- Email: [email protected]
Effects of a NiFe co-deposited layer on α-Al2O3 formation by oxidation of a β-NiAl alloy
Surface & Coatings Technology 325 (2017) 673–681
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Effects of a NiFe co-deposited layer on α-Al2O3 formation by oxidation of a β-NiAl alloy Ali Shaaban a,b,⁎, Shigenari Hayashi c, Kazuhisa Azumi d a
School of Materials and Chemical Technology, Tokyo Institute of Technology, Tokyo, Japan Surface Protection and Corrosion Control Lab., Central Metallurgical Research and Development Institute (CMRDI), Helwan, Egypt Advanced High Temperature Oxidation Materials Laboratory, Graduate School of Engineering, Hokkaido University, Sapporo, Japan d Electronic Materials Chemistry Laboratory, Graduate School of Engineering, Hokkaido University, Sapporo, Japan b c
a r t i c l e
i n f o
Article history: Received 6 June 2017 Revised 11 July 2017 Accepted in revised form 13 July 2017 Available online 13 July 2017 Keywords: NiAl alloy α-Al2O3 Oxidation NiFe co-deposition Oxide scale Spinel oxide
a b s t r a c t Effects of Ni metal, Fe metal or NiFe alloy deposits on the oxidation behavior and oxidation products of a β-NiAl alloy at 1000 °C in air were studied. The surface morphologies of the deposited layers and oxide scales were examined by using field emission scanning electron microscope (FE-SEM) and transmission electron microscopy (TEM). The chemical compositions of the deposited layers were determined before and after oxidation by using energy dispersive X-ray (EDX). The deposited layer and the resultant oxidation products were identified by using X-ray diffractometer (XRD). The chemical composition of the deposited layers was found to affect the morphology of the deposits and the final oxidation products. By oxidation at 1000 °C in air, θ-Al2O3 was detected for bare, Ni-coated and Ni-rich coated (Ni17.7 at.%Fe) samples, but was not detected for Fe-coated and Fe-rich coated (Ni72 at.%Fe) samples at all oxidation times. The oxidation mass gain after 100 h on Fe-coated sample was the highest among the coated samples. TEM Cross-sectional images revealed that the grain size of α-Al2O3 on both Fe coated and Fe-rich NiFe coated samples were the smallest among the coated samples. XRD and EDX point analysis confirmed the formation of different multilayered oxide scales on oxidized samples, NiO/ NiAl2O4/Al2O3 on Ni coated, complex spinel Ni(Fe)Al2O4/Al2O3 on Ni-rich NiFe coated, Fe2O3/(Fe, Ni, Al)2O3/αAl2O3 on Fe-rich NiFe coated and Fe2O3/α-Al2O3 on Fe-coated samples, respectively. Introducing Fe or Fe-rich NiFe coating layer prior to the oxidation of a NiAl alloy resulted in suppression of θ-Al2O3 formation and finer grains of α-Al2O3, which accelerated the growth rate of α-Al2O3 scale. Up to 72 at.% of Fe in NiFe coating was found to be beneficial as an oxidation pretreatment for NiAl alloys to obtain a better high temperature oxidation resistance. © 2017 Published by Elsevier B.V.
1. Introduction Oxidation resistance of alumina-forming alloys is promoted by the formation of stable Al2O3 scale in which the diffusion of both oxygen and Al is slowed down. Various kinds of Al2O3 phases form on these alloys such as metastable γ-, δ- and θ-Al2O3 and stable α-Al2O3 phases. Among them α-Al2O3 oxide scale provides an excellent protective property because of its thermodynamic stability and slow growth rate [1,2]. However, at lower temperature and/or at the early stages of oxidation, metastable alumina phases form first then transform to stable αAl2O3. Because of the rapid growth rates of metastable alumina phases are higher than that of α-Al2O3 by about two orders of magnitude [2– 4], direct formation of α-Al2O3 or rapid transformation of metastable
⁎ Corresponding author at: School of Materials and Chemical Technology, Tokyo Institute of Technology, Tokyo, Japan. E-mail address: [email protected] (A. Shaaban).
http://dx.doi.org/10.1016/j.surfcoat.2017.07.031 0257-8972/© 2017 Published by Elsevier B.V.
alumina to stable α-Al2O3 is required to improve the high temperature oxidation resistance. Kitajima et al. reported that the deposition of Cr or Fe coating on Al containing alloys suppressed the formation of metastable θ-Al2O3 phase by oxidation at 900 °C in air, while Ni coating delayed the transformation of metastable θ-Al2O3 to stable α-Al2O3 phase [5]. They proposed that the suppression of the θ-Al2O3 formation was caused by rapid formation of α-Al2O3 phase. The rapid formation of α-Al2O3 was attributed to the formation of coated metal oxides, Cr2O3 and Fe2O3, which have an isomorphous corundum structure with α-Al2O3 [5]. These metal oxides may provide a higher density of nucleation sites for α-Al2O3. We also investigated the growth rate of Al2O3 scales in the oxidation of β-Ni50Al alloy at 1000 °C in air and confirmed that metastable Al2O3 phase was not formed on NiAl alloy with Fe or Cr coatings [6]. These coatings also affected the growth kinetics of Al2O3 scale via affecting the size of α-Al2O3 grains. The α-Al2O3 scale that transformed earlier always consisted of smaller grains, whereas the size of grains
674
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
Table 1 Composition of electrodeposition bath [g dm−3]. Bath
Coating
[NiSO4·6H2O]
[FeSO4·7H2O]
[Na2SO4]
[H3BO3]
H2SO4
a) b) c) d)
Ni Ni-Fe Ni-Fe Fe
15.5 15.5 15.5 0
0 1.4 27.8 27.8
35, 50 35, 50 35, 50 35
12.5 12.5 12.5 12.5
0 0 0 0.1 M
increased with delay in the transformation from θ-Al2O3 to α-Al2O3. Generally the growth rate of alumina scale decreases with increasing the size of Al2O3 grain. An Fe coating promoted the rapid formation of α-Al2O3 as mentioned above resulting in a small grain size. On the other hand Ni coating was found to delay the transformation and thus enlarged the grain size of α-Al2O3 reducing the growth rate of oxide scale compared with the Fe or Cr coatings. From the other viewpoint, however, the rapid transformation of θ-Al2O3 to α-Al2O3 phase is desirable since the shrinkage of the oxide accompanying the transformation causes stresses and thus cracking of the scales particularly in the cyclic oxidation. A composition of coatings also affects the oxidation kinetics. Ni coating did not suppress the formation of θ-Al2O3 phase while Ni3.1 at.%Fe2O3 nano powder composite coating accelerated the transformation of θ- to α-Al2O3 phase resulting in an increased grain size of Al2O3 and a decreased growth rate of Al2O3 scale. Further addition of Fe2O3 of 5.2 and 7.4 at.% to Ni coatings suppressed completely the formation of θ-Al2O3 phase or promoted direct formation of α-Al2O3
phase resulting in a slow growth rate of scales. However, addition of Fe2O3 more than 7.4 at.% did not show any improvement because of agglomeration of Fe2O3 particles. Therefore, an Fe/Ni ratio in the NiFe coatings is a meaningful parameter because higher ratio of Fe content corresponds to higher concentration of Fe2O3 in the oxide that may affect the oxidation behavior. The purpose of this study is therefore to investigate a suitable Fe/Ni ratio of the NiFe precoating that may improve the oxidation resistance of a NiAl alloy at 1000 °C in air. For this purpose, NiFe alloy coatings were electrodeposited on a NiAl alloy samples via pulsed electrodeposition (PED) prior to the oxidation test. Ni and Fe have different electrodeposition rates [8–12] and thus Fe/Ni ratio can be adjusted by controlling the electrodeposition parameters. 2. Materials and methods A β-Ni50 at.%Al alloy ingot was prepared from Al and Ni metals (~ 99.99% in purity) by using the Ar-arc melting technique followed by homogenization at 1200 °C for 48 h in vacuum (5 × 10−3 Pa) and was cut into specimens of ca. 1 mm thickness. These specimens were ground with SiC paper up to 4000 grit and finished with 3 μm diamond paste to obtain a mirror finish surface and then ultrasonically degreased in acetone for 10 min. Ni, NiFe or Fe coating was electrodeposited on the NiAl alloy from different baths adjusted to pH 3 listed in Table 1. The pulsed electrodeposition (PED) was conducted by using the following conditions; current density of icd = − 10 mA·cm− 2, deposition time ton = 2 s,
Fig. 1. Current waveform for galvanostatic PED and potential response measured for electrodeposition of coatings on Ni50Al samples at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C, pH 3, for 600 s in the baths (a)–(d) listed in Table 1 with agitation at 400 rpm.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
resting time toff = 4 s and temperature T = 25 °C for 600 s with stirring rate at 400 rpm by using a potentiostat (Ivium Technologies Co., model Iviumstat) in an electrochemical cell with a Ag/AgCl reference electrode, Ni sheet counter electrode and a working electrode of NiAl sheet with ca. 2 cm2 in area. After electrodeposition, these samples were subjected to high temperature oxidation tests. Samples were placed in a furnace exposed to air and a temperature was elevated from room temperature (RT) to 1000 °C with a rate of 10 °C min−1 and kept at 1000 °C for tox = 0.5, 1, 4, 7, 9, 25, 49 and 100 h and then cooled down to RT. Weight of samples were measured before and after the oxidation tests to determine the mass gain.
