- Email: [email protected]
Residual stresses in oxide scale formed on Fe–17Cr stainless steel
Applied Surface Science 316 (2014) 108–113
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Residual stresses in oxide scale formed on Fe–17Cr stainless steel Ning Li a,∗ , Ji Xiao a , Nathalie Prud’homme a , Zhe Chen b , Vincent Ji a a b
ICMMO/SP2M UMR CNRS 8182, Université Paris-Sud, bat. 410, 91405 Orsay Cédex, France State Key Laboratory of Metal Composites, Shanghai Jiao Tong University, 200240 Shanghai, China
a r t i c l e
i n f o
Article history: Received 28 April 2014 Received in revised form 12 July 2014 Accepted 31 July 2014 Available online 12 August 2014 Keywords: Solid oxide fuel cells (SOFCs) Stainless steels Oxidation High temperature Residual stresses X ray diffraction
a b s t r a c t The purpose of this work was to investigate residual stresses in the oxide scale formed on ferritic stainless steel, which is proposed to be used as interconnector in the planar solid oxide fuel cells (SOFCs). The oxidation of the alloy has been conducted at 700 ◦ C, 800 ◦ C and 900 ◦ C for 12–96 h by thermal gravimetric analysis (TGA) system. The oxide surface morphology, cross-section microstructure and the chemical composition of the oxide scale were studied after oxidation, and the residual stresses distribution of the oxide scale were determined at room temperature. It has been found that the oxide scale composed of an inner Cr2 O3 layer and an outer Mn1.5 Cr1.5 O4 spinel layer, the residual stresses in both oxide layers are compressive and the growth stresses plays an important role. The competition of the stresses generation and relaxation during oxidation and cooling affects the residual stresses level. The evolution of residual stresses in the two layers is different according to the oxidation temperature, and the stresses in the two layers are interactional. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Solid oxide fuel cells (SOFCs) are electrochemical devices that convert fuel and oxidant gases into electricity without harmful emissions [1], the reduction of the SOFCs operating temperature from 1000 ◦ C to about 800 ◦ C in recent years widens the materials choice for stack component such as the interconnects [2–6]. SOFC interconnect materials should have high electronic conductivity, good stability in oxidizing and reducing atmospheres and thermal expansion coefficient (TEC) that matches other components of the fuel cell [3]. Ferritic stainless steels became the most promising candidate as interconnects since it exhibits significant advantages, such as good electrical conductivity, appropriate thermal expansion coefficient and low cost [2,3]. Plenty of research works based on the study of oxidation for ferritic steel at high temperature have been reported before, but all of those works were mainly focused on oxidation kinetics [2–5]. However, in the oxide scale systematically accompanied the development of growth stresses during isothermal oxidation and of thermal stresses during cooling, which also limit the lifetime of the alloy [7–10]. Although some research works already performed for Ni–Cr alloy shows that residual stresses mainly come from thermal stresses and the oxide growth stresses were negligible [11–13],
∗ Corresponding author. Tel.: +33 0 169153202. E-mail address: [email protected] (N. Li). http://dx.doi.org/10.1016/j.apsusc.2014.07.195 0169-4332/© 2014 Elsevier B.V. All rights reserved.
very few works focused on internal stresses for ferritic stainless steels till now. The aim of the present study is to investigate experimentally the residual stresses distribution in the oxide layers after high temperature oxidation of AISI 430 ferritic stainless steel.
2. Experimental method The AISI 430 ferritic stainless steel used in this work was supplied by PX Precimet SA, the chemical composition is given in Table 1. For the thermal gravimetric analysis (TGA), the samples were cut into a dimension of 10 mm × 10 mm × 1 mm, and a small hole of 0.8 mm in diameter was made near an edge to hang them by Platinum cricket in the TGA system. All samples were polished by SiC paper followed by silicon paste (up to 2 m) to make sure that the samples have the same surface condition. The TGA (model SETARAM 92-16.18) was used to study the oxidation kinetics, the experiments were carried out at 700 ◦ C, 800 ◦ C and 900 ◦ C for 12–96 h in artificial air (20%O2 + 80%N2 ). Each oxidation condition has been preformed with at least 2 samples, the results were reproducible. After oxidation, the surface morphology, composition of the oxide scale and the cross-section microstructure were examined by using FEG-SEM (ZEISS SUPRA 55VP) equipped with an EDX system (Energy-dispersive X-ray spectroscopy). The oxide phase identification were carried out by grazing incidence X-ray diffraction (GIXRD) technique using Panalytical X’Pert under Copper radiation
N. Li et al. / Applied Surface Science 316 (2014) 108–113
(K␣ = 0.154 nm) with fixed incident angle of 0.5◦ , 2◦ and by normal XRD. For residual stresses analysis in oxide layers, European standard NF EN 15305 (version of April 2009) [14] has been applied with XRD method [15]. The {1 0 4} peaks for Cr2 O3 , {3 1 1} peaks for Mn1.5 Cr1.5 O4 and {1 1 0} peaks for substrate were selected to determine the residual stresses level because these peaks provide sufficient intensities, and more than 13 peaks with different Psi angles (varying from −60◦ to +60◦ ) were recorded. Because of high resolution configuration of XRD system, the determination of peak position error is less than 0.005%, irrespective to the 2 position. 3. Results and discussion 3.1. Oxidation kinetics Fig. 1a shows the mass gain curve as a function of isothermal oxidation duration at 700 ◦ C, 800 ◦ C and 900 ◦ C, it is clear that all the oxidation rates obeyed a parabolic law. The parabolic rate constant, kp , were 3.7 × 10−15 g2 cm−4 s−1 , 7.5 × 10−14 g2 cm−4 s−1 and 1.4 × 10−12 g2 cm−4 s−1 for 700 ◦ C, 800 ◦ C and 900 ◦ C, respectively. According to Wagner theory, a parabolic oxidation rate indicates the formation of a protective scale acting as a diffusion barrier and the oxidation kinetics is controlled by ions diffusion through the oxide scale [16]. And Fig. 1b shows a linear relationship between
109
Table 1 Chemical composition of AISI 430 stainless steel (weight %). Fe
Cr
Mn
Si
P
S
C
Bal.
