Initial stage oxidation corrosion of commercial ferritic stainless steels with different Cr contents at 650 °C for solid oxide fuel cells

Abstract

Selecting adequate ferritic stainless steel (FSS) with a high corrosion resistance and a low cost is critical for solid oxide fuel cells (SOFCs) operating at intermediate temperature. In this study, the corrosion behaviors of four commercial FSSs involving TS430, TY441, YG442, and TY445 with a Cr content ranging from 16.18 wt.% to 21.73 wt.% are investigated at 650 °C. The oxidation mass gains, microstructures of surface oxide scale, and electrical conductivities are measured. The effects of grain size as well as doped elements are estimated together with the Cr volatilization. Flaky Cr₂O₃ particles are formed on TS430 and TY441 dominated by the outward migration of Cr³⁺. In comparison, a thin and dense layer of chromia is observed on YG442 and TY445. A high Cr content and a uniformly distributed grain size are conducive to the formation of a thin and dense chromia scale on the FSS surface during the initial oxidation process. On the other hand, the addition of Nb, Ti, and Mo weakens the outward diffusion of Cr³⁺ and reduces the particle size of chromia. After oxidation at 650 °C for 120 h, scattered (Mn, Cr)₃O₄ spinel particles occur on TS430, YG442, and TY445. TY445 and YG442 exhibit a higher conductivity although all the results of area specific resistance (ASR) are less than 6 mΩ·cm². Meanwhile, the effect of Cr volatilization is enlarged on the estimation of mass gain at 650 °C compared with even higher temperatures.

1. Introduction

With the development of material design and preparation technology, the operating temperature of SOFC has declined from above 800 °C to below 700 °C, where ferritic stainless steel (FSS) is appropriate as interconnects thanks to its low cost, good thermal stability, high mechanical strength, and well matching of thermal expansion coefficient with ceramics. FSS is also the preferred alloy of choice for metal-supported SOFCs [1] and solid oxide electrolysis cells (SOECs) [2]. The Solid State Energy Conversion Alliance (SECA) program required a life of SOFC stacks is more than 40, 000 h with a degradation rate of <0.1 %·kh⁻¹ and an area specific resistance (ASR) of <100 mΩ·cm² [3]. Coatings are often employed to meet the requirements of SECA [4]. Protective coatings reduce the overgrowth of low-conductivity oxide scales (mainly in Cr₂O₃) on FSS substrates, acting as a barrier to the volatilization of Cr^(VI) and enhancing electron conduction, which is necessary to achieve long-term stability of SOFC stacks [5]. However, pores and pinholes may exist in these coatings, and there are bonding problems at the interface between coating and substrate, leading to unavoidable oxidation of the metal substrate. Therefore, the oxidation behavior of metal substrate is primarily important for long-term stability. Selection of a suitable FSS before applying the coating is critical and the continuous growth of chromia scale and the associated Cr volatilization must be estimated in advance [6].

Some FSSs are often recommended for SOFCs such as Crofer 22 APU [7], Crofer 22H [8], Sanergy HT [9], E-Brite [10], and ZMG232 [11]. However, these particularly developed FSSs are expensive. Ordinary commercial FSSs like AISI 430 and 441 are more attractive [12]. The corrosion resistance of Fe–Cr alloy originates from a continuous and dense chromia scale with enough thickness and good adhesion. The content of Cr in FSS matrix is the key factor for forming a protective Cr₂O₃ scale. An adequate original Cr content (i.e. an abundant Cr reservoir) is a necessary condition for the long life of the alloy [13]. The element diffusion and oxidation kinetics of steel matrix during the long-term working process are usually positively dependent on temperature [[14], [15]]. At reduced oxidation temperature, the Cr content requirements of FSS need to be further explored.

The oxidation kinetics and Cr diffusion are affected by doped elements. The effects of these doped elements on the oxidation behaviors under low and medium temperatures still need to be investigated. A small amount of Mn (0.3–0.5 wt.%) was added to form (Mn, Cr)₃O₄ spinel [16]. (Mn, Cr)₃O₄ spinel shows better corrosion resistance than Cr₂O₃. The beneficial effect of adding 0.4 wt.% Mn at 600 °C was confirmed [17]. Meanwhile, strengthening elements such as Nb, Ti, Mo, and Cu are doped to improve the oxidation resistance and reduce the creep phenomenon. The Laves phases precipitate along the grain boundaries and impede the rapid outward diffusion of cations. This can eliminate the negative effects of Si as well [18]. There is a synergistic effect between Si and Nb, and the addition of a small amount of Si can also eliminate the adverse effect of Nb on oxidation [19]. However, Talic et al. found that excessive Ti was not beneficial to the oxidation resistance. Ti⁴⁺ entered into the Cr₂O₃ layer, increasing Cr vacancies and enhancing the diffusion rate of Cr³⁺ [20]. Mo is added to interconnect alloys such as Sanergy HT (0.96 wt.% Mo) [21], ZMG232J3 (1.98 wt.% Mo), and ZMG232G10 (1.8 wt.% Mo) [22], which contributes to Laves-type precipitates and enhance the creep strength [23]. Nevertheless, the addition of Mo is normally less than 4 wt.%, otherwise the oxidation is aggravated [24]. Meanwhile, the probability of forming σ phase escalated and the stability of the alloy declined [25]. Cu is another additive to promote the densification of the oxide scale. For instance, 1 wt.% Cu was added to ZMG232G10. After 40,000 h oxidation at 850 °C, Cu appeared in the surface spinel layer and formed (Mn, Cr, Cu)₃O₄ phase, which impeded Cr volatilization effectively [26]. Swaminathan et al. [27] found that the best oxidation resistance was achieved by doping 1.57 wt.% Cu into an FSS with 22 wt.% Cr. Due to the weaker oxygen affinity of Cu than Cr, after 2000 h oxidation at 800 °C, the metallic Cu in FSS was precipitated below Cr₂O₃. The Cu segregation at the interface of scale/alloy reduced the oxygen activity and inhibited the internal growth of oxide scale by suppressing the ingress of oxygen. However, most of the above literatures on the influence of doped elements are based on the high-temperature oxidation process of about 800 °C. There are differences in the diffusion behavior of low-temperature elements below 700 °C, and further research is necessary.