675
Surface morphology and chemical composition of electrodeposited coatings were observed by using FE-SEM with EDX (JEOL Co., model 6500F). Phase analysis of coatings was conducted by using X-ray diffractometer XRD (JEOL Co., model JDX 3500, operated at 20 mA and 40 kV with CuKα radiation). TEM (JEOL Co., model JEM 2000FX, operated at 200 kV) was employed to investigate the cross-section morphology, elemental concentrations of Al, Ni, Fe and O in oxide scales, and to identify the oxides. Multi-beam FIB-SEM (JEOL Co., model JIB-4600F/HDK) was used for TEM cross-sectional samples preparation of oxide scales. During sample preparation, a carbon layer or a tungsten layer was pre-deposited on the sample to protect the surface from damage or contamination by Ga+ beam. Then a cross-section sample was lifted
Fig. 2. Surface and cross-sectional morphologies FE-SEM images of (a) Ni, (b) Ni-rich NiFe, (c) Fe-rich NiFe and (d) Fe on a Ni50Al samples, which was deposited for 600 s with pulse conditions at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C.
676
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
out and attached to a Mo support grid that fits into a specimen holder of TEM. 3. Results and discussion 3.1. Properties of Ni, NiFe and Fe coatings Fig. 1 shows the current waveform and potential response at the initial stage of PED from Ni-bath (a), Ni-Fe baths (b)–(c) and Fe-bath (d). When icd was applied at ton the potential was dropped suddenly in a less noble direction to initiate electrodeposition. The potential shift during ton decreased gradually due to increase in a surface area of the deposits. When icd was turned to off at toff, the potential shifted in a noble direction due to recovery of the concentration of Ni2+ or Fe2+ ions near the sample surface in the bath to improve the uniform electrodeposition in the next electrodeposition period ton. Potential at toff in bath (d) was considerably different from those in baths (a)–(c) because of the difference in coating composition as Fe coating deposited from bath (d) while Ni containing coatings deposited from baths (a)–(c). EDX analysis showed the Fe content of 17.7 at.% in a coating electrodeposited from bath (b) and 72 at.% from bath (c). The coatings deposited from baths (a)–(d) are, therefore, named to be Ni coating, Ni-rich NiFe coating, Fe-rich NiFe coating, and Fe coating, respectively, in this paper. Surface and cross-sectional SEM observation confirmed that compact, dense and crack-free coatings with different microstructure were formed on the NiAl substrates as shown in Fig. 2. Ni coating consisted of small facet-structured particles, Ni-rich NiFe coating consisted of fine nodular-structured, Fe-rich NiFe coating consisted of fine fluffy-structured and Fe coating consisted of coarse tetrahedralstructured particles. Average particle size in the coatings was in the order of Fe coating N Ni coating N Ni-rich NiFe coating N Fe-rich NiFe coating, i.e., the particle size of NiFe coatings were smaller than those of Ni and Fe coatings. Thickness of Ni or Fe coating was 0.7–1 μm and 1 μm, respectively, as shown in Fig. 2(a) and (d) while the thickness of NiFe coatings was about 0.8 μm for Ni-rich coating and 0.6 μm for Ferich coating and tended to be thinner than that of Ni or Fe coating as shown in Fig. 2(b) and (c). Although the electrodeposition charge for all coatings was the same. Particle size and thickness of NiFe coatings smaller than those of Ni and Fe coatings could be attributed to the anomalous co-deposition of NiFe alloys as reported in Ref. [8–9,13–16]. Various mechanisms have been reported for NiFe alloy co-deposition. Metal hydroxide formation mechanisms of NiFe alloys were reported in ref. [8–9,13] and confirmed by direct measurements in ref. [14–16]. Hydroxide suppression mechanism by Dahms et al. in which the nobler (Ni) deposition was suppressed by the formation of less noble hydroxide (Fe(OH)2) during electrodeposition process. Fe(OH)2 formation is preferentially occurs in the cathode layer and adsorbed on the cathode [8,9]. Hessami et al. reported another mathematical model in which they focused on the difference in the dissociation or hydrolysis constants between intermediate compounds FeOH+ and NiOH+ [13]. The electrochemical reactions of Ni and Fe at the surface of the electrode (cathode) are given as NiðOHÞþ þ 2e− →Ni þ OH−
ð1Þ
FeðOHÞþ þ 2e− →Fe þ OH−
ð2Þ
Fig. 3. XRD patterns of Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe coated Ni50Al samples, which were deposited for 600 s with pulse conditions at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C.
dissociation constant than Fe(OH)+ (5.8 × 10−8) in the cathode layer. Fe(OH)+ is preferentially formed at the expense of Ni(OH)+, resulting in preferential deposition of Fe to Ni.
The homogeneous equilibrium reactions of Ni and Fe in the electrodeposition system are given as Ni2þ þ OH− →NiðOHÞþ
ð3Þ
Fe2þ þ OH− →FeðOHÞþ
ð4Þ
According to these equations and based on the dissociation mechanism, Ni(OH)+ (4.5 × 10−5) has three order of magnitude higher
Fig. 4. Change of oxidation mass gain of bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fecoated Ni50Al samples in the oxidation tests at 1000 °C in air up to (a) short exposure up to 1 h and (b) long exposure up to 100 h.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
Fig. 3 shows the XRD patterns of electrodeposited samples. Ni coating showed three strong peaks, Ni(111), Ni(200) and Ni(220) of Ni fcc structure. The strongest Ni(111) peak corresponded to the preferential growth orientation. Ni-rich NiFe coating also showed a Ni(111) peak, however, the intensity of Ni(200) and Ni(220) peaks was considerably decreased and shifted slightly to a lower 2θ°. For Fe-rich NiFe coating, the Ni(200) and Ni(220) peaks almost disappeared while strong Fe(110) peak and small Fe(200) and Fe(211) peaks of bcc Fe structure evolved with increasing Fe concentration. These peaks of Fe were observed clearer for Fe coating. 3.2. High temperature oxidation test The bare NiAl alloy, NiAl alloys coated with Ni, Ni-rich NiFe, Fe-rich NiFe and Fe coatings were subjected to high temperature oxidation tests at 1000 °C in air. Oxidation mass gain (Δ m) was obtained from weight change measurement before and after the test for tox = 0.5, 1, 4, 7, 9, 25, 49 and 100 h as shown in Fig. 4(a). Two oxidation stages
677
were observed, i.e., the initial stage with rapid growth rate followed by the second stage with slow and steady growth rate. Fig. 4(b) shows the details of the initial stage. Since the growth rate of the oxide on the coated samples at the initial stage were considerably faster than that of bare NiAl alloy, the growth of NiO on Ni or Ni-rich NiFe coatings and FeO, Fe3O4 or Fe2O3 on Fe-rich NiFe or Fe coatings seemed to be responsible for the rapid growth of scales at the initial stage of oxidation. The oxidation mass gain on the coated samples after 1 h of oxidation was in the order of Fe N Fe-rich N Ni-rich NiFe N Ni-coated NiAl. In the further oxidation up to 100 h shown in Fig. 4(a), growth rate of the oxide scale on Ni, Ni-rich NiFe and Fe-rich NiFe coated samples reduced comparing to the bare NiAl alloy and Fe coated NiAl alloy. Fig. 5 shows the surface morphologies of samples oxidized at 1000 °C for tox = 9 h. An oxide scale of blade-like morphology covered the surface of the bare Ni50Al alloy (Fig. 5(a)). This kind of morphology was typically observed for transient phases of alumina such as θ-Al2O3 [6–7,17–21]. An oxide scale of faceted morphology covered the Ni coated Ni50Al sample (Fig. 5(b)). Mixture of faceted and tetrahedral
Fig. 5. Surface morphology FE-SEM images of (a) bare Ni50Al, (b) Ni-, (c) Ni-rich NiFe, (d) Fe-rich NiFe, and (e) Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
678
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