16–18
<1.00
<1.00
<0.04
<0.03
<0.08
log kp and 1/T at different oxidation temperature, and the activation energy is 279.9 kJ/mol which is comparable to that obtained by M. Palcut et al. [17]. 3.2. Microstructure and composition of the scales The scale composition and microstructure were studied at room temperature by using XRD/GIXRD and FEG-SEM/EDX. Fig. 2a shows the 2◦ GIXRD pattern of AISI 430 stainless steel oxidized for 96 h at 700 ◦ C, 800 ◦ C and 900 ◦ C, which indicates the formation of Cr2 O3 (JCPDS 38-1479) and spinel-type Mn1.5 Cr1.5 O4 (JCPDS 33-0892) on the alloy surface under different temperature. Fig. 2b shows 0.5◦ , 2◦ GIXRD and normal XRD pattern of AISI 430 after 48 h oxidation at 900 ◦ C. In the 0.5◦ pattern, only spinel phase was detected. Along with this, Cr2 O3 was detected in 2◦ and normal pattern, so we can assume that the spinel phase is in the outer part of the scale and the Cr2 O3 is in the inner part of the scale. And there are no obvious texture effects, because all XRD intensities correspond to theoretical proportions
Fig. 1. Weight gain curves (a) and log kp as a function of 1000/T (b) of AISI 430 during exposure at. 700 ◦ C, 800 ◦ C and 900 ◦ C
Fig. 2. 2o GIXRD pattern for AISI 430 oxidized at 700 ◦ C, 800 ◦ C and 900 ◦ C for 96 h (a), 0.5◦ , 2◦ GIXRD and normal XRD pattern of AISI 430 after 48 h oxidation at 900 ◦ C (b).
110
N. Li et al. / Applied Surface Science 316 (2014) 108–113
Fig. 3. Surface morphology of scale on the substrate after oxidation.
Fig. 4. Cross section of AISI 430 after oxidation at 900 ◦ C for 48 h.
Fig. 3 displays the surface morphology of the oxide scale on the substrate after oxidation. With the EDX analyses, we observed that the scale is mainly consisted of Cr, Mn and oxygen. After oxidation at 700 ◦ C for 96 h (Fig. 3a), some small grains formed on the surface, especially at the grain boundaries. After oxidation at 800 ◦ C for 96 h (Fig. 3b) and 900 ◦ C for 12 h (Fig. 3c), the surface was nearly covered by the small grains, and after oxidation at 900 ◦ C for 24 h (Fig. 3d) continuous oxide scale was formed, undulations were also observed after 48 h (Fig. 3e) and 96 h (Fig. 3f) oxidation.
Observation of the cross section after oxidation at 900 ◦ C for 48 h by FEG-SEM (Fig. 4) shows that the oxide scale consisted of two layers, EDX analyses indicates that the outer layer is Mn1.5 Cr1.5 O4 and the inner layer is Cr2 O3 . This is because the oxidation procedure was followed by three steps, Fig. 5 schematically illustrates the steps of the oxidation. At the step 1, the oxidation was mainly controlled by the outward diffusion of Cr3+ because of selective oxidation, and a protective Cr2 O3 oxide layer was formed. After that at step 2, the oxidation was mainly controlled by the outward diffusion of Mn3+
Fig. 5. Schematic illustration of oxidation stages of AISI 430.
N. Li et al. / Applied Surface Science 316 (2014) 108–113
111
Fig. 6. Surface morphology of scale on the substrate after oxidation at 700 ◦ C.
which result into the formation of an outer spinel layer, Fig. 6 shows that with the increase of oxidation time the spinel grains increased on the surface, this occurs as the Mn ion diffuses faster than Cr3+ via Cr3+ lattice in Cr2 O3 [2]. Since the percentage composition of Mn in the alloy is small, that’s why at step 3 the oxidation was controlled by the diffusion of Cr3+ again, which makes the Cr2 O3 layer thicker. In the meantime, this can also explain that the relative high GIXRD intensity of Mn1.5 Cr1.5 O4 peaks occurs due to the presence of the Mn1.5 Cr1.5 O4 phase in outer part of the scale instead of the higher proportion of Cr2 O3 in the scale.