The element diffusion in the oxide scale and the kinetics of the surface reaction escalate with the increase in temperature. Most investigations focused on SOFCs operating under a high temperature above 750 °C [[28], [29]]. With the development of ceramic technology, the thickness of electrolyte decreases to <10 μm such that the operating temperature of SOFC drops to <750 °C [30]. For these intermediate-temperature SOFCs, FSS is normally employed as interconnects and metal supports. The oxidation kinetics of FSS accelerated at high temperatures is often inconsistent with the actual oxidation behavior of SOFC operating at low to medium temperatures. Tucker et al. [31] simulated the oxidation of P434L porous stainless steel matrix in SOFC cathode air for 500 h at 650 °C or below, and observed the non-parabolic trend of oxidation weight gain and even negative weight change. This differs from parabolic oxidation kinetics at 700 °C and above, possibly due to competition between increased oxidation weight and mass loss from Cr evaporation [32]. A similar oxidation phenomenon has been observed in air at 550 °C–650 °C [33]. At lower temperature, Cr evaporation is the dominant factor, which increases the complexity of oxidation kinetics. In addition, Cr species migrate to the air electrode [[34], [35]] through vapor transport (or deposition) of Cr species or solid state diffusion, which reduces the electrochemical activity. Therefore, the impact of Cr volatilization at medium and low temperatures should not be underestimated. Froitzheim et al. have done detailed work on the application of AISI441, Crofer 22 APU, and other alloys at 600 °C, such as pre-oxidation treatment [36], dual-atmosphere exposure [37], etc. However, the oxidation behavior of alloys with different element compositions at low-temperature is different, which should be analyzed in detail. Therefore, oxidation behaviors and corrosion mechanisms of various commercial FSSs under low and intermediate temperatures still need to be revealed.

In addition, the grain size affects the diffusion process of elements, the oxidation resistance, and the mechanical properties of alloys. In the FSS matrix, grain boundary is the preferred channel for Cr element diffusion. Smaller grain sizes tend to promote the outward diffusion of Cr, resulting in faster formation of chromia scale [38]. Chyrkin et al. [39] investigated the high-temperature oxidation behavior of additively manufactured (AM) Ni-based alloy IN625. It was found that the grain size affected the overall oxidation kinetics, but not the intergranular oxidation morphology. The grain size is affected by the manufacturing process, which increases the complexity of FSS oxidation behavior. However, most investigations on the oxidation behavior of interconnects only compared the oxide structure, surface resistance, and other characteristics, but did not make a comprehensive and in-depth analysis of the influencing factors such as grain size and roughness.

According to the above analysis, it is necessary to investigate the oxidation behavior of alloys with different element compositions at 650 °C, to supplement the oxidation kinetics analysis of FSS at medium and low temperatures. At 650 °C, the effects of grain size, Cr content, and different doping elements on the oxidation kinetics of FSS need to be re-estimated. The effect of Cr volatilization on the initial oxidation behavior should also be taken into account. In this study, corrosion behaviors of four commercial FSSs with a Cr content from 16.18 wt.% to 21.73 wt.% are compared at 650 °C during the initial oxidation stage. The surface morphology of the bare FSSs and the oxide scales are characterized. The mass gains are measured while the Cr volatilization is estimated. The electrical conductivity is evaluated as well by an ASR test. The growth mechanism of thermally grown oxides is analyzed. Special attention is paid to the influences of grain size and elemental composition on the formation of the initial oxide scale. The results provide some insights into FSS alloy design and manufacturing process optimization for intermediate-temperature SOFC applications.

2. Experimental

Four FSSs labeled TS430, TY441, YG442, and TY445 are used in the oxidation test. The compositions of the samples (in wt.%) are listed in Table 1. The steel samples were cut into 2 cm × 2 cm via a laser cutting process. The thicknesses of TS430, TY441, YG442, and TY445 are 0.3 mm, 0.5 mm, 0.4 mm, and 1.0 mm, respectively, according to the availability from the market. First, oil and impurities were cleared from the surfaces. Samples were placed in ethanol for ultrasonic cleaning and then dried. The oxidation experiment was carried out in static air in a muffle furnace (KSL-1500X–S, Hefei Kejing Material Technology). The temperature was set to 650 °C at a heating rate of 5 °C·min⁻¹. The discontinuous weighing method [40] was employed in this study. After the prescribed time intervals (5 min, 15 min, 30 min, 1 h, 4 h, 24 h, 48 h, 84 h, 120 h), the samples were taken out to measure the oxidation mass gains. Each sample was weighed using an analytical balance (MS105DU, Mettler Toledo).

Table 1.

Chemical compositions (in wt.%) of the four FSSs.

FSS	Fe	Cr	Mn	Si	Ni	C	P	S	N	Nb	Ti	Mo	Cu
TS430	Bal.	16.18	0.38	0.38	0.086	0.0534	0.018	0.002	0.036	–	–	–	–
TY441	Bal.	17.9	0.10	0.27	–	0.01	0.021	0.001	–	0.43	0.18	–	–
YG442	Bal.	19.36	0.10	0.48	–	0.009	0.021	0.001	0.01	0.52	–	–	0.44
TY445	Bal.	21.73	0.23	0.19	–	0.011	0.018	0.001	0.015	0.14	0.22	1.27	–

A white light interferometer with phase shift interferometry was adopted to measure the surface roughness of bare FSS samples without surface polishing. Then, the surfaces of four FSSs were polished by argon ion and characterized by electron ion beam diffraction (EBSD) to obtain the distributions of grain size and orientation. The EBSD test was performed with a field emission scanning electron microscope (Zeiss Merlin, Carl Zeiss AG, Oberkochen, Germany) in an acceleration voltage of 20 kV and a sample tilt of 70° together with a step size of 0.75 μm. The scanning electron microscope (SEM) was also used to characterize the micromorphology. The energy dispersive spectrometer (EDS) was employed to analyze the distribution of constituent elements. The acceleration voltage was set to 15 kV. The phase composition of the oxide scale was detected by X-ray diffractometer (XRD) using Cu Kα radiation.