morphologies covered the NiFe coated Ni50Al samples (Fig. 5(c) and (d)). A tetrahedral-shaped oxide covered the Fe coated Ni50Al (Fig. 5(e)). XRD patterns of bare and coated samples for tox = 9 h oxidation and their phase identification were shown in Fig. 6. Al, Ni and/or Fe were oxidized simultaneously to form Al2O3, NiO and/or Fe-oxide phases. Peaks of metastable θ-Al2O3 and α-Al2O3 phases were observed for bare NiAl sample. Peaks of NiO, spinel NiAl2O4, θ-Al2O3 and α-Al2O3 phases were observed for Ni and Ni-rich NiFe coated samples. Peaks of FeO, α-Fe2O3 and α-Al2O3 phases were observed for Fe and Fe-rich NiFe coated samples. Peaks of θ-Al2O3 phase were detected for bare, Ni and Ni-rich NiFe coated samples, but were not for Fe or Fe-rich NiFe coated samples. Fe3O4 phase was not detected because it could be transformed to Fe2O3 or reacted with Al2O3/NiO to form spinel FeAl2O4 and/or inverse spinel NiFe2O4. XRD patterns of the oxidized Ni-rich NiFe and Fe-rich NiFe coated samples are shown in Fig. 7 as a function of oxidation time tox from 0.5 h to 7 h. At the early stage of oxidation at tox = 0.5 h, NiO and FeO/Fe2O3 predominated the oxide scale on Ni-rich and Fe-rich NiFe coated samples, respectively. By further oxidation up to tox = 7 h, not only single oxide phases but also composite oxide phases were detected. Peaks of α-Al2O3, θ-Al2O3, NiO, spinel FeAl2O4 and inverse spinel NiFe2O4 were observed on Ni-rich NiFe coated sample and peaks of FeO, α-Fe2O3, α-Al2O3, (Fe, Al)2O3 and spinel FeAl2O4 were observed on Fe-rich NiFe coated sample. Metastable Al2O3 phase was not detected for Fe-rich NiFe coated sample. Fig. 8 shows the cross-sectional TEM images of the oxide scale formed on NiAl alloys with/without coating of Ni-rich NiFe, Fe-rich NiFe and Fe layers after oxidation test at 1000 °C for tox = 9 h. On bare NiAl sample, ca. 0.9 μm thick single layer of θ-Al2O3 with needlelike morphology including columnar grains of α-Al2O3 phase at the interface between the oxide layer and substrate alloy was observed as shown in Fig. 8(a). On the other hand a multilayer structure of the scales was confirmed for all the coated samples shown in Fig. 8 (b–e). These scales were thicker (1.5–2.0 μm) than that on the bare NiAl sample due to the fast growth of oxides of the coatings at the initial stage of oxidation as shown in Fig. 4. Composition of each layer of scales was analyzed by EDX and summarized in Table 2 with their multilayer phase. In Fig. 8(b–e), the grain size of α-Al2O3 depended on the kind of coating, i.e., the largest size was found on Ni coating, reduced with increasing in the Fe content and reached the smallest size on Fe coating. In order to investigate the correlation between the overall structure of the oxide scales and the oxidation kinetics of NiAl alloys with/without coating, two stages of oxidation composed of the initial stage (tox = 1 → 9 h) controlled by fast growing oxides formation and the late stage (tox = 9 → 100 h) controlled by slower growing oxides will be discussed below. Fig. 9 shows the distribution of the oxygen uptake (mass gain) to each phase in the oxide scale formed on Ni50Al samples with/without coating of Ni, Ni-rich NiFe, Fe-rich NiFe or Fe after the oxidation at 1000 °C for 9 h in air that was calculated from the results of XRD, TEM and EDS presented in this paper. Fe-coated sample showed the largest oxidation mass gain among all samples, i.e., about 67% of oxygen was taken in Fe oxide. On Fe-rich NiFe coated sample, about 47% of oxygen was taken in Fe oxides and 16% in spinel phases. On Ni-rich coated sample, 67% of oxygen was taken in spinel phase. On Ni coated sample, 25% and 29% of oxygen was taken in NiO and spinel NiAl2O4, respectively. Different distribution of oxygen to the coatings of different Fe/Ni ratios resulted from the difference in the initial oxidation rate and thus affected the total oxidation mass gain in the entire oxidation time. At the initial stage of oxidation (tox = 1 → 9 h) and even during heating, the composition of the coatings plays a strong role in the oxidation kinetics. During heating, oxide phases such as NiO and/or FeO, Fe3O4, Fe2O3 grew until Ni or Fe was consumed. Increase in the Fe/Ni ratio of the coatings resulted in increase in the oxidation mass gain due to the large Gibbs formation energy of Fe2O3 (− 380 kJ mol−1)
Fig. 6. XRD patterns of bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
compared with that of NiO (−280 kJ mol−1). This suggests that Fe2O3 was preferred to be formed than NiO and provided faster growth in the outer layer of the oxide scale at 1000 °C as presented in Fig. 8. The order of oxidation mass gain was thus in the order of Fe N Fe-rich NiFe N Ni-rich NiFe N Ni NiAl alloy as shown in Figs. 4(b) and 9. Due to the slower growth rate of θ-Al2O3 than the growth rate of coating oxides,
Fig. 7. XRD patterns of (a) Ni-rich NiFe and (b) Fe-rich NiFe coated Ni50Al alloy for 0.5 h up to 7 h at 1000 °C in air.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
the oxidation mass gain of the bare Ni50Al was smaller than that on the coated samples. At tox = 9 h, θ-Al2O3 still continued to grow on bare Ni50Al surface accompanied by slight formation of α-Al2O3 grains at θ-Al2O3/NiAl
679
interface. On the other hand, thin layer of α-Al2O3 scales with different grain size was formed on all coated samples. Although the oxide scale formed on Ni or Ni-rich NiFe coated samples contained residual untransformed θ-Al2O3 as detected in XRD (Fig. 6), it could not be observed
Fig. 8. Cross-sectional TEM images of oxide scales formed on (a) bare Ni50Al, (b) Ni-, (c) Ni-rich NiFe, (d) Fe-rich NiFe, and (e) Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
680
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
by cross-sectional TEM observation (Fig. 8(b) and (c)). Therefore, the oxidation rate remained faster, particularly on Ni coated sample on which θ-Al2O3 still growing (Fig. 4(a)). On Ni or Ni-rich NiFe coated samples, solid phase reactions of NiO with Al2O3 might form an intermediate layer of NiAl2O4 spinel between the outer layer of NiO and the inner layer of Al2O3 on samples coated with Ni and Ni-rich NiFe layers as shown in Figs. 6 and 8(b) and (c). On Fe or Fe-rich NiFe coated samples, Al2O3 may react directly with FeO or may dissolve into Fe3O4 resulting in formation of FeAl2O4 spinel as expected from the phase diagram [22]. Beside the NiAl2O4 spinel formation, (Fe, Ni, Al)2O4 spinel may also be formed on the inner layer of Al2O3 on Ni-rich NiFe coated sample as shown in Fig. 8(c). Because of the solubility limit of Fe in α-Al2O3 and of Al in α-Fe2O3, a (Fe, Al)2O3 phase was formed as an independent layer between the α-Fe2O3 and α-Al2O3 layers on Fe-rich NiFe coated sample as shown in Fig. 8(d). Al2O3 could appear in two different forms, i.e. (Fe, Al)2O3 and α-Al2O3, due to formation of (Fe, Al)2O3 solid solution as confirmed from XRD in Ref. [23]. Formation of a duplex Fe2O3 scale composed of an outer layer of relatively pure Fe2O3 and an inner layer of Al saturated Fe2O3 was found by Hayashi et al. [24] and it was also confirmed from TEM observation Fig. 8(d) and EDX point analysis in this study. Since neither spinel FeAl2O4 nor solid solution of (Fe, Al)2O3 layer was observed by TEM observation of oxidized Fe coated sample shown in Fig. 8(e), it can be suggested that the presence of Ni in the coating promotes formation of spinel phase. Based on cationic configurations inside the oxygen sub-lattice, θ-Al2O3 and NiO, FeO (Fe3O4) and/or NiAl2O4 (FeAl2O4) phases are related to the similar oxygen framework of fcc structure. In the O2− framework of θ-Al2O3 phase, ½ of the Al3+ ions occupy octahedral interstitial sites and another ½ occupies tetrahedral sites. Meanwhile, the Ni2+/Fe2+ ions occupy all the octahedral interstitial sites of oxygen framework in NiO/FeO, respectively. In Fe3O4 (magnetite) the oxygen anion O2− form a close-packed fcc sub-lattice with Fe2 + and Fe3+ cations located in interstitial sites. Two different kinds of cation sites exist in the magnetite crystal: tetrahedrally coordinated A sites occupied by Fe3+ and octahedrally coordinated B sites occupied by ½ Fe2+ and ½ Fe3+. The reaction(s) between θ-Al2O3 and NiO/FeO could be represented as, Spinel NiAl2 O4 : ½Ni2þ O þ Al
3þ
½Al
3þ
O3 →Al
3þ
Spinel FeAl2 O4 : ½Fe2þ O þ Al
3þ
½Al
3þ
O3 →Al
3þ
h h
Ni2þ Al
3þ
Fe2þ Al
3þ
i i
O4
ð5Þ
O4
ð6Þ
And the reaction between Fe3O4 and θ-Al2O3 could be represented as, h i Spinel FeAl2 O4 : Fe3þ Fe2þ Fe3þ O4 þ Al
3þ
½Al
3þ
3þ
O3 →7=4Al
h
Fe2þ Al
3þ
i
O4
ð7Þ
