Influence of different aspects of the SOFC anode environment on the oxidation behavior of porous samples made of crofer

1 FC-10-1019 Antepara Influence of Different Aspects of the SOFC Anode Environment on the Oxidation Behaviour of Porous Samples made of Crofer I. Anteparaa*, M. Rivasa, I. Villarreala, N. Burgosb, F. Castrob a) IkerlanEnergía, Parque Tecnológico de Álava, Juan de La Cierva,1 E01510 Miñano, Álava, Spain b) CEIT Materials Department, Pº Manuel Lardizabal, 15 E-20018 San Sebastian Spain *corresponding author: I. Antepara*, IkerlanEnergía, Parque Tecnológico de Álava, 01510 Miñano, Spain tel. +34 945 297032, fax: +34 945 296926, e-mail: [email protected] ABSTRACT Crofer can be considered as the reference interconnect material in SOFCs working under 800ºC. Thanks to its thermal expansion coefficient (TEC), it is suitable to replace ceramic components, such as the interconnect and the metal support, and it can be cost effective. Several research groups, including Ikerlan, have used porous substrates with the same composition as Crofer (PM from H.C. Starck GmbH) as the metal support for their SOFC This is the accepted manuscript of the article that appeared in final form in Journal of Fuel Cell Science and Technology 7(6) : (2010) // Article ID 061010 , which has been published in final form at https:// doi.org/10.1115/1.4001764. © 2010 American Society of Mechanical Engineers under CC-BY distribution license (http://creativecommons.org/licenses/by/4.0/) 2 FC-10-1019 Antepara cells. The aim of this study is to determine the effect of certain variables (time, temperature, vapour content, cycling, porosity and current flow), while other aspects are constant (sample composition and particle size and shape). KEYWORDS: SOFC, metal support, oxidation, porosity ABBREVIATIONS: TEC, thermal expansion coefficient; PM, powder metallurgy; FU, fuel utilization; APU, auxiliary power unit. 1 INTRODUCTION The approach of Ikerlan’s SOFC project [1] is similar to those of other research units [2], although it has planned a tubular instead of a planar geometry for the prototype and a different final application in domestic generators that produce HSW and electricity. The basis is an operating temperature of below 800ºC, so that certain expensive ceramic components, such as the support, can be replaced with metal components. There are a many references on dense metallic materials in cathode or anode oxidizing atmospheres, but very little on the oxidation of porous materials [3], and even less on porous materials in SOFC anode conditions. The materials used in some of these studies were Crofer (Thyssenkrupp VDM GmbH) [2, 4], and ITM (Plansee) [5]. From the first study, it was concluded that for a life time of 5000h at 800ºC the surface area of the porous Crofer must be 0.04 m2/g or less [2]. At Ikerlan, preliminary tests were carried out in air for dense and porous samples of Fe70/Cr30 [3]. One of the conclusions from this study was that porosity is one of the main issues for porous samples in oxidizing atmospheres. There is currently no universally accepted fuel cell industry standard on how porosity should be measured. Archimedes 3 FC-10-1019 Antepara measurements typically result in slightly lower porosity values. Mercury porosimetry results in significantly lower values [6]. Porosity calculation through geometry and weight was used, although also mercury porosimetry data was available. The aim of the first part of the present work is to determine the effect on the behaviour of porous substrates in anode oxidizing conditions of certain variables (time, temperature, vapour content, cycling), while other aspects are constant (sample composition, particle size, porosity or no current flowing). For safety reasons, many research groups use H2 diluted in Ar as the oxidation atmosphere, because the partial pressure of O2 (pO2) is the same [7] and metal behaviour against oxidation is usually a thermodynamic function of pO2 [8]. Additional tests in H2 diluted in Ar were performed to compare with the results from the first part with pure H2 mixed with H2O. Fuel gas supply to the anode through the metal support must remain adequate during any oxidation that occurs during operation [6] and surface area is a key variable in the oxidation process of a porous sample [2]. Thus, the permeability and surface area of all the samples were characterized using different techniques, and data was compared in order to define the most suitable. In the second part, porosity and current flow were taken into account as new variables. One sample with a lower porosity was tested. Like every electrochemical reaction, corrosion consists of the interchange of electrons. Electric current has minimal effect on the composition of the oxide scales. However, because the metal ion diffusion rate will be accelerated by the electrical field, the growth rate of oxide scales on the negative side were higher than those on the positive side [9]. Thus, the current in a fuel cell tends to slow down the growth rate of the oxide on the 4 FC-10-1019 Antepara cathode side and accelerate it on the anode side, but the effects are small for an electronic conductor [10]. 