Both sides of the sample were painted with silver paste, and one silver mesh in 1 cm × 1 cm was placed on each side. The samples were heated in a tube furnace to 650 °C in the air atmosphere. The ASR was measured using an electrochemical workstation in direct current (DC) mode with a four-point probe. A current of 100 mA·cm⁻² was imposed by the electrochemical workstation and the voltage drop was measured. The ASR is determined as the product of the resistance and the nominal contact surface area between the oxide scale and the steel substrate, and expressed by

$$ASR=\frac{US}{2I}$$ (1)

where U and I are the measured voltage (V) and current (mA), respectively, and S is the effective contact area (cm²).

3. Results

3.1. Grain size and surface roughness of bare FSSs

The FSS sheets are produced by cold rolling. The grain size and orientation of each FSS sample are characterized by EBSD. In Fig. 1, the left column shows the inverse pole diagram (IPF-Z) perpendicular to the FSS surface, and the statistical bar diagram of grain size is displayed on the right. The corresponding statistical results are listed in Table 2. In the IPF-Z diagram, the red grain has a [001] crystal axis, the blue grain has a [111] crystal axis, and the green grain has a [101] crystal axis. The grains of the four FSSs manifest a random orientation. The grain size of TS430 is the smallest, which is normally distributed with an average grain size of 7.4 ± 4.4 μm. The average grain sizes of TY441 and YG442 are 11.7 ± 10.8 μm and 11.3 ± 8.2 μm, respectively. Meanwhile, the grains of YG442 are smaller than TY441. The grain size of TY445 is the largest and distributed most unevenly. Small grains exist in the gap of large grains. The maximum grain size of TY445 is 76.17 μm, whereas the average grain size obtained statistically is 5.8 ± 10.6 μm. The difference in grain size highly depends on the processing technology and will affect the corrosion behaviors of the FSSs.

Fig. 1.

Results of grain size distribution of bare FSSs: (a) TS430; (b) TY441; (c) YG442; (d) TY445.

Table 2.

Statistics of grain sizes of the four FSSs.

FSS	TS430	TY441	YG442	TY445
Ave (μm)	7.4	11.7	11.3	5.8
Max (μm)	23.3	41.1	36.1	76.2
Min (μm)	0.9	0.9	0.9	0.9

The surface roughness was measured by a white light interferometer. The pictures are shown in Fig. 2. Table 3 lists the averaged values of surface roughness, where Ra is the arithmetic mean deviation of the roughness profile, Rp is the maximum contour peak height, and Rq is the root-mean-square deviation of the profile. The surfaces of TS430 and YG442 are basically mirrored and the Ra values are much lower at 0.070 μm and 0.031 μm, respectively. The surface roughnesses of TY441 and TY445 are relatively larger and the Ra values are 0.372 μm and 0.421 μm, respectively. Investigations indicated that surface roughness had an impact on the oxidation resistance and Cr volatilization [41]. A rougher surface is exposed to the ambient with a larger surface area which results in a higher oxidation rate and Cr vaporization rate. However, in this study, the effect of surface roughness on initial oxidation is obscure compared with that under high termperature conditions.

Fig. 2.

Surface roughnesses of the four FSSs: (a) TS430; (b) TY441; (c) YG442; (d) TY445.

Table 3.

Results of surface roughness of the four FSSs.

FSS	TS430	TY441	YG442	TY445
Ra (μm)	0.070	0.372	0.031	0.421
Rp (μm)	2.698	4.049	0.16	1.47
Rq (μm)	0.096	0.665	0.039	0.518

3.2. Morphology of oxide scales

Fig. 3(a) shows the macroscopic morphology of the four FSSs at different stages of oxidation at 650 °C. After a short-term oxidation, the surfaces of the four FSSs appear slightly gold or blue, indicating that the surfaces are covered with a nano-scale oxide layer partially transparent to light. Thus, visible interference colors are observed under the white light [42].

Fig. 3.

Macroscopic morphology and surface microstructure of the samples after oxidation at 650 °C: (a) surface after oxidation at different time intervals; (b) SEM view of TS430 after 120 h; (c) SEM view of TY441 after 120 h; (d) SEM view of YG442 after 120 h; (e) SEM view of TY445 after 120 h.

The XRD patterns of the original and oxidized samples are compared in Fig. 4. The black profile refers to the original one. It can be seen that the original bare FFSs have typical Fe–Cr characteristic diffraction peaks corresponding to (110), (200), and (211) crystallographic planes at about 45°, 65°, and 82° diffraction angles, respectively. The diffraction peak at 45° in the original TS430 is weak, which may be attributed to the texture structure. Compared with the original samples, the characteristic peaks of Cr₂O₃ and (Mn, Cr)₃O₄ are detected on the oxidized TS430 and YG442. The characteristic peak of Cr₂O₃ on TS430 is stronger. Whereas, the characteristic peak of (Mn, Cr)₃O₄ spinel on YG442 is higher. Only diffraction peaks of Cr₂O₃ are found on TY441 and TY445 and the peak intensity of Cr₂O₃ is larger on TY441.

Fig. 4.

XRD images after 120 h oxidation at 650 °C: (a) TS430; (b) TY441; (c) YG442; (d) TY445. (Upper: oxidized samples; Lower: original bare samples.)

The surface microstructures are displayed in Fig. 3(b) through (e). Fig. 5 shows the distributions of elements on the surface oxide scale. The flaky Cr₂O₃ particles form with a corundum structure, while the particles with prism and diamond shapes are (Mn, Cr)₃O₄ spinel. It can be seen that TS430 is corroded most seriously, on which the size of lamellar Cr₂O₃ grains is as large as 3 μm, and (Mn, Cr)₃O₄ is precipitated in the gap. The surface of TY441 is dominated by lamellar Cr₂O₃ oxides with a size of about 1 μm, significantly smaller than that of TS430. No apparent (Mn, Cr)₃O₄ spinel phase is observed on TY441, mainly because the Mn content in TY441 is merely 0.1 wt%, and the diffusion rate of Mn is apparently slow at 650 °C. The surface of YG442 is covered by a dense Cr₂O₃ layer and cubic (Mn, Cr)₃O₄ spinel oxides appear on the top. In comparison, the surface of TY445 is primarily covered by fine Cr₂O₃ particles and only a few of scattered (Mn, Cr)₃O₄ particles with almost the same size as that on YG442.

