation of Kanthal A1 In situ high temperature X-ray diffraction patterns obtained on Kanthal A1 specimens are given in Fig. 4. The relatively high intensity of the main characteristic diffraction peaks of the FeCrAl alloys during the whole experiment suggests that the
1820
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 3. GAXRD patterns performed at various glancing angles on yttrium-implanted Kanthal A1 at 278 K in air.
sample is slowly oxidized. The diffractograms clearly show that a-Al2 O3 (JCPDS 461212) is formed during the first hours. The typical diffraction peaks of a-Al2 O3 are detected after 2 h and their intensity increases continuously up to 30 h. The presence of transition aluminas, d-Al2 O3 (JCPDS 46-1131) and h-Al2 O3 (JCPDS 35-0121), is also detected during the oxidation test after 7 h. The high intensity of the peak near 2h ¼ 37:7° leads to the assumption that this peak does not contain only the aAl2 O3 (1 1 0) peak. It probably corresponds to the overlapping of the d-Al2 O3 (0 2 5) (JCPDS file relative intensity: 43%), d-Al2 O3 (1 2 3) (JCPDS file relative intensity: 38%), d-Al2 O3 (1 0 9) (JCPDS file relative intensity: 34%) and the a-Al2 O3 (1 1 0) (JCPDS file relative intensity: 21%) diffraction peaks. All the experimental peaks which allow identification of h-Al2 O3 correspond to the overlapping of h-Al2 O3 and
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1821
Fig. 4. Selected in situ XRD patterns performed on Kanthal A1 at 1173 K in air (0–30 h).
d-Al2 O3 characteristic diffraction peaks. Among them, the peak located near 2h ¼ 32:8° exhibits the greatest intensity. It can be attributed to the overlapping of the h-Al2 O3 (2 0 2) (JCPDS file relative intensity: 100%), h-Al2 O3 (2 0 0) (JCPDS file relative intensity: 100%), d-Al2 O3 (0 2 2) (JCPDS file relative intensity: 71%) diffraction peaks. This peak intensity increases significantly from the 18 to 30 h oxidation whereas the intensity of the d-Al2 O3 (2 2 0) peak located at 2h ¼ 45:6° remains roughly constant during the same period. This relative evolution of the transition alumina characteristic diffraction peaks underlines the more pronounced evidence of the h-Al2 O3 presence during the last 10 h oxidation. 3.3.2. Oxidation of yttrium-implanted Kanthal A1 The diffractograms of Fig. 5 clearly indicate that the composition of the oxide layer formed on yttrium-implanted Kanthal A1 is completely different from the one determined for the yttrium-free alloy. Peaks derived from a-Al2 O3 , (Fe,Cr)2 O3 (JCPDS 34-0412) and YAlO3 (JCPDS 33-0041) were detected. The relative intensities of the characteristic diffraction peaks of YAlO3 and (Fe,Cr)2 O3 , which do not show any significant evolution during the first 8 oxidation hours of the test, tend to decrease slightly during the last 20 h exposure. This indicates that the formation of these two mixed oxides is probably achieved after the very first oxidation hours. However, the yttrium–aluminium mixed oxide YAlO3 seems to be the prevalent oxidation compound whereas the iron–chromium mixed oxide (Fe,Cr)2 O3 is present in very low amount. The stable phase of alumina (a-Al2 O3 ) is clearly identified in the oxide layer after 1 h exposure and exhibits a gradually growth during the whole test.
1822
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 5. Selected in situ XRD patterns performed on yttrium-implanted Kanthal A1 at 1173 K in air (0–30 h).
