- Email: [email protected]
electrolysis applications by simple interconnect aluminization
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Enhancing the long-term stability of Ag based seals for solid oxide fuel/electrolysis applications by simple interconnect aluminization Xiaoqing Si a,b, Jian Cao b,*, Ilaria Ritucci a, Belma Talic a, Jicai Feng b, Ragnar Kiebach a,** a
Department of Energy Conversion and Storage, Technical University of Denmark, Risø Campus, Roskilde 4000, Denmark b State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, China
article info
abstract
Article history:
Ag-(0e8 mol%) CuO is used to successfully join aluminized ferritic stainless steel inter-
Received 14 September 2018
connect to the ceria-gadolinia (CGO) barrier layer of a solid oxide fuel/electrolysis cell by
Received in revised form
reactive air brazing at 1000 C in air. The wetting of AgeCuO on CGO is tailored by varying
6 November 2018
the CuO content. The effects of the CuO content on the joint microstructure are discussed.
Accepted 10 November 2018
The long-term stability of brazed joints is evaluated by aging in oxidizing (air) and reducing
Available online 6 January 2019
(4% H2e50% H2OeN2) atmospheres at 800 C for 250 h. An Ag-2mol% CuO braze results in the best joint stability during aging. Aluminization of the steel to create an alumina surface
Keywords:
layer provides excellent protection of the steel both during the joining process and aging in
Reactive air brazing
the 2 atm. No degradation related to steel corrosion and outward diffusion of elements
Aluminization
from the steel can be observed.
AgeCuO braze
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Solid oxide fuel cell Solid oxide electrolysis cell High temperature corrosion
Introduction Solid oxide fuel/electrolysis cells (SOFC/SOEC) have attracted considerable interest because of their high efficiency, fuel flexibility, quiet operation and ultra-low pollution [1e4]. To generate power or produce hydrogen on a large scale, several cells are combined together in a stack with interconnects between the individual cells. For planar stacks operating at intermediate temperatures (~650e900 C), ferritic stainless steels are preferred as the interconnect material [5,6]. Stable,
gas tight sealing between the cell and the interconnect is required to ensure fuel purity, a high efficiency and to maintain mechanical stability [7e10]. Over the past decade, several sealing concepts have been tested for SOFC/SOEC application, among which glass/glass-ceramic bonding [11e18] and reactive air brazing (RAB) [19e25] are the most studied. Glass/glass-ceramic materials are cost-effective and relatively easy to process, and therefore a popular choice for sealing SOFC/SOEC stacks. However, many glass/glass ceramics are prone to devitrification during operation, which affects the long-term stability in terms of microstructure and
* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (J. Cao), [email protected] (R. Kiebach). https://doi.org/10.1016/j.ijhydene.2018.11.071 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
3064
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
mechanical properties [11e16]. Furthermore, glass/glassceramic sealants typically have poor resistance to stressinduced cracking, which is especially an issue for use in mobile applications where the stresses tend to be larger [17,18]. RAB is another approach for sealing SOFC/SOEC stacks. One of the preferred materials for RAB is Ag, which is often mixed with a certain amount of Cu or CuO for improved wettability on ceramic components [19,20]. The excellent oxidation resistance of brazes based on noble metal such as Ag allows RAB to be performed directly in air, thus eliminating the need for fluxing agents or inert gases [20e22]. Although RAB has great potential for sealing SOFC/SOEC stacks, there are still some challenges currently hindering its application. The most serious is the excessive reactivity between the steel interconnect and Cu/CuO [23e25]. Formation of a more than 20 mm thick Cu/Cr/Mn/Fe-oxide scale along the braze/interconnect interface has been observed, resulting in a mechanically weak part in the joint [22e25]. The high mobility of oxygen in Ag means that oxygen easily reaches the braze/interconnect interface, resulting in a further oxidation of the steel interconnect [26]. Future development of RAB for SOFC/SOEC stacks requires development of means to control the reaction at the braze/ interconnect interface. The basic reason for corrosion at this interface is that the steel interconnect is in direct contact with the Cu/CuO-containing braze [22e26]. Thus, to mitigate corrosion, a protective coating may be fabricated on the interconnect surface to protect the steel against oxidation and block interdiffusion across the interconnect/braze interface [27,28]. A promising approach is aluminization of the steel. This creates an Al2O3 scale on the steel surface that has been shown to mitigate reaction between the steel constituents and a glass sealing material [29,30]. Because the Al2O3 scale is insulating, it will be applied only in the sealing area of the interconnect in real applications. In this paper, RAB is used to join aluminized Crofer 22 H to anode-supported half-cells. The long-term stability of these joints is tested under exposure to SOFC/SOEC relevant conditions. The evolution of the alumina layer on the aluminized steel during the RAB process and long-term exposure is studied in detail. We show that by tailoring the composition of the AgeCuO braze a balance between sufficient wetting and alleviating overreaction with the aluminized steel can be achieved.
Experimental procedure Materials and sample preparation Crofer 22 H with a thickness of 0.3 mm was used as the interconnect alloy in this study. Crofer 22 H (Thyssenkrupp, Germany) is a ferritic stainless steel containing (in wt.%) 22.3 Cr, 0.43 Mn, 0.48 Nb, 0.27 Si, 2.0 W, 0.06 Ti, 0.06 La and balance Fe. The steel was cut into 2 2 cm2 coupons and ultrasonically cleaned in ethanol for 10 min before aluminization. The procedure for reactive air aluminization consisted of three steps: (a) application and drying of an Al paste, (2) heat treatment in air, (3) removal of unreacted loose material [29,30]. A coating slurry was prepared by mixing Al powder (325 mesh, Alfa
Aesar) with a commercial binder system. The slurry was brush painted on the Crofer 22 H, applying approximately 4e5 mg cm2 of slurry. After drying the samples were heated at 1000 C for 1 h with a heating and cooling rate of 180 C h1. Finally, the samples were ultrasonically cleaned for 10 min in ethanol in order to remove any loose/unreacted Al2O3 scale on the surface. The planar anode-supported half-cells consisted of a NiO-YSZ support layer, a thin NiO-YSZ active anode layer, a dense YSZ electrolyte layer and a porous ceria-gadolinia (CGO) barrier layer. The half-cells were developed at Technical University of Denmark, and the procedure for making half-cells was described in detail elsewhere [31]. The half-cells were laser cut into 2 2 cm2 specimens, and then ultrasonically cleaned in ethanol for 10 min. For the braze, Cu powder (1 mm, 99.9%, Sigma-Aldrich) was mixed with Ag powder (1 mm, 99.9%, Sigma-Aldrich) and milled in a planetary ball mill (Retsch PM 1000) for 10 h at 150 rpm. As-milled powders were pressed (2 t, uniaxial pressing) into bars with a dimension of 20 5 2 mm3. The pressed bar was first heated uniformly by a flame gun for 1 min in order to improve its plastic deformation ability. Then the sample was rolled by a rolling press. The thickness of the pressed bar was ~2 mm, so the roller gap was first set to ~1.90 mm. After twice or more rolling at the same thickness, the sample was reheated. Each time reduction of the roller gap must be less than 0.10 mm. Finally the thickness of sample is ~100 mm, and the whole rolling process took ~20 times. Four kinds of braze sheets (~100 mm thick) with 0 mol%, 2 mol%, 4 mol% and 8 mol % of Cu were produced. An image of prepared Age2CuO braze sheet was shown in Fig. S1. The Cu is oxidized to CuO during RAB. Therefore, the four brazes are in the following denoted as Ag, Age2CuO, Age4CuO and Age8CuO, respectively. The braze sheets were cut into 2 2 cm2 pieces, and sandwiched between the aluminized Crofer 22 H and the half-cells. The CGO barrier layer was chosen as the joining surface of the half-cells. A normal load of 1 kg cm2 was applied on the sandwich to maintain proper contact. The assemblies were heated under load in air with 100 C h1 to the sealing temperature of 1000 C where they were kept for 20 min before cooling down to room temperature with 60 C h1. Wetting of the molten braze on the ceramic substrate plays an important role in the joining process [31e34]. Especially, the CGO layer used in this study has a porous structure, which may affect the wetting of AgeCuO braze. To study the wetting of AgeCuO brazes, small pieces of the AgeCuO sheets with different CuO amounts were individually placed on top of the CGO barrier layer on the half-cells and heat treated at 1000 C for 20 min using the same heating (100 C h1) and cooling (60 C h1) rates. That is, no load and no Crofer 22 H was used for the wetting experiments. After cooling the samples were sectioned and the contact angle between the AgeCuO and the CGO layer was evaluated from images obtained by scanning electron microscopy (SEM, TM3000, Hitachi).
Long-term test in reducing and oxidizing atmospheres To investigate the stability of joints in reducing and oxidizing atmospheres two sets of aging experiments were performed. For this purpose, two samples of each braze composition were
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
tested in both atmospheres. The sealed joints were placed inside an alumina crucible and exposed to 800 C for 250 h in a horizontal tube furnace. The heating and cooling rates were 100 C h1. For the long-term test in reducing atmosphere (4% H2, 50% H2O, 46% N2), the H2 was diluted by an inert gas (N2) to keep a safe concentration. The steam was supplied by flowing H2eN2 through a water bubbler heated to 80 C for 50% humidity. The gas flow was kept constant at 6 L h1 during the long-term test in reducing atmosphere. The long-term test in oxidizing atmosphere was conducted in stagnant air.
Characterization of joint properties test The microstructure and composition of the as-joined and aged samples were characterized employing a tabletop scanning electron microscope (SEM, TM3000, Hitachi) equipped with an energy dispersive spectrometer (EDS). Samples for SEM analysis were vacuum embedded in epoxy resin (Epofix, Struers) and cut along the center of the assembly. The cross-sections were ground by SiC paper and diamonds in suspension in successive steps down to 1 mm. Phase analysis of sample surfaces was conducted by an X-ray diffraction spectrometer (XRD, D8-Advance, Bruker) with Cu-Ka radiation. The scans were made between 20 and 90 with a step size of 0.02 and a speed of 1 s per step. The shear test was carried out at roomtemperature using a bond tester (Dage 4000, Nordson dage, UK) at a constant speed of 0.1 mm/min. Five samples were tested for each process parameter, and the average value of the shear strength was obtained.
3065
Results and discussion Characterization of aluminized Crofer 22 H Representative SEM images and EDS elemental maps of the Crofer 22 H steel surface and cross section after reactive air aluminization are shown in Fig. 1. The steel surface is covered by a continuous, dense and crack free layer of Al2O3 having a uniform thickness of ~2 mm. The surface is wavy with an up to 12 mm distance between the highest peak and lowest valley, and the average roughness (Ra) is 4.2 mm. A mechanical interlocking structure can be formed between the braze and the rough aluminized Crofer 22 H, which may improve the interfacial bonding [35]. No other oxidation products (e.g. Cr2O3) are observed on the SEM cross sections. XRD analysis of the aluminized Crofer 22 H indicates that the surface layer consists of pure a-Al2O3, as shown in Fig. 1c. No Cr2O3 or any secondary phase indicating a reaction between the Al2O3 and the Crofer 22 H is detected. The surface morphology of the Al2O3 protective layer is displayed in the top-view image in Fig. 1d. The surface is relatively coarse with many staggered protrusions. To investigate the thermal stability of the aluminized Crofer 22 H, a sample was heated following the same procedure used for the RAB process (1000 C/20 min). Inspections after this exposure showed that the Al2O3 layer remains intact and that outward diffusion of Cr and Fe from the steel is effectively blocked (Fig. S2 and Fig. S3 in the supplementary
Fig. 1 e Typical results after aluminization of Crofer 22 H: (a) cross-sectional SEM image (b) high magnification view from Fig. 1a with EDS maps, (The magnification of the EDS maps is the same as the corresponding SEM image) (c) XRD scan of the aluminized Crofer 22 H, (d) surface morphology of the Al2O3 layer.