Fig. 5.

Results of EDS elemental analysis for the FSSs after oxidation at 650 °C for 120 h: (a) TS430; (b) TY441; (c) YG442; (d) TY445.

The corresponding images of the cross sections are shown in Fig. 6. The samples were embedded in epoxy resin and then polished before SEM testing. The thickness of the oxide scale was obtained by Image J software which is the average of 10 evenly distributed points. The oxide scale on TS430 is the thickest, with the average value of about 1 μm In contrast, the oxide scales on TY441 and YG442 are smoother, with a thickness of 0.4 μm and 0.2 μm, respectively. The thickness of the oxide scale on TY445 is close to YG442. Notably, it is found that Mn diffuses from the steel matrix to the oxide layer in all four FSSs although the relatively low temperature of 650 °C is used in this study.

Fig. 6.

SEM images and EDS scanning results on the cross sections after oxidation at 650 °C for 120 h: (a) TS430; (b) TY441; (c) YG442; (d) TY445.

Some white intermetallic precipitates appear inside TY445 as shown in Fig. 6(d), which is a Laves phase composed of (Fe, Cr)₂(Nb, Mo, Ti). The formation of Laves phase can improve the high temperature strength and creep resistance. In comparison, only a small amount of Laves phase is observed in TY441 and YG442. It is worth noting that no obvious SiO₂ grains are found beneath the oxide scale for the four FSSs. In addition, no internal rutile TiO₂ oxidation zone is observed in TY441 and TY445 although they contain 0.18 wt.% and 0.22 wt.% of Ti, respectively. Meanwhile, no obvious Cu-rich region is found below the inner oxide layer of YG442, and no Cu is detected in the surface spinel phase. This may be due to a small Cu content, low operation temperature, and short oxidation time.

3.3. Oxidation mass gain

The oxidation mass gain is often employed as an index representing the oxidation resistance. The parabolic rate constant can be determined according to the Wagner equation [43] and is expressed by

$${\left(\frac{\Delta \mathit{m}}{\mathit{S}}\right)}^{\mathbf{2}}={\mathit{k}}_{\mathit{p}}\mathit{t}+\mathit{C}$$ (2)

where $\Delta m/S$ is the oxidation mass gain per unit area (mg·cm⁻²), $k_p$ is the parabolic rate constant (mg²·cm⁻⁴·s⁻¹), $t$ is the oxidation time (s), and $C$ is the integral constant related to the start of the parabolic kinetics. The parabolic rate $k_p$ can be correlated as the slope of the previous linear function with respect to time. The measured mass gains and the associated oxidation rate constants are listed in Table 4 (the left column). The results of the four FSSs are obtained under the same conditions, i.e., oxidation at 650 °C for 120 h in air. The four FSSs generally manifest a parabolic trend, meaning that the control step of oxidation at 650 °C is the diffusion of ions into the oxide scale. The mass gain of TS430 is greater by almost one order of magnitude compared with the others. Accordingly, the oxidation rate constant of TS430 is apparently larger than the others. The oxidation rate constants of YG442 and TY445 are approximate and about one-third of TY441. The corrosion resistance primarily follows the sequence of TS430 < TY441 < YG442 < TY445.

Table 4.

Results of mass gains and oxidation rate constants for the four FSSs with/without Cr volatilization correction*.

FSS	Mass gain		Oxidation rate constant		Cr2O3 scale thickness
	(mg·cm−2)		(mg2·cm−4·s−1)		(μm)
TS430	0.110	0.120	7.68 × 10−9	8.80 × 10−9	0.67	0.73
TY441	0.029	0.038	8.07 × 10−10	2.01 × 10−9	0.17	0.23
YG442	0.013	0.023	3.03 × 10−10	8.05 × 10−10	0.08	0.14
TY445	0.011	0.021	3.01 × 10−10	7.25 × 10−10	0.07	0.13
*Data without correction are listed in the left column while the right column are those with Cr volatilization correction.

The measured mass gains of TY441 (Cr 17.9 wt.%) and YG442 (Cr 19.36 wt.%) are compared with AISI441 (Cr 17.8 wt.%) [44] and Sanergy HT (Cr 20.4 wt.%) [45] as shown in Fig. 7(a). AISI441 was exposed to humidified air in a flow of 40 ml·min⁻¹·cm⁻² and Sanergy HT was oxidized in the air/3 % H₂O environment with a flow rate of 6000 sml·min⁻¹. The trend of TY441 in this study is similar to AISI441 at 650 °C while the mass gains of YG442 and Sanergy HT are approximate. Meanwhile, the mass gains of the other two FSSs under 750 °C and 850 °C are also plotted, which shows a significant influence of working temperature on the mass gain. The oxidation rate constant of AISI441 is 1.8 × 10⁻⁷ mg²·cm⁻⁴·s⁻¹ at 850 °C, 1.4 × 10⁻⁸ mg²·cm⁻⁴·s⁻¹ at 750 °C, and 8.3 × 10⁻¹⁰ mg²·cm⁻⁴·s⁻¹ at 650 °C, respectively. The oxidation rate constant of Sanergy HT is 2.0 × 10⁻⁷ mg²·cm⁻⁴·s⁻¹ at 850 °C, 7.0 × 10⁻⁹ mg²·cm⁻⁴·s⁻¹ at 750 °C, and 4.5 × 10⁻¹⁰ mg²·cm⁻⁴·s⁻¹ at 650 °C, respectively. The oxidation rate constant descends by about 18 times for every 100 °C reduction of the operation temperature.

Fig. 7.

Results of mass gains and electrical properties: (a) Comparison of mass gains of TY441 and YG442 with AISI441 [44] and Sanergy HT [45]; (b) Mass gains of the four FSSs with Cr volatilization correction (the dashed lines represent the net mass gains and the solid lines denote the gross mass gains); (c) ASR values measured at different temperatures for the four FSSs after 120 h; (d) ASR values for the four FSSs during the early oxidation process at 650 °C.