The relative intensities of the characteristic diffraction peaks of a-alumina indicate some orientation effects during the initial growth of the oxide. It is interesting to note the absence of transition aluminas, which were coexisting with a-Al2 O3 in the oxide scale formed on Kanthal A1. 3.3.3. Oxidation of Kanthal AF Fig. 6 reveals that the growth of a-alumina starts from the beginning of the test, at least after the first hour exposure. The diffraction peaks of a-Al2 O3 are clearly detected all along the 30 h oxidation test. As observed in the case of yttrium-implanted Kanthal A1, transition aluminas are not detected. Yttrium seems to suppress the formation of transition alumina and favours early a-Al2 O3 formation and growth. However, the in situ X-ray diffraction analysis has not allowed the identification of any yttrium or yttrium–aluminium oxides. Nevertheless, we cannot reliably conclude on the absence of the reactive element in the oxide scale since its concentration can be below the detection limit of XRD. It would not be surprising since the yttrium concentration in Kanthal AF is very low compared to that in yttrium-implanted Kanthal A1. 3.4. Oxide scales morphology The scale surface morphology is strongly affected by the scale composition. On the one hand, Fig. 7(a) shows the blade-like whiskers morphology of the outer surfaces of the oxide scale developed on the yttrium-free alloy. This typical morphology appears to be indicative of h-Al2 O3 formation [12,18,19]. On the other hand, Fig. 7(c)
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1823
Fig. 6. Selected in situ XRD patterns performed on Kanthal AF at 1173 K in air (0–30 h).
Fig. 7. Surface scale morphology (SEM secondary electron images) of Kanthal A1 (a), yttrium-implanted Kanthal A1 (b), Kanthal AF (c), oxidized during 120 h in air at 1173 K.
reveals that equiaxed oxide grains (with diameter ranging from 1 to 2 lm) entirely cover the surface of the Kanthal AF. This microstructure is typical of a-Al2 O3 [13,20]. The surface morphology of the oxidized yttrium-implanted alloy (Fig. 7(b)) is slightly different. The metal is covered by a thin oxide layer (polishing stripes have not disappeared) composed of very small grains (diameter less than 0.5 lm), without any whisker. These SEM observations are consistent with X-ray diffraction patterns and confirm that the fast oxide growth is related to the presence of transition alumina and the low kinetic rate corresponds to a-alumina. 3.5. SEM cross-sections and EDXS analysis The SEM cross-sections of the oxidized specimens (120 h isothermal exposure) reveals that the oxide layer formed on the yttrium-free alloy (Fig. 8(a)) is thicker than the one developed on the yttrium-containing alloys (Fig. 8(b) and (c)). The
1824
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 8. SEM backscattered electron cross-section images of the oxide scales formed on Kanthal A1 (a), yttrium-implanted Kanthal A1 (b), Kanthal AF (c), in air at 1173 K (120 h exposure).
thickness of the oxide layer formed on Kanthal A1 is equal to 2 lm, which can be compared to the approximately 0.5 and 1 lm thick oxide scales formed respectively on yttrium-implanted Kanthal A1 and Kanthal AF. These SEM observations can be related to the kinetic results, i.e. to the higher mass gains obtained after 120 h oxidation for the yttrium-free specimen than for the yttrium-containing alloys. The very low thickness of the oxide layer formed on yttrium-implanted Kanthal A1 is also consistent with its surface aspect. The EDXS analyses of the oxide scales formed on Kanthal A1 and Kanthal AF (Figs. 9 and 10) show that mainly aluminium and oxygen are detected in the intermediate and the outer parts of the layer. In the inner part, low intensity peaks of chromium and iron are detected, which is consistent with a contribution of the metal to the X-ray emission in the spotted region. Within the detection limit of EDXS (about 0.5 at.%), yttrium was not detected in the oxide scale formed on Kanthal AF. This corroborates the XRD results and suggests that yttrium compounds, if exist in Kanthal AF, are present in very low amounts. The EDXS analyses performed on the yttrium-implanted alloy (Fig. 11) clearly reveal the presence of the reactive element in the oxide scale. The most intense YLa and YLb peaks are relative to X-ray emission from the inner part of the oxide layer. This is consistent with a more likely presence of yttrium at the metal–oxide interface. Peaks attributed to iron and chromium are detected whatever the spotted region of the scale. Although the presence of such peaks derived from EDXS analyses in the inner and the intermediate parts of the oxide may partly be explained by the significant contribution of the metal to the X-ray emission (related to the low oxide scale thickness), their identification in the outermost part of the scale cannot primarily be accounted by the influence of the underlying matrix. This iron and chromium detection at the oxide– gas interface, together with a poor alumina concentration (the AlKa peak intensity is strongly reduced in this outermost oxide region) is consistent with an outer part of the oxide layer mostly composed of iron–chromium mixed oxides.