3066
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
material). The good stability of the aluminized steel is the key for preventing damage of the Crofer 22 H substrate during RAB and subsequent long-term exposure, as will be discussed in Sections Reactive air brazing of SOFC/SOEC components and Stability of joints in oxidizing and reducing atmospheres.
Wetting behavior of AgeCuO braze on the CGO Fig. 2 displays cross-sectional micrographs of the braze droplet/CGO interface and the EDS Ag elemental map with higher magnification images showing details of the wetting
Fig. 2 e Micrographs of cross-sectional droplet/CGO with element Ag mapping analysis after cooling to room temperature from 1000 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (The magnification of the EDS maps is the same as the corresponding SEM images).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
triple line (zone 1, Fig. 2a) and the central interface (zone 2, Fig. 2a). Although pure Ag apparently bonds to the CGO, the contact angle is high (~72 ) and the Ag braze does not penetrate into the pores of the CGO (Fig. 2a). Fig. 2b illustrates that addition of Cu improves the wettability of the Ag braze. Accordingly, the Age2CuO braze has a lower contact angle on CGO (~35 ) compared to pure Ag. Characterization of the wetting triple line (zone 1, Fig. 2b) reveals that the Age2CuO braze partially penetrates into the pores of the CGO. Apparently, the Age2CuO barely spreads in the pores of the CGO and the penetration front does not reach the CGO/YSZ interface. The higher magnification SEM image of the braze droplet (zone 2, Fig. 2b) shows the presence of CuO precipitates inside the braze. The high-magnification SEM image of the central interface (zone 3, Fig. 2b) and the corresponding Ag map further confirms that the Age2CuO braze only partially penetrates into the CGO. SEM/EDS images of the sample with the Age4CuO braze are shown in Fig. 2c. The wetting on CGO is further improved and the contact angle is decreased to ~25 . The Age4CuO braze spreads into the pores of CGO as well as on the CGO surface. The spreading distance along the surface was measured to ~0.5 mm, as shown in the high-magnification image of zone 1 (Fig. 2c). The high-magnification images of zones 2 and 3 (Fig. 2c) show that Ag fully penetrates the porous parts of the CGO, and that the braze reaches all the way to the CGO/YSZ interface. Similar results were obtained with the Age8CuO braze (Fig. 2d). The Age8CuO braze infiltrates the pores of the CGO all the way to the CGO/YSZ interface (zones 2 and 3, Fig. 2d), and apparently adheres to the YSZ electrolyte. The Age8CuO braze also spreads on the CGO surface, in this case with a longer spreading distance of ~0.8 mm (zone 1, Fig. 2d). A longer spreading distance implicates a risk of cathode contamination if the cathode is too close to the sealing area. It is noted that even in the case of a high CuO concentration (Age8CuO) no CuO precipitates are detected in the pores of CGO, although Cu obviously plays an important role in facilitating wetting. These results are consistent with those reported by Meier et al.
3067
[34] for AgeCuO/Al2O3 and Seeharaj et al. [36] for AgeCuO/ CGO, in which the CuO precipitates were found only at the interface and within the Ag matrix. Fig. 3 shows the contact angle of AgeCuO braze on the porous CGO substrate as a function of CuO content. Increasing the CuO content from 0 to 8 mol% promotes wetting of the AgeCuO braze, corresponding to a significant reduction in the contact angle from ~72 (Ag) to ~18 (Age8CuO). Seeharaj et al. [36] studied the wetting of an AgeCuO braze on a denser CGO substrate. A comparable correlation between the wetting angle and the CuO content was observed, as shown in Fig. 3. Comparing our results with those of Seeharaj et al. [36] shows that in all cases the AgeCuO brazes have a smaller contact angle on the porous CGO than on the denser CGO. This difference may be due to the differences in the experiment condition and samples. However, a same trend is presented on two kinds of CGO, that is, the improved wetting with increasing CuO content improves the penetration of AgeCuO braze into the CGO pores, as shown in Fig. 3. The large surface tension of pure liquid Ag and the large contact angle on CGO impedes the penetration of pure Ag into CGO [36]. Adding CuO lowers the surface energy/tension of liquid Ag, which is partially responsible for the improved wetting of AgeCuO on CGO. Consequentially, liquid Ag penetrates into the pores of the CGO owning to the capillary force imposed on the Ag. In case of the Ag-4/8CuO brazes, the penetrate depth exceeded 7.8 mm (the thickness of the CGO layer).
Reactive air brazing of SOFC/SOEC components The Age2CuO braze was used to join aluminized Crofer 22 H to the CGO barrier layer of a NiO-YSZ anode-supported half-cell. This kind of joint reflects a relevant assembly in planar SOFC/ SOEC stacks. Fig. 4 shows overview SEM images and corresponding EDS maps of the joined interfaces after the RAB process. A defect free joint between aluminized Crofer 22 H and the CGO is obtained. The Age2CuO braze wets the surface of the aluminized Crofer 22 H and CGO sufficiently, and no defects such as cracks or pores can be found (Fig. 4a). High
Fig. 3 e Contact angle of AgeCuO brazes on the CGO with pores and the dense CGO [36] as a function of CuO content, and the schematic diagrams of the penetration.
3068
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
Fig. 4 e (a) SEM image of aluminized Crofer 22 H brazed to the CGO layer of a NiO-YSZ anode-supported half-cell using Age 2CuO. High magnification SEM images and corresponding EDS maps of region 1, 2 and 3 are shown in (b), (c) and (d), respectively. (The EDS maps have the same magnification as the corresponding SEM image). Points A, B, C and D mark EDS point analysis in Table 1.
magnification images of zones 1 and 2 from the CGO side are shown in Fig. 4b and c, respectively. EDS analysis (Table 1) (pt A, Fig. 4b, pt B, Fig. 4c) confirms that the light-grey phase along the CGO interface is CuO. The elemental Ag map reveals that in some places the braze fully penetrates into the pores of the CGO (Fig. 4c). This is in contrast to the observations of the wetting experiment (Fig. 2b) and can be attributed to the pressure applied during the RAB joining process (1 kg cm2), which promotes infiltration into the porous structure [36]. No reaction between the CGO and the braze can be found in Fig. 4b and c. A high magnification image of zone 3 from the
Table 1 e EDS composition (at. %) for the points highlighted in Fig. 4. (Important values marked in bold). Points
Cu
Al
O
pt A pt B pt C pt D
50.4 51.6 16.1 5.8
0 0 28.5 36.1
41.4 40.6 48.6 50.5
Ag Fe 8.2 7.8 5.7 6.3
e e 0.3 0.4
Cr Mn Possible phase e e 0.7 0.7
e e 0.1 0.2
CuO CuO CuAl2O4 Al2O3
aluminized Crofer 22 H side is shown in Fig. 4d. Based on EDS analysis (pt C, Fig. 4d), the dark-grey phase along the Al2O3/ braze interface is assigned to be CuAl2O4, which likely has formed through the reaction of Al2O3 þ CuO / CuAl2O4. This would be in agreement with results from Kim et al. [22] and Cao et al. [37], who also reported the formation of CuAl2O4 when CuO reacted with Al2O3 during RAB. Nevertheless, the Al2O3 layer remains continuous and fully covers the steel surface (pt D, Fig. 4d). Elemental maps of Cu, Al and O indicate the distribution of the CuAl2O4 phase, and further confirm the integrity of the Al2O3 layer. Additionally, elemental maps of Fe, Cr and Mn indicate that the outward diffusion of these elements from the steel is completely blocked by the Al2O3 layer. Thus, the Crofer 22 H steel is very well protected during RAB. In the following, the effect of the CuO concentration on the interfacial microstructure is discussed. Overview SEM images of assemblies with different CuO concentrations in the braze after joining at 1000 C under a weight load of 1 kg cm2 are presented in Fig. 5. Good/sufficient bonding to the aluminized Crofer 22 H and the CGO was found in general, irrespective of
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
3069
Fig. 5 e SEM images of joints brazed using different brazes: (a) Ag, (b) Age4CuO, (c) high magnification of the red marked region in Fig. 5b, (d) Age8CuO, (e) high magnification and corresponding EDS maps of the red mark region in Fig. 5d. (The magnification of the EDS maps is the same as the corresponding SEM image). Points A, B, C, D, E and F mark EDS point analysis in Table 2. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
the CuO amount in the braze. Pure Ag does not penetrate into the pores of the CGO (Fig. 5a) even under the applied pressure (as discussed in Section Stability of joints in oxidizing and reducing atmospheres, see later). The Age4CuO braze fully penetrates into the CGO, and apparently adheres to the halfcell (Fig. 5b). The high-magnification SEM image of the braze/aluminized Crofer 22 H interface indicates the presence of CuO between the CuAl2O4 reaction layer and the braze (EDS analysis Table 2, pt A, Fig. 5c). It appears that the CuO is not fully consumed through the reaction with the Al2O3 scale on the Crofer 22 H (Al2O3 þ CuO / CuAl2O4) during the joining process (1000 C, 20 min). Compared to the Age2CuO joint (Fig. 3), more CuAl2O4 (pt B, Fig. 5c) is observed. The thickness of the composite oxide layer containing CuO and CuAl2O4 is nearly 10 mm. Although the Al2O3 layer to a certain extent is consumed by the reaction with CuO, it still remains continuous and thus effectively protects the Crofer 22 H substrate. No Fe, Cr or Mn diffusion from the steel into the braze is found on the basis of the EDS analysis (pt A, B and C, Fig. 5c). When the joining temperature for the Age4CuO braze is increased to
1020 C, the CuO near the aluminized Crofer 22 H is fully consumed through the reaction with Al2O3. The results of these experiments are described in detail in the supplementary material (Fig. S4) for the interested readers. It may be expected that also in case of a joint prepared at 1000 C all of the CuO will, with time, react with the Al2O3. Fig. 5d displays a SEM cross section image of the joint brazed using Age8CuO. Also in this case, the Age8CuO braze fully penetrates into the pores of the CGO, in consistence with the results of the wetting test (Fig. 2d). As shown in Fig. 5e, the high CuO content results in the formation of an oxide layer with a thickness of ~15 mm between the braze and the aluminized Crofer 22 H. This oxide layer consists of CuO (pt D, Fig. 5e) and CuAl2O4 (pt E, Fig. 5e), indicating a partly reaction between CuO and Al2O3. The CuO, CuAl2O4 and Al2O3 phases are clearly distinguishable in the corresponding EDS maps of Al, O and Cu. The Al2O3 layer remains continuous, although it is further thinned due to the greater concentration of CuO in the braze. Nevertheless, the complete coverage of Al2O3 on the steel is good enough to provide a sufficient protection. The
Table 2 e EDS composition (at. %) of points highlighted in Fig. 5. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt pt pt pt pt pt
46.5 18.8 1.7 44.8 20.3 2.7
5.6 29.2 0.8 6.8 31.7 0.4
40.3 45.6 5.3 41.3 40.6 5.2
6.9 5.2 91.1 6.2 6.2 90.4
0.1 0.3 0.3 0.2 0.3 0.2
0.4 0.7 0.7 0.6 0.7 0.8
0.2 0.2 0.1 0.1 0.2 0.3
CuO CuAl2O4 Ag CuO CuAl2O4 Ag
A B C D E F
3070
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
EDS maps of elements Fe, Cr and Mn and EDS results of pt F (Fig. 5e) indicate no outward diffusion of Fe, Cr and Mn from the steel into the braze.