3.3. Residual stresses analysis The residual stresses (RS) in the chromia layer and in the spinel layer were determined after oxidation at room temperature. The Young’s modulus and Poisson’s ratio used for the residual stresses calculation were 280 GPa and 0.29 for Cr2 O3 [13,15], 250 GPa and 0.27 for Mn1.5 Cr1.5 O4 [18], 220 GPa and 0.28 for substrate, respectively. The d vs. f (sin2 ) plots were shown in Fig. 7. For oxide phases, the anisotropic factor has been fixed to be one because of the lack of information in bibliography. The results of the RS in oxide scale after oxidation from 12 h to 96 h at 900 ◦ C were shown in Fig. 8a, this indicates that the RS are compressive, and as the time goes on, the changing trends of the RS in the spinal layer and the chromia layer are similar at 900 ◦ C, this may occur due to the good adherence between the two oxide layers. It is known that the residual stresses in the oxide scale are mainly come from growth stresses during oxidation and thermal stresses during cooling after oxidation. If ignoring the small changes of the oxide layers thickness (the thickness of the two oxide layers after oxidation at 900 ◦ C were given in Fig. 8a), the thermal stresses should be close under same temperature after different oxidation duration according to the equation [19,20], th =
E − ˛substrate ) T (˛ 1 − scale
(1)
where E is the Young’s modulus, is the Poisson’s ratio, ˛scale and ˛substrate are the thermal expansion coefficients of the scale and the substrate, respectively, and T is the temperature difference of the oxidation temperature and the room temperature. For this two layer and one substrate system, because of the great difference of thickness between oxide layers and substrate, and according to elastic mechanic assumption, the elastic deformation is considered identical between the two oxide layer, the deformation in substrate is of same quantity but in opposite direction, and the associated stress in each oxide layer is a function of its own elastic modulus. In the present case, ˛chromia = 9.6 × 10−6 K−1 [2], ˛spinel = 7.5 × 10−6 K−1 [2] and ˛substrate = 12.6 × 10−6 K−1 [21]. So the fluctuation of the stresses value at the same oxidation temperature indicates the competition of stresses generation and relaxation during oxidation [22]. From 12 h to 24 h the RS increased, especially in the spinel layer, this is because after 12 h oxidation the continuous spinel layer didn’t form as we mentioned before, and there was no contribution of thermal stresses to the total stresses, so the −477 ± 38 MPa stresses is mainly come from growth stresses. After the formation of continuous spinel layer at 24 h, the thermal stresses play a part in the total stresses, thus lead to the increase of stresses obviously. The calculated thermal stress in the spinel layer is −1528 MPa, but not all the calculated stresses can appear in the total residual stresses level, because there is a accommodation problem between the two oxide layer. The good adherence between the two oxide layers arise the stresses in the Cr2 O3 layer simultaneously. The increase of RS from 48 h to 96 h is mainly originated from growth stresses in the duration when the Cr2 O3 layer thicken after the stresses relaxation at 48 h. All this indicates that the growth stresses generated during oxidation plays an important role in RS at room temperature for AISI 430 stainless steel, it is in contrary to the previous studies of RS in Cr2 O3 for Ni-Cr alloy [11–13]. The Rhines-Wolf model suggests that the growth stresses is generated by the formation of new oxide inside the scale due to counter diffusion of the oxidizing component and oxygen during high temperature oxidation [22–24]. Although oxidation of AISI 430 is mainly controlled by the outward diffusion of Cr and Mn, the
Fig. 7. The d = f (sin2 ) plot of oxide layers after oxidation for 96 h at 900 ◦ C.
112
N. Li et al. / Applied Surface Science 316 (2014) 108–113
Fig. 8. Residual stresses in oxide scale after oxidation at 900 ◦ C for 12–96 h (a), and 700–900 ◦ C for 96 h (b).
inward diffusion of oxygen happens inevitable, thus generating the growth stresses in the oxide scale. Moreover, from 24 h to 48 h, the RS is decrease which indicates that the stresses are relaxed as the oxide scale thicken. The elastic strain energy stored in the scale per unit area is (1 − v) 2 h/E, and the criterion for spallation failure is given in equation [25]:
the author thought that was the result of internal oxidation of the substrate, but in our experiments we didn’t observe internal oxidation zone. Moreover, considering the small compressive stresses still existed after the stresses relief by the undulation of the substrate, we attribute that to the good adherence of the oxide scale and the substrate.
(1 − ) 2 h > Gc E
4. Conclusions
(2)
where E and are the elastic modulus and Poisson’s ratio of the scale, h is the scale thickness, is the equal biaxial residual stresses in the scale and Gc is fracture resistance of the interface. If Gc is high, the compressive stresses can be accommodated by simultaneous deformation of the scale and alloy without spallation. No spallation was found in the oxide surface morphology in present experiments (Fig. 3), the stresses relief may arise mainly due to the undulation of the scale and the substrate. Fig. 8b shows the effect of oxidation temperature on the RS in chromia layer and in spinel layer. It can be seen that the oxidation temperature has different effect on two layers. For the chromia layer, after 96 h oxidation from 700 ◦ C to 800 ◦ C, the RS increased a little, this is mainly due to the increase of thermal stresses. Since the stresses relaxation happens at 900 ◦ C 48 h, the stresses decreased from 800 ◦ C to 900 ◦ C. But for the spinel layer, the compressive RS increased obviously, from 700 ◦ C to 800 ◦ C only the growth stresses lead to increase of stresses, because the continuous oxide layer did not form. While from 800 ◦ C to 900 ◦ C, the stresses increased was due to the combination of growth stresses, thermal stresses and also stresses relaxation. So the influences of the temperature on RS in the two layers are definitely different, especially before the formation of the continuous spinel layer. However, seldom previous theoretical and experimental studies considered this difference, just saw the oxide layers as a whole during calculate and measure the RS [18]. Table 2 lists the RS in substrate after oxidation at 900 ◦ C. The stresses are small and compressive, there is no much difference between the RS in the substrate with increasing oxidation time. It’s in agreement with the result of A.M. Huntz et al. [12] on Nickel, and Table 2 Residual stresses in substrate after oxidation at 900 ◦ C. Condition 900 ◦ C 12 h 900 ◦ C 24 h 900 ◦ C 48 h 900 ◦ C 96 h
RS in substrate (MPa) −80 −90 −95 −75
± ± ± ±
10 10 15 20
The residual stresses analysis on oxidation scales were carried out after oxidation at 700 ◦ C, 800 ◦ C and 900 ◦ C of AISI 430 stainless steel, the basic conclusions from present work can be made as follows: (1) The oxide scale is consisted of a inner Cr2 O3 layer and an outer Mn1.5 Cr1.5 O4 layer, and the residual stresses in both oxide layers are compressive; (2) The competition of the stresses generation and relaxation during oxidation and cooling affects the residual stresses level. The growth stresses during oxidation plays an important role in residual stresses obtained in oxide scale for AISI 430 after oxidation; (3) The influences of the oxidation temperature on residual stresses levels in two oxide layers after oxidation are different, which was unnoticed before, and the two layers can effect each other because of good adherence. For the spinel layer, the thermal stresses contribute to the residual stresses only after the formation of continuous layer. Acknowledgement The authors are grateful for the financial support from the Chinese Scholarship Council. References [1] J. Larminie, A. Dicks, Fuel Cell Systems Explained, 2nd ed., Wiley, 2003. [2] S.J. Geng, J.H. Zhu, Promising alloys for intermediate-temperature solid oxide fuel cell interconnect application, J. Power Sources 160 (2006) 1009–1016. [3] B. Hua, J. Pu, F.S. Lu, J.F. Zhang, B. Chi, L. Jian, Development of a Fe–Cr alloy for interconnect application in intermediate temperature solid oxide fuel cells, J. Power Sources 195 (2010) 2782–2788. ˜ [4] V. Miguel-Pérez, A. Martínez-Amesti, M.L. Nó, A. Larranaga, M.I. Arriortua, Oxide scale formation on different metallic interconnects for solid fuel cells, Corros. Sci. 60 (2012) 38–49. [5] P.P. Edwars, V.L. Kuznetsov, W.I.F. David, N.P. Brandon, Hydrogen and fuel cells: towards a sustainable energy future, Energy Policy 36 (2008) 4356–4362.