4. Discussion GAXRD analyses performed on the yttrium-implanted Kanthal A1 indicate, besides some preferred orientation of the metal, the presence of a nanocrystalline
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1825
Fig. 9. EDXS spectra of the oxide scale formed on Kanthal A1: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
phase. This phase, which is present throughout the implanted region, is not due to a surface oxide. A previous study on yttrium-implanted Ni–20Cr reported the presence of a nanocrystalline phase upon initial nucleation [21]. Its author identified, by XRD and EXAFS analyses, this phase to be Y2 O3 . Other studies relating to the yttriumimplantation effects using XRD, GAXRD and RBS also reported the presence of Y2 O3 as the main compound promoted by yttrium implantation on pure iron and various steels [22,23]. These conclusions are somehow different from those reported in a study performed on yttrium-implanted chromium [24]. Their authors stated that a large proportion of yttrium is not contained in the Y2 O3 phase and is thus in ‘‘random’’ positions in the implanted region. With our experimental implantation conditions, the possible formation of yttria in the implanted-specimen would be
1826
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
Fig. 10. EDXS spectra of the oxide scale formed on Kanthal AF: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
related to the presence of oxygen which can appear in the sample by the combination of two phenomena: the sputtering of the naturally oxidized sample surfaces (the sample was not cleaned by sputtering before implantation) during the implantation process and the low oxygen partial pressure existing in the implantation chamber [25,26]. Kinetic results suggest that the oxidation behaviour during the initial stage of the oxidation process is strongly dependent on the introduction mode of yttrium in the alloy: the oxidation rate is only significantly reduced when yttrium is implanted in the alloy, probable consequence of the formation of an oxide scale immediately acting as an effective diffusion barrier. In situ X-ray diffraction analyses performed during this oxidation period revealed drastic changes in the nature of the main oxide
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1827
Fig. 11. EDXS spectra of the oxide scale formed on yttrium-implanted Kanthal A1: metal–oxide interface (a), intermediate part of the scale (b), oxide–gas interface (c).
compounds formed, consecutive to yttrium implantation in the alloy: the oxide scales developed in the first 6 h on Kanthal A1 and Kanthal AF are mainly composed of aAl2 O3 whereas oxidation of yttrium-implanted Kanthal A1 leads to the formation of YAlO3 , (Fe,Cr)2 O3 and a-Al2 O3 , with a much greater amount of the yttrium– aluminium mixed oxide. These results must be approached by taking into accounts general considerations regarding the transient stage of oxide development on FeCrAl alloys. During the early stage of oxidation, transient oxides, rich in iron and chromium are usually formed before the development of the a-Al2 O3 layer at the inner interface. The formation of the a-Al2 O3 layer beneath the transient oxides arises from the selective oxidation of aluminium which occurs with the resulting reduction in oxygen potential at the oxide–alloy interface [27]. In the present study, the (Fe,Cr)2 O3 phase probably forms during the very early stage of oxidation and
1828
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
remains on the surface of the growing oxide layer. The formation of these (Fe,Cr)2 O3 oxides during the initial transient oxidation stage of yttrium-bearing aluminaforming alloys has been previously reported in other studies [28,29]. As the iron– chromium mixed oxide is only detected in the yttrium-implanted specimen, we can suppose that the (Fe,Cr)2 O3 enrichment and the formation of the yttrium–aluminium mixed oxides (YAlO3 ) are closely linked. Indeed, the yttrium–aluminate formation may causes the thin alloy surface layer to be aluminium depleted in the early stage of oxidation [2]. However, there is no reliable indication suggesting any possible influence of this (Fe,Cr)2 O3 formation on the oxidation kinetic in the early stage of oxidation. We only observed that the scale growth kinetic of the yttriumimplanted alloy, in the initial oxidation stage, is dominated by diffusion of reactants through a growing compact scale. In the presence of oxygen, yttrium has a very high affinity to aluminium, leading to the formation of yttrium aluminates (YAlO3 and/or Y3 Al5 O12 ), which have a high thermodynamic stability. In the present study, the implanted yttrium was found to incorporate in the scale in the perovskite structure (YAlO3 ), in the alumina-rich part of the scale (at the metal–oxide interface). This yttrium–aluminium mixed oxide, which forms during the very first oxidation hours, provides from aluminium reaction with oxygen in the presence of yttria, according to the reaction [30]: Y2 O3 þ Al2 O3 ! 2YAlO3
ð1Þ