Stability of joints in oxidizing and reducing atmospheres For studying the long-term stability in oxidizing and reducing atmospheres, the aluminized Crofer 22 H/AgeCuO/half-cell joints were aged for 250 h at 800 C in air and 4%H2e50%H2Oe N2, respectively. In the following, the results of aging in the 2 atm will be discussed separately.
Stability in oxidizing atmosphere Cross sectional SEM images of joints aged in air for 250 h at 800 C are shown in Fig. 6. With the pure Ag braze, the exposure to aging in air does not lead to any delamination at the braze/aluminized Crofer 22 H interface (Fig. 6a). The Al2O3 layer remains intact, continuous and tightly bonded to the steel substrate even after 250 h of aging. EDS analysis of pt. A in Fig. 6a (Table 3) indicates that the Al2O3 layer effectively blocks outward diffusion of Fe, Cr and Mn from the steel. However, serious delamination has occurred at the braze/CGO interface. A continuous crack is formed at this interface, and the CGO barrier layer is broken in several places. This suggests that the bonding between pure Ag and the CGO is too weak, probably due to poor wetting of the pure Ag braze. Kim et al. reported that for AgeCuO brazes with a CuO content less than 8 mol% poor wetting leads to a reduced bonding strength in the systems AgeCuO/YSZ and AgeCuO/Al2O3 [20e22]. Considering the large thermal expansion coefficient mismatch (CTE) between Ag (19 106 K1) [38] and the CGO
(12.54 ± 0.10 106 K1) [36], stresses are readily formed along the Ag/CGO interface during heating and cooling. As a result, the joint cracks along the CGO. SEM images after aging of the joint brazed using the Age 2CuO are shown in Fig. 6b. The microstructure is similar to that of the as-brazed joint with good adhesion of the braze to both the aluminized steel and the CGO of the half-cell (Fig. 4). The long-term aging in air causes no detectable delamination at the CGO side. This is likely due to the penetration of Ag into the pores of the CGO, forming an integrated whole, which keeps the joint intact despite the brittleness of the CGO. At higher magnification (region 2) it is apparent that the Al2O3 layer is intact. EDS analysis (pt. B, Fig. 6b) confirms that outward diffusion of Fe, Cr and Mn is effectively blocked. Also in case of the assembly brazed using Age4CuO no significant delamination is observed after aging in air, as seen in Fig. 6c. No indications of cracking or spalling of the CGO layer are found, indicating that the infiltration of braze in the pores of the CGO is good for maintaining the stability of the joint. Based on the high magnification SEM image of region 3 in Fig. 6c, it can be assumed that more CuAl2O4 (pt C, Fig. 6c) is formed after aging compared to the as-brazed joint (Fig. 5b), and only small amounts of CuO remain unreacted. These results suggest that during exposure to oxidizing conditions (air) the reaction between the residual CuO in the braze and the Al2O3 layer on aluminized Crofer 22 H continues. Although the Al2O3 layer is consumed by this reaction, its integrity has not been destroyed. Fig. 6d shows the microstructure of the Age8CuO joint after long-term oxidization. No evidence of cracks or other microstructural degradation can be observed. After 250 h in air
Fig. 6 e SEM images of joints after aging in air for 250 h at 800 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (High magnification SEM images 1, 2, 3, and 4 correspond to red regions in Fig. 6aed, respectively). Points A, B, C, D and E mark EDS point analysis in Table 3. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
3071
Table 3 e EDS composition (at. %) for points highlighted in Fig. 6. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt pt pt pt pt
e 19.0 20.1 19.7 43.6
0.6 29.4 30.7 31.7 e
6.6 45.3 41.9 40.8 47.2
91.7 5.6 6.1 6.6 9.2
0.3 0.4 0.3 0.3 e
0.6 0.8 0.8 0.7 e
0.2 0.2 0.1 0.2 e
Ag CuAl2O4 CuAl2O4 CuAl2O4 CuO
A B C D E
the Al2O3 layer still effectively blocks the outward diffusion of elements Fe, Cr and Mn based on the EDS analysis of pt D. It is apparent that the CuAl2O4 layer (pt D, Fig. 6d) formed with Age 8CuO is thicker compared to that formed with brazes having a lower CuO concentration. Furthermore, larger amounts of unreacted CuO are found as secondary phase precipitates in the braze (point E, Fig. 6d). Therefore, it may be expected that the Al2O3 layer will continue to be consumed with further oxidation. Based on our findings, the reaction between the CuO and Al2O3 layer is found to be the only factor that consumes the Al2O3 layer. Therefore, in the case of low-CuOcontaining brazes (Age2CuO), it can be inferred that the Al2O3 protective layer should still be intact after a longer service (e.g. 40000 h).
In reducing atmosphere Fig. 7 displays the cross sectional SEM images of the joints after exposure for 250 h at 800 C in 4%H2e50%H2OeN2. Based on the high-magnification SEM images of regions 1e4 (Fig. 7) and corresponding EDS analyses of pt A-D (Table 4), it is shown that in all cases the Al2O3 layer is intact after aging, and
the outward diffusion of elements Fe, Cr and Mn is blocked. For the pure Ag braze a delamination of the braze from the half-cell is found (Fig. 7a), similar to the result obtained with aging in an oxidizing atmosphere (Fig. 6a). As discussed above in Section Stability in oxidizing atmosphere, the formation of the crack along the CGO is mainly due to the poor bonding at the Ag/CGO interface and/or the large CTE mismatch between them. The bonding between the braze and the aluminized Crofer 22 H is still intact. When using the Age2CuO braze, exposure to reducing atmosphere does not lead to any delamination at the braze/CGO interface or the braze/aluminized Crofer 22 H interface, as shown in Fig. 7b. A reaction layer is formed between the braze and the Al2O3 layer on the aluminized Crofer 22 H. EDS analysis (pt B, region 2, Fig. 7b) indicates that the Cu concentration in this reaction layer is much smaller compared to the concentration in the reaction layer formed after oxidation in air. This may be due to hydrogen-inducted decomposition of the CuAl2O4 and CuO phases through the following two reactions: CuAl2O4 þ H2 / Cu þ Al2O3 þ H2O and CuO þ H2 / Cu þ H2O. A similar finding was reported by Kim et al. [19,22] and Hu
Fig. 7 e SEM images of joints after exposure in 4%H2e50%H2O-N2 for 250 h at 800 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (High magnification SEM images 1, 2, 3, and 4 correspond to red regions in Fig. 7aed, respectively). Points A, B, C, D and E mark EDS point analysis in Table 4. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
3072
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
Table 4 e EDS composition (at. %) of points highlighted in Fig. 7. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt A pt B pt C pt D pt E
e 4.1 5.3 6.1 10.2
0.7 52.4 51.8 51.0 e
4.3 30.1 29.4 29.5 7.2
93.8 12.3 12.4 11.9 82.2
0.3 0.3 0.2 0.4 0.1
0.7 0.6 0.8 0 0.2
0.2 0.2 0.1 0.3 0.1
Ag Al2O3 Al2O3 Al2O3 Ag
Table 5 e Room-temperature shear strength of joints asbrazed and after aging in different atmospheres (MPa). Sample
As-brazed
Air (800 C/250 h)
4H2e50H2OeN2 (800 C/250 h)
Age2CuO
39 ± 3
38 ± 2
35 ± 3
et al. [41]. Hydrogen diffusing through the braze will induce the decomposition of CuAl2O4 and CuO phases to the metallic copper, which can alloys with the surrounding silver forming a solid solution in the reducing atmosphere. In case of the Age4CuO joint shown in Fig. 7c, no delamination between the braze and CGO or the braze and aluminized Crofer 22 H is observed. However, at the interface between the braze and aluminized Crofer 22 H, some porosity is present. EDS analysis of the reaction layer formed at this interface (pt C, region 3, Fig. 7c) indicates a possible decomposition of the CuAl2O4 phase formed during RAB (Fig. 4). Therefore, the observed porosity may be attributable to hydrogen effect and the decomposition of CuO precipitates [19,22]. Fig. 7d shows the microstructure of the joint brazed using Age8CuO after long-term exposure to reducing conditions. Some small pores appear at the braze/CGO interface, which is comparable to the results of Kim et al. about the influence of reducing atmospheres [19,20,22]. More seriously, widely extending delamination is observed along interface between the braze and aluminized Crofer 22 H. As discussed above, it may be assumed that hydrogen effect and the decomposition of CuO precipitates results in formation of pores along this interface. In the case of Age8CuO braze, the above two effects would be more severe due to the higher CuO concentration. Meanwhile, the large CTE mismatch between the braze (19 106 K1) and Crofer 22 H (12 106 K1) gives rise to large stresses at this interface, which may accelerate the delamination [38]. EDS analysis of pt E (Fig. 7d) indicates that there is 10.2 at% copper in the silver matrix, which demonstrates that the original CuO phase underwent decomposition,
and that the metallic copper was alloyed with the surrounding silver forming a solid solution. The shear strength was also determined to evaluate the effect of aging test on the joint properties. Only the joints brazed using Age2CuO were tested because other joints already broken after aging. The measured room-temperature shear strength is shown in Table 5. The Age2CuO joints obtain a shear strength of 39 ± 3 MPa. After aging in air at 800 C for 250 h, there is almost no change in the shear strength of joints (38 ± 2 MPa). This is attributed to the high quality Al2O3 protective layer, which maintains the stable joint microstructure. After aging in 4H2e50H2OeN2 at 800 C for 250 h, the shear strength of joints experiences a limited reduction (35 ± 3 MPa). In addition to the stable joint microstructure after aging, the mechanical interlocking structures on both sides (the penetration of braze into the pores of the CGO and the tight bonding with the rough aluminized Crofer 22 H) also play a key role in maintaining the joint strength.
The edges of the Age2CuO joints after aging tests Previous studies [26,39,42] have reported that when joining Ag based brazes to Crofer 22 APU steel without a protective layer the two main degradation mechanisms after aging in oxidizing atmospheres are evaporation of Ag and the formation of AgeCr-oxides at the joint edges (pure oxygen-850 C [26], ambient air (~55% humidity)-700/800/900 C [39] and dry air-625 C [40]). Therefore, the sample edges were inspected to identify whether similar phenomena occur in this study. Fig. 8 shows SEM images of the edges of the Age2CuO joints after aging in oxidizing (air) and reducing (4%H2e50%H2OeN2) atmospheres for 250 h at 800 C. No evaporation of Ag or formation of AgeCr-oxides can be identified either atmosphere. It is also worth noting that there are no FeeCreMn-oxides on the surface of aluminized Crofer 22 H. These encouraging results can be attributed to the high quality Al2O3 protective layer fabricated by reactive aluminization. The Al2O3 layer provides good protection of the Crofer 22 H steel and
Fig. 8 e SEM images of Age2CuO joint edges after aging in (a) oxidizing and (b) reducing atmospheres for 250 h at 800 C.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
effectively blocks outward migration of Fe, Cr and Mn from the steel. This prevents exposure of the Ag based braze to Cr. As the evaporation of Ag and the formation of AgeCr-oxides are due to the reaction between the Ag and Cr [39,42], the Al2O3 layer effectively mitigates both degradation mechanisms. The joints should be tested longer and in dual atmosphere/in a stack, but the results here are very encouraging.