Another reaction is expected to occur between the initially formed Fe3O4 and NiO phases, in which Ni may substitute Fe2+ in Fe3O4 (magnetite) to form NiFe2O4 (nickel ferrite) in the cubic inverse spinel structure. The reaction between Fe3O4 and NiO could be represented as, h i Inverse spinel NiFe2 O4 : Fe3þ Fe2þ Fe3þ O4
h i þ ½Ni2þ O→5=4Fe3þ Ni2þ Fe3þ O4
coated sample. Not only α-Al2O3 grain size, but also spinel formation should be considered. Although the growth and its grain size α-Al2O3 controls the oxidation kinetics at the late stage of oxidation, effect of spinel or formation of spinel complex phases cannot be neglected. Growth rates of these spinel oxides were slower than those of the initially formed oxides and as fast as to those of metastable alumina phases, resulting in the reduction of the growth rate of oxides at tox = 25– 100 h (Fig. 4(a)). It was also found in our previous work that the formation of spinel phases such as NiAl2O4 could alter the oxidation kinetics. Existence of such intermediate layer could reduce the oxidation rate regardless the transformation of θ-Al2O3 phase was completed or not as discussed in Ref. [6,7]. In this case, reduction of the oxidation mass gain could be attributed to two factors; (i) slowing down the mobility of cations and anions and/or (ii) reducing the chemical potential gradient in this intermediate spinel layer. Also, the counter diffusion of Ni2+ or Al3+ cations through these spinel oxides is much slower than that through the initially formed oxides. However, after longer oxidation (tox = 100 h), development of a uniform α-Al2O3 layer might suppress the outward diffusion of Ni2+ and thus prevent further growth of NiAl2O4. In such case the oxidation kinetics will be totally dominated only by α-Al2O3 growth rate. Formation of at least one of NiO, FeO, Fe3O4, NiAl2O4, and/or FeAl2O4 phases is believed to delay the transformation of θ- to α-Al2O3 phase [6, 7] while coexistence of Fe2O3 nano-particle in a composite coating [7] or oxidized Fe coating [5,6,21] is believed to promote the formation of αAl2O3 phase. The beneficial effect of α-Fe2O3 phase formation on direct formation of α-Al2O3 might be resulted from (i) providing preferential nucleation sites for α-Al2O3 phase formation rather than θ-Al2O3 [7], and/or (ii) precipitation of α-Al2O3 phase but not metastable Al2O3 phase from Al saturated Fe2O3 [23,24]. According to the results obtained in this study, existence of NiO, FeO or Fe3O4 was confirmed to stabilize θ-Al2O3 and delay the transformation of θ-Al2O3 to α-Al2O3 phase in the long oxidation process. This effect might be beneficial to reduce the oxidation rate. Addition of Fe to Ni was found to promote the transformation of θ-Al2O3 to α-Al2O3 phase in the oxidation process by forming Fe2O3 with a suitable grain size for nucleation of α-Al2O3 phase. Fe coating on NiAl alloy was not beneficial because αAl2O3 phase with a too small grain size was formed that could not reduce the oxidation rate. However, coexisting of Ni and Fe in the coating provided a better high temperature resistance. Optimization of Fe/Ni ratio should be considered the suitable balance between the time for transformation of θ-Al2O3 to α-Al2O3 phase and the time for growth of α-Al2O3. For example, at high temperature, a low Fe/Ni ratio is advantageous to obtain the α-Al2O3 phase with large grain size that reduces effectively the oxidation rate, and, at low temperature, high Fe/Ni ratio is favoured to accelerate the phase transformation from metastable θ-Al2O3 to α-Al2O3 phase. 4. Conclusion PED Ni, Fe or NiFe alloys (Ni-rich NiFe and Fe-rich NiFe) coating layer was introduced as a pretreatment coating prior the oxidation of a NiAl Table 2 Composition in [at.%] and predicted multilayer phases of electrodeposited NiFe alloy coatings shown in Fig. 8. (b)
ð8Þ (c)
where square brackets represent cationic octahedral sites. At the late stage of oxidation (tox = 9–100 h), overall microstructure of the oxide scale especially α-Al2O3 grain size takes over the role in controlling the oxidation kinetics. Growth rate of the oxide scale on bare alloy was faster than those on coated NiAl samples because of the rapid growth of θ-Al2O3 rather than α-Al2O3. The oxidation mass gain of bare alloy was greater than those on coated samples except for Fe-
(d)
(e)
Coating Composition Phase Coating Composition Phase Coating Composition Phase Coating Composition Phase
Ni Ni23.78Al0.65O75.57/Ni5.13Al29.56O65.31/Ni0.07Al38.83O61.10 NiO/NiAl2O4/Al2O3 Ni-rich NiFe Ni0.1Fe3.9Al28.7O67.3 or Ni1.9Fe0.4Al26.6O71.1/Ni0.1Fe0.1Al42.8O57 (Fe, Ni, Al)2O4/(Fe, Ni, Al)2O4/Al2O3 Fe-rich NiFe Fe17.3Al5.3O77.4/Ni2.3Fe9.5Al12.2O76.0/Ni0.1Fe0.1Al39.6O60.2 α-Fe2O3/(Fe, Ni, Al)2O3/α-Al2O3 Fe Fe19.38Al0.95O79.67/Fe0.05Al40.72O59.23 α-Fe2O3/α-Al2O3
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
681
4. The existence of α-Fe2O3 phase results in direct formation of α-Al2O3 because of template effect or because of α-Al2O3 precipitation from Al saturated Fe2O3, without metastable Al2O3 formation. 5. Grain size of α-Al2O3 affects the suppression of θ-Al2O3 formation. The grain size of α-Al2O3 was smaller than those on samples on which θ-Al2O3 still grow. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
Fig. 9. Amount of the oxygen uptake (in mg·cm−2) calculated for oxide phases on bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air based on the results of XRD, TEM and EDS analysis.
[11] [12]
[13] [14] [15]
alloy. Their effects on the oxidation products and the growth kinetics of Al2O3 scale formation on NiAl at 1000 °C in air were investigated. The results are summarized as follows;
[16]
1. The oxidation kinetics depends initially on the growth rate of the oxidized coating, then on the spinel phases formation and finally on the α-Al2O3 scale microstructure.
[18] [19] [20]
2. The formation of either metastable or stable Al2O3 phase depends on the initially formed oxide phases from coating layer. 3. The coexistence of the initially formed NiO, FeO and/or Fe3O4 phases with Al2O3 phase enhance the formation of spinel NiAl2O4 and/or spinel FeAl2O4 that resulted in a slowing of the oxidation kinetics and a delay in the θ → α-Al2O3 phase transformation.
[21] [22]
[17]
[23] [24]
H.J. Grabke, Intermetallics 7 (1999) 1153–1158. G.C. Rybicki, J.L. Smialek, Oxid. Met. 275–304 (1989). H.J. Grabke, M.W. Brumm, B. Wagemann, Mater. Corros. 47 (1996) 675. I. Rommerskirchen, B. Eltester, H.J. Grabke, Mater. Corros. 47 (1996) 646. Y. Kitajima, S. Hayashi, T. Nishimoto, T. Narita, S. Ukai, Oxid. Met. 73 (2010) 375–388. A. Shaaban, S. Hayashi, K. Azumi, Oxid. Met. 82 (2014) 85. A. Shaaban, S. Hayashi, K. Azumi, Surf. Coat. Technol. 266 (2015) 113–121. H. Dahms, J. Electroanal. Chem. Interfacial Electrochem. 8 (1964) 5. H. Dahms, I.M. Croll, Surf. Coat. Technol. 112 (1965) 771. P.C. Andricacos, C. Arana, J. Tabib, J. Dukovic, L.T. Romankiw, J. Electrochem. Soc. 136 (1989) 1336. P.C. Andricacos, J. Tabib, L.T. Romankiw, J. Electrochem. Soc. 135 (1988) 1172. K. Ming Yin, B.N. Popov, R.E. White, Electrodeposition Modeling of Nickel-Iron Alloys in the Presence of Organic Additives, AESF SUR/FIN ’92 International Technical Conference Proceedings, June 22, Session B, Research 1992, p. 103. S. Hessami, C.W. Tobias, J. Electrochem. Soc. 136 (1989) 3611–3616. N. Zech, E.J. Podlaha, D. Landolt, J. Electrochem. Soc. 146 (8) (1999) 2892–2900. H. Nakano, M. Matsuno, S. Oue, M. Yano, S. Kobayashi, H. Fukushima, Mater. Trans. 45 (11) (2004) 3130–3135. M. Ramasubromanian, S.N. Popova, B.N. Popov, R.E. White, K.-M. Yin, J. Electrochem. Soc. 143 (7) (1996). S.R. Allahkaram, M. Honarvar Nazari, S. Mamaghani, A. Zarebidaki, Mater. Des. 32 (2011) 750. L. Chen, L.P. Wang, Z.X. Zeng, J.Y. Zhang, Mater. Sci. Eng. A 434 (2006) 319. J. Doychak, J.L. Smialek, T.E. Mitchell, Metall. Trans. A. 20 (1989) 499. P. van Manen, E.W.A. Young, D. Schalkoord, C.J. van der Wekken, J.H.W. de Wit, Surf. Interface Anal. 12 (1988) 391. Y. Kitajima, S. Hayashi, T. Nishimoto, T. Narita, S. Ukai, Oxid. Met. 75 (2011) 41. L.M. Atlas, W.K. Sumida, Solidus, subsolidus and sub dissociation phase equilibria in the system Fe-Al-O, J. Am. Ceram. Soc. 4l (1958) 150–160. S. Hayashi, I. Saeki, Y. Nishiyama, T. Doi, S. Kyo, M. Segawa, Mater. Sci. Forum 696 (2011) 63. S. Hayashi, Y. Takada, I. Saeki, A. Yamauchi, Y. Nishiyama, T. Doi, S. Kyo, M. Sato, Mater. Corros. 63 (10) (2012).