N. Li et al. / Applied Surface Science 316 (2014) 108–113 [6] J. Froitzheim, G.H. Meier, L. Niewolak, P.J. Ennis, H. Hattendorf, L. Singheiser, W.J. Quadakkers, Development of high strength ferritic steel for interconnect application in SOFCs, J. Power Sources 178 (2009) 163–173. [7] A.M. Huntz, L. Marcéchal, B. Lesage, R. Molins, Thermal expansion coefficient of alumina films developed by oxidation of a FeCrAl alloy determined by a deflection technique, Appl. Surf. Sci. 252 (2006) 7781–7787. [8] B.R. Barnard, T.R. Watkins, P.K. Liaw, An evaluation of the use of X-ray residual stress determination as a means of characterizing oxidation damage of nickelBased, Cr2 O3 -forming superalloys subjected to various oxidizing conditions, Oxid. Met. 74 (2010) 305–318. [9] S.J. Bull, Modeling of residual stress in oxide scales, Oxid. Met. 49 (1/2) (1998). [10] P.Y. Hou, A.P. Paulikas, B.W. Veal, J.L. Smialek, Thermally grown Al2 O3 on a H2 annealed Fe3 Al alloy: stress evolution and film adhesion, Acta Mater. 55 (2007) 5601–5613. [11] S. Daghigh, J.L. Lebrun, A.M. Huntz, Stresses in Cr2 O3 Scales Developed on Ni30Cr, Trans Tech Publication, Switzerland, 1997. [12] A.M. Huntz, C. Liu, M. Kornmeier, J.L. Lebrun, The determination of stresses during oxidation of Ni: in situ measurements by XRD at high temperature, Corros. Sci. 35 (1993) 989–997. [13] A.M. Huntz, Stresses in NiO, Cr2 O3 and Al2 O3 oxide scales, Mater. Sci. Eng. A 201 (1995) 211–228. [14] European Standard No NF15305, Test Method for Residual Stresses Analysis by X-ray diffraction, 2009. [15] J. Xiao, N. Prud’homme, N. Li, V. Ji, Influence of humidity on high temperature oxidation of Inconel 600 alloy: oxide layers and residual stress study, Appl. Surf. Sci. 284 (2013) 446–452.
113
[16] P. Kofstad, High Temperature Corrosion, Elsevier, Essex, England, 1988. [17] M. Palcut, L. Mikkelsen, K. Neufeld, M. Chen, R. Knibbe, P.V. Hendriksen, Corrosion stability of ferritic stainless steels for solid oxide electrolyser cell interconnects, Corros. Sci. 52 (2010) 3309–3320. [18] W.N. Liu, X. Sun, E. Stephens, M.A. Khaleel, Life prediction of coated and uncoated metallic interconnect for solid oxide fuel cell applications, J. Power Sources 189 (2009) 1044–1050. [19] D. Burgreen, Elements of Thermal Stresses Analysis, C.P. Press, New York, 1971. [20] H.M. Tung, J.F. Stubbins, Incipient oxidation kinetics and residual stress of the oxide scale grown on Haynes 230 at high temperatures, Mater. Sci. Eng. A 538 (2012) 1–6. [21] B. Hua, Y.H. Kong, W.Y. Zhang, B. Chi, L. Jian, The effect of Mn on the oxidation behavior and electrical conductivity of Fe–17Cr alloys in solid fuel cell cathode atmosphere, J. Power Sources 196 (2011) 7627–7638. [22] A.M. Limarga, D.S. Wilkinson, G.C. Weatherly, Modeling of oxidation-induced growth stresses, Scripta Mater. 50 (2004) 1475–1479. [23] B. Panicaud, J.L. Grosseau-Poussard, J.F. Dinhut, Comparison of growth stress measurements with modelling in thin iron oxide films, Appl. Surf. Sci. 252 (2006) 5700–5713. [24] F.N. Rhines, J.S. Wolf, The role of oxide microstructure and growth stresses in the high-temperature scaling of nickel, Metall. Trans. 1 (1970) 1701–1710. [25] N. Birks, G.H. Meier, F.S. Pettit, Introduction to the High-temperature Oxidation of Metals, 2nd ed., C.U. Press, UK, 2006.