2 EXPERIMENTAL Porous substrates were consolidated at high temperature in hydrogen until enough mechanical properties without breaking the samples were obtained (70% porosity). Crofer powder (H.C. Starck GmbH) was used as a raw material; no other sample composition was tested. In addition, the following aspects were constant; particle shape (irregular) and particle size. Long tubes were consolidated and samples of 1cm (H) x 1,4cm (Øext) x 1cm (Øint) were cut for the oxidation study. The same type of samples were used for isothermal and cyclic experiments and for H2 diluted in Ar. Oxidation tests were defined following a standard procedure for SOFC interconnect [11]. After measuring and weighing (GR-200 from A&D Instruments LTD), the samples were introduced into the oven (oven and controller from Carbolite). When air was used as the oxidizing atmosphere, no special equipment was needed. But when using H2-H2O as the oxidizing atmosphere, the samples were introduced into the oven inside a top-end alumina tube (4 cm inner diam., CoorsTek, Inc.). A flow of 100ml/min of pure H2 (H2 flow velocity of 4 mm/s at 800ºC, 10 times the recommended maximum [11], even higher if humidity is taken into account) was used, while in the case of Ar-10%H2, 25ml/min was used. Pure H2 was forced to bubble into the humidification system (HS-AWF from FuelCellStore.com). This system was fed with water from a purifier (AP-3 model from Puragua Systems). The flow at the outlet of the humidification system was supposed to be saturated, so, by controlling the temperature of the water, it was possible to control the H2/H2O ratio. The final humidity at the outlet of the humidification system was not 5 FC-10-1019 Antepara measured. The whole pipe from this point to the alumina tube inlet was heated over 100ºC to avoid condensation. The equipment for mercury porosimetry (AutoPore IV from Micromeritics and AutoPore IV 9500 software) was working from 50 to 60000psi, so that pores between 0.003 and 360µm could be measured. SEM and EDAX analyses were carried out with a Quanta 200 F microscope (FEI Company). The equipment model 1PW1825 from PHILIPS was used for XRD analysis. The parameters were; CuKα1 line (λ= 1.542 A) as radiation, an X-ray tube potential of 40 kV and a current of 40 mA. A voltage and current generation system based on Field Point commercial modules and LabWindows/CVI software (National Instruments, NI) were used to study the influence of current flow on the oxidation. Samples with porosities as low as 30% were sintered to study the influence of a lower porosity on the oxidation of porous crofer. 3 RESULTS 3.1 Effect of temperature and water vapour content Isothermal mass gain determination tests were performed in humidified hydrogen at temperatures of 600ºC, 700ºC and 800ªC, and the following water vapour contents; 3, 7, 25, 50 and 70%. 3 points (2, 24 and 72 hours) described the oxidation kinetics curves. In this study, only one sample was considered enough to determine the oxidation kinetics, as will be explained later. It was found that in air at 800ºC with less than 3% of water vapour (the oven is open and ambient humidity is not controlled but varies between 2 and 3%), the oxidation was catastrophic, whereas in a hydrogen atmosphere with 3% water vapour content, the oxide 6 FC-10-1019 Antepara scale formed is protective (Fig 1). Higher temperatures cause higher mass gains. A test of 144 hours in hydrogen with 3% of water vapour (H2-3%H2O) was performed on one of the samples in order to confirm the behaviour over time. A SEM analysis (Fig 2 a)) showed that from 600 to 800ºC the samples were oxidized uniformly in H2-3%H2O. The oxide layers were not thick enough for an EDAX analysis, although higher Cr concentrations were measured. From the study in H2-7%H2O the same conclusions could be drawn, except that thicker oxide layers are formed. Thus, the EDAX analysis suggested more categorically that the stoichiometry of the (Fe, Cr) spinel was very close to that of FeCr2O4. It was not so clear that in H2-25%H2O the higher the temperature, the higher