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Effects of a NiFe co-deposited layer on α-Al2O3 formation by oxidation of a β-NiAl alloy Ali Shaaban a,b,⁎, Shigenari Hayashi c, Kazuhisa Azumi d a
School of Materials and Chemical Technology, Tokyo Institute of Technology, Tokyo, Japan Surface Protection and Corrosion Control Lab., Central Metallurgical Research and Development Institute (CMRDI), Helwan, Egypt Advanced High Temperature Oxidation Materials Laboratory, Graduate School of Engineering, Hokkaido University, Sapporo, Japan d Electronic Materials Chemistry Laboratory, Graduate School of Engineering, Hokkaido University, Sapporo, Japan b c
a r t i c l e
i n f o
Article history: Received 6 June 2017 Revised 11 July 2017 Accepted in revised form 13 July 2017 Available online 13 July 2017 Keywords: NiAl alloy α-Al2O3 Oxidation NiFe co-deposition Oxide scale Spinel oxide
a b s t r a c t Effects of Ni metal, Fe metal or NiFe alloy deposits on the oxidation behavior and oxidation products of a β-NiAl alloy at 1000 °C in air were studied. The surface morphologies of the deposited layers and oxide scales were examined by using field emission scanning electron microscope (FE-SEM) and transmission electron microscopy (TEM). The chemical compositions of the deposited layers were determined before and after oxidation by using energy dispersive X-ray (EDX). The deposited layer and the resultant oxidation products were identified by using X-ray diffractometer (XRD). The chemical composition of the deposited layers was found to affect the morphology of the deposits and the final oxidation products. By oxidation at 1000 °C in air, θ-Al2O3 was detected for bare, Ni-coated and Ni-rich coated (Ni17.7 at.%Fe) samples, but was not detected for Fe-coated and Fe-rich coated (Ni72 at.%Fe) samples at all oxidation times. The oxidation mass gain after 100 h on Fe-coated sample was the highest among the coated samples. TEM Cross-sectional images revealed that the grain size of α-Al2O3 on both Fe coated and Fe-rich NiFe coated samples were the smallest among the coated samples. XRD and EDX point analysis confirmed the formation of different multilayered oxide scales on oxidized samples, NiO/ NiAl2O4/Al2O3 on Ni coated, complex spinel Ni(Fe)Al2O4/Al2O3 on Ni-rich NiFe coated, Fe2O3/(Fe, Ni, Al)2O3/αAl2O3 on Fe-rich NiFe coated and Fe2O3/α-Al2O3 on Fe-coated samples, respectively. Introducing Fe or Fe-rich NiFe coating layer prior to the oxidation of a NiAl alloy resulted in suppression of θ-Al2O3 formation and finer grains of α-Al2O3, which accelerated the growth rate of α-Al2O3 scale. Up to 72 at.% of Fe in NiFe coating was found to be beneficial as an oxidation pretreatment for NiAl alloys to obtain a better high temperature oxidation resistance. © 2017 Published by Elsevier B.V.
1. Introduction Oxidation resistance of alumina-forming alloys is promoted by the formation of stable Al2O3 scale in which the diffusion of both oxygen and Al is slowed down. Various kinds of Al2O3 phases form on these alloys such as metastable γ-, δ- and θ-Al2O3 and stable α-Al2O3 phases. Among them α-Al2O3 oxide scale provides an excellent protective property because of its thermodynamic stability and slow growth rate [1,2]. However, at lower temperature and/or at the early stages of oxidation, metastable alumina phases form first then transform to stable αAl2O3. Because of the rapid growth rates of metastable alumina phases are higher than that of α-Al2O3 by about two orders of magnitude [2– 4], direct formation of α-Al2O3 or rapid transformation of metastable
⁎ Corresponding author at: School of Materials and Chemical Technology, Tokyo Institute of Technology, Tokyo, Japan. E-mail address: [email protected] (A. Shaaban).
http://dx.doi.org/10.1016/j.surfcoat.2017.07.031 0257-8972/© 2017 Published by Elsevier B.V.
alumina to stable α-Al2O3 is required to improve the high temperature oxidation resistance. Kitajima et al. reported that the deposition of Cr or Fe coating on Al containing alloys suppressed the formation of metastable θ-Al2O3 phase by oxidation at 900 °C in air, while Ni coating delayed the transformation of metastable θ-Al2O3 to stable α-Al2O3 phase [5]. They proposed that the suppression of the θ-Al2O3 formation was caused by rapid formation of α-Al2O3 phase. The rapid formation of α-Al2O3 was attributed to the formation of coated metal oxides, Cr2O3 and Fe2O3, which have an isomorphous corundum structure with α-Al2O3 [5]. These metal oxides may provide a higher density of nucleation sites for α-Al2O3. We also investigated the growth rate of Al2O3 scales in the oxidation of β-Ni50Al alloy at 1000 °C in air and confirmed that metastable Al2O3 phase was not formed on NiAl alloy with Fe or Cr coatings [6]. These coatings also affected the growth kinetics of Al2O3 scale via affecting the size of α-Al2O3 grains. The α-Al2O3 scale that transformed earlier always consisted of smaller grains, whereas the size of grains
674
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
Table 1 Composition of electrodeposition bath [g dm−3]. Bath
Coating
[NiSO4·6H2O]
[FeSO4·7H2O]
[Na2SO4]
[H3BO3]
H2SO4
a) b) c) d)
Ni Ni-Fe Ni-Fe Fe
15.5 15.5 15.5 0
0 1.4 27.8 27.8
35, 50 35, 50 35, 50 35
12.5 12.5 12.5 12.5
0 0 0 0.1 M
increased with delay in the transformation from θ-Al2O3 to α-Al2O3. Generally the growth rate of alumina scale decreases with increasing the size of Al2O3 grain. An Fe coating promoted the rapid formation of α-Al2O3 as mentioned above resulting in a small grain size. On the other hand Ni coating was found to delay the transformation and thus enlarged the grain size of α-Al2O3 reducing the growth rate of oxide scale compared with the Fe or Cr coatings. From the other viewpoint, however, the rapid transformation of θ-Al2O3 to α-Al2O3 phase is desirable since the shrinkage of the oxide accompanying the transformation causes stresses and thus cracking of the scales particularly in the cyclic oxidation. A composition of coatings also affects the oxidation kinetics. Ni coating did not suppress the formation of θ-Al2O3 phase while Ni3.1 at.%Fe2O3 nano powder composite coating accelerated the transformation of θ- to α-Al2O3 phase resulting in an increased grain size of Al2O3 and a decreased growth rate of Al2O3 scale. Further addition of Fe2O3 of 5.2 and 7.4 at.% to Ni coatings suppressed completely the formation of θ-Al2O3 phase or promoted direct formation of α-Al2O3
phase resulting in a slow growth rate of scales. However, addition of Fe2O3 more than 7.4 at.% did not show any improvement because of agglomeration of Fe2O3 particles. Therefore, an Fe/Ni ratio in the NiFe coatings is a meaningful parameter because higher ratio of Fe content corresponds to higher concentration of Fe2O3 in the oxide that may affect the oxidation behavior. The purpose of this study is therefore to investigate a suitable Fe/Ni ratio of the NiFe precoating that may improve the oxidation resistance of a NiAl alloy at 1000 °C in air. For this purpose, NiFe alloy coatings were electrodeposited on a NiAl alloy samples via pulsed electrodeposition (PED) prior to the oxidation test. Ni and Fe have different electrodeposition rates [8–12] and thus Fe/Ni ratio can be adjusted by controlling the electrodeposition parameters. 2. Materials and methods A β-Ni50 at.%Al alloy ingot was prepared from Al and Ni metals (~ 99.99% in purity) by using the Ar-arc melting technique followed by homogenization at 1200 °C for 48 h in vacuum (5 × 10−3 Pa) and was cut into specimens of ca. 1 mm thickness. These specimens were ground with SiC paper up to 4000 grit and finished with 3 μm diamond paste to obtain a mirror finish surface and then ultrasonically degreased in acetone for 10 min. Ni, NiFe or Fe coating was electrodeposited on the NiAl alloy from different baths adjusted to pH 3 listed in Table 1. The pulsed electrodeposition (PED) was conducted by using the following conditions; current density of icd = − 10 mA·cm− 2, deposition time ton = 2 s,
Fig. 1. Current waveform for galvanostatic PED and potential response measured for electrodeposition of coatings on Ni50Al samples at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C, pH 3, for 600 s in the baths (a)–(d) listed in Table 1 with agitation at 400 rpm.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
resting time toff = 4 s and temperature T = 25 °C for 600 s with stirring rate at 400 rpm by using a potentiostat (Ivium Technologies Co., model Iviumstat) in an electrochemical cell with a Ag/AgCl reference electrode, Ni sheet counter electrode and a working electrode of NiAl sheet with ca. 2 cm2 in area. After electrodeposition, these samples were subjected to high temperature oxidation tests. Samples were placed in a furnace exposed to air and a temperature was elevated from room temperature (RT) to 1000 °C with a rate of 10 °C min−1 and kept at 1000 °C for tox = 0.5, 1, 4, 7, 9, 25, 49 and 100 h and then cooled down to RT. Weight of samples were measured before and after the oxidation tests to determine the mass gain.