The measured mass gains from the oxidation test for the four FSSs are shown as the solid lines in Fig. 7(b). Uncertainties are expressed by the error bars, which are determined by a method similar to that in Ref. [46]. During the test, no obvious spalling was observed. The oxidation mass gains of all steels are positive and Cr evaporation is relatively small due to the exposure to the static air in the muffle furnace. The mass gain of TS430 increases evidently as the prolongation of time. The mass gain of TY441 ascends as well. In comparison, the mass gains of YG442 and TY445 first increase and then maintain at a low level.

A thin and continuous oxide scale occurs on the surfaces of FSSs after oxidation at 650 °C for 120 h seen as in the SEM images of both surfaces (Fig. 3) and cross-sections (Fig. 6). It is known that the chemical stability of thermally grown oxide (TGO) on steel surface is related to the minimum concentration of Cr in alloy [47]. When the operating temperature was 850 °C, a minimum concentration of Cr ranged from 9.1 wt.% to 16.8 wt.% to obtain a stable TGO layer. Crofer 22 APU with a Cr content of about 22 wt.% has been developed particularly for SOFCs. However, the cost is very high. The Cr contents of the four commercial FSSs range from 16.18 wt.% to 21.73 wt.%, exceeding the minimum threshold and covers for most of the ordinary scope of Cr content for SOFC applications.

The thickness of oxide scale l is determined based on the measured mass gain assuming the continuous oxide layer is dense Cr₂O₃ according to Tsai et al. [48].

$$\mathit{l}=\frac{\mathit{w}{\mathit{M}}_{\mathit{o}\mathit{x}}}{\mathbf{3}{\mathit{M}}_{\mathit{o}}{\mathit{\rho }}_{\mathit{o}\mathit{x}}}$$ (3)

where w is the measured mass gain (g⋅cm⁻²); M_ox and M_o are the molecular weight of Cr₂O₃ (g⋅mol⁻¹) and oxygen (g⋅mol⁻¹), respectively; ρ_ox is the density of Cr₂O₃ (g⋅cm⁻³). Herein, the scale thickness is determined with an assumption of pure Cr₂O₃ and no (Mn, Cr)₃O₄. Therefore, certain deviations exist compared with the measured scale thicknesses. After 120 h oxidation at 650 °C, the thickness of Cr₂O₃ layer is about 0.67 μm for TS430 while it is much less for the remaining three FSSs.

The release of volatile Cr substances will affect the measured mass gains. Therefore, the amount of Cr volatilization is estimated. The considered reaction is

$$\frac{1}{2}{Cr}_2{O}_3\left(s\right)+H_{2}O\left(g\right)+\frac{1}{4}{O}_2\left(g\right)=Cr{\left(OH\right)}_2{O}_2\left(g\right)$$ (4)

The Cr evaporation rate is estimated using the method of Holcomb et al. [49]. Volatility is constrained to the transport of volatile species through the gas phase boundary layer and the involved gases are assumed ideal gases. For the laminar flow along a flat plate, the Cr volatilization rate $k_e$ (kg·m⁻²·s⁻¹) is determined by

$$k_e=0.664{Re}^{0.5}{Sc}^{0.343}\frac{D_{AB}{M}_i}{LRT}{P}_{\text{Cr}{\mathrm{O}}_2{\left(\text{OH}\right)}_2}$$ (5)

where Re and Sc are dimensionless Reynolds and Schmidt numbers, respectively, D_AB is the gas diffusion coefficient (m²·s⁻¹) between gaseous Cr(OH)₂O₂ and the solvent gas, M_i is the molecular mass (kg·mol⁻¹) of Cr(OH)₂O₂, L is the length (m) in the flow direction, R is the gas constant and P_CrO2(OH)2 is the partial pressure (atm) of CrO₂(OH)₂.

The estimated thicknesses of the Cr₂O₃ scales on the four FSSs are listed in Table 4, where the left column shows results determined by the measured oxidation gains and the right column gives the corresponding values corrected by the above method of Cr volatilization. Cr volatilization has a great impact on the estimated thickness of oxide scale. An average increase of 51.2 % is obtained for the thickness after considering the Cr volatilization. It is more prominent as the Cr content increases. The thickness increment ranges from 8.96 % to 85.7 % as the Cr content of FSS rises from 16.18 wt.% to 21.73 wt.%. The estimated Cr evaporation rates are compared with other studies in Table 5. It can be seen that the results of this study approach to the evaporation rates of the other relevant studies. The volatilizing rate in flowing air is higher than in a saturated static air atmosphere according to Sachitanand et al. [50]. The results of this study are slightly greater than those in Ref. [45]. The evaporation rate reduces by half for every 100 °C reduction of the working temperature. Similar results have been reported in Ref. [51]. However, the evaporation rate of Cr under 650 °C is still the same order of magnitude with those under 750 °C and 850 °C. Therefore, the effect of Cr volatilization on the estimation of the mass gain is more important at 650 °C.

Table 5.

Comparison of Cr evaporation rates in different studies.

Alloy	Air flow rate	ke (kg·m−2·s−1)			Ref.
		650 °C	750 °C	850 °C
This work	0.02 m⋅s−1	1.6 × 10−10	3.4 × 10−10	6.5 × 10−10
Crofer 22H	6000 smL⋅min−1	1.1 × 10−10	2.5 × 10−10	6.67 × 10−10	[45]
Sanergy HT	6000 smL⋅min−1	2.62 × 10−10	5.69 × 10−10	8.64 × 10−10	[52]
AISI 441	40 ml⋅min−1⋅cm−2	2.0 × 10−8	1.3 × 10−7	3.1 × 10−7	[44]
–	–	4 × 10−8	1.5 × 10−7	3 × 10−7	[53]

The gross mass gains of the four FSSs are determined taking into account the Cr volatilization. The results are also displayed in Fig. 7(b). It can be seen that the effects of Cr volatilization are more prominent as the increase of Cr content. This is because the mass gains are lower for the FSS with a higher Cr content while the mass losses due to Cr evaporation are the same. Meanwhile, the parabolic oxidation rate decreases significantly as the temperature drops whereas the decline of the evaporation rate is much less. In addition, the Cr volatilization rates of FSSs are in the same order of magnitude as the oxidation rates at 650 °C. If the Cr volatilization rate is higher than the oxidation rate, a continuous passivation layer of Cr₂O₃ cannot be formed, which will cause breakaway oxidation and the occurrence of a non-protective iron-rich oxide [54]. Therefore, the negative effects caused by Cr volatilization under intermediate temperatures such as 650 °C can not be ignored and coatings must be applied to effectively suppress the Cr volatilization although the rate is much lower [55].