Conclusions The results show that AgeCuO braze can be used to successfully join aluminized Crofer 22 H to the ceria-gadolinia layer of a SOFC/SOEC half-cell at 1000 C in air. The main conclusions are summarized as follows: (1) The high contact angle of pure Ag on the ceria-gadolinia implies the poor wettability. Increasing the CuO content improves the wettability of the AgeCuO braze by lowering the contact angle. (2) A dense and uniform Al2O3 surface layer (~2 mm) can be formed on Crofer 22 H by reactive air aluminization. The Al2O3 provides excellent protection of the Crofer 22 H substrate during RAB and hinders the outward diffusion of Fe, Cr and Mn from the steel during aging. Aluminization of the steel thereby overcomes two major shortcomings of AgeCuO brazes (interconnect corrosion by CuO and formation of AgeCr-oxides). (3) Brazed joints are aged in oxidizing (air) and reducing (4%H2e50%H2OeN2) atmospheres for 250 h at 800 C. In both atmospheres, addition of CuO to the braze is necessary to avoid delamination between the braze and the ceria-gadolinia. In oxidizing atmosphere, CuO in the braze reacts with and consumes the Al2O3 layer on aluminized Crofer 22 H, forming a CuAl2O4 reaction layer. In reducing atmosphere, hydrogen diffused through the braze leads to decomposition of CuAl2O4 and CuO, leaving pores along the braze/steel interface. (4) The recommended braze is Age2CuO, which provides a good balance between sufficient wetting and minimal reaction/degradation.
Acknowledgment The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China under Grant Nos. 51622503 and U1537206. Xiaoqing Si acknowledges the support from the China Scholarship Council for one year study at the Technical University of Denmark. The work on Al coatings is a part of the project DTU stack platform.
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijhydene.2018.11.071.
3073
references
[1] Tucker MC. Progress in metal-support solid oxide fuel cells: a review. J Power Sources 2010;195:4570e82. [2] Kennouche D, Chen-Wiegart YK, Yakal-Kremski KJ, Wang J, Gibbs JW, Voorhees PW, et al. Observing the microstructural evolution of Ni-Yttria-stabilized zirconia solid oxide fuel cell anodes. Acta Mater 2016;103:204e10. [3] Zheng Y, Zhang CM, Ran R, Cai R, Shao ZP, Farrusseng D. A new symmetric solid-oxide fuel cell with La0.8Sr0.2Sc0.2Mn0.8O3-d perovskite oxide as both the anode and cathode. Acta Mater 2009;57:1165e75. [4] Haile SM. Fuel cell materials and components. Acta Mater 2003;51:5981e6000. [5] Linderoth S, Hendriksen PV, Mogensen M, Langvad N. Investigations of metallic alloys for use as interconnects in solid oxide fuel cell stack. J Mater Sci 1996;31:5077e82. [6] Kofstad P, Bredesen R. High temperature corrosion in SOFC environment. Solid State Ionics 1992;52:69e75. [7] Singh RN. High-temperature seals for solid oxide fuel cell (SOFC). J Mater Eng Perform 2006;15:422e6. [8] Yan D, Liang LJ, Yang JJ, Zhang T, Pu J, Chi B, et al. Performance degradation and analysis of 10-cell anodesupported SOFC stack with external manifold structure. Energy 2017;125:663e70. [9] Lin KL, Singh M, Asthana R, Lin CH. Interfacial and mechanical characterization of yttria-stabilized zirconia (YSZ) to stainless steel joints fabricated using Ag-Cu-Ti interlayer. Ceram Int 2014;40:2063e71. [10] Si XQ, Cao J, Liu S, Song XG, Qi JL, Huang YX, et al. Fabrication of 3D Ni nanosheet array on Crofer22APU interconnect and NiO-YSZ anode support to sinter with small-size Ag nanoparticles for low-temperature sealing SOFCs. Int J Hydrogen Energy 2018;43:2977e89. pez SR, Haanappel VAC, Dura n A, Mun ~ oz F, Mather GC, [11] Lo Pascual MJ, et al. Glass-ceramics seals in the system MgOBaO-B2O3-SiO2 operation under simulated SOFC conditions. Int J Hydrogen Energy 2016;41:15335e45. [12] Wang SF, Hsu YF, Cheng CS, Hsieh YC. SiO2-Al2O3-Y2O3-ZnO glass sealants for intermediate temperature solid oxide fuel cell application. Int J Hydrogen Energy 2013;38:14779e90. [13] Ritucci I, Agersted K, Zielke P, Wulff AC, Khajavi R, Smeacetto F, et al. A Ba-free sealing glass with a high coefficient of thermal expansion and excellent interface stability optimized for SOFC/SOEC stack applications. Int J Appl Ceram Technol 2018;15:1011e22. [14] Peng L, Zhu QS, Xie ZH, Wang P. Interface reactions between sealing glass and metal interconnect under static and dynamic heat treatment conditions. J Fuel Cell Sci Technol 2015;12:0610091e7. [15] Hsu JH, Kim CW, Brow RK, Szabo J, Crouch R, Baird R. An alkali-free barium borosilicate viscous sealing glass for solid oxide fuel cells. J Power Sources 2014;270:14e20. [16] Kiebach R, Agersted K, Zielke P, Ritucci I, Brock M, Hendriksen PV. A novel SOFC/SOEC sealing glass with a low SiO2 content and a high thermal expansion coefficient. ECS Trans 2017;78:1739e47. [17] Malzbender J, Zhao Y, Beck T. Fracture and creep of glassceramic solid oxide fuel cell sealant materials. J Power Sources 2014;246:574e80. [18] Liu W, Sun X, Khaleel MA. Predicting Young's modulus of glass/ceramic sealant for solid oxide fuel cell considering the combined effects of aging, micro-voids and self-healing. J Power Sources 2008;185:1193e200. [19] Kim JY, Hardy JS, Weil KS. High-temperature tolerance of the silver-copper oxide braze in reducing and oxidizing atmospheres. J Mater Res Vel 2006;21:1434e42.
3074
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
[20] Kim JY, Hardy JS, Weil KS. High-temperature tolerance of the silver-copper oxide braze in reducing and oxidizing atmospheres. J Mater Res 2006;21:1431e42. [21] Kati R, Muksin, Seyoung K, Kwang-sup S, Ji HY, Dang-Hyok Y. Joining of metal-ceramic using reactive air brazing for oxygen transport membrane applications. Mater Des 2016;109:233e41. [22] Kim JY, Hardy JS, Weil S. Dual-atmosphere tolerance of AgCuO-based air braze. Int J Hydrogen Energy 2007;32:3655e63. [23] Kuhn B, Wetzel FJ, Malzbender J, Steinbrech RW, Singheiser L. Mechanical performance of reactive-air-braze (RAB) ceramic/metal joints for solid oxide fuel cells at ambient temperature. J Power Sources 2009;193:199e202. [24] Kuhn B, Wessel E, Malzbender J, Steinbrech RW, Singheiser L. Effect of isothermal aging on the mechanical performance of brazed ceramic/metal joints for planar SOFC-stacks. Int J Hydrogen Energy 2010;35:9158e65. [25] Wahid MFR, Dang-Hyok Y, Kati R, Seyoung K, Kwang-sup S, Ji HY. Interfacial microstructure and shear strength of reactive air brazed oxygen transport membrane ceramicmetal alloy joints. Met Mater Int 2018;24:157e69. [26] Kiebach R, Engelbrecht K, Grahl-Madsen L, Sieborg B, Chen M, Hjelm J, et al. An Ag based brazing system with tunable thermal expansion for the use as sealant for solid oxide cells. J Power Sources 2016;315:339e50. [27] Choi JP, Weil KS, Chou YM, Stevenson JW, Yang ZG. Development of MnCoO coating with new aluminizing process for planar SOFC stack. Int J Hydrogen Energy 2011;36:4549e56. [28] Stevenson JW, Yang ZG, Xia GG, Nie Z, Templeton JD. Longterm oxidation behavior of spinel-coated ferritic stainless steel for solid oxide fuel cell interconnect applications. J Power Sources 2013;231:256e63. [29] Choi JP, Chou YS, Stevenson JW. Reactive air aluminization. 2011. PNNL-20859. [30] Hsu JH, Kim CW, Brow RK. Interfacial interactions between an alkali-free borosilicate viscous sealing glass and aluminized ferritic stainless steel. J Power Sources 2014;250:236e41. [31] Hauch A, Brodersen K, Chen M, Mogensen MB. Ni/YSZ electrodes structures optimized for increased electrolysis
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
performance and durability. Solid State Ionics 2016;293:27e36. Song XG, Zhao YX, Hu SP, Cao J, Fu W, Feng JC. Wetting of AgCu-Ti filler on porous Si3N4 ceramic and brazing of the ceramic to TiAl alloy. Cerm Int 2018;44:4622e9. Fu W, Hu SP, Song XG, Li JX, Cao J, Feng JC, et al. Wettability and bonding of graphite by Sn0.3Ag0.7Cu-Ti alloys. Carbon 2017;121:536e43. Eustathopoulos N. Wetting by liquid metals e application in materials processing: the contribution of the Grenoble group. Metals 2015;5:350e70. Zhang KX, Xia LH, Zhang FQ, He LL. Active brazing of C/C composite to copper by AgCuTi filer meta. Metall Mater Trans A 2016;47:2162e76. Seeharaj P, Atkinson A. Diffusion and conductivity of mixedconducting Ag/CGO composites. Solid State Ionics 2011;204e205:46e52. Cao J, Si XQ, Li WJ, Song XG, Feng JC. Reactive air brazing of YSZ-electrolyte and Al2O3-substrate for gas sensor sealing: interfacial microstructure and mechanical properties. Int J Hydrogen Energy 2017;42:10683e94. Wang LS, Barnett SA. Deposition, structure, and properties of cermet thin films composed of Ag and Y-stabilized zirconia. J Electrochem Soc 1992;139:1134e40. Sofie SW, Gannon P, Gorokhovsky V. Silver-chromium oxide interactions in SOFC environments. J Power Sources 2009;191:465e72. Abernathy HW, Koep E, Compson C, Cheng Z, Liu ML. Monitoring Ag-Cr interactions in SOFC cathodes using Raman spectroscopy. J Phys Chem C 2008;112:13299e303. Hu WT, Donat F, Scott SA, Dennis JS. The interaction between CuO and Al2O3 and the reactivity of copper aluminates below 1000 C and their implication on the use of the Cu-Al-O system for oxygen storage and production. RSC Adv 2016;6:113016e24. € M, Wiesner S, Kaletsch A, Broeckmann C. Bobzin K, Ote Characterization of reactive air brazed ceramic/metal joints unadapted thermal expansion behavior. Adv Eng Mater 2014;16:1490e7.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Enhancing the long-term stability of Ag based seals for solid oxide fuel/electrolysis applications by simple interconnect aluminization Xiaoqing Si a,b, Jian Cao b,*, Ilaria Ritucci a, Belma Talic a, Jicai Feng b, Ragnar Kiebach a,** a
Department of Energy Conversion and Storage, Technical University of Denmark, Risø Campus, Roskilde 4000, Denmark b State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, China
article info
abstract
Article history:
Ag-(0e8 mol%) CuO is used to successfully join aluminized ferritic stainless steel inter-
Received 14 September 2018
connect to the ceria-gadolinia (CGO) barrier layer of a solid oxide fuel/electrolysis cell by
Received in revised form
reactive air brazing at 1000 C in air. The wetting of AgeCuO on CGO is tailored by varying
6 November 2018
the CuO content. The effects of the CuO content on the joint microstructure are discussed.