675
Surface morphology and chemical composition of electrodeposited coatings were observed by using FE-SEM with EDX (JEOL Co., model 6500F). Phase analysis of coatings was conducted by using X-ray diffractometer XRD (JEOL Co., model JDX 3500, operated at 20 mA and 40 kV with CuKα radiation). TEM (JEOL Co., model JEM 2000FX, operated at 200 kV) was employed to investigate the cross-section morphology, elemental concentrations of Al, Ni, Fe and O in oxide scales, and to identify the oxides. Multi-beam FIB-SEM (JEOL Co., model JIB-4600F/HDK) was used for TEM cross-sectional samples preparation of oxide scales. During sample preparation, a carbon layer or a tungsten layer was pre-deposited on the sample to protect the surface from damage or contamination by Ga+ beam. Then a cross-section sample was lifted
Fig. 2. Surface and cross-sectional morphologies FE-SEM images of (a) Ni, (b) Ni-rich NiFe, (c) Fe-rich NiFe and (d) Fe on a Ni50Al samples, which was deposited for 600 s with pulse conditions at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C.
676
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
out and attached to a Mo support grid that fits into a specimen holder of TEM. 3. Results and discussion 3.1. Properties of Ni, NiFe and Fe coatings Fig. 1 shows the current waveform and potential response at the initial stage of PED from Ni-bath (a), Ni-Fe baths (b)–(c) and Fe-bath (d). When icd was applied at ton the potential was dropped suddenly in a less noble direction to initiate electrodeposition. The potential shift during ton decreased gradually due to increase in a surface area of the deposits. When icd was turned to off at toff, the potential shifted in a noble direction due to recovery of the concentration of Ni2+ or Fe2+ ions near the sample surface in the bath to improve the uniform electrodeposition in the next electrodeposition period ton. Potential at toff in bath (d) was considerably different from those in baths (a)–(c) because of the difference in coating composition as Fe coating deposited from bath (d) while Ni containing coatings deposited from baths (a)–(c). EDX analysis showed the Fe content of 17.7 at.% in a coating electrodeposited from bath (b) and 72 at.% from bath (c). The coatings deposited from baths (a)–(d) are, therefore, named to be Ni coating, Ni-rich NiFe coating, Fe-rich NiFe coating, and Fe coating, respectively, in this paper. Surface and cross-sectional SEM observation confirmed that compact, dense and crack-free coatings with different microstructure were formed on the NiAl substrates as shown in Fig. 2. Ni coating consisted of small facet-structured particles, Ni-rich NiFe coating consisted of fine nodular-structured, Fe-rich NiFe coating consisted of fine fluffy-structured and Fe coating consisted of coarse tetrahedralstructured particles. Average particle size in the coatings was in the order of Fe coating N Ni coating N Ni-rich NiFe coating N Fe-rich NiFe coating, i.e., the particle size of NiFe coatings were smaller than those of Ni and Fe coatings. Thickness of Ni or Fe coating was 0.7–1 μm and 1 μm, respectively, as shown in Fig. 2(a) and (d) while the thickness of NiFe coatings was about 0.8 μm for Ni-rich coating and 0.6 μm for Ferich coating and tended to be thinner than that of Ni or Fe coating as shown in Fig. 2(b) and (c). Although the electrodeposition charge for all coatings was the same. Particle size and thickness of NiFe coatings smaller than those of Ni and Fe coatings could be attributed to the anomalous co-deposition of NiFe alloys as reported in Ref. [8–9,13–16]. Various mechanisms have been reported for NiFe alloy co-deposition. Metal hydroxide formation mechanisms of NiFe alloys were reported in ref. [8–9,13] and confirmed by direct measurements in ref. [14–16]. Hydroxide suppression mechanism by Dahms et al. in which the nobler (Ni) deposition was suppressed by the formation of less noble hydroxide (Fe(OH)2) during electrodeposition process. Fe(OH)2 formation is preferentially occurs in the cathode layer and adsorbed on the cathode [8,9]. Hessami et al. reported another mathematical model in which they focused on the difference in the dissociation or hydrolysis constants between intermediate compounds FeOH+ and NiOH+ [13]. The electrochemical reactions of Ni and Fe at the surface of the electrode (cathode) are given as NiðOHÞþ þ 2e− →Ni þ OH−
ð1Þ
FeðOHÞþ þ 2e− →Fe þ OH−
ð2Þ
Fig. 3. XRD patterns of Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe coated Ni50Al samples, which were deposited for 600 s with pulse conditions at icd = −10 mA·cm−2, ton = 2 s, toff = 4 s and T = 25 °C.
dissociation constant than Fe(OH)+ (5.8 × 10−8) in the cathode layer. Fe(OH)+ is preferentially formed at the expense of Ni(OH)+, resulting in preferential deposition of Fe to Ni.
The homogeneous equilibrium reactions of Ni and Fe in the electrodeposition system are given as Ni2þ þ OH− →NiðOHÞþ
ð3Þ
Fe2þ þ OH− →FeðOHÞþ
ð4Þ
According to these equations and based on the dissociation mechanism, Ni(OH)+ (4.5 × 10−5) has three order of magnitude higher
Fig. 4. Change of oxidation mass gain of bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fecoated Ni50Al samples in the oxidation tests at 1000 °C in air up to (a) short exposure up to 1 h and (b) long exposure up to 100 h.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
Fig. 3 shows the XRD patterns of electrodeposited samples. Ni coating showed three strong peaks, Ni(111), Ni(200) and Ni(220) of Ni fcc structure. The strongest Ni(111) peak corresponded to the preferential growth orientation. Ni-rich NiFe coating also showed a Ni(111) peak, however, the intensity of Ni(200) and Ni(220) peaks was considerably decreased and shifted slightly to a lower 2θ°. For Fe-rich NiFe coating, the Ni(200) and Ni(220) peaks almost disappeared while strong Fe(110) peak and small Fe(200) and Fe(211) peaks of bcc Fe structure evolved with increasing Fe concentration. These peaks of Fe were observed clearer for Fe coating. 3.2. High temperature oxidation test The bare NiAl alloy, NiAl alloys coated with Ni, Ni-rich NiFe, Fe-rich NiFe and Fe coatings were subjected to high temperature oxidation tests at 1000 °C in air. Oxidation mass gain (Δ m) was obtained from weight change measurement before and after the test for tox = 0.5, 1, 4, 7, 9, 25, 49 and 100 h as shown in Fig. 4(a). Two oxidation stages
677
were observed, i.e., the initial stage with rapid growth rate followed by the second stage with slow and steady growth rate. Fig. 4(b) shows the details of the initial stage. Since the growth rate of the oxide on the coated samples at the initial stage were considerably faster than that of bare NiAl alloy, the growth of NiO on Ni or Ni-rich NiFe coatings and FeO, Fe3O4 or Fe2O3 on Fe-rich NiFe or Fe coatings seemed to be responsible for the rapid growth of scales at the initial stage of oxidation. The oxidation mass gain on the coated samples after 1 h of oxidation was in the order of Fe N Fe-rich N Ni-rich NiFe N Ni-coated NiAl. In the further oxidation up to 100 h shown in Fig. 4(a), growth rate of the oxide scale on Ni, Ni-rich NiFe and Fe-rich NiFe coated samples reduced comparing to the bare NiAl alloy and Fe coated NiAl alloy. Fig. 5 shows the surface morphologies of samples oxidized at 1000 °C for tox = 9 h. An oxide scale of blade-like morphology covered the surface of the bare Ni50Al alloy (Fig. 5(a)). This kind of morphology was typically observed for transient phases of alumina such as θ-Al2O3 [6–7,17–21]. An oxide scale of faceted morphology covered the Ni coated Ni50Al sample (Fig. 5(b)). Mixture of faceted and tetrahedral
Fig. 5. Surface morphology FE-SEM images of (a) bare Ni50Al, (b) Ni-, (c) Ni-rich NiFe, (d) Fe-rich NiFe, and (e) Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
678
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
