SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

High-performance solid oxide fuel cells require a thin and gas-tight electrolyte membrane that must be coated on a porous and relatively rough support. A pretreatment of the delivered submicronmeter electrolyte powder of 8 mol%-yttria-stabilized zirconia (8YSZ) yielded a reduced sintering mismatch between the anode substrate made from NiO/8YSZ and the electrolyte coating. Furthermore, it also enhanced the powder packing inside the green film. Constrained sintering usually leads to inadequate film density and an unfavorable pore deformation and orientation. It was demonstrated that these limitations can be resolved by using a coshrinking substrate in a planar cell design. Relative densities of >97% were achieved, which are higher than those for free-standing layers. Additionally, the camber behavior was investigated in dependence of the temperature program with and without gravity effects, giving an overall suggestion for the cofiring parameters of the electrolyte.


I. Introduction

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

Solid oxide fuel cells (SOFC) constitute potential, promising candidates for future electrical power generation. The capability of running a variety of fuels, e.g. diesel reformat or natural gas, its high efficiency, and low emissions represent distinguished features of this technology, which can be utilized in stationary, mobile, and portable applications. Thin electrolytes lower the electrical resistance of the cell and facilitate the use of high-performance cathodes, e.g. with lanthanum strontium cobalt ferrite (LSCF), improving the cell performance.1,2 An anode-supported SOFC with a thin electrolyte (<10 μm) achieves areal power densities approximately five times higher than electrolyte-supported cells with a thick electrolyte (150 μm).3 However, the cell-manufacturing process is more sophisticated for anode-supported cells.

In the cell design developed at Forschungszentrum Jülich, a presintered anode substrate (NiO/8 mol% yttria-stabilized zirconia (8YSZ) is coated with an anode (NiO/8YSZ) and an electrolyte layer (8YSZ), each with a final thickness of 5–10 μm.4 Whereas large pore sizes of up to 50 μm in the anode substrate provide sufficient gas transport during later operation, the anode layer exhibits a finer microstructure in order to increase the length of the active three-phase boundaries and, therefore, the electrochemical performance of the cell.5 The three components, constituting a half-cell, are processed as oxide ceramics during manufacturing and cosintered in a single step (1400°C/5 h). Complete anode-supported cells, including a lanthanum calcium manganite cathode (LCM), were reported to be manufactured by a single cofiring step (1300°C/2 h using sintering additives)6 and to yield maximum current densities of 0.9 A/cm2 running on H2/3% H2O and air at a cell voltage of 0.7 V and a temperature of 700°C (no cell size was reported).7 However, cathodes with higher performance (e.g., LSCF) require lower sintering temperatures, typically in the range of 1000°–1100°C.2 Thus, they are processed in a separate step and yield, in combination with the half-cells described in this work, approximately 1.7 A/cm2 under the above conditions and a 16 cm2 active cell area.8

The functional microstructure of all layers of the half-cell has to evolve during the cofiring procedure. For the electrolyte, this means the absence of any open porosity, i.e. gas tightness. Constrained sintering on rigid substrates usually leads to inadequate film densities.9 Furthermore, a pore structure develops that opposes gas-tight layers, i.e. elongated pores form, which are oriented perpendicular to the film plane.10 The group of Guillon and Rödel investigated these phenomena for alumina films sintered on a rigid substrate by performing advanced image analysis of layer cross sections.10,11 Whereas it is generally known that a coshrinking substrate enhances the densification of the applied film, there are few quantitative analyses available for coshrinking substrates. Kanters et al.12 simulated the densification behavior of nanocrystalline zirconia and found that the density of the cosintered film was higher than for free-standing layers. However, this was not examined experimentally. Yamaguchi et al.13 found that a coshrinking anode support was necessary to obtain a dense ceria electrolyte in their microtubular cell design. In the present work, screen-printed zirconia (8YSZ) films that were applied on coshrinking anode supports were analyzed and compared with both free-standing layers and layers on rigid substrates (dense and porous). Thereby, density, pore size, pore form, and pore orientation were considered.