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Residual stresses in oxide scale formed on Fe–17Cr stainless steel Ning Li a,∗ , Ji Xiao a , Nathalie Prud’homme a , Zhe Chen b , Vincent Ji a a b
ICMMO/SP2M UMR CNRS 8182, Université Paris-Sud, bat. 410, 91405 Orsay Cédex, France State Key Laboratory of Metal Composites, Shanghai Jiao Tong University, 200240 Shanghai, China
a r t i c l e
i n f o
Article history: Received 28 April 2014 Received in revised form 12 July 2014 Accepted 31 July 2014 Available online 12 August 2014 Keywords: Solid oxide fuel cells (SOFCs) Stainless steels Oxidation High temperature Residual stresses X ray diffraction
a b s t r a c t The purpose of this work was to investigate residual stresses in the oxide scale formed on ferritic stainless steel, which is proposed to be used as interconnector in the planar solid oxide fuel cells (SOFCs). The oxidation of the alloy has been conducted at 700 ◦ C, 800 ◦ C and 900 ◦ C for 12–96 h by thermal gravimetric analysis (TGA) system. The oxide surface morphology, cross-section microstructure and the chemical composition of the oxide scale were studied after oxidation, and the residual stresses distribution of the oxide scale were determined at room temperature. It has been found that the oxide scale composed of an inner Cr2 O3 layer and an outer Mn1.5 Cr1.5 O4 spinel layer, the residual stresses in both oxide layers are compressive and the growth stresses plays an important role. The competition of the stresses generation and relaxation during oxidation and cooling affects the residual stresses level. The evolution of residual stresses in the two layers is different according to the oxidation temperature, and the stresses in the two layers are interactional. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Solid oxide fuel cells (SOFCs) are electrochemical devices that convert fuel and oxidant gases into electricity without harmful emissions [1], the reduction of the SOFCs operating temperature from 1000 ◦ C to about 800 ◦ C in recent years widens the materials choice for stack component such as the interconnects [2–6]. SOFC interconnect materials should have high electronic conductivity, good stability in oxidizing and reducing atmospheres and thermal expansion coefficient (TEC) that matches other components of the fuel cell [3]. Ferritic stainless steels became the most promising candidate as interconnects since it exhibits significant advantages, such as good electrical conductivity, appropriate thermal expansion coefficient and low cost [2,3]. Plenty of research works based on the study of oxidation for ferritic steel at high temperature have been reported before, but all of those works were mainly focused on oxidation kinetics [2–5]. However, in the oxide scale systematically accompanied the development of growth stresses during isothermal oxidation and of thermal stresses during cooling, which also limit the lifetime of the alloy [7–10]. Although some research works already performed for Ni–Cr alloy shows that residual stresses mainly come from thermal stresses and the oxide growth stresses were negligible [11–13],
∗ Corresponding author. Tel.: +33 0 169153202. E-mail address: [email protected] (N. Li). http://dx.doi.org/10.1016/j.apsusc.2014.07.195 0169-4332/© 2014 Elsevier B.V. All rights reserved.
very few works focused on internal stresses for ferritic stainless steels till now. The aim of the present study is to investigate experimentally the residual stresses distribution in the oxide layers after high temperature oxidation of AISI 430 ferritic stainless steel.
2. Experimental method The AISI 430 ferritic stainless steel used in this work was supplied by PX Precimet SA, the chemical composition is given in Table 1. For the thermal gravimetric analysis (TGA), the samples were cut into a dimension of 10 mm × 10 mm × 1 mm, and a small hole of 0.8 mm in diameter was made near an edge to hang them by Platinum cricket in the TGA system. All samples were polished by SiC paper followed by silicon paste (up to 2 m) to make sure that the samples have the same surface condition. The TGA (model SETARAM 92-16.18) was used to study the oxidation kinetics, the experiments were carried out at 700 ◦ C, 800 ◦ C and 900 ◦ C for 12–96 h in artificial air (20%O2 + 80%N2 ). Each oxidation condition has been preformed with at least 2 samples, the results were reproducible. After oxidation, the surface morphology, composition of the oxide scale and the cross-section microstructure were examined by using FEG-SEM (ZEISS SUPRA 55VP) equipped with an EDX system (Energy-dispersive X-ray spectroscopy). The oxide phase identification were carried out by grazing incidence X-ray diffraction (GIXRD) technique using Panalytical X’Pert under Copper radiation
N. Li et al. / Applied Surface Science 316 (2014) 108–113
(K␣ = 0.154 nm) with fixed incident angle of 0.5◦ , 2◦ and by normal XRD. For residual stresses analysis in oxide layers, European standard NF EN 15305 (version of April 2009) [14] has been applied with XRD method [15]. The {1 0 4} peaks for Cr2 O3 , {3 1 1} peaks for Mn1.5 Cr1.5 O4 and {1 1 0} peaks for substrate were selected to determine the residual stresses level because these peaks provide sufficient intensities, and more than 13 peaks with different Psi angles (varying from −60◦ to +60◦ ) were recorded. Because of high resolution configuration of XRD system, the determination of peak position error is less than 0.005%, irrespective to the 2 position. 3. Results and discussion 3.1. Oxidation kinetics Fig. 1a shows the mass gain curve as a function of isothermal oxidation duration at 700 ◦ C, 800 ◦ C and 900 ◦ C, it is clear that all the oxidation rates obeyed a parabolic law. The parabolic rate constant, kp , were 3.7 × 10−15 g2 cm−4 s−1 , 7.5 × 10−14 g2 cm−4 s−1 and 1.4 × 10−12 g2 cm−4 s−1 for 700 ◦ C, 800 ◦ C and 900 ◦ C, respectively. According to Wagner theory, a parabolic oxidation rate indicates the formation of a protective scale acting as a diffusion barrier and the oxidation kinetics is controlled by ions diffusion through the oxide scale [16]. And Fig. 1b shows a linear relationship between
109
Table 1 Chemical composition of AISI 430 stainless steel (weight %). Fe
Cr
Mn
Si
P
S
C
Bal.