The presence of YAlO3 particles within the scale of oxidized yttrium-doped aluminaforming alloys is in agreement with other studies [31–34]. It is often suggested [31,35] that yttrium diffuses rapidly along the metal–oxide interface and segregates to the alumina grain boundaries. Other authors have detected the presence of this perovskite, together with yttria, in the outermost region of an oxidized yttrium-doped alumina-forming alloy [33]. This was attributed to internal oxidation which occurred below the alumina–alloy interface, particularly in the alloy grain boundaries. Other studies concluded on the internal oxidation of yttrium-doped NiAl [36] or yttriumimplanted pure iron [37], leading to the formation of yttrium-containing oxide particles (respectively Al2 Y4 O9 and Fe2 YO4 ) located inside the metal. In the present work, such an internal oxidation of yttrium in the alloy during the oxidation process is possible, owing to the very high intensity observed for the YAlO3 (1 0 0) peak (JCPDS file relative intensity: 100%) during in situ X-ray diffraction. The steady-state stage of oxidation corresponds, for both alloys, to an oxidation process limited by diffusion through the growing oxide scale. The mass gain curves clearly show a marked influence of yttrium on the scale growth rate during this oxidation stage. The addition of the reactive element leads to a reduction of the parabolic rate constant by a factor of 10. Furthermore, the in situ X-ray diffraction analysis revealed that transition alumina is not detected in the oxide scales developed on the yttrium-doped alloys. These structural analyses correlated with kinetic studies suggest that the effect of the reactive element, by hampering the formation of transition aluminas or by leading to an acceleration of the phase transformation of transition aluminas into stable alumina, promotes the formation of a protective aAl2 O3 scale. This influence of the reactive element on the rate of transformation of
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
1829
metastable alumina into a-alumina at temperatures below 1273 K is still a matter of discussion. Some authors have reported that yttrium addition delays the phase transformation [12,38] whereas other investigations led to opposite conclusions, i.e. an acceleration of the rate of transformation [15,39]. A possible explanation of these various existing controversies is proposed by Jedlinski [40]. This author suggests that the influence of the reactive element can be accounted for in terms of several mechanisms, which lead to an acceleration or a retardation of the phase transformation. The relative contribution of each of these mechanisms depends on the amount and form of the reactive element [40]. Our results support the assumption that the involved mechanisms favour the formation of a-Al2 O3 . This leads to a reduced mass gain since transition aluminas are less protective and faster growing than a-alumina [6–9]. Yttrium may affect the properties of unstable aluminas and their transformation into a-alumina by virtue of microstructural effects, involving additional sites for heterogeneous nucleation of the oxide as well as hampering the grain growth of unstable aluminas [40]. Earlier nucleation of the oxide layer results in accelerated evolution of the scale, and consequently, in earlier formation of the stable alumina [40]. Chemical effects involving injection of vacancies into unstable aluminas, due to the formation of mixed yttrium–aluminium oxides in the scale, can also account for the influence of yttrium in promoting the growth of stable aalumina [40]. Some authors also suggested that yttrium promotes the selective oxidation of aluminium and facilitates the development of a-Al2 O3 scales, probably by acting as an additional getter for oxygen in the alloy [1]. Other studies have reported that the presence of titanium in alumina forming alloys could play a role in the phase transformation of transition alumina to a-alumina [6,12,13]. Small additions of this element to the alloy would cause an acceleration of the phase transformation resulting in an early formation of stable a-Al2 O3 . Since our study results are in agreement with the above assumption (Kanthal AF alloy contains the greatest amount of Ti and does not reveal the presence of transition alumina in the oxide scale), a possible effect of Ti on the phase transformation of alumina cannot be neglected. It is well established that the growth mechanism of a-alumina involves both inward oxygen diffusion and outward cation diffusion via short-circuit paths such as grain boundaries [41]. Some authors proposed that yttrium modifies the transport properties along the grain boundaries in a-alumina scales [42,43]. These studies suggest that the lower parabolic rate constant observed with addition of yttrium is related to the reduction of the aluminium outward transport through the oxide scale. The formation on the yttrium-free specimen surface of whiskers protruding towards the atmosphere may likely be due to a relatively fast diffusion of aluminium ions through the preformed layer towards the surface. It has been also reported that reactive elements have an important influence on the morphology of the scale. Several works have indicated that these elements can modify the grain size in the oxide scale. In many cases, it is mentioned that yttrium induces a decrease of the grain size of the alumina scale [20,44,45]. This indicates a change in the oxide growth process, which can affect the oxide growth rate and the mechanical properties (spallation resistance) of the oxide scale. Recently, the ‘‘dynamic segregation model’’
1830
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831
[46] established that the reactive element particles are not stagnant dopants in the grain boundaries but diffuse outward during scale growth, thus suggesting that the effect of the reactive element on the growth mechanism and on the scale microstructure is attenuated as the temperature increases. In the present study, the above theory of physical blocking of cation diffusion may rather apply to the effect of implanted-yttrium since the presence of the reactive element in the oxide scale formed on the yttrium-alloyed specimen is probably very limited.