3.4. Electrical conductivity property of the oxide scale

In order to evaluate the electron conduction properties of the oxide, ASR tests were performed on steel samples oxidized at 650 °C for 120 h. Due to the high conductivity of the alloy material, the resistance of the FSS substrate can be ignored compared with the oxide scale. The results of ASR measured under different temperatures are shown in Fig. 7(c). The oxide scale exhibits semiconductor characteristics. As the temperature increases from 400 °C to 650 °C, the ASR decreases, indicating that the electrical conductivity is enhanced. At 650 °C, the ASR values are in the range of 3.11–5.81 mΩ·cm². The ASR decreases with the increase of Cr content primarily. TS430 exhibits the largest ASR in 5.81 mΩ·cm², followed by TY441 in 5.00 mΩ·cm². The ASR values of YG442 and TY445 are evidently less than TS430 and TY441. Accordingly, the electronic conductivity of the oxide scale follows the order of TS430<TY441<YG442<TY445 and agrees well with the oxidation resistances. The elevation of Cr content is beneficial to the oxidation resistance and electronic conductivity. A higher Cr content yields a denser and thinner Cr₂O₃ layer that determines the electronic conductivity principally.

A linear function is manifested between ln (ASR/T) and T⁻¹. Hence, the thermally-activated conductivity follows the Arrhenius equation [56].

$$\frac{\mathit{A}\mathit{S}\mathit{R}}{\mathit{T}}=\mathit{A}\mathit{exp}\left(\frac{{\mathit{E}}_{\mathit{a}}}{\mathit{k}\mathit{T}}\right)$$ (8)

where A is the pre-exponential constant (Ω·cm²·K⁻¹), E_a is the activation energy (eV), and k is the Boltzmann's constant (eV·K⁻¹). According to the above equation, the least square method is used to determine the activation energy of the conductivity. The results are listed in Table 6. The activation energies of Cr₂O₃ range from 0.24 to 0.82 eV, which is the same order of magnitude as that in Ref. [57]. TY445 exhibits the lowest E_a owing to the thinnest Cr₂O₃ scale and the electronic conductivity is the highest.

Table 6.

Results of the activation energies E_a and the pre-exponential constants A.

FSS	TS430	TY441	YG442	TY445
Ea (eV)	0.82	0.54	0.57	0.24
A (Ω·cm2·K−1)	2.05 × 10−9	5.64 × 10−8	1.93 × 10−8	1.98 × 10−6

The measured results of ASR for the four FSSs are compared during the oxidation process and shown in Fig. 7(d). The ASR test was repeated at different oxidation intervals (5 min, 15 min, 30 min, 1 h, 4 h, 48 h, 84 h and 120 h). Abrupt increments occur such as TY441 at 24 h, TY445 at 84 h, and YG442 at 84 h. This may be attributed to the deviations caused by the discontinuous oxidation method. Meanwhile, the ASR fluctuates obviously during the first hour, which is because of the variation of the test environment and the nucleation and growth of Cr oxide in the early oxidation stage. The ASR of each FSS exhibits a trend of slow increase with the extension of time. This is similar to that of mass gain. The ASR results of the four FSSs are much lower at 650 °C and less than 6 mΩ·cm² after 120 h.

4. Discussion

Factors incorporating the chemical composition, grain size, surface roughness, and sample thickness will influence the oxidation resistance and electronic conductivity. The thicknesses of the four FSSs are different in this study due to market restrictions. However, all are greater than 0.3 mm. It is usually believed that the crosstalk phenomenon is minor when the thickness exceeds 0.1 mm. Meanwhile, a continuous and dense oxide layer is formed and no breakaway oxidation is observed for the four FSSs during the early 120 h oxidation at 650 °C. It shows that there is no Cr depletion caused by high Cr consumption [58]. The electronic conductivity of metal bulk is greater than the oxide scale by magnitude of orders. Therefore, the effects of thickness variation of the FSSs on the oxidation behavior and electrical properties can be neglected in this study.

The average surface roughnesses follow the order of TY445 > TY441 > TS430 > YG442. Generally, the higher the surface roughness, the larger the contact area with oxygen and the greater surface defects, hence the more prone to oxidation corrosion. The low roughnesses of TS430 and YG442 are beneficial for the improvement of corrosion resistance. However, the grain sizes of the four steels are essentially greater than 0.9 μm, at least 2 times greater than that of the surface roughnesses Ra. Therefore, the effect of surface roughness on the oxidation behavior is minor compared with the most important factors grain size and chemical composition.

The grain size of FSS matrix affects the diffusion rates of Cr, Mn, and other elements from the bulk to the oxide scale, and thus the oxidation behavior of FSS. According to the formation process of a passivation film on the FSS surface [59], oxides nucleate along the metal surface and grow laterally until a continuous oxide layer is formed after oxygen is adsorbed on the surface. Grain boundaries often provide a highly diffusive path for ions transport. Therefore, oxides preferentially nucleate along grain boundaries. At this stage, the grain size affects the formation rate of a continuous and dense Cr₂O₃ layer. With the decrease in grain size, the increase of grain boundary area facilitates the outward diffusion of Cr, resulting in a decrease of the critical concentration of chromium to form a dense Cr₂O₃ layer [60]. Ralston et al. [61] reviewed the effects of grain size on the corrosion behaviors and found a positive correlation with the overall grain boundary length. In a passivation environment, nanocrystalline material with a narrow grain size distribution exhibits better corrosion resistance [62]. An investigation from Zhang et al. [63] also indicated that an inverse proportional to the grain size existed, showing a Hall-Petch relationship. Due to the smallest grain size of TS430, the Cr₂O₃ layer is formed rapidly during the initial oxidation stage. Subsequently, metal cations such as Cr and Mn and oxygen anions react with each other through solid state diffusion, and the element diffusion process becomes the determining step. All four FSSs formed a continuous Cr₂O₃ oxide scale very soon at 650 °C, and therefore, the oxidation process is controlled by the solid diffusion of metal ions [64]. The outward migration of Cr is dominated by grain boundary diffusion [34], while the dominant mechanism of Mn transport is volume diffusion [65]. Larger grain sizes will result in fewer grain boundaries and longer transport distances, thereby reducing the element diffusion rate. Stainless steels with moderate grain size and uniform distribution, as well as few grain boundaries, exhibit the best passivation film quality and excellent corrosion resistance [66]. TY441 and YG442 steels have similar average grain sizes, but the grain size distribution of YG442 is more uniform, hence the Cr₂O₃ film on YG442 is denser and more uniform. Meanwhile, more (Mn, Cr)₃O₄ spinel particles appear on YG442. For TY445, although the average grain size is small, the grain distribution is not uniform with a maximum grain diameter of about 76.2 μm. Additionally, the addition of Nb, Ti, and Mo will form Laves phases at the grain boundaries (Fig. 6(d)), which hinder the diffusion of elements along the grain boundaries. Therefore, TY445 has the thinnest oxide layer with a few (Mn, Cr)₃O₄ spinel particles on the surface.