16–18
<1.00
<1.00
<0.04
<0.03
<0.08
log kp and 1/T at different oxidation temperature, and the activation energy is 279.9 kJ/mol which is comparable to that obtained by M. Palcut et al. [17]. 3.2. Microstructure and composition of the scales The scale composition and microstructure were studied at room temperature by using XRD/GIXRD and FEG-SEM/EDX. Fig. 2a shows the 2◦ GIXRD pattern of AISI 430 stainless steel oxidized for 96 h at 700 ◦ C, 800 ◦ C and 900 ◦ C, which indicates the formation of Cr2 O3 (JCPDS 38-1479) and spinel-type Mn1.5 Cr1.5 O4 (JCPDS 33-0892) on the alloy surface under different temperature. Fig. 2b shows 0.5◦ , 2◦ GIXRD and normal XRD pattern of AISI 430 after 48 h oxidation at 900 ◦ C. In the 0.5◦ pattern, only spinel phase was detected. Along with this, Cr2 O3 was detected in 2◦ and normal pattern, so we can assume that the spinel phase is in the outer part of the scale and the Cr2 O3 is in the inner part of the scale. And there are no obvious texture effects, because all XRD intensities correspond to theoretical proportions
Fig. 1. Weight gain curves (a) and log kp as a function of 1000/T (b) of AISI 430 during exposure at. 700 ◦ C, 800 ◦ C and 900 ◦ C
Fig. 2. 2o GIXRD pattern for AISI 430 oxidized at 700 ◦ C, 800 ◦ C and 900 ◦ C for 96 h (a), 0.5◦ , 2◦ GIXRD and normal XRD pattern of AISI 430 after 48 h oxidation at 900 ◦ C (b).
110
N. Li et al. / Applied Surface Science 316 (2014) 108–113
Fig. 3. Surface morphology of scale on the substrate after oxidation.
Fig. 4. Cross section of AISI 430 after oxidation at 900 ◦ C for 48 h.
Fig. 3 displays the surface morphology of the oxide scale on the substrate after oxidation. With the EDX analyses, we observed that the scale is mainly consisted of Cr, Mn and oxygen. After oxidation at 700 ◦ C for 96 h (Fig. 3a), some small grains formed on the surface, especially at the grain boundaries. After oxidation at 800 ◦ C for 96 h (Fig. 3b) and 900 ◦ C for 12 h (Fig. 3c), the surface was nearly covered by the small grains, and after oxidation at 900 ◦ C for 24 h (Fig. 3d) continuous oxide scale was formed, undulations were also observed after 48 h (Fig. 3e) and 96 h (Fig. 3f) oxidation.
Observation of the cross section after oxidation at 900 ◦ C for 48 h by FEG-SEM (Fig. 4) shows that the oxide scale consisted of two layers, EDX analyses indicates that the outer layer is Mn1.5 Cr1.5 O4 and the inner layer is Cr2 O3 . This is because the oxidation procedure was followed by three steps, Fig. 5 schematically illustrates the steps of the oxidation. At the step 1, the oxidation was mainly controlled by the outward diffusion of Cr3+ because of selective oxidation, and a protective Cr2 O3 oxide layer was formed. After that at step 2, the oxidation was mainly controlled by the outward diffusion of Mn3+
Fig. 5. Schematic illustration of oxidation stages of AISI 430.
N. Li et al. / Applied Surface Science 316 (2014) 108–113
111
Fig. 6. Surface morphology of scale on the substrate after oxidation at 700 ◦ C.
which result into the formation of an outer spinel layer, Fig. 6 shows that with the increase of oxidation time the spinel grains increased on the surface, this occurs as the Mn ion diffuses faster than Cr3+ via Cr3+ lattice in Cr2 O3 [2]. Since the percentage composition of Mn in the alloy is small, that’s why at step 3 the oxidation was controlled by the diffusion of Cr3+ again, which makes the Cr2 O3 layer thicker. In the meantime, this can also explain that the relative high GIXRD intensity of Mn1.5 Cr1.5 O4 peaks occurs due to the presence of the Mn1.5 Cr1.5 O4 phase in outer part of the scale instead of the higher proportion of Cr2 O3 in the scale.