Furthermore, the multilayer compound suffers from compatibility stresses that lead to viscous deformation and camber during sintering.14–16 For successful stack assembly, the sintered fuel cells need to provide a certain flatness. Because camber-free-layered structures are of common interest, many groups carried out a variety of bending measurements of various layer compounds.16–18 The observations were correlated with the free-sintering behavior and elastic or viscous properties of the layers19 and sometimes the application of pressure was proposed to avoid warpage.20 Some reported experiments were carried out for vertically hanging samples,17 in order to exclude the influence of gravity on compounds with a low sintering viscosity and relate the observations directly to the sintering mismatch. In this work, the curvature of the compound was studied experimentally during the course of sintering and for varying heating and cooling rates. This behavior was investigated for both vertically hanging and normally placed specimens. Together with the investigation of the microstructure, the sintering regime could be optimized and tested with typical half-cells.

In the past, mainly vacuum slip casting21 was used for electrolyte coatings in Jülich. This method required the use of calcined and milled electrolyte powder, to avoid drying and sintering cracks.22 This pretreated powder was also used in the main part of this work, although layers were now screen printed, which is more feasible for mass production. A short comparison of as-received powder grades from Tosoh and the pretreated powder with respect to packing density in screen-printed films and sintering behavior will document the benefits of the pretreated powder.

II. Experimental Procedure

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

(1) Films and Substrates

Electrolyte films were prepared from 8YSZ (theoretical density 5.96 g/cm3) by screen printing. As-received grades of TZ-8Y and TZ-8YS from Tosoh (Tosoh Corp., Tokyo, Japan) and a calcined and milled variety of TZ-8Y (1230°–1240°C/3 h; approximately 60 h of wet milling with 3–5 mm grinding balls on a roller bank) served as starting powders (Fig. 1, Table I). Screen-printing pastes included terpineol as the solvent (Fluka, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany; or DP 8250, Du Pont de Nemours Deutschland GmbH, Cologne, Germany) and ethyl cellulose as the binder (4 wt% per ceramic content; Sigma, 45 cps, Sigma-Aldrich Chemie GmbH). A dispersing agent, Hypermer KD-1 (Uniqema GmbH & Co. KG, Emmerich, Germany), was added at a concentration of 1 mg/m2 of 8YSZ powders to some of the pastes. Furthermore, free-standing layers of lateral dimension 70 mm × 70 mm and dried thickness of 50–80 μm were prepared using a waterslide transfer paper (Trucal Plus, Tullis Russel Coaters Ltd., Macclesfield, U.K.). Very thick layers (up to 700 μm for dilatometric analysis) were obtained by joining several dried free-standing layers together with a thin film of the same paste.

image

Figure 1.  Micrographs of the 8YSZ electrolyte powders used: (a) TZ-8Y as delivered, (b) TZ-8YS as delivered, (c) TZ-8Y calcined at 1230°C/3 h and ball milled. 8YSZ, 8 mol% yttria-stabilized zirconia.

Download figure to PowerPoint

Table I.    Comparison between Delivered Grades, TZ-8Y and TZ-8YS, and the Calcined and Milled TZ-8Y
 Grain size from SEM (estimation μm)Particle size d10/d50/d90 (μm)Specific surface (m2/g)Paste solid content (wt%)Leak rate (hPa dm3)/(s cm2)Green density (% theoretical density)
  1. The leak-rate of screen printed electrolytes was measured after heat-treatment of 1400°C/5 h, 120 mg (± 5%) 8YSZ were deposited on an area of 71 mm × 71 mm resulting in a sintered layer of approximately 5 μm.