Accepted 10 November 2018
The long-term stability of brazed joints is evaluated by aging in oxidizing (air) and reducing
Available online 6 January 2019
(4% H2e50% H2OeN2) atmospheres at 800 C for 250 h. An Ag-2mol% CuO braze results in the best joint stability during aging. Aluminization of the steel to create an alumina surface
Keywords:
layer provides excellent protection of the steel both during the joining process and aging in
Reactive air brazing
the 2 atm. No degradation related to steel corrosion and outward diffusion of elements
Aluminization
from the steel can be observed.
AgeCuO braze
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Solid oxide fuel cell Solid oxide electrolysis cell High temperature corrosion
Introduction Solid oxide fuel/electrolysis cells (SOFC/SOEC) have attracted considerable interest because of their high efficiency, fuel flexibility, quiet operation and ultra-low pollution [1e4]. To generate power or produce hydrogen on a large scale, several cells are combined together in a stack with interconnects between the individual cells. For planar stacks operating at intermediate temperatures (~650e900 C), ferritic stainless steels are preferred as the interconnect material [5,6]. Stable,
gas tight sealing between the cell and the interconnect is required to ensure fuel purity, a high efficiency and to maintain mechanical stability [7e10]. Over the past decade, several sealing concepts have been tested for SOFC/SOEC application, among which glass/glass-ceramic bonding [11e18] and reactive air brazing (RAB) [19e25] are the most studied. Glass/glass-ceramic materials are cost-effective and relatively easy to process, and therefore a popular choice for sealing SOFC/SOEC stacks. However, many glass/glass ceramics are prone to devitrification during operation, which affects the long-term stability in terms of microstructure and
* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (J. Cao), [email protected] (R. Kiebach). https://doi.org/10.1016/j.ijhydene.2018.11.071 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
3064
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
mechanical properties [11e16]. Furthermore, glass/glassceramic sealants typically have poor resistance to stressinduced cracking, which is especially an issue for use in mobile applications where the stresses tend to be larger [17,18]. RAB is another approach for sealing SOFC/SOEC stacks. One of the preferred materials for RAB is Ag, which is often mixed with a certain amount of Cu or CuO for improved wettability on ceramic components [19,20]. The excellent oxidation resistance of brazes based on noble metal such as Ag allows RAB to be performed directly in air, thus eliminating the need for fluxing agents or inert gases [20e22]. Although RAB has great potential for sealing SOFC/SOEC stacks, there are still some challenges currently hindering its application. The most serious is the excessive reactivity between the steel interconnect and Cu/CuO [23e25]. Formation of a more than 20 mm thick Cu/Cr/Mn/Fe-oxide scale along the braze/interconnect interface has been observed, resulting in a mechanically weak part in the joint [22e25]. The high mobility of oxygen in Ag means that oxygen easily reaches the braze/interconnect interface, resulting in a further oxidation of the steel interconnect [26]. Future development of RAB for SOFC/SOEC stacks requires development of means to control the reaction at the braze/ interconnect interface. The basic reason for corrosion at this interface is that the steel interconnect is in direct contact with the Cu/CuO-containing braze [22e26]. Thus, to mitigate corrosion, a protective coating may be fabricated on the interconnect surface to protect the steel against oxidation and block interdiffusion across the interconnect/braze interface [27,28]. A promising approach is aluminization of the steel. This creates an Al2O3 scale on the steel surface that has been shown to mitigate reaction between the steel constituents and a glass sealing material [29,30]. Because the Al2O3 scale is insulating, it will be applied only in the sealing area of the interconnect in real applications. In this paper, RAB is used to join aluminized Crofer 22 H to anode-supported half-cells. The long-term stability of these joints is tested under exposure to SOFC/SOEC relevant conditions. The evolution of the alumina layer on the aluminized steel during the RAB process and long-term exposure is studied in detail. We show that by tailoring the composition of the AgeCuO braze a balance between sufficient wetting and alleviating overreaction with the aluminized steel can be achieved.
Experimental procedure Materials and sample preparation Crofer 22 H with a thickness of 0.3 mm was used as the interconnect alloy in this study. Crofer 22 H (Thyssenkrupp, Germany) is a ferritic stainless steel containing (in wt.%) 22.3 Cr, 0.43 Mn, 0.48 Nb, 0.27 Si, 2.0 W, 0.06 Ti, 0.06 La and balance Fe. The steel was cut into 2 2 cm2 coupons and ultrasonically cleaned in ethanol for 10 min before aluminization. The procedure for reactive air aluminization consisted of three steps: (a) application and drying of an Al paste, (2) heat treatment in air, (3) removal of unreacted loose material [29,30]. A coating slurry was prepared by mixing Al powder (325 mesh, Alfa
Aesar) with a commercial binder system. The slurry was brush painted on the Crofer 22 H, applying approximately 4e5 mg cm2 of slurry. After drying the samples were heated at 1000 C for 1 h with a heating and cooling rate of 180 C h1. Finally, the samples were ultrasonically cleaned for 10 min in ethanol in order to remove any loose/unreacted Al2O3 scale on the surface. The planar anode-supported half-cells consisted of a NiO-YSZ support layer, a thin NiO-YSZ active anode layer, a dense YSZ electrolyte layer and a porous ceria-gadolinia (CGO) barrier layer. The half-cells were developed at Technical University of Denmark, and the procedure for making half-cells was described in detail elsewhere [31]. The half-cells were laser cut into 2 2 cm2 specimens, and then ultrasonically cleaned in ethanol for 10 min. For the braze, Cu powder (1 mm, 99.9%, Sigma-Aldrich) was mixed with Ag powder (1 mm, 99.9%, Sigma-Aldrich) and milled in a planetary ball mill (Retsch PM 1000) for 10 h at 150 rpm. As-milled powders were pressed (2 t, uniaxial pressing) into bars with a dimension of 20 5 2 mm3. The pressed bar was first heated uniformly by a flame gun for 1 min in order to improve its plastic deformation ability. Then the sample was rolled by a rolling press. The thickness of the pressed bar was ~2 mm, so the roller gap was first set to ~1.90 mm. After twice or more rolling at the same thickness, the sample was reheated. Each time reduction of the roller gap must be less than 0.10 mm. Finally the thickness of sample is ~100 mm, and the whole rolling process took ~20 times. Four kinds of braze sheets (~100 mm thick) with 0 mol%, 2 mol%, 4 mol% and 8 mol % of Cu were produced. An image of prepared Age2CuO braze sheet was shown in Fig. S1. The Cu is oxidized to CuO during RAB. Therefore, the four brazes are in the following denoted as Ag, Age2CuO, Age4CuO and Age8CuO, respectively. The braze sheets were cut into 2 2 cm2 pieces, and sandwiched between the aluminized Crofer 22 H and the half-cells. The CGO barrier layer was chosen as the joining surface of the half-cells. A normal load of 1 kg cm2 was applied on the sandwich to maintain proper contact. The assemblies were heated under load in air with 100 C h1 to the sealing temperature of 1000 C where they were kept for 20 min before cooling down to room temperature with 60 C h1. Wetting of the molten braze on the ceramic substrate plays an important role in the joining process [31e34]. Especially, the CGO layer used in this study has a porous structure, which may affect the wetting of AgeCuO braze. To study the wetting of AgeCuO brazes, small pieces of the AgeCuO sheets with different CuO amounts were individually placed on top of the CGO barrier layer on the half-cells and heat treated at 1000 C for 20 min using the same heating (100 C h1) and cooling (60 C h1) rates. That is, no load and no Crofer 22 H was used for the wetting experiments. After cooling the samples were sectioned and the contact angle between the AgeCuO and the CGO layer was evaluated from images obtained by scanning electron microscopy (SEM, TM3000, Hitachi).
Long-term test in reducing and oxidizing atmospheres To investigate the stability of joints in reducing and oxidizing atmospheres two sets of aging experiments were performed. For this purpose, two samples of each braze composition were
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
tested in both atmospheres. The sealed joints were placed inside an alumina crucible and exposed to 800 C for 250 h in a horizontal tube furnace. The heating and cooling rates were 100 C h1. For the long-term test in reducing atmosphere (4% H2, 50% H2O, 46% N2), the H2 was diluted by an inert gas (N2) to keep a safe concentration. The steam was supplied by flowing H2eN2 through a water bubbler heated to 80 C for 50% humidity. The gas flow was kept constant at 6 L h1 during the long-term test in reducing atmosphere. The long-term test in oxidizing atmosphere was conducted in stagnant air.
Characterization of joint properties test The microstructure and composition of the as-joined and aged samples were characterized employing a tabletop scanning electron microscope (SEM, TM3000, Hitachi) equipped with an energy dispersive spectrometer (EDS). Samples for SEM analysis were vacuum embedded in epoxy resin (Epofix, Struers) and cut along the center of the assembly. The cross-sections were ground by SiC paper and diamonds in suspension in successive steps down to 1 mm. Phase analysis of sample surfaces was conducted by an X-ray diffraction spectrometer (XRD, D8-Advance, Bruker) with Cu-Ka radiation. The scans were made between 20 and 90 with a step size of 0.02 and a speed of 1 s per step. The shear test was carried out at roomtemperature using a bond tester (Dage 4000, Nordson dage, UK) at a constant speed of 0.1 mm/min. Five samples were tested for each process parameter, and the average value of the shear strength was obtained.
3065
Results and discussion Characterization of aluminized Crofer 22 H Representative SEM images and EDS elemental maps of the Crofer 22 H steel surface and cross section after reactive air aluminization are shown in Fig. 1. The steel surface is covered by a continuous, dense and crack free layer of Al2O3 having a uniform thickness of ~2 mm. The surface is wavy with an up to 12 mm distance between the highest peak and lowest valley, and the average roughness (Ra) is 4.2 mm. A mechanical interlocking structure can be formed between the braze and the rough aluminized Crofer 22 H, which may improve the interfacial bonding [35]. No other oxidation products (e.g. Cr2O3) are observed on the SEM cross sections. XRD analysis of the aluminized Crofer 22 H indicates that the surface layer consists of pure a-Al2O3, as shown in Fig. 1c. No Cr2O3 or any secondary phase indicating a reaction between the Al2O3 and the Crofer 22 H is detected. The surface morphology of the Al2O3 protective layer is displayed in the top-view image in Fig. 1d. The surface is relatively coarse with many staggered protrusions. To investigate the thermal stability of the aluminized Crofer 22 H, a sample was heated following the same procedure used for the RAB process (1000 C/20 min). Inspections after this exposure showed that the Al2O3 layer remains intact and that outward diffusion of Cr and Fe from the steel is effectively blocked (Fig. S2 and Fig. S3 in the supplementary
Fig. 1 e Typical results after aluminization of Crofer 22 H: (a) cross-sectional SEM image (b) high magnification view from Fig. 1a with EDS maps, (The magnification of the EDS maps is the same as the corresponding SEM image) (c) XRD scan of the aluminized Crofer 22 H, (d) surface morphology of the Al2O3 layer.