3.3. Residual stresses analysis The residual stresses (RS) in the chromia layer and in the spinel layer were determined after oxidation at room temperature. The Young’s modulus and Poisson’s ratio used for the residual stresses calculation were 280 GPa and 0.29 for Cr2 O3 [13,15], 250 GPa and 0.27 for Mn1.5 Cr1.5 O4 [18], 220 GPa and 0.28 for substrate, respectively. The d vs. f (sin2 ) plots were shown in Fig. 7. For oxide phases, the anisotropic factor has been fixed to be one because of the lack of information in bibliography. The results of the RS in oxide scale after oxidation from 12 h to 96 h at 900 ◦ C were shown in Fig. 8a, this indicates that the RS are compressive, and as the time goes on, the changing trends of the RS in the spinal layer and the chromia layer are similar at 900 ◦ C, this may occur due to the good adherence between the two oxide layers. It is known that the residual stresses in the oxide scale are mainly come from growth stresses during oxidation and thermal stresses during cooling after oxidation. If ignoring the small changes of the oxide layers thickness (the thickness of the two oxide layers after oxidation at 900 ◦ C were given in Fig. 8a), the thermal stresses should be close under same temperature after different oxidation duration according to the equation [19,20], th =
E − ˛substrate ) T (˛ 1 − scale
(1)
where E is the Young’s modulus, is the Poisson’s ratio, ˛scale and ˛substrate are the thermal expansion coefficients of the scale and the substrate, respectively, and T is the temperature difference of the oxidation temperature and the room temperature. For this two layer and one substrate system, because of the great difference of thickness between oxide layers and substrate, and according to elastic mechanic assumption, the elastic deformation is considered identical between the two oxide layer, the deformation in substrate is of same quantity but in opposite direction, and the associated stress in each oxide layer is a function of its own elastic modulus. In the present case, ˛chromia = 9.6 × 10−6 K−1 [2], ˛spinel = 7.5 × 10−6 K−1 [2] and ˛substrate = 12.6 × 10−6 K−1 [21]. So the fluctuation of the stresses value at the same oxidation temperature indicates the competition of stresses generation and relaxation during oxidation [22]. From 12 h to 24 h the RS increased, especially in the spinel layer, this is because after 12 h oxidation the continuous spinel layer didn’t form as we mentioned before, and there was no contribution of thermal stresses to the total stresses, so the −477 ± 38 MPa stresses is mainly come from growth stresses. After the formation of continuous spinel layer at 24 h, the thermal stresses play a part in the total stresses, thus lead to the increase of stresses obviously. The calculated thermal stress in the spinel layer is −1528 MPa, but not all the calculated stresses can appear in the total residual stresses level, because there is a accommodation problem between the two oxide layer. The good adherence between the two oxide layers arise the stresses in the Cr2 O3 layer simultaneously. The increase of RS from 48 h to 96 h is mainly originated from growth stresses in the duration when the Cr2 O3 layer thicken after the stresses relaxation at 48 h. All this indicates that the growth stresses generated during oxidation plays an important role in RS at room temperature for AISI 430 stainless steel, it is in contrary to the previous studies of RS in Cr2 O3 for Ni-Cr alloy [11–13]. The Rhines-Wolf model suggests that the growth stresses is generated by the formation of new oxide inside the scale due to counter diffusion of the oxidizing component and oxygen during high temperature oxidation [22–24]. Although oxidation of AISI 430 is mainly controlled by the outward diffusion of Cr and Mn, the
Fig. 7. The d = f (sin2 ) plot of oxide layers after oxidation for 96 h at 900 ◦ C.
112
N. Li et al. / Applied Surface Science 316 (2014) 108–113
Fig. 8. Residual stresses in oxide scale after oxidation at 900 ◦ C for 12–96 h (a), and 700–900 ◦ C for 96 h (b).
inward diffusion of oxygen happens inevitable, thus generating the growth stresses in the oxide scale. Moreover, from 24 h to 48 h, the RS is decrease which indicates that the stresses are relaxed as the oxide scale thicken. The elastic strain energy stored in the scale per unit area is (1 − v) 2 h/E, and the criterion for spallation failure is given in equation [25]:
the author thought that was the result of internal oxidation of the substrate, but in our experiments we didn’t observe internal oxidation zone. Moreover, considering the small compressive stresses still existed after the stresses relief by the undulation of the substrate, we attribute that to the good adherence of the oxide scale and the substrate.
(1 − ) 2 h > Gc E
4. Conclusions
(2)
where E and are the elastic modulus and Poisson’s ratio of the scale, h is the scale thickness, is the equal biaxial residual stresses in the scale and Gc is fracture resistance of the interface. If Gc is high, the compressive stresses can be accommodated by simultaneous deformation of the scale and alloy without spallation. No spallation was found in the oxide surface morphology in present experiments (Fig. 3), the stresses relief may arise mainly due to the undulation of the scale and the substrate. Fig. 8b shows the effect of oxidation temperature on the RS in chromia layer and in spinel layer. It can be seen that the oxidation temperature has different effect on two layers. For the chromia layer, after 96 h oxidation from 700 ◦ C to 800 ◦ C, the RS increased a little, this is mainly due to the increase of thermal stresses. Since the stresses relaxation happens at 900 ◦ C 48 h, the stresses decreased from 800 ◦ C to 900 ◦ C. But for the spinel layer, the compressive RS increased obviously, from 700 ◦ C to 800 ◦ C only the growth stresses lead to increase of stresses, because the continuous oxide layer did not form. While from 800 ◦ C to 900 ◦ C, the stresses increased was due to the combination of growth stresses, thermal stresses and also stresses relaxation. So the influences of the temperature on RS in the two layers are definitely different, especially before the formation of the continuous spinel layer. However, seldom previous theoretical and experimental studies considered this difference, just saw the oxide layers as a whole during calculate and measure the RS [18]. Table 2 lists the RS in substrate after oxidation at 900 ◦ C. The stresses are small and compressive, there is no much difference between the RS in the substrate with increasing oxidation time. It’s in agreement with the result of A.M. Huntz et al. [12] on Nickel, and Table 2 Residual stresses in substrate after oxidation at 900 ◦ C. Condition 900 ◦ C 12 h 900 ◦ C 24 h 900 ◦ C 48 h 900 ◦ C 96 h
RS in substrate (MPa) −80 −90 −95 −75
± ± ± ±
10 10 15 20
The residual stresses analysis on oxidation scales were carried out after oxidation at 700 ◦ C, 800 ◦ C and 900 ◦ C of AISI 430 stainless steel, the basic conclusions from present work can be made as follows: (1) The oxide scale is consisted of a inner Cr2 O3 layer and an outer Mn1.5 Cr1.5 O4 layer, and the residual stresses in both oxide layers are compressive; (2) The competition of the stresses generation and relaxation during oxidation and cooling affects the residual stresses level. The growth stresses during oxidation plays an important role in residual stresses obtained in oxide scale for AISI 430 after oxidation; (3) The influences of the oxidation temperature on residual stresses levels in two oxide layers after oxidation are different, which was unnoticed before, and the two layers can effect each other because of good adherence. For the spinel layer, the thermal stresses contribute to the residual stresses only after the formation of continuous layer. Acknowledgement The authors are grateful for the financial support from the Chinese Scholarship Council. References [1] J. Larminie, A. Dicks, Fuel Cell Systems Explained, 2nd ed., Wiley, 2003. [2] S.J. Geng, J.H. Zhu, Promising alloys for intermediate-temperature solid oxide fuel cell interconnect application, J. Power Sources 160 (2006) 1009–1016. [3] B. Hua, J. Pu, F.S. Lu, J.F. Zhang, B. Chi, L. Jian, Development of a Fe–Cr alloy for interconnect application in intermediate temperature solid oxide fuel cells, J. Power Sources 195 (2010) 2782–2788. ˜ [4] V. Miguel-Pérez, A. Martínez-Amesti, M.L. Nó, A. Larranaga, M.I. Arriortua, Oxide scale formation on different metallic interconnects for solid fuel cells, Corros. Sci. 60 (2012) 38–49. [5] P.P. Edwars, V.L. Kuznetsov, W.I.F. David, N.P. Brandon, Hydrogen and fuel cells: towards a sustainable energy future, Energy Policy 36 (2008) 4356–4362.