  2. SEM, Scanning electron microscopy; 8YSZ, 8 mol% yttria-stabilized zirconia.

TZ-8Y as delivered0.10.16/0.25/0.4011.9502.9 ± 1.1·10−444.0 ± 0.4
TZ-8YS as delivered0.20.26/0.46/0.965.9574.5 ± 2.5·10−547.3 ± 0.4
TZ-8Y calcined/milled0.50.33/0.44/0.574.6659.9 ± 3.7·10−658.7 ± 0.5

The density of green coatings was calculated from film volume, mass, and ceramic content. The volume of layers with a dried thickness of 30–40 μm (40 mm × 40 mm) that were printed on polished steel substrates (50 mm × 50 mm × 1.5 mm) was measured using differential laser topography (CT200 with triangular sensor DRS-500, Cyber Technologies GmbH, Ingolstadt, Germany).23 Coated and blank substrates were scanned with a step size of 500 μm × 100 μm and the substrate bending caused by drying stresses was compensated.

The coatings were applied and sintered on dense 99.6% alumina (51 mm × 51 mm, Rubalit® 710, CeramTec AG, Marktredwitz, Germany), porous 8YSZ (50 mm × 50 mm), and presintered NiO/8YSZ anode substrates (50 mm × 50 mm or 75 mm × 75 mm; NiO: Mallinckrodt Baker Inc., Phillipsburg, NJ; 8YSZ: Unitec Ceramics Ltd., Stafford, U.K.). The porous 8YSZ and anode substrates were manufactured by warm pressing resin-coated powders (Coat-Mix® process24). Except for curvature test samples, the anode substrates were coated with a 7-μm-thick anode layer by vacuum slip casting made from 56 wt% 8YSZ (TZ-8Y, calcined/milled) and milled NiO (Mallinckrodt Baker, Phillipsburg, NJ) before the application of the electrolyte. In order to study the influence of the substrate, the same screen (theoretical paint volume of 29 cm3/m2) was used for printing the layers, resulting in a dried thickness of 6–8 μm. The characteristics of the substrates used are summarized in Table II. Particle sizes were measured with dynamic light scattering (UPA 150, Microtrac Inc., Montgomeryville, PA), specific surfaces by nitrogen adsorption, and mechanical and optical roughness values were obtained using a perthometer (Perthometer M2 with probe NHT 6-100, Mahr GmbH, Göttingen, Germany, in compliance with DIN EN ISO 3274) and an optical confocal inspection system (CT200 with sensor LT9010M, Cyber Technologies GmbH, Ingolstadt, Germany), respectively. The particle size obtained by dynamic light scattering was affected by the strong tendency of powder agglomeration even after ultrasonic treatment. As steep pores are smoothed physically by the mechanical system, the roughness values obtained by two devices differ.

Table II.    Characteristics of Used Substrates
SubstrateMaterial/processHeat pretreatmentThicknessPorosity (%)Roughness (mechanical) Ra/Rq/Rz (μm)Roughness (optical) Ra/Rq/Rz (μm)
  1. Roughness values according to DIN EN ISO 4287.

  2. 8YSZ, 8 mol% yttria-stabilized zirconia.

Coshrinking anode substrate57 wt% NiO/43 wt% 8YSZ Coat-Mix®1230°C/3 h1.1 or 1.75 mm550.69/0.89/5.11.2/1.6/11
With anode layerVia vacuum slip casting1000°C/1 h7 μm (layer)0.45/0.60/3.80.9/1.2/5.9
Rigid, denseAl2O3 (Rubalit® 710 CeramTec AG, Marktredwitz, Germany)Unknown0.63 mm0.09/0.11/0,780.21/0.26/1.6
Rigid, porous8YSZ, Coat-Mix®1500°C/12 h1.5 mm230.26/0.35/2.30.56/0.73/5.6

(2) Densification and Microstructure

Free shrinkage data of pressed pellets and the lateral densification of free-standing layers and substrates25 were acquired in air by a vertical push-rod dilatometer (Setsys 16/18, Setaram Inc., Caluire, France) applying a force of 10 mN or smaller. The thermal expansion of the sample and of the measurement system was corrected by blank and standard measurements. Taking into account the large deformation during sintering, true strains were used in this paper.