3066
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
material). The good stability of the aluminized steel is the key for preventing damage of the Crofer 22 H substrate during RAB and subsequent long-term exposure, as will be discussed in Sections Reactive air brazing of SOFC/SOEC components and Stability of joints in oxidizing and reducing atmospheres.
Wetting behavior of AgeCuO braze on the CGO Fig. 2 displays cross-sectional micrographs of the braze droplet/CGO interface and the EDS Ag elemental map with higher magnification images showing details of the wetting
Fig. 2 e Micrographs of cross-sectional droplet/CGO with element Ag mapping analysis after cooling to room temperature from 1000 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (The magnification of the EDS maps is the same as the corresponding SEM images).
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
triple line (zone 1, Fig. 2a) and the central interface (zone 2, Fig. 2a). Although pure Ag apparently bonds to the CGO, the contact angle is high (~72 ) and the Ag braze does not penetrate into the pores of the CGO (Fig. 2a). Fig. 2b illustrates that addition of Cu improves the wettability of the Ag braze. Accordingly, the Age2CuO braze has a lower contact angle on CGO (~35 ) compared to pure Ag. Characterization of the wetting triple line (zone 1, Fig. 2b) reveals that the Age2CuO braze partially penetrates into the pores of the CGO. Apparently, the Age2CuO barely spreads in the pores of the CGO and the penetration front does not reach the CGO/YSZ interface. The higher magnification SEM image of the braze droplet (zone 2, Fig. 2b) shows the presence of CuO precipitates inside the braze. The high-magnification SEM image of the central interface (zone 3, Fig. 2b) and the corresponding Ag map further confirms that the Age2CuO braze only partially penetrates into the CGO. SEM/EDS images of the sample with the Age4CuO braze are shown in Fig. 2c. The wetting on CGO is further improved and the contact angle is decreased to ~25 . The Age4CuO braze spreads into the pores of CGO as well as on the CGO surface. The spreading distance along the surface was measured to ~0.5 mm, as shown in the high-magnification image of zone 1 (Fig. 2c). The high-magnification images of zones 2 and 3 (Fig. 2c) show that Ag fully penetrates the porous parts of the CGO, and that the braze reaches all the way to the CGO/YSZ interface. Similar results were obtained with the Age8CuO braze (Fig. 2d). The Age8CuO braze infiltrates the pores of the CGO all the way to the CGO/YSZ interface (zones 2 and 3, Fig. 2d), and apparently adheres to the YSZ electrolyte. The Age8CuO braze also spreads on the CGO surface, in this case with a longer spreading distance of ~0.8 mm (zone 1, Fig. 2d). A longer spreading distance implicates a risk of cathode contamination if the cathode is too close to the sealing area. It is noted that even in the case of a high CuO concentration (Age8CuO) no CuO precipitates are detected in the pores of CGO, although Cu obviously plays an important role in facilitating wetting. These results are consistent with those reported by Meier et al.
3067
[34] for AgeCuO/Al2O3 and Seeharaj et al. [36] for AgeCuO/ CGO, in which the CuO precipitates were found only at the interface and within the Ag matrix. Fig. 3 shows the contact angle of AgeCuO braze on the porous CGO substrate as a function of CuO content. Increasing the CuO content from 0 to 8 mol% promotes wetting of the AgeCuO braze, corresponding to a significant reduction in the contact angle from ~72 (Ag) to ~18 (Age8CuO). Seeharaj et al. [36] studied the wetting of an AgeCuO braze on a denser CGO substrate. A comparable correlation between the wetting angle and the CuO content was observed, as shown in Fig. 3. Comparing our results with those of Seeharaj et al. [36] shows that in all cases the AgeCuO brazes have a smaller contact angle on the porous CGO than on the denser CGO. This difference may be due to the differences in the experiment condition and samples. However, a same trend is presented on two kinds of CGO, that is, the improved wetting with increasing CuO content improves the penetration of AgeCuO braze into the CGO pores, as shown in Fig. 3. The large surface tension of pure liquid Ag and the large contact angle on CGO impedes the penetration of pure Ag into CGO [36]. Adding CuO lowers the surface energy/tension of liquid Ag, which is partially responsible for the improved wetting of AgeCuO on CGO. Consequentially, liquid Ag penetrates into the pores of the CGO owning to the capillary force imposed on the Ag. In case of the Ag-4/8CuO brazes, the penetrate depth exceeded 7.8 mm (the thickness of the CGO layer).
Reactive air brazing of SOFC/SOEC components The Age2CuO braze was used to join aluminized Crofer 22 H to the CGO barrier layer of a NiO-YSZ anode-supported half-cell. This kind of joint reflects a relevant assembly in planar SOFC/ SOEC stacks. Fig. 4 shows overview SEM images and corresponding EDS maps of the joined interfaces after the RAB process. A defect free joint between aluminized Crofer 22 H and the CGO is obtained. The Age2CuO braze wets the surface of the aluminized Crofer 22 H and CGO sufficiently, and no defects such as cracks or pores can be found (Fig. 4a). High
Fig. 3 e Contact angle of AgeCuO brazes on the CGO with pores and the dense CGO [36] as a function of CuO content, and the schematic diagrams of the penetration.
3068
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
Fig. 4 e (a) SEM image of aluminized Crofer 22 H brazed to the CGO layer of a NiO-YSZ anode-supported half-cell using Age 2CuO. High magnification SEM images and corresponding EDS maps of region 1, 2 and 3 are shown in (b), (c) and (d), respectively. (The EDS maps have the same magnification as the corresponding SEM image). Points A, B, C and D mark EDS point analysis in Table 1.
magnification images of zones 1 and 2 from the CGO side are shown in Fig. 4b and c, respectively. EDS analysis (Table 1) (pt A, Fig. 4b, pt B, Fig. 4c) confirms that the light-grey phase along the CGO interface is CuO. The elemental Ag map reveals that in some places the braze fully penetrates into the pores of the CGO (Fig. 4c). This is in contrast to the observations of the wetting experiment (Fig. 2b) and can be attributed to the pressure applied during the RAB joining process (1 kg cm2), which promotes infiltration into the porous structure [36]. No reaction between the CGO and the braze can be found in Fig. 4b and c. A high magnification image of zone 3 from the
Table 1 e EDS composition (at. %) for the points highlighted in Fig. 4. (Important values marked in bold). Points
Cu
Al
O
pt A pt B pt C pt D
50.4 51.6 16.1 5.8
0 0 28.5 36.1
41.4 40.6 48.6 50.5
Ag Fe 8.2 7.8 5.7 6.3
e e 0.3 0.4
Cr Mn Possible phase e e 0.7 0.7
e e 0.1 0.2
CuO CuO CuAl2O4 Al2O3
aluminized Crofer 22 H side is shown in Fig. 4d. Based on EDS analysis (pt C, Fig. 4d), the dark-grey phase along the Al2O3/ braze interface is assigned to be CuAl2O4, which likely has formed through the reaction of Al2O3 þ CuO / CuAl2O4. This would be in agreement with results from Kim et al. [22] and Cao et al. [37], who also reported the formation of CuAl2O4 when CuO reacted with Al2O3 during RAB. Nevertheless, the Al2O3 layer remains continuous and fully covers the steel surface (pt D, Fig. 4d). Elemental maps of Cu, Al and O indicate the distribution of the CuAl2O4 phase, and further confirm the integrity of the Al2O3 layer. Additionally, elemental maps of Fe, Cr and Mn indicate that the outward diffusion of these elements from the steel is completely blocked by the Al2O3 layer. Thus, the Crofer 22 H steel is very well protected during RAB. In the following, the effect of the CuO concentration on the interfacial microstructure is discussed. Overview SEM images of assemblies with different CuO concentrations in the braze after joining at 1000 C under a weight load of 1 kg cm2 are presented in Fig. 5. Good/sufficient bonding to the aluminized Crofer 22 H and the CGO was found in general, irrespective of
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
3069
Fig. 5 e SEM images of joints brazed using different brazes: (a) Ag, (b) Age4CuO, (c) high magnification of the red marked region in Fig. 5b, (d) Age8CuO, (e) high magnification and corresponding EDS maps of the red mark region in Fig. 5d. (The magnification of the EDS maps is the same as the corresponding SEM image). Points A, B, C, D, E and F mark EDS point analysis in Table 2. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
the CuO amount in the braze. Pure Ag does not penetrate into the pores of the CGO (Fig. 5a) even under the applied pressure (as discussed in Section Stability of joints in oxidizing and reducing atmospheres, see later). The Age4CuO braze fully penetrates into the CGO, and apparently adheres to the halfcell (Fig. 5b). The high-magnification SEM image of the braze/aluminized Crofer 22 H interface indicates the presence of CuO between the CuAl2O4 reaction layer and the braze (EDS analysis Table 2, pt A, Fig. 5c). It appears that the CuO is not fully consumed through the reaction with the Al2O3 scale on the Crofer 22 H (Al2O3 þ CuO / CuAl2O4) during the joining process (1000 C, 20 min). Compared to the Age2CuO joint (Fig. 3), more CuAl2O4 (pt B, Fig. 5c) is observed. The thickness of the composite oxide layer containing CuO and CuAl2O4 is nearly 10 mm. Although the Al2O3 layer to a certain extent is consumed by the reaction with CuO, it still remains continuous and thus effectively protects the Crofer 22 H substrate. No Fe, Cr or Mn diffusion from the steel into the braze is found on the basis of the EDS analysis (pt A, B and C, Fig. 5c). When the joining temperature for the Age4CuO braze is increased to
1020 C, the CuO near the aluminized Crofer 22 H is fully consumed through the reaction with Al2O3. The results of these experiments are described in detail in the supplementary material (Fig. S4) for the interested readers. It may be expected that also in case of a joint prepared at 1000 C all of the CuO will, with time, react with the Al2O3. Fig. 5d displays a SEM cross section image of the joint brazed using Age8CuO. Also in this case, the Age8CuO braze fully penetrates into the pores of the CGO, in consistence with the results of the wetting test (Fig. 2d). As shown in Fig. 5e, the high CuO content results in the formation of an oxide layer with a thickness of ~15 mm between the braze and the aluminized Crofer 22 H. This oxide layer consists of CuO (pt D, Fig. 5e) and CuAl2O4 (pt E, Fig. 5e), indicating a partly reaction between CuO and Al2O3. The CuO, CuAl2O4 and Al2O3 phases are clearly distinguishable in the corresponding EDS maps of Al, O and Cu. The Al2O3 layer remains continuous, although it is further thinned due to the greater concentration of CuO in the braze. Nevertheless, the complete coverage of Al2O3 on the steel is good enough to provide a sufficient protection. The
Table 2 e EDS composition (at. %) of points highlighted in Fig. 5. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt pt pt pt pt pt
46.5 18.8 1.7 44.8 20.3 2.7
5.6 29.2 0.8 6.8 31.7 0.4
40.3 45.6 5.3 41.3 40.6 5.2
6.9 5.2 91.1 6.2 6.2 90.4
0.1 0.3 0.3 0.2 0.3 0.2
0.4 0.7 0.7 0.6 0.7 0.8
0.2 0.2 0.1 0.1 0.2 0.3
CuO CuAl2O4 Ag CuO CuAl2O4 Ag
A B C D E F
3070
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
EDS maps of elements Fe, Cr and Mn and EDS results of pt F (Fig. 5e) indicate no outward diffusion of Fe, Cr and Mn from the steel into the braze.