N. Li et al. / Applied Surface Science 316 (2014) 108–113 [6] J. Froitzheim, G.H. Meier, L. Niewolak, P.J. Ennis, H. Hattendorf, L. Singheiser, W.J. Quadakkers, Development of high strength ferritic steel for interconnect application in SOFCs, J. Power Sources 178 (2009) 163–173. [7] A.M. Huntz, L. Marcéchal, B. Lesage, R. Molins, Thermal expansion coefficient of alumina films developed by oxidation of a FeCrAl alloy determined by a deflection technique, Appl. Surf. Sci. 252 (2006) 7781–7787. [8] B.R. Barnard, T.R. Watkins, P.K. Liaw, An evaluation of the use of X-ray residual stress determination as a means of characterizing oxidation damage of nickelBased, Cr2 O3 -forming superalloys subjected to various oxidizing conditions, Oxid. Met. 74 (2010) 305–318. [9] S.J. Bull, Modeling of residual stress in oxide scales, Oxid. Met. 49 (1/2) (1998). [10] P.Y. Hou, A.P. Paulikas, B.W. Veal, J.L. Smialek, Thermally grown Al2 O3 on a H2 annealed Fe3 Al alloy: stress evolution and film adhesion, Acta Mater. 55 (2007) 5601–5613. [11] S. Daghigh, J.L. Lebrun, A.M. Huntz, Stresses in Cr2 O3 Scales Developed on Ni30Cr, Trans Tech Publication, Switzerland, 1997. [12] A.M. Huntz, C. Liu, M. Kornmeier, J.L. Lebrun, The determination of stresses during oxidation of Ni: in situ measurements by XRD at high temperature, Corros. Sci. 35 (1993) 989–997. [13] A.M. Huntz, Stresses in NiO, Cr2 O3 and Al2 O3 oxide scales, Mater. Sci. Eng. A 201 (1995) 211–228. [14] European Standard No NF15305, Test Method for Residual Stresses Analysis by X-ray diffraction, 2009. [15] J. Xiao, N. Prud’homme, N. Li, V. Ji, Influence of humidity on high temperature oxidation of Inconel 600 alloy: oxide layers and residual stress study, Appl. Surf. Sci. 284 (2013) 446–452.
113
[16] P. Kofstad, High Temperature Corrosion, Elsevier, Essex, England, 1988. [17] M. Palcut, L. Mikkelsen, K. Neufeld, M. Chen, R. Knibbe, P.V. Hendriksen, Corrosion stability of ferritic stainless steels for solid oxide electrolyser cell interconnects, Corros. Sci. 52 (2010) 3309–3320. [18] W.N. Liu, X. Sun, E. Stephens, M.A. Khaleel, Life prediction of coated and uncoated metallic interconnect for solid oxide fuel cell applications, J. Power Sources 189 (2009) 1044–1050. [19] D. Burgreen, Elements of Thermal Stresses Analysis, C.P. Press, New York, 1971. [20] H.M. Tung, J.F. Stubbins, Incipient oxidation kinetics and residual stress of the oxide scale grown on Haynes 230 at high temperatures, Mater. Sci. Eng. A 538 (2012) 1–6. [21] B. Hua, Y.H. Kong, W.Y. Zhang, B. Chi, L. Jian, The effect of Mn on the oxidation behavior and electrical conductivity of Fe–17Cr alloys in solid fuel cell cathode atmosphere, J. Power Sources 196 (2011) 7627–7638. [22] A.M. Limarga, D.S. Wilkinson, G.C. Weatherly, Modeling of oxidation-induced growth stresses, Scripta Mater. 50 (2004) 1475–1479. [23] B. Panicaud, J.L. Grosseau-Poussard, J.F. Dinhut, Comparison of growth stress measurements with modelling in thin iron oxide films, Appl. Surf. Sci. 252 (2006) 5700–5713. [24] F.N. Rhines, J.S. Wolf, The role of oxide microstructure and growth stresses in the high-temperature scaling of nickel, Metall. Trans. 1 (1970) 1701–1710. [25] N. Birks, G.H. Meier, F.S. Pettit, Introduction to the High-temperature Oxidation of Metals, 2nd ed., C.U. Press, UK, 2006.