The density of the applied, sintered films (5–8 μm in thickness) was determined from image analysis of SEM micrographs using polished cross sections (Zeiss Gemini Ultra 55, Carl Zeiss NTS GmbH, Oberkochen, Germany). Depending on the pore frequency, 10–30 images per sample were processed. For free-standing layers of larger thickness (50 or 320 μm), 5–20 images were taken per sample. Pores were approximated by ellipses, in order to quantify their properties. At least 400 pores, maximum 5000 pores, were investigated per sample. The average pore elongation factor was calculated by the arithmetic mean of the aspect ratio of the approximated ellipses, ɛi (greater than or equal to unity), weighted by their larger diameter di. Furthermore, the pore orientation was quantified using the concept of cumulated pore lengths10 in which the weighted pore length, ɛi·di, is analyzed as a function of the pore orientation angle θi. This angle is measured between the film plane and the larger diameter of the ellipses. Guillon and colleagues10,26 utilized the ratio of vertical and horizontal cumulated pore length to define the pore orientation factor k. Thereby, pores with θi from 0° to 30° and from 150° to 180° were considered as horizontal, and pores with θi from 60° to 120° as vertical.26 In the present work, the pore orientation factor was defined as

  • image(1)

which provides a more general form, as all pores are included and their precise orientation is taken into account. If the cumulated pore length plots, in polar coordinates, exhibit a semi-elliptical shape, this value equals to the ratio of the cumulated pore length of completely vertical (90°) and horizontal (0°, 180°) pores and, therefore, it resembles the previous definition. For the sake of uniform scaling, the logarithm ln k is used, which introduces a mathematical symmetry between vertically and horizontally aligned pores. Before performing the described quantitative analyses, the acquired images were processed in the following way: contrast enhancement, background subtraction, anisotropic diffusion filtering,27 and binarization.

In order to plot the results achieved at different temperatures and dwelling times on the same scale, the concept of master sintering curves28 and the related master variable log (Θ), “sintering work,”29 was used

  • image(2)

where QMSC denotes the effective activation energy, T the temperature, t the time, and R the molar gas constant. The effective activation energy of free-standing electrolyte layers was determined from dilatometric shrinkage data of four different heating rates (+1, +3, +5, and +8 K/min) and subsequent dwelling at 1500°C/2 h to inline image.30 As most of the sample groups were sintered under the same or similar conditions, they can be compared without validating the concept's underlying assumption of a dominating sintering mechanism.

The grain size of the sintered layers was analyzed for one sintering program: +3 K/min, 1400°C/5 h, –10 K/min. After the samples had been thermally etched at 1310°C/1 h, seven to nine SEM images per sample were processed by the linear intercept method and a correction factor of 1.56 was used. The presented values for the gas tightness of sintered electrolyte layers (half-cell dimension: 65 mm × 65 mm) were measured by a helium leak finder (HLT260, Pfeiffer Vacuum, Asslar, Germany). The helium flow through the half-cell was determined with a mass spectrometer at a pressure difference of 1000 hPa. The values were normalized to measured area (16 cm2) and to a pressure difference of 100 hPa, which is typical for an SOFC stack.