the mass gain. At 600, 700 and 800ºC the mass gain after 2 hours and 72 hours was almost the same, but always higher than in H2-7%H2O. The concentration of Fe in the oxide layer was higher. Both this result from EDAX and the very similar mass gains that were measured were ratified by XRD data (Fig3 a). Mass gains with a 50% water vapour content were more than double the previous case (Fig 4). This can be explained thermodynamically because of the formation of different iron oxides depending on the water vapour content [2]; at a certain temperature, as the percentage of water vapour is greater, higher pO2 are thermodynamically in equilibrium. In H2-25%H2O at 800ºC only Cr and (Cr,Fe) oxides are thermodynamically stable while no Fe oxide is stable and FeO is stable in H2-50%H2O. In H2-50%H2O at 700 and 800ºC, the main mass gains occurred during the first two hours, while 24 hours were needed at 600ºC. Only the sample at 800ºC continued to gain mass after the first two hours, up to a mass gain of 35%. This value is considered as the highest oxidation state for Crofer with the said pO2 [4]. Photos from SEM analysis confirmed that samples at 800ºC were almost fully oxidized. SEM images from samples in 7 FC-10-1019 Antepara H2-50%H2O looked like those in H2-70%H2O (Fig 2 b)). At 600 and 700ºC, non-oxidized FeCr particles could be found easily and the Cr content in these particles was lower than in Crofer. The peak of Crofer from XRD data (Fig3 b)) varied from being significant after 2 hours at 600ºC, to zero after 72 hours at 800ºC. In this case, Fe3O4 and (Fe,Cr)3O4 oxides were the main phases detected. Mass gains in H2-70%H2O for all the temperatures were a little higher than in H250%H2O, but they produced the same curves. In addition, SEM and EDAX analyses were very similar to those of the samples in H2-50%H2O. Except that at 600ºC after two hours, the XRD peak for Crofer was very weak. Mass gains at 600ºC were the lowest among 600ºC, 700ºC and 800ºC for 3 and 7% of humidity, and at 700ºC for 25% and higher water vapour contents (Fig 4). This can be explained by different iron oxides formed depending on the water vapour content and by a different dependence on temperature of the diffusion coefficients of Fe and Cr. With Crofer in air, high mass gains were also obtained at temperatures of around 600ºC. Nonprotective iron oxides were detected outside the cromia at 650ºC [12]. 3.2 Effect of cycles. Two cyclic oxidation tests were performed in H2-3%H2O; the first, three cycles of 8 hours to compare with the isothermal mass gain after 24 hours; the second, another three cycles of 24 hours to compare with the data after 72 hours. The heating/cooling slope is 10ºC/min, the recommended maximum for the top-end alumina tube. In the figure showing all the data with 3% vapour (Fig. 1), the mass gains were almost coincident. Although in the real case steeper heating/cooling and a larger number of cycles will have to be considered, they seem not to have an additional detrimental effect on oxidation resistance. 8 FC-10-1019 Antepara 3.3 Effect of dilution in Ar. Two kind of diluted atmospheres were investigated. The same Ar-10%H2 gas mix was used in both cases, but the temperature of the water of the humidification system bath was different. Based on the assumption that the Ar-10%H2 gas mix was saturated at the outlet of the humidification system (the final dew point of the gas mix was not measured), the following two gas compositions were obtained: - With the temperature of the humidification system bath at 25ºC, if the gas reached saturation (low flows were used, so the assumption of saturation was more feasible), absolute humidity was 3%. Thus, the final composition was Ar – 9%H2 – 3%H2O (H2/H2O ratio 9/3 = 3) - With the bath at 44ºC, the final composition was Ar – 9%H2 – 9%H2O (H2/H2O ratio 9/9 = 1) Figure 5 shows results from this oxidation test and the correlation between a pure H2 – H2O atmosphere and the same H2/H2O ratio diluted in Ar. Lower kinetics were obtained because the oxidizing flow used was not enough. In the end, the same oxidation level was reached, as can be theoretically expected [7]. After 72 hours, all samples oxidized in an atmosphere with the same H2/H2O ratio followed the same XRD pattern. 3.4 Use of surface area from Hg porosimetry to represent mass gain data. Different techniques were used to obtain the surface area; determination of permeability [13] and calculation of surface area through the Kozeny-Carman equation [14], BET, and Hg porosimetry. The best results were obtained with Hg porosimetry. The samples for the study to determine the oxidation kinetics were cut from a long PM tube consolidated at high temperature in H2. The permeability of a portion of this long tube was measured using Hg porosimetry, as it is a destructive analysis and after the Hg porosimetry, the properties of the sample changed. The porosimeter reports the surface 9 FC-10-1019 Antepara area directly. This value was used to report the mass gain as area-specific mass gain and compare it with the mass gain reported as a percentage of mass gained from initial mass. As shown in figure 4, they looked identical and both graphs could be used to report the oxidation kinetics. The mass gain reported as a percentage of mass gained from initial mass is more reliable as it only considers the error of two mass measurements. The areaspecific mass gain has the additional error of the surface area measurement. With a single sample, it was not possible to perform the Hg porosimetry test first and then try to use the sample in the oxidation test. This is why specific mass gains referred to the surface area of a different, non-oxidized sample (although both samples were cut from the same tube). Hg porosimetry tests were carried out on oxidized samples, but further investigation is required to reach definitive conclusions. The average surface area of the samples used in this study was 500 times the recommended area for dense samples in thermogravimetric experiments [11], which is why repeatability was surprisingly so good in this study. 3.5 Effect of porosity. When considering the oxidation of the metal support of an SOFC, its porosity is a very important variable. In the beginning of the manufacturing process of the SOFC cells, the support has a high porosity, but it shrinks during sinterization. This is why both the porosity and surface area of the metal-support decrease, as shown in figure 6. Compared to the specific surface area of spherical particles, irregular ones of the same size have a higher specific surface area. The oxidation of porous samples depends on specific surface area [2]. Thus, irregular particles have a weaker oxidation resistance. The sintering process transforms the initial shape of the particles, it makes the irregular particles more round, and the porosity and the surface area decrease at the same time. These two aspects enhance the oxidation resistance of the metal support after sintering. 16 FC-10-1019 Antepara FIGURE CAPTIONS Fig1. Mass gains after 2, 24 and 72 hours at 600ºC, 700ºC and 800ºC. In addition, 3 cycles of 8 and 24 hours in H2-3%H2O at 800ºC. Fig2. SEM images for samples oxidized at 800ºC for 72h with a) 3% of vapour, and b) 70% of vapour Fig3. XRD patterns for samples oxidized at 600, 700 and 800ºC for 2, 24 and 72 hours with a vapour content of a) 25% and b) 50%. Fig4. Mass gains of the porous samples after 72 hours in H2 with vapour contents of 3, 7, 25, 50 y 70% a) referred to initial mass, and b) referred to surface area measured by Hg porosimetry before oxidation Fig5. Mass gains after 2, 24 and 72 hours at 600ºC, 700ºC and 800ºC in H2/H2O atmospheres (ratios 1 and 3) diluted in Ar. Fig6. SEM images of the surface of three non-oxidized samples a) consolidated at high temperature in hydrogen (70% porosity) b) 20% of shrinkage c) 25% of shrinkage (~30% porosity) Fig7. SEM images for samples oxidized at 800ºC in H2/H2O=1 for 70h a) consolidated at high temperature in hydrogen (70% porosity) b) 25% of shrinkage (~30% porosity) Fig8. SEM images for samples oxidized at 800ºC in H2/H2O=1 for 70h with a current flow of a) 0mA/cm2 b) 800mA/cm2 17 FC-10-1019 Antepara Fig1. Mass gains after 2, 24 and 72 hours at 600ºC, 700ºC and 800ºC. In addition, 3 cycles of 8 and 24 hours in H2-3%H2O at 800ºC. a) 18 FC-10-1019 Antepara b) 19 FC-10-1019 Antepara Fig2. SEM images for samples oxidized at 800ºC for 72h with a) 3% of vapour, and b) 70% of vapour a) 20 FC-10-1019 Antepara b) Fig3. XRD patterns for samples oxidized at 600, 700 and 800ºC for 2, 24 and 72 hours with a vapour content of a) 25% and b) 50%. a) 21 FC-10-1019 Antepara b) 22 FC-10-1019 Antepara Fig4. Mass gains of the porous samples after 72 hours in H2 with vapour contents of 3, 7, 25, 50 y 70% a) referred to initial mass, and b) referred to surface area measured by Hg porosimetry before oxidation Fig5. Mass gains after 2, 24 and 72 hours at 600ºC, 700ºC and 800ºC in H2/H2O atmospheres (ratios 1 and 3) diluted in Ar. a) 23 FC-10-1019 Antepara b) 24 FC-10-1019 Antepara c) 25 FC-10-1019 Antepara Fig6. SEM images of the surface of three non-oxidized samples a) consolidated at high temperature in hydrogen (70% porosity) b) 20% of shrinkage c) 25% of shrinkage (~30% porosity) a)