5. Conclusions This study suggests that the oxidation behaviour of the yttrium-doped FeCrAl alloys closely depends on the introduction mode of yttrium in the alloy. On the one hand, the kinetic curves of the yttrium-free and the yttrium-alloyed specimens exhibit an initial transient stage during the first 6 h, followed by a parabolic regime. On the other hand, the scale growth kinetic of the yttrium-implanted alloy obeys a parabolic rate law with two subsequent stages with a much lower rate constant during the second stage. Thus, during the early stage of oxidation, the oxide scale formed on the yttrium-implanted alloy immediately acts as a diffusion barrier and the oxidation rate is significantly reduced. In situ X-ray diffraction analyses performed during this oxidation period revealed drastic changes in the nature of the main oxide compounds formed, consecutive to yttrium implantation in the alloy: besides the formation of a-Al2 O3 , which is also effective on Kanthal A1 and Kanthal AF, oxidation of yttrium-implanted Kanthal A1 leads to the formation of YAlO3 and (Fe,Cr)2 O3 . The steady-state stage of oxidation corresponds, for both alloys, to an oxidation process limited by diffusion through the growing oxide scale. The mass gain curves clearly show a marked influence of yttrium on the scale growth rate during this oxidation stage. The addition of the reactive element by ion implantation or alloying addition leads to a reduction of the parabolic rate constant by a factor of 10. The in situ X-ray diffraction study revealed that this reduction of the oxidation rate is related to the absence of transition alumina in the oxide layer formed on the yttrium-containing alloys. This phenomenon, which has been established as part of the reactive element effect on the oxidation of alumina forming alloys, directly reflects the influence of yttrium in promoting the nucleation and growth of a-alumina. As a consequence, the formation of a protective oxide scale occurs earlier on the yttrium-containing alloys than on the yttrium-free alloy.
References [1] F.A. Golightly, F.H. Stott, G.C. Wood, Oxid. Met. 10 (1976) 163. [2] W.J. Quadakkers, J. Jedlinski, K. Schmidt, M. Krasovec, G. Borchardt, H. Nickel, Appl. Surf. Sci. 47 (1991) 261. [3] H.J. Schmutzler, H. Viefhaus, H.J. Grabke, Surf. Interf. Anal. 18 (1992) 581. [4] P.Y. Hou, J. Stringer, Oxid. Met. 38 (1992) 323. [5] C. Mennicke, E. Schumann, C. Ulrich, M. Ruhle, Mater. Sci. Forum 251–254 (1997) 389.