morphologies covered the NiFe coated Ni50Al samples (Fig. 5(c) and (d)). A tetrahedral-shaped oxide covered the Fe coated Ni50Al (Fig. 5(e)). XRD patterns of bare and coated samples for tox = 9 h oxidation and their phase identification were shown in Fig. 6. Al, Ni and/or Fe were oxidized simultaneously to form Al2O3, NiO and/or Fe-oxide phases. Peaks of metastable θ-Al2O3 and α-Al2O3 phases were observed for bare NiAl sample. Peaks of NiO, spinel NiAl2O4, θ-Al2O3 and α-Al2O3 phases were observed for Ni and Ni-rich NiFe coated samples. Peaks of FeO, α-Fe2O3 and α-Al2O3 phases were observed for Fe and Fe-rich NiFe coated samples. Peaks of θ-Al2O3 phase were detected for bare, Ni and Ni-rich NiFe coated samples, but were not for Fe or Fe-rich NiFe coated samples. Fe3O4 phase was not detected because it could be transformed to Fe2O3 or reacted with Al2O3/NiO to form spinel FeAl2O4 and/or inverse spinel NiFe2O4. XRD patterns of the oxidized Ni-rich NiFe and Fe-rich NiFe coated samples are shown in Fig. 7 as a function of oxidation time tox from 0.5 h to 7 h. At the early stage of oxidation at tox = 0.5 h, NiO and FeO/Fe2O3 predominated the oxide scale on Ni-rich and Fe-rich NiFe coated samples, respectively. By further oxidation up to tox = 7 h, not only single oxide phases but also composite oxide phases were detected. Peaks of α-Al2O3, θ-Al2O3, NiO, spinel FeAl2O4 and inverse spinel NiFe2O4 were observed on Ni-rich NiFe coated sample and peaks of FeO, α-Fe2O3, α-Al2O3, (Fe, Al)2O3 and spinel FeAl2O4 were observed on Fe-rich NiFe coated sample. Metastable Al2O3 phase was not detected for Fe-rich NiFe coated sample. Fig. 8 shows the cross-sectional TEM images of the oxide scale formed on NiAl alloys with/without coating of Ni-rich NiFe, Fe-rich NiFe and Fe layers after oxidation test at 1000 °C for tox = 9 h. On bare NiAl sample, ca. 0.9 μm thick single layer of θ-Al2O3 with needlelike morphology including columnar grains of α-Al2O3 phase at the interface between the oxide layer and substrate alloy was observed as shown in Fig. 8(a). On the other hand a multilayer structure of the scales was confirmed for all the coated samples shown in Fig. 8 (b–e). These scales were thicker (1.5–2.0 μm) than that on the bare NiAl sample due to the fast growth of oxides of the coatings at the initial stage of oxidation as shown in Fig. 4. Composition of each layer of scales was analyzed by EDX and summarized in Table 2 with their multilayer phase. In Fig. 8(b–e), the grain size of α-Al2O3 depended on the kind of coating, i.e., the largest size was found on Ni coating, reduced with increasing in the Fe content and reached the smallest size on Fe coating. In order to investigate the correlation between the overall structure of the oxide scales and the oxidation kinetics of NiAl alloys with/without coating, two stages of oxidation composed of the initial stage (tox = 1 → 9 h) controlled by fast growing oxides formation and the late stage (tox = 9 → 100 h) controlled by slower growing oxides will be discussed below. Fig. 9 shows the distribution of the oxygen uptake (mass gain) to each phase in the oxide scale formed on Ni50Al samples with/without coating of Ni, Ni-rich NiFe, Fe-rich NiFe or Fe after the oxidation at 1000 °C for 9 h in air that was calculated from the results of XRD, TEM and EDS presented in this paper. Fe-coated sample showed the largest oxidation mass gain among all samples, i.e., about 67% of oxygen was taken in Fe oxide. On Fe-rich NiFe coated sample, about 47% of oxygen was taken in Fe oxides and 16% in spinel phases. On Ni-rich coated sample, 67% of oxygen was taken in spinel phase. On Ni coated sample, 25% and 29% of oxygen was taken in NiO and spinel NiAl2O4, respectively. Different distribution of oxygen to the coatings of different Fe/Ni ratios resulted from the difference in the initial oxidation rate and thus affected the total oxidation mass gain in the entire oxidation time. At the initial stage of oxidation (tox = 1 → 9 h) and even during heating, the composition of the coatings plays a strong role in the oxidation kinetics. During heating, oxide phases such as NiO and/or FeO, Fe3O4, Fe2O3 grew until Ni or Fe was consumed. Increase in the Fe/Ni ratio of the coatings resulted in increase in the oxidation mass gain due to the large Gibbs formation energy of Fe2O3 (− 380 kJ mol−1)
Fig. 6. XRD patterns of bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
compared with that of NiO (−280 kJ mol−1). This suggests that Fe2O3 was preferred to be formed than NiO and provided faster growth in the outer layer of the oxide scale at 1000 °C as presented in Fig. 8. The order of oxidation mass gain was thus in the order of Fe N Fe-rich NiFe N Ni-rich NiFe N Ni NiAl alloy as shown in Figs. 4(b) and 9. Due to the slower growth rate of θ-Al2O3 than the growth rate of coating oxides,
Fig. 7. XRD patterns of (a) Ni-rich NiFe and (b) Fe-rich NiFe coated Ni50Al alloy for 0.5 h up to 7 h at 1000 °C in air.
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
the oxidation mass gain of the bare Ni50Al was smaller than that on the coated samples. At tox = 9 h, θ-Al2O3 still continued to grow on bare Ni50Al surface accompanied by slight formation of α-Al2O3 grains at θ-Al2O3/NiAl
679
interface. On the other hand, thin layer of α-Al2O3 scales with different grain size was formed on all coated samples. Although the oxide scale formed on Ni or Ni-rich NiFe coated samples contained residual untransformed θ-Al2O3 as detected in XRD (Fig. 6), it could not be observed
Fig. 8. Cross-sectional TEM images of oxide scales formed on (a) bare Ni50Al, (b) Ni-, (c) Ni-rich NiFe, (d) Fe-rich NiFe, and (e) Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air.
680
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
by cross-sectional TEM observation (Fig. 8(b) and (c)). Therefore, the oxidation rate remained faster, particularly on Ni coated sample on which θ-Al2O3 still growing (Fig. 4(a)). On Ni or Ni-rich NiFe coated samples, solid phase reactions of NiO with Al2O3 might form an intermediate layer of NiAl2O4 spinel between the outer layer of NiO and the inner layer of Al2O3 on samples coated with Ni and Ni-rich NiFe layers as shown in Figs. 6 and 8(b) and (c). On Fe or Fe-rich NiFe coated samples, Al2O3 may react directly with FeO or may dissolve into Fe3O4 resulting in formation of FeAl2O4 spinel as expected from the phase diagram [22]. Beside the NiAl2O4 spinel formation, (Fe, Ni, Al)2O4 spinel may also be formed on the inner layer of Al2O3 on Ni-rich NiFe coated sample as shown in Fig. 8(c). Because of the solubility limit of Fe in α-Al2O3 and of Al in α-Fe2O3, a (Fe, Al)2O3 phase was formed as an independent layer between the α-Fe2O3 and α-Al2O3 layers on Fe-rich NiFe coated sample as shown in Fig. 8(d). Al2O3 could appear in two different forms, i.e. (Fe, Al)2O3 and α-Al2O3, due to formation of (Fe, Al)2O3 solid solution as confirmed from XRD in Ref. [23]. Formation of a duplex Fe2O3 scale composed of an outer layer of relatively pure Fe2O3 and an inner layer of Al saturated Fe2O3 was found by Hayashi et al. [24] and it was also confirmed from TEM observation Fig. 8(d) and EDX point analysis in this study. Since neither spinel FeAl2O4 nor solid solution of (Fe, Al)2O3 layer was observed by TEM observation of oxidized Fe coated sample shown in Fig. 8(e), it can be suggested that the presence of Ni in the coating promotes formation of spinel phase. Based on cationic configurations inside the oxygen sub-lattice, θ-Al2O3 and NiO, FeO (Fe3O4) and/or NiAl2O4 (FeAl2O4) phases are related to the similar oxygen framework of fcc structure. In the O2− framework of θ-Al2O3 phase, ½ of the Al3+ ions occupy octahedral interstitial sites and another ½ occupies tetrahedral sites. Meanwhile, the Ni2+/Fe2+ ions occupy all the octahedral interstitial sites of oxygen framework in NiO/FeO, respectively. In Fe3O4 (magnetite) the oxygen anion O2− form a close-packed fcc sub-lattice with Fe2 + and Fe3+ cations located in interstitial sites. Two different kinds of cation sites exist in the magnetite crystal: tetrahedrally coordinated A sites occupied by Fe3+ and octahedrally coordinated B sites occupied by ½ Fe2+ and ½ Fe3+. The reaction(s) between θ-Al2O3 and NiO/FeO could be represented as, Spinel NiAl2 O4 : ½Ni2þ O þ Al
3þ
½Al
3þ
O3 →Al
3þ
Spinel FeAl2 O4 : ½Fe2þ O þ Al
3þ
½Al
3þ
O3 →Al
3þ
h h
Ni2þ Al
3þ
Fe2þ Al
3þ
i i
O4
ð5Þ
O4
ð6Þ
And the reaction between Fe3O4 and θ-Al2O3 could be represented as, h i Spinel FeAl2 O4 : Fe3þ Fe2þ Fe3þ O4 þ Al
3þ
½Al
3þ
3þ
O3 →7=4Al
h
Fe2þ Al
3þ
i
O4
ð7Þ
Another reaction is expected to occur between the initially formed Fe3O4 and NiO phases, in which Ni may substitute Fe2+ in Fe3O4 (magnetite) to form NiFe2O4 (nickel ferrite) in the cubic inverse spinel structure. The reaction between Fe3O4 and NiO could be represented as, h i Inverse spinel NiFe2 O4 : Fe3þ Fe2þ Fe3þ O4