Stability of joints in oxidizing and reducing atmospheres For studying the long-term stability in oxidizing and reducing atmospheres, the aluminized Crofer 22 H/AgeCuO/half-cell joints were aged for 250 h at 800 C in air and 4%H2e50%H2Oe N2, respectively. In the following, the results of aging in the 2 atm will be discussed separately.
Stability in oxidizing atmosphere Cross sectional SEM images of joints aged in air for 250 h at 800 C are shown in Fig. 6. With the pure Ag braze, the exposure to aging in air does not lead to any delamination at the braze/aluminized Crofer 22 H interface (Fig. 6a). The Al2O3 layer remains intact, continuous and tightly bonded to the steel substrate even after 250 h of aging. EDS analysis of pt. A in Fig. 6a (Table 3) indicates that the Al2O3 layer effectively blocks outward diffusion of Fe, Cr and Mn from the steel. However, serious delamination has occurred at the braze/CGO interface. A continuous crack is formed at this interface, and the CGO barrier layer is broken in several places. This suggests that the bonding between pure Ag and the CGO is too weak, probably due to poor wetting of the pure Ag braze. Kim et al. reported that for AgeCuO brazes with a CuO content less than 8 mol% poor wetting leads to a reduced bonding strength in the systems AgeCuO/YSZ and AgeCuO/Al2O3 [20e22]. Considering the large thermal expansion coefficient mismatch (CTE) between Ag (19 106 K1) [38] and the CGO
(12.54 ± 0.10 106 K1) [36], stresses are readily formed along the Ag/CGO interface during heating and cooling. As a result, the joint cracks along the CGO. SEM images after aging of the joint brazed using the Age 2CuO are shown in Fig. 6b. The microstructure is similar to that of the as-brazed joint with good adhesion of the braze to both the aluminized steel and the CGO of the half-cell (Fig. 4). The long-term aging in air causes no detectable delamination at the CGO side. This is likely due to the penetration of Ag into the pores of the CGO, forming an integrated whole, which keeps the joint intact despite the brittleness of the CGO. At higher magnification (region 2) it is apparent that the Al2O3 layer is intact. EDS analysis (pt. B, Fig. 6b) confirms that outward diffusion of Fe, Cr and Mn is effectively blocked. Also in case of the assembly brazed using Age4CuO no significant delamination is observed after aging in air, as seen in Fig. 6c. No indications of cracking or spalling of the CGO layer are found, indicating that the infiltration of braze in the pores of the CGO is good for maintaining the stability of the joint. Based on the high magnification SEM image of region 3 in Fig. 6c, it can be assumed that more CuAl2O4 (pt C, Fig. 6c) is formed after aging compared to the as-brazed joint (Fig. 5b), and only small amounts of CuO remain unreacted. These results suggest that during exposure to oxidizing conditions (air) the reaction between the residual CuO in the braze and the Al2O3 layer on aluminized Crofer 22 H continues. Although the Al2O3 layer is consumed by this reaction, its integrity has not been destroyed. Fig. 6d shows the microstructure of the Age8CuO joint after long-term oxidization. No evidence of cracks or other microstructural degradation can be observed. After 250 h in air
Fig. 6 e SEM images of joints after aging in air for 250 h at 800 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (High magnification SEM images 1, 2, 3, and 4 correspond to red regions in Fig. 6aed, respectively). Points A, B, C, D and E mark EDS point analysis in Table 3. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
3071
Table 3 e EDS composition (at. %) for points highlighted in Fig. 6. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt pt pt pt pt
e 19.0 20.1 19.7 43.6
0.6 29.4 30.7 31.7 e
6.6 45.3 41.9 40.8 47.2
91.7 5.6 6.1 6.6 9.2
0.3 0.4 0.3 0.3 e
0.6 0.8 0.8 0.7 e
0.2 0.2 0.1 0.2 e
Ag CuAl2O4 CuAl2O4 CuAl2O4 CuO
A B C D E
the Al2O3 layer still effectively blocks the outward diffusion of elements Fe, Cr and Mn based on the EDS analysis of pt D. It is apparent that the CuAl2O4 layer (pt D, Fig. 6d) formed with Age 8CuO is thicker compared to that formed with brazes having a lower CuO concentration. Furthermore, larger amounts of unreacted CuO are found as secondary phase precipitates in the braze (point E, Fig. 6d). Therefore, it may be expected that the Al2O3 layer will continue to be consumed with further oxidation. Based on our findings, the reaction between the CuO and Al2O3 layer is found to be the only factor that consumes the Al2O3 layer. Therefore, in the case of low-CuOcontaining brazes (Age2CuO), it can be inferred that the Al2O3 protective layer should still be intact after a longer service (e.g. 40000 h).
In reducing atmosphere Fig. 7 displays the cross sectional SEM images of the joints after exposure for 250 h at 800 C in 4%H2e50%H2OeN2. Based on the high-magnification SEM images of regions 1e4 (Fig. 7) and corresponding EDS analyses of pt A-D (Table 4), it is shown that in all cases the Al2O3 layer is intact after aging, and
the outward diffusion of elements Fe, Cr and Mn is blocked. For the pure Ag braze a delamination of the braze from the half-cell is found (Fig. 7a), similar to the result obtained with aging in an oxidizing atmosphere (Fig. 6a). As discussed above in Section Stability in oxidizing atmosphere, the formation of the crack along the CGO is mainly due to the poor bonding at the Ag/CGO interface and/or the large CTE mismatch between them. The bonding between the braze and the aluminized Crofer 22 H is still intact. When using the Age2CuO braze, exposure to reducing atmosphere does not lead to any delamination at the braze/CGO interface or the braze/aluminized Crofer 22 H interface, as shown in Fig. 7b. A reaction layer is formed between the braze and the Al2O3 layer on the aluminized Crofer 22 H. EDS analysis (pt B, region 2, Fig. 7b) indicates that the Cu concentration in this reaction layer is much smaller compared to the concentration in the reaction layer formed after oxidation in air. This may be due to hydrogen-inducted decomposition of the CuAl2O4 and CuO phases through the following two reactions: CuAl2O4 þ H2 / Cu þ Al2O3 þ H2O and CuO þ H2 / Cu þ H2O. A similar finding was reported by Kim et al. [19,22] and Hu
Fig. 7 e SEM images of joints after exposure in 4%H2e50%H2O-N2 for 250 h at 800 C: (a) Ag, (b) Age2CuO, (c) Age4CuO and (d) Age8CuO. (High magnification SEM images 1, 2, 3, and 4 correspond to red regions in Fig. 7aed, respectively). Points A, B, C, D and E mark EDS point analysis in Table 4. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
3072
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
Table 4 e EDS composition (at. %) of points highlighted in Fig. 7. (Important values marked in bold). Points
Cu
Al
O
Ag
Fe
Cr
Mn
Possible phase
pt A pt B pt C pt D pt E
e 4.1 5.3 6.1 10.2
0.7 52.4 51.8 51.0 e
4.3 30.1 29.4 29.5 7.2
93.8 12.3 12.4 11.9 82.2
0.3 0.3 0.2 0.4 0.1
0.7 0.6 0.8 0 0.2
0.2 0.2 0.1 0.3 0.1
Ag Al2O3 Al2O3 Al2O3 Ag
Table 5 e Room-temperature shear strength of joints asbrazed and after aging in different atmospheres (MPa). Sample
As-brazed
Air (800 C/250 h)
4H2e50H2OeN2 (800 C/250 h)
Age2CuO
39 ± 3
38 ± 2
35 ± 3
et al. [41]. Hydrogen diffusing through the braze will induce the decomposition of CuAl2O4 and CuO phases to the metallic copper, which can alloys with the surrounding silver forming a solid solution in the reducing atmosphere. In case of the Age4CuO joint shown in Fig. 7c, no delamination between the braze and CGO or the braze and aluminized Crofer 22 H is observed. However, at the interface between the braze and aluminized Crofer 22 H, some porosity is present. EDS analysis of the reaction layer formed at this interface (pt C, region 3, Fig. 7c) indicates a possible decomposition of the CuAl2O4 phase formed during RAB (Fig. 4). Therefore, the observed porosity may be attributable to hydrogen effect and the decomposition of CuO precipitates [19,22]. Fig. 7d shows the microstructure of the joint brazed using Age8CuO after long-term exposure to reducing conditions. Some small pores appear at the braze/CGO interface, which is comparable to the results of Kim et al. about the influence of reducing atmospheres [19,20,22]. More seriously, widely extending delamination is observed along interface between the braze and aluminized Crofer 22 H. As discussed above, it may be assumed that hydrogen effect and the decomposition of CuO precipitates results in formation of pores along this interface. In the case of Age8CuO braze, the above two effects would be more severe due to the higher CuO concentration. Meanwhile, the large CTE mismatch between the braze (19 106 K1) and Crofer 22 H (12 106 K1) gives rise to large stresses at this interface, which may accelerate the delamination [38]. EDS analysis of pt E (Fig. 7d) indicates that there is 10.2 at% copper in the silver matrix, which demonstrates that the original CuO phase underwent decomposition,
and that the metallic copper was alloyed with the surrounding silver forming a solid solution. The shear strength was also determined to evaluate the effect of aging test on the joint properties. Only the joints brazed using Age2CuO were tested because other joints already broken after aging. The measured room-temperature shear strength is shown in Table 5. The Age2CuO joints obtain a shear strength of 39 ± 3 MPa. After aging in air at 800 C for 250 h, there is almost no change in the shear strength of joints (38 ± 2 MPa). This is attributed to the high quality Al2O3 protective layer, which maintains the stable joint microstructure. After aging in 4H2e50H2OeN2 at 800 C for 250 h, the shear strength of joints experiences a limited reduction (35 ± 3 MPa). In addition to the stable joint microstructure after aging, the mechanical interlocking structures on both sides (the penetration of braze into the pores of the CGO and the tight bonding with the rough aluminized Crofer 22 H) also play a key role in maintaining the joint strength.
The edges of the Age2CuO joints after aging tests Previous studies [26,39,42] have reported that when joining Ag based brazes to Crofer 22 APU steel without a protective layer the two main degradation mechanisms after aging in oxidizing atmospheres are evaporation of Ag and the formation of AgeCr-oxides at the joint edges (pure oxygen-850 C [26], ambient air (~55% humidity)-700/800/900 C [39] and dry air-625 C [40]). Therefore, the sample edges were inspected to identify whether similar phenomena occur in this study. Fig. 8 shows SEM images of the edges of the Age2CuO joints after aging in oxidizing (air) and reducing (4%H2e50%H2OeN2) atmospheres for 250 h at 800 C. No evaporation of Ag or formation of AgeCr-oxides can be identified either atmosphere. It is also worth noting that there are no FeeCreMn-oxides on the surface of aluminized Crofer 22 H. These encouraging results can be attributed to the high quality Al2O3 protective layer fabricated by reactive aluminization. The Al2O3 layer provides good protection of the Crofer 22 H steel and
Fig. 8 e SEM images of Age2CuO joint edges after aging in (a) oxidizing and (b) reducing atmospheres for 250 h at 800 C.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
effectively blocks outward migration of Fe, Cr and Mn from the steel. This prevents exposure of the Ag based braze to Cr. As the evaporation of Ag and the formation of AgeCr-oxides are due to the reaction between the Ag and Cr [39,42], the Al2O3 layer effectively mitigates both degradation mechanisms. The joints should be tested longer and in dual atmosphere/in a stack, but the results here are very encouraging.