Elemental composition is another critical factor affecting the oxidation kinetics of FSS. The Cr contents of the four FSSs range from 16.18 wt.% to 21.73 wt.%. The ternary phase diagram of Fe–Cr–O is shown in Fig. 8(a) at 650 °C with a pressure of 1 atm. The corresponding stability diagram under different partial pressures of oxygen is displayed in Fig. 8(b). The composition range of the four FSSs is labeled in the red region in Fig. 8(a) and the blue dashed lines in Fig. 8(b). For the Fe-rich region 8 in Fig. 8(a), the Cr concentration is lower than 25 at.%. When the Cr concentration is higher than 25 at.%, σ phase will be formed corresponding to the regions 9, 10, and 11 in Fig. 8(a). A brittle intermetallic compound σ phase will be formed in the temperature range of 500 °C–750 °C [67]. The nucleation and growth of σ phase in traditional FSSs are very slow. During the initial stage of oxidation at 650 °C, the σ phase is not observed in the four FSSs. However, σ phase might occur during the long-term oxidation process, especially for FSSs TY441 and TY445 that are doped with elements such as Mo, W, Si, or Ti [68]. At the beginning of oxidation under 650 °C, the body-centered cubic (BCC) phase is in equilibrium with Cr₂O₃ (region 8). As the oxidation continues, it goes through regions 7, 6, 5, 3, 2, 1 and the spinel (Fe, Cr)₃O₄, the halite FeO, and the corundum (Fe, Cr)₂O₃ will occur in sequence. The corundum is the most stable phase at 650 °C when the oxygen partial pressure is greater than 10⁻¹³ atm. The oxygen concentration in the solid solution corundum phase formed by the rhombic Fe₂O₃ and Cr₂O₃ is constant at 60 at.% (region 1 in Fig. 8(a)). As can be seen from Fig. 8(b), the equilibrium window of each phase moves towards a lower $p_{{\mathrm{O}}_2}$ at 650 °C compared with the stability diagrams given by Voramon et al. [69] at 800–1000 °C. During the oxidation process, oxygen diffuses inward and there is a gradient of oxygen concentration through the oxide scale. At the interface in direct contact with the alloy matrix, the Cr₂O₃ will be formed under a very low oxygen partial pressure. From the XRD results in Fig. 4 and surface morphology in Fig. 3, only Cr₂O₃ corundum phase and (Mn, Cr)₃O₄ spinel phase are observed. This may be attributed to the doped elements and impurities as well as the disturbance of the matrix due to a very thin oxide scale. The Cr content of FSS affects the relative fractions of the surface oxides [55]. A greater Cr content in the (Fe, Cr)₂O₃ corundum phase occurs with a higher Cr content in the matrix, and a stronger protection of the oxide layer is obtained.

Fig. 8.

Phase diagram of Fe–Cr–O at 650 °C: (a) isothermal ternary phase diagram; (b) stability diagram (the blue dashed lines represent the Cr contents of the four FSSs). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

A high Cr content is critical for the formation of a continuous and dense Cr₂O₃ scale. FSS with a higher Cr content will form the Cr₂O₃ passivation layer more rapidly. The Cr content of TY445 is the highest at 21.73 wt.%, which approaches Crofer 22 APU (22.92 wt.%), ZMG232L (22.0 wt.%), and Sanergy HT (22.4 wt.%). In addition, compared with the other three steels, TY445 is dominated by large grains, and the Cr₂O₃ granules formed on the surface are smaller and thinner after undergoing the same oxidation process. Therefore, the electrical properties of TY445 are better. The top surface of YG442 is dominated by cubic (Mn, Cr)₃O₄ spinel. The conductivity of (Mn, Cr)₃O₄ is greater than Cr₂O₃, providing a certain degree of Cr blocking and improving the oxidation resistance.

430 steel is a typical FSS and has been widely investigated as interconnects of SOFC. Compared with TS430, 0.43 wt.% Nb and 0.18 wt.% Ti are added in TY441, 0.52 wt.% Nb and 0.44 wt.% Cu are doped in YG442, 0.14 wt.% Nb, 0.22wt.% Ti, and 1.27 wt.% Mo are added to TY445. Nb, Ti, Mo, and Cu often act as strengthening elements. The influences of these elements on oxidation behavior are complex in the presence of other additive elements or impurities. It mainly lies in the difference between diffusion and oxidation kinetics and the interactive synergistic effect among them.

In this study, the Mn content in TS430 is 0.38 wt.%, and the cubic (Mn, Cr)₃O₄ spinel phase is observed among the Cr₂O₃ particles. After oxidation at 650 °C for 120 h, few Mn elements have diffused outward to the surface. Therefore, discontinuous (Mn, Cr)₃O₄ particles formed on TY445. In comparison, the Mn content in TY441 is only 0.1 wt.%, and no apparent (Mn, Cr)₃O₄ spinel phase is observed. The oxidation resistance of TY441 is better than TS430, mainly due to a larger grain size and a higher Cr content, resulting in a thin and dense Cr₂O₃ layer. In addition, Nb and Ti in TY441 tend to oxidize internally, forming Laves phase precipitation at the grain boundaries, thus slowing down the oxidation kinetics dominated by element diffusion.