h i þ ½Ni2þ O→5=4Fe3þ Ni2þ Fe3þ O4
coated sample. Not only α-Al2O3 grain size, but also spinel formation should be considered. Although the growth and its grain size α-Al2O3 controls the oxidation kinetics at the late stage of oxidation, effect of spinel or formation of spinel complex phases cannot be neglected. Growth rates of these spinel oxides were slower than those of the initially formed oxides and as fast as to those of metastable alumina phases, resulting in the reduction of the growth rate of oxides at tox = 25– 100 h (Fig. 4(a)). It was also found in our previous work that the formation of spinel phases such as NiAl2O4 could alter the oxidation kinetics. Existence of such intermediate layer could reduce the oxidation rate regardless the transformation of θ-Al2O3 phase was completed or not as discussed in Ref. [6,7]. In this case, reduction of the oxidation mass gain could be attributed to two factors; (i) slowing down the mobility of cations and anions and/or (ii) reducing the chemical potential gradient in this intermediate spinel layer. Also, the counter diffusion of Ni2+ or Al3+ cations through these spinel oxides is much slower than that through the initially formed oxides. However, after longer oxidation (tox = 100 h), development of a uniform α-Al2O3 layer might suppress the outward diffusion of Ni2+ and thus prevent further growth of NiAl2O4. In such case the oxidation kinetics will be totally dominated only by α-Al2O3 growth rate. Formation of at least one of NiO, FeO, Fe3O4, NiAl2O4, and/or FeAl2O4 phases is believed to delay the transformation of θ- to α-Al2O3 phase [6, 7] while coexistence of Fe2O3 nano-particle in a composite coating [7] or oxidized Fe coating [5,6,21] is believed to promote the formation of αAl2O3 phase. The beneficial effect of α-Fe2O3 phase formation on direct formation of α-Al2O3 might be resulted from (i) providing preferential nucleation sites for α-Al2O3 phase formation rather than θ-Al2O3 [7], and/or (ii) precipitation of α-Al2O3 phase but not metastable Al2O3 phase from Al saturated Fe2O3 [23,24]. According to the results obtained in this study, existence of NiO, FeO or Fe3O4 was confirmed to stabilize θ-Al2O3 and delay the transformation of θ-Al2O3 to α-Al2O3 phase in the long oxidation process. This effect might be beneficial to reduce the oxidation rate. Addition of Fe to Ni was found to promote the transformation of θ-Al2O3 to α-Al2O3 phase in the oxidation process by forming Fe2O3 with a suitable grain size for nucleation of α-Al2O3 phase. Fe coating on NiAl alloy was not beneficial because αAl2O3 phase with a too small grain size was formed that could not reduce the oxidation rate. However, coexisting of Ni and Fe in the coating provided a better high temperature resistance. Optimization of Fe/Ni ratio should be considered the suitable balance between the time for transformation of θ-Al2O3 to α-Al2O3 phase and the time for growth of α-Al2O3. For example, at high temperature, a low Fe/Ni ratio is advantageous to obtain the α-Al2O3 phase with large grain size that reduces effectively the oxidation rate, and, at low temperature, high Fe/Ni ratio is favoured to accelerate the phase transformation from metastable θ-Al2O3 to α-Al2O3 phase. 4. Conclusion PED Ni, Fe or NiFe alloys (Ni-rich NiFe and Fe-rich NiFe) coating layer was introduced as a pretreatment coating prior the oxidation of a NiAl Table 2 Composition in [at.%] and predicted multilayer phases of electrodeposited NiFe alloy coatings shown in Fig. 8. (b)
ð8Þ (c)
where square brackets represent cationic octahedral sites. At the late stage of oxidation (tox = 9–100 h), overall microstructure of the oxide scale especially α-Al2O3 grain size takes over the role in controlling the oxidation kinetics. Growth rate of the oxide scale on bare alloy was faster than those on coated NiAl samples because of the rapid growth of θ-Al2O3 rather than α-Al2O3. The oxidation mass gain of bare alloy was greater than those on coated samples except for Fe-
(d)
(e)
Coating Composition Phase Coating Composition Phase Coating Composition Phase Coating Composition Phase
Ni Ni23.78Al0.65O75.57/Ni5.13Al29.56O65.31/Ni0.07Al38.83O61.10 NiO/NiAl2O4/Al2O3 Ni-rich NiFe Ni0.1Fe3.9Al28.7O67.3 or Ni1.9Fe0.4Al26.6O71.1/Ni0.1Fe0.1Al42.8O57 (Fe, Ni, Al)2O4/(Fe, Ni, Al)2O4/Al2O3 Fe-rich NiFe Fe17.3Al5.3O77.4/Ni2.3Fe9.5Al12.2O76.0/Ni0.1Fe0.1Al39.6O60.2 α-Fe2O3/(Fe, Ni, Al)2O3/α-Al2O3 Fe Fe19.38Al0.95O79.67/Fe0.05Al40.72O59.23 α-Fe2O3/α-Al2O3
A. Shaaban et al. / Surface & Coatings Technology 325 (2017) 673–681
681
4. The existence of α-Fe2O3 phase results in direct formation of α-Al2O3 because of template effect or because of α-Al2O3 precipitation from Al saturated Fe2O3, without metastable Al2O3 formation. 5. Grain size of α-Al2O3 affects the suppression of θ-Al2O3 formation. The grain size of α-Al2O3 was smaller than those on samples on which θ-Al2O3 still grow. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
Fig. 9. Amount of the oxygen uptake (in mg·cm−2) calculated for oxide phases on bare Ni50Al, Ni, Ni-rich NiFe, Fe-rich NiFe alloy, Fe-coated Ni50Al samples after oxidation for 9 h at 1000 °C in air based on the results of XRD, TEM and EDS analysis.
[11] [12]
[13] [14] [15]
alloy. Their effects on the oxidation products and the growth kinetics of Al2O3 scale formation on NiAl at 1000 °C in air were investigated. The results are summarized as follows;
[16]
1. The oxidation kinetics depends initially on the growth rate of the oxidized coating, then on the spinel phases formation and finally on the α-Al2O3 scale microstructure.
[18] [19] [20]
2. The formation of either metastable or stable Al2O3 phase depends on the initially formed oxide phases from coating layer. 3. The coexistence of the initially formed NiO, FeO and/or Fe3O4 phases with Al2O3 phase enhance the formation of spinel NiAl2O4 and/or spinel FeAl2O4 that resulted in a slowing of the oxidation kinetics and a delay in the θ → α-Al2O3 phase transformation.
[21] [22]
[17]
[23] [24]
H.J. Grabke, Intermetallics 7 (1999) 1153–1158. G.C. Rybicki, J.L. Smialek, Oxid. Met. 275–304 (1989). H.J. Grabke, M.W. Brumm, B. Wagemann, Mater. Corros. 47 (1996) 675. I. Rommerskirchen, B. Eltester, H.J. Grabke, Mater. Corros. 47 (1996) 646. Y. Kitajima, S. Hayashi, T. Nishimoto, T. Narita, S. Ukai, Oxid. Met. 73 (2010) 375–388. A. Shaaban, S. Hayashi, K. Azumi, Oxid. Met. 82 (2014) 85. A. Shaaban, S. Hayashi, K. Azumi, Surf. Coat. Technol. 266 (2015) 113–121. H. Dahms, J. Electroanal. Chem. Interfacial Electrochem. 8 (1964) 5. H. Dahms, I.M. Croll, Surf. Coat. Technol. 112 (1965) 771. P.C. Andricacos, C. Arana, J. Tabib, J. Dukovic, L.T. Romankiw, J. Electrochem. Soc. 136 (1989) 1336. P.C. Andricacos, J. Tabib, L.T. Romankiw, J. Electrochem. Soc. 135 (1988) 1172. K. Ming Yin, B.N. Popov, R.E. White, Electrodeposition Modeling of Nickel-Iron Alloys in the Presence of Organic Additives, AESF SUR/FIN ’92 International Technical Conference Proceedings, June 22, Session B, Research 1992, p. 103. S. Hessami, C.W. Tobias, J. Electrochem. Soc. 136 (1989) 3611–3616. N. Zech, E.J. Podlaha, D. Landolt, J. Electrochem. Soc. 146 (8) (1999) 2892–2900. H. Nakano, M. Matsuno, S. Oue, M. Yano, S. Kobayashi, H. Fukushima, Mater. Trans. 45 (11) (2004) 3130–3135. M. Ramasubromanian, S.N. Popova, B.N. Popov, R.E. White, K.-M. Yin, J. Electrochem. Soc. 143 (7) (1996). S.R. Allahkaram, M. Honarvar Nazari, S. Mamaghani, A. Zarebidaki, Mater. Des. 32 (2011) 750. L. Chen, L.P. Wang, Z.X. Zeng, J.Y. Zhang, Mater. Sci. Eng. A 434 (2006) 319. J. Doychak, J.L. Smialek, T.E. Mitchell, Metall. Trans. A. 20 (1989) 499. P. van Manen, E.W.A. Young, D. Schalkoord, C.J. van der Wekken, J.H.W. de Wit, Surf. Interface Anal. 12 (1988) 391. Y. Kitajima, S. Hayashi, T. Nishimoto, T. Narita, S. Ukai, Oxid. Met. 75 (2011) 41. L.M. Atlas, W.K. Sumida, Solidus, subsolidus and sub dissociation phase equilibria in the system Fe-Al-O, J. Am. Ceram. Soc. 4l (1958) 150–160. S. Hayashi, I. Saeki, Y. Nishiyama, T. Doi, S. Kyo, M. Segawa, Mater. Sci. Forum 696 (2011) 63. S. Hayashi, Y. Takada, I. Saeki, A. Yamauchi, Y. Nishiyama, T. Doi, S. Kyo, M. Sato, Mater. Corros. 63 (10) (2012).