Conclusions The results show that AgeCuO braze can be used to successfully join aluminized Crofer 22 H to the ceria-gadolinia layer of a SOFC/SOEC half-cell at 1000 C in air. The main conclusions are summarized as follows: (1) The high contact angle of pure Ag on the ceria-gadolinia implies the poor wettability. Increasing the CuO content improves the wettability of the AgeCuO braze by lowering the contact angle. (2) A dense and uniform Al2O3 surface layer (~2 mm) can be formed on Crofer 22 H by reactive air aluminization. The Al2O3 provides excellent protection of the Crofer 22 H substrate during RAB and hinders the outward diffusion of Fe, Cr and Mn from the steel during aging. Aluminization of the steel thereby overcomes two major shortcomings of AgeCuO brazes (interconnect corrosion by CuO and formation of AgeCr-oxides). (3) Brazed joints are aged in oxidizing (air) and reducing (4%H2e50%H2OeN2) atmospheres for 250 h at 800 C. In both atmospheres, addition of CuO to the braze is necessary to avoid delamination between the braze and the ceria-gadolinia. In oxidizing atmosphere, CuO in the braze reacts with and consumes the Al2O3 layer on aluminized Crofer 22 H, forming a CuAl2O4 reaction layer. In reducing atmosphere, hydrogen diffused through the braze leads to decomposition of CuAl2O4 and CuO, leaving pores along the braze/steel interface. (4) The recommended braze is Age2CuO, which provides a good balance between sufficient wetting and minimal reaction/degradation.
Acknowledgment The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China under Grant Nos. 51622503 and U1537206. Xiaoqing Si acknowledges the support from the China Scholarship Council for one year study at the Technical University of Denmark. The work on Al coatings is a part of the project DTU stack platform.
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijhydene.2018.11.071.
3073
references
[1] Tucker MC. Progress in metal-support solid oxide fuel cells: a review. J Power Sources 2010;195:4570e82. [2] Kennouche D, Chen-Wiegart YK, Yakal-Kremski KJ, Wang J, Gibbs JW, Voorhees PW, et al. Observing the microstructural evolution of Ni-Yttria-stabilized zirconia solid oxide fuel cell anodes. Acta Mater 2016;103:204e10. [3] Zheng Y, Zhang CM, Ran R, Cai R, Shao ZP, Farrusseng D. A new symmetric solid-oxide fuel cell with La0.8Sr0.2Sc0.2Mn0.8O3-d perovskite oxide as both the anode and cathode. Acta Mater 2009;57:1165e75. [4] Haile SM. Fuel cell materials and components. Acta Mater 2003;51:5981e6000. [5] Linderoth S, Hendriksen PV, Mogensen M, Langvad N. Investigations of metallic alloys for use as interconnects in solid oxide fuel cell stack. J Mater Sci 1996;31:5077e82. [6] Kofstad P, Bredesen R. High temperature corrosion in SOFC environment. Solid State Ionics 1992;52:69e75. [7] Singh RN. High-temperature seals for solid oxide fuel cell (SOFC). J Mater Eng Perform 2006;15:422e6. [8] Yan D, Liang LJ, Yang JJ, Zhang T, Pu J, Chi B, et al. Performance degradation and analysis of 10-cell anodesupported SOFC stack with external manifold structure. Energy 2017;125:663e70. [9] Lin KL, Singh M, Asthana R, Lin CH. Interfacial and mechanical characterization of yttria-stabilized zirconia (YSZ) to stainless steel joints fabricated using Ag-Cu-Ti interlayer. Ceram Int 2014;40:2063e71. [10] Si XQ, Cao J, Liu S, Song XG, Qi JL, Huang YX, et al. Fabrication of 3D Ni nanosheet array on Crofer22APU interconnect and NiO-YSZ anode support to sinter with small-size Ag nanoparticles for low-temperature sealing SOFCs. Int J Hydrogen Energy 2018;43:2977e89. pez SR, Haanappel VAC, Dura n A, Mun ~ oz F, Mather GC, [11] Lo Pascual MJ, et al. Glass-ceramics seals in the system MgOBaO-B2O3-SiO2 operation under simulated SOFC conditions. Int J Hydrogen Energy 2016;41:15335e45. [12] Wang SF, Hsu YF, Cheng CS, Hsieh YC. SiO2-Al2O3-Y2O3-ZnO glass sealants for intermediate temperature solid oxide fuel cell application. Int J Hydrogen Energy 2013;38:14779e90. [13] Ritucci I, Agersted K, Zielke P, Wulff AC, Khajavi R, Smeacetto F, et al. A Ba-free sealing glass with a high coefficient of thermal expansion and excellent interface stability optimized for SOFC/SOEC stack applications. Int J Appl Ceram Technol 2018;15:1011e22. [14] Peng L, Zhu QS, Xie ZH, Wang P. Interface reactions between sealing glass and metal interconnect under static and dynamic heat treatment conditions. J Fuel Cell Sci Technol 2015;12:0610091e7. [15] Hsu JH, Kim CW, Brow RK, Szabo J, Crouch R, Baird R. An alkali-free barium borosilicate viscous sealing glass for solid oxide fuel cells. J Power Sources 2014;270:14e20. [16] Kiebach R, Agersted K, Zielke P, Ritucci I, Brock M, Hendriksen PV. A novel SOFC/SOEC sealing glass with a low SiO2 content and a high thermal expansion coefficient. ECS Trans 2017;78:1739e47. [17] Malzbender J, Zhao Y, Beck T. Fracture and creep of glassceramic solid oxide fuel cell sealant materials. J Power Sources 2014;246:574e80. [18] Liu W, Sun X, Khaleel MA. Predicting Young's modulus of glass/ceramic sealant for solid oxide fuel cell considering the combined effects of aging, micro-voids and self-healing. J Power Sources 2008;185:1193e200. [19] Kim JY, Hardy JS, Weil KS. High-temperature tolerance of the silver-copper oxide braze in reducing and oxidizing atmospheres. J Mater Res Vel 2006;21:1434e42.
3074
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 3 0 6 3 e3 0 7 4
[20] Kim JY, Hardy JS, Weil KS. High-temperature tolerance of the silver-copper oxide braze in reducing and oxidizing atmospheres. J Mater Res 2006;21:1431e42. [21] Kati R, Muksin, Seyoung K, Kwang-sup S, Ji HY, Dang-Hyok Y. Joining of metal-ceramic using reactive air brazing for oxygen transport membrane applications. Mater Des 2016;109:233e41. [22] Kim JY, Hardy JS, Weil S. Dual-atmosphere tolerance of AgCuO-based air braze. Int J Hydrogen Energy 2007;32:3655e63. [23] Kuhn B, Wetzel FJ, Malzbender J, Steinbrech RW, Singheiser L. Mechanical performance of reactive-air-braze (RAB) ceramic/metal joints for solid oxide fuel cells at ambient temperature. J Power Sources 2009;193:199e202. [24] Kuhn B, Wessel E, Malzbender J, Steinbrech RW, Singheiser L. Effect of isothermal aging on the mechanical performance of brazed ceramic/metal joints for planar SOFC-stacks. Int J Hydrogen Energy 2010;35:9158e65. [25] Wahid MFR, Dang-Hyok Y, Kati R, Seyoung K, Kwang-sup S, Ji HY. Interfacial microstructure and shear strength of reactive air brazed oxygen transport membrane ceramicmetal alloy joints. Met Mater Int 2018;24:157e69. [26] Kiebach R, Engelbrecht K, Grahl-Madsen L, Sieborg B, Chen M, Hjelm J, et al. An Ag based brazing system with tunable thermal expansion for the use as sealant for solid oxide cells. J Power Sources 2016;315:339e50. [27] Choi JP, Weil KS, Chou YM, Stevenson JW, Yang ZG. Development of MnCoO coating with new aluminizing process for planar SOFC stack. Int J Hydrogen Energy 2011;36:4549e56. [28] Stevenson JW, Yang ZG, Xia GG, Nie Z, Templeton JD. Longterm oxidation behavior of spinel-coated ferritic stainless steel for solid oxide fuel cell interconnect applications. J Power Sources 2013;231:256e63. [29] Choi JP, Chou YS, Stevenson JW. Reactive air aluminization. 2011. PNNL-20859. [30] Hsu JH, Kim CW, Brow RK. Interfacial interactions between an alkali-free borosilicate viscous sealing glass and aluminized ferritic stainless steel. J Power Sources 2014;250:236e41. [31] Hauch A, Brodersen K, Chen M, Mogensen MB. Ni/YSZ electrodes structures optimized for increased electrolysis
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
performance and durability. Solid State Ionics 2016;293:27e36. Song XG, Zhao YX, Hu SP, Cao J, Fu W, Feng JC. Wetting of AgCu-Ti filler on porous Si3N4 ceramic and brazing of the ceramic to TiAl alloy. Cerm Int 2018;44:4622e9. Fu W, Hu SP, Song XG, Li JX, Cao J, Feng JC, et al. Wettability and bonding of graphite by Sn0.3Ag0.7Cu-Ti alloys. Carbon 2017;121:536e43. Eustathopoulos N. Wetting by liquid metals e application in materials processing: the contribution of the Grenoble group. Metals 2015;5:350e70. Zhang KX, Xia LH, Zhang FQ, He LL. Active brazing of C/C composite to copper by AgCuTi filer meta. Metall Mater Trans A 2016;47:2162e76. Seeharaj P, Atkinson A. Diffusion and conductivity of mixedconducting Ag/CGO composites. Solid State Ionics 2011;204e205:46e52. Cao J, Si XQ, Li WJ, Song XG, Feng JC. Reactive air brazing of YSZ-electrolyte and Al2O3-substrate for gas sensor sealing: interfacial microstructure and mechanical properties. Int J Hydrogen Energy 2017;42:10683e94. Wang LS, Barnett SA. Deposition, structure, and properties of cermet thin films composed of Ag and Y-stabilized zirconia. J Electrochem Soc 1992;139:1134e40. Sofie SW, Gannon P, Gorokhovsky V. Silver-chromium oxide interactions in SOFC environments. J Power Sources 2009;191:465e72. Abernathy HW, Koep E, Compson C, Cheng Z, Liu ML. Monitoring Ag-Cr interactions in SOFC cathodes using Raman spectroscopy. J Phys Chem C 2008;112:13299e303. Hu WT, Donat F, Scott SA, Dennis JS. The interaction between CuO and Al2O3 and the reactivity of copper aluminates below 1000 C and their implication on the use of the Cu-Al-O system for oxygen storage and production. RSC Adv 2016;6:113016e24. € M, Wiesner S, Kaletsch A, Broeckmann C. Bobzin K, Ote Characterization of reactive air brazed ceramic/metal joints unadapted thermal expansion behavior. Adv Eng Mater 2014;16:1490e7.