(3) Camber Development

The curvature of sintered samples was studied after cooling with optical laser profilometry (CT200 with sensor DRS-8000). Electrolyte-coated anode substrates (Coat-Mix®, 46 mm × 23 mm × 1.1 mm) without the anode layer were used, in order to study the two dominating layers. The electrolyte layer of calcined and milled TZ-8Y (60 wt% solid content in the paste, no dispersing agent) displayed an average dried thickness of 7.5 μm measured using a micrometer screw and 11 μm including roughness valleys of the supporting substrate. Two samples were placed horizontally on kiln furniture and two samples were hung vertically,17 for each temperature program, in order to study the influence of gravity. A small hole was centered and drilled close to one of the short edges for vertical fixture. The curvature was calculated by fitting a circle to the long center profile. This method yields high-precision curvature data with a typical error of 0.02 m−1, in contrast to in situ characterizations, which display typical errors12,17 of approximately 0.5–5 m–1. However, the investigated samples exhibited some local deviation from perfect spherical bending, which was found to be approximately 5% for free-hanging samples. For the normally placed samples, the maximum curvature was included in the results. It was obtained by searching for the most bent part with a length of 25% of the total profile. The results from these small test samples were verified on larger half-cells (75 mm × 75 mm × 1.1 mm) with anode layers, placed horizontally with the electrolyte facing upwards.

III. Results and Discussion

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

(1) Sintering of Free Components and Effect of Powder Conditioning

Utilizing free-standing layers, precise data of free lateral shrinkage for every layer of the multilayer compound—anode substrate, anode, and electrolyte layer—were obtained (Fig. 2). Hereby, the calcined and milled variety of the electrolyte powder was used. Taking into account the sintering rate, the anode layer exhibits two phases: below approximately 1200°C, its NiO component dominates the sintering rate, whereas above this temperature the 8YSZ component, which was processed in the same manner as the electrolyte powder, is governing. Although the sintering mismatch is most pronounced between the anode layer and the anode substrate, the mismatch between the electrolyte and the substrate governs the cofiring behavior due to the large viscosity of the electrolyte throughout the heat treatment.31 The variance is most distinct during heat-up; it cannot be changed significantly by the heating rate or the green density of the electrolyte,25 but is mainly a consequence of intrinsic powder properties. The free sintering rate of the electrolyte exceeds the rate of the substrate during the heating ramp (+3 K/min) and the first 15 min of the isothermal period at 1400°C. Assuming comparable kinetics in the multi-layer compound, this corresponds to tensile stresses of the electrolyte, which can potentially affect its integrity.32 After this point, the shrinking substrate compresses the film.

image

Figure 2.  Free sintering of the three half-cell components (NiO/8YSZ anode substrate, NiO/8YSZ anode layer, and 8YSZ electrolyte layer) during the sintering regime +3 K/min, 1400°C 5 h,−5 K/min. 8YSZ, 8 mol% yttria-stabilized zirconia

Download figure to PowerPoint

As the comparison in Table I shows, the as-received powders from Tosoh lead to significantly higher leak rates of similarly prepared screen-printed and sintered layers. The quality control threshold of the cell manufacturing at the institute lies at 4.0 × 10−5 (hPa·dm3)/(s·cm2), which is only safely achieved by the calcined and milled powder. Whereas some potential in further decrease of the leak rate from larger layer thicknesses and additional process optimization exists, the layer quality using calcined and milled powder could yet not be achieved the as-received grades. One advantage of the calcined and milled TZ-8Y represents its high green density, which enhances later densification. The calcined grains exert only little agglomeration forces; this allows the preparation of pastes without a dispersing agent23 and use in suspensions with little binder content such as slip casting ethanol-based suspensions.21 Furthermore, the sintering behavior of this powder exhibits less mismatch to the substrate and the range of common sintering activity is augmented (Fig. 3). For a precise treatment, the investigation of the grain-size-dependent viscous properties and sintering stresses of the layer system would be required.33 In the further course of this work, only calcined and milled TZ-8Y was used.

image

Figure 3.  Sintering of different electrolyte powders (8YSZ, pellets with equal green density 50.6±0.8% theoretical density) in comparison with NiO/8YSZ anode substrate; constant heating rate: +3 K/min. 8YSZ, 8 mol% yttria-stabilized zirconia.