The diffusion of elements is thermally activated and has a strong temperature dependence. At 650 °C, the oxidation and interaction of Si, Nb, Ti, Mo, and Cu are weak. No obvious SiO₂ particles are found in the four FSSs after 120 h. However, apparent white Laves phase precipitates occur in TY445 (Fig. 7(d)). In addition, no TiO₂ oxidation zone is observed in TY441 and TY445 in Fig. 7(b)(d), while it appears obviously at 750 °C [20]. Although the Mn content in YG442 is only 0.1 wt.%, the uniform distribution of moderate grain size enables the rapid formation of a thin and dense layer of Cr₂O₃ and promotes the formation of (Mn, Cr)₃O₄ spinel. Therefore, YG442 exhibits a lower ASR. After oxidation at 650 °C for 120 h, no Cu deposition was detected in the top (Mn, Cr)₃O₄ spinel layer. However, in the long-term oxidation process, the incorporation of Cu into the (Mn, Cr)₃O₄ layer has been found, which is conducive to the improvement of the conductivity of the oxide scale and the inhibition of Cr volatilization. Further experiments are required for long-term oxidation tests.

The influences of doping elements and grain size on the oxidation behavior are directly reflected in the morphology of oxides. After 120 h oxidation at 650 °C, flaky Cr₂O₃ particles with corundum phase on TS430 have a size of about 3 μm. Compared with TS430, the flaky Cr₂O₃ particles on TY441 are smaller by about 1 μm. This is not only because of a higher Cr content and a larger grain size of TY441, but also the addition of 0.43 wt.% Nb and 0.18 wt.% Ti. The affinity of Nb to oxygen is higher than Cr³⁺. Nb and Ti form Laves precipitates at the grain boundaries, which impedes the outward migration of Cr³⁺. The Cr₂O₃ layer on YG442 is thin and dense, and more cubic (Mn, Cr)₃O₄ spinel particles are generated, which is mainly because of a more evenly distributed of grain size. For TY445, the surface is dominated by a fine and dense Cr₂O₃ layer, and cubic (Mn, Cr)₃O₄ particles are scattered with almost the same size as those on YG442. The additions of Mo and Nb lead to more Laves phase precipitating in TY445 compared with TY441 and YG442 merely containing Nb [70]. The higher proportion of Laves phase precipitation will weaken the diffusion rate of Mn to the oxide layer. In addition, FSS dominated by a larger grain size exhibit a slower element diffusion rate. Therefore, the formation of the (Mn, Cr)₃O₄ phase in TY445 is slower than in YG442.

Finally, the cost is an important factor for the commercialization of SOFC. In the market, the prices of cold-rolled TS430, TY441, YG442, and TY445 with a thickness of 2 mm are 8,000, 12,500, 17,000, and 25,000 ¥ per ton, respectively. If the thickness decreases to 0.3 mm, customized processing is required, and the cost will increase by at least 3000 ¥. Taking into account the corrosion resistance and the cost, YG442 is recommended as interconnect or metal support for SOFCs operating below 700 °C.

5. Conclusions

The oxidation behaviors of four commercial FSSs including TS430, TY441, YG442, and TY445 with a Cr content ranging from 16.18 wt.% to 21.73 wt.% were studied. The samples were oxidized at 650 °C in air for 120 h. The mass gain and micromorphology of the oxide scale as well as electrical conductivity were estimated. The main conclusions are summarized as follows.

• 1)The oxidation resistance and electrical conductivity of the four steels follow the sequence of TY445>YG442>TY441>TS430. The ASR values of the four FSSs are all less than 6 mΩ·cm². YG442 exhibits a better overall performance and is recommended for SOFCs operating at intermediate temperature.
• 2)The oxidation scales formed on the surfaces of TS430 and TY441 are flaky Cr₂O₃ corundum phase due to the outward diffusion of Cr³⁺ at 650 °C and the chromia particles are smaller on TY441. The outward diffusion of Mn occurs although the rate is low at 650 °C. A dense Cr₂O₃ layer is observed on YG442 and TY445 with few (Mn, Cr)₃O₄ spinel particles scattered on the top.
• 3)Chemical composition and grain size are the main factors affecting the initial oxidation kinetics of FSSs at 650 °C. A thin and dense Cr₂O₃ oxide scale is formed and more (Mn, Cr)₃O₄ spinel particles are found on the surface of YG442 owing to a high Cr content and an even distribution of grain size. The oxidation kinetics and interaction of doping elements are weak in the initial stage under 650 °C, and no adverse oxidation products such as SiO₂ and TiO₂ are found. However, Laves phase due to the addition of Mo, Nb, and Ti is observed in TY441, YG442, and TY445. Long-term oxidation tests are required to further reveal the mechanisms of these additive elements.
• 4)The estimated rate constant of Cr volatilization at 650 °C is about one-half of that at 750 °C. Meanwhile, the parabolic rate constant of mass gain is significantly lower by one order of magnitude. Therefore, the estimation of Cr volatilization is more important for the measurement of mass gain at 650 °C. To meet the stack life requirement of 80, 000 h, further investigations are needed to validate the performances of these FSSs. Meanwhile, coatings must be deposited to suppress the Cr volatilization.

Data availability statement

Data will be made available on request. Further information and requests for resources should be directed to the lead contact, Enhua Wang (wangenhua@bit.edu.cn).

Ethics Declarations

Review and/or approval by an ethics committee was not needed for this study because our research work did not involve human or animal participation.

CRediT authorship contribution statement

Jingwen Mao: Writing – original draft, Visualization, Software, Investigation, Formal analysis, Data curation. Enhua Wang: Writing – review & editing, Validation, Formal analysis, Conceptualization. Youpeng Chen: Resources, Methodology, Data curation. Yadi Liu: Validation, Software, Investigation. Hewu Wang: Validation, Project administration. Minggao Ouyang: Supervision, Resources. Haoran Hu: Project administration, Funding acquisition, Conceptualization. Languang Lu: Visualization, Resources. Dongsheng Ren: Resources, Data curation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.