R. Cueff et al. / Corrosion Science 45 (2003) 1815–1831 [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46]
1831
B.A. Pint, M. Treska, L.W. Hobbs, Oxid. Met. 47 (1997) 1. K.D.D. Lagerlof, B.J. Pletka, T.E. Mitchell, A.H. Heuer, Radiat. Defects 74 (1983) 87. R. Prescott, M.J. Graham, Oxid. Met. 38 (1992) 73. K. Messaoudi, A.M. Huntz, B. Lesage, Mater. Sci. Eng. A 247 (1998) 248. R. Prescott, D.F. Mitchell, J.W. Fraser, M.J. Graham, J. Phys. IV 3 (1993) 301. E. Schumann, Oxid. Met. 43 (1995) 157. B.A. Pint, J.R. Martin, L.W. Hobbs, Solid State Ionics 78 (1995) 99. K.M.N. Prasanna, A.S. Khanna, R. Chandra, W.J. Quadakkers, Oxid. Met. 46 (1996) 465. H.J. Grabke, Intermetallics 7 (1999) 1153. I. Rommerskirchen, V. Kolarik, Mater. Corros. 47 (1996) 625. J.M. Herbelin, M. Mantel, Mater. Sci. Forum 251–254 (1997) 349. V. Kolarik, H. Fietzek, W. Engel, B. Eltester, H.J. Grabke, Mater. Sci. Forum 251–254 (1997) 251. C. Mennicke, E. Schumann, M. Ruhle, R.J. Hussey, G.I. Sproule, M.J. Graham, Oxid. Met. 49 (1998) 455. B.A. Pint, L.W. Hobbs, J. Electrochem. Soc. 141 (1994) 2443. M.W. Brumm, H.J. Grabke, Corros. Sci. 33 (1992) 1677. T.A. Crabb, Ph.D. Thesis, University of Strathclyde, 1993. E. Caudron, H. Buscail, R. Cueff, Y.P. Jacob, M.F. Stroosnijder, Thin Solid Films 350 (1999) 168. E. Caudron, H. Buscail, V.A.C. Haanapel, Y.P. Jacob, C. Josse-Courty, M.F. Stroosnijder, Appl. Surf. Sci. 153 (2000) 156. M.J. Cristobal, P.N. Gibson, M.F. Stroosnijder, Corros. Sci. 38 (1996) 805. E. Caudron, H. Buscail, Corros. Sci. 43 (2001) 1477. E. Caudron, F. Riffard, H. Buscail, R. Cueff, C. Issartel, S. Perrier, Eur. Phys. J. AP 17 (2002) 87. F.A. Golightly, G.C. Wood, F.H. Stott, Oxid. Met. 14 (1980) 217. S. Weinbruch, A. Anastassiadis, H.M. Ortner, H.P. Martinz, P. Wilhartitz, Oxid. Met. 51 (1999) 111. P. Perez, V.A.C. Haanapel, M.F. Stroosnijder, Surf. Coat. Technol. 139 (2001) 207. H.G. Jung, K.Y. Kim, Oxid. Met. 49 (1998) 403. T.A. Ramanarayanan, M. Raghavan, R.P. Luton, J. Electrochem. Soc. 131 (1984) 923. F. Clemendot, J.M. Gras, J.C. van Duysen, J. Phys. IV 3 (1993) 291. P. Choquet, R. Mevrel, Mater. Sci. Eng. A 120 (1989) 153. N. Czech, M. Juez-Lorenzo, V. Kolarik, W. Stamm, Surf. Coat. Technol. 108–109 (1998) 36. M.J. Graham, J.I. Eldrige, D.F. Mitchell, R.J. Hussey, Mater. Sci. Forum 43 (1989) 207. E. Schumann, J.C. Yang, M.J. Graham, Script. Mater. 9 (1996) 1365. C. Josse-Courty, H. Buscail, M.F. Stroosnijder, P. Dufour, J.P. Larpin, Oxid. Met. 52 (1999) 321. Y. Saito, B. Onay, T. Maruyama, J. Phys. IV 3 (1993) 217. H.J. Grabke, Mater. Sci. Forum 251–254 (1997) 149. J. Jedlinski, Oxid. Met. 39 (1993) 55. F.H. Stott, Mater. Sci. Forum 251–254 (1997) 19. B.A. Pint, J.R. Martin, L.W. Hobbs, Oxid. Met. 39 (1993) 167. B.A. Pint, A.J. Garratt-Reed, L.W. Hobbs, Mater. High Temp. 13 (1995) 3. M.K. Loudjani, C. Haut, J. Eur. Ceram. Soc. 16 (1996) 1099. M.K. Loudjani, C. Haut, S. Parisot, Radiat. Effects Defects Solids 134 (1995) 233. B.A. Pint, Oxid. Met. 45 (1996) 1.