Download figure to PowerPoint

Another way to minimize the sintering mismatch of the components is to modify the sintering behavior of the substrate and to use green substrates for the cofiring. Considering a constant heating rate, Moon et al.34 were able to adapt the sintering behavior of tape-cast anode substrates to match the electrolyte's shrinkage by replacing 50% of the coarse zirconia by a fine zirconia powder (TZ-8Y) in the substrate and cofiring the half-cells at 1350°C/3 h. For small button cells with LSM/8YSZ cathodes, they achieved modest current densities of 0.5–0.6 A/cm2 at 700°C (0.7 V, H2/3%H2O). The developed anode substrate and active layer remain to be examined for sufficient gas diffusion and electro-chemical activity at higher current densities, e.g. at higher temperatures or other cathodes; furthermore, no comment on the final camber was given.

(2) Substrate Influence on Densification and Microstructure

The influence of the substrate during cofiring was investigated for the supports given in Table II. Additionally, free-standing layers were prepared and sintered on a powder bed. Below, “free-standing 1” refers to layers of dried thickness of approximately 320 μm (which were designed and used for bending viscosity measurements in Mücke30) and “free-standing 2” refers to layers of dried thickness of 50 μm. The relative densities in Fig. 4 reveal the large influence of the substrates. After 1400°C/5 h, the free-standing layers, 1 and 2, achieve values of 96.5% and 96.6% theoretical density, respectively. In contrast, no gas-tight layers could be obtained on rigid substrates. Relative densities were as low as 86.6% and 85.5% for the dense alumina and porous zirconia (8YSZ) substrates, respectively (compare Fig. 5). This limitation in densification is typical for films of all kinds of materials on rigid substrates11,35,36 and relates to the viscous Poisson's ratio, the evolution of the sintering potential, the accompanying anisotropic grain growth, and the developing anisotropic pore structure.16 The last two points are particularly challenging in that they limit the understanding of the densification of constrained films and restrict model predictions.37

image

Figure 4.  Sintered density of free-standing and substrate-bound layers prepared from calcined/milled TZ-8Y as a function of the sintering work, log Θ (Eq. (2)), of the free-standing electrolyte. Used substrates: rigid, dense alumina; rigid, porous zirconia (8YSZ); anode support with an anode layer (NiO/8YSZ). The corresponding sintering programs are given on top; additionally, the first four data points of “free-standing 1” were obtained after 1 h of sintering at 1318°, 1359°, and 1386°C, and after 2 h at 1400°C, respectively. Common heating rate, +3K/min; cooling rate, −10K/min. 8YSZ, 8 mol% yttria-stabilized zirconia.

Download figure to PowerPoint

image

Figure 5.  Scanning electron microscopy images of 8YSZ layers sintered at +3 K/min, 1400°C/5 h, and −10 K/min (a) on a rigid, dense alumina substrate, (b) on a coshrinking anode substrate with an anode layer (NiO/8YSZ), (c) on a rigid, porous zirconia substrate (8YSZ), (d) freestanding (same magnification, c and d after thermal etching). 8YSZ, 8 mol% yttria-stabilized zirconia.

Download figure to PowerPoint

The coshrinking anode support (NiO/8YSZ) changes the densification behavior significantly. In the beginning, the layers applied on these substrates are subjected to the same constraints as the layers on rigid substrates; after 1400°C/0.25 h, the densification accelerates significantly, which correlates to the free-sintering rates of Fig. 2. The coshrinking substrate compressed the coating to higher densities and increased the thickness compared with the rigid supports (Figs. 5(a) and (b)). The density of the cosintered layer exceeded the density of the free layers after approximately 1400°C/3 h (95.9% theoretical density) and increased to 97.1% after 1400°C/5 h, i.e. 0.5% points higher than for free-standing layers. A similar effect was simulated by Kanters et al.12 for nanocrystalline zirconia layers. Yamagushi et al.13 observed improved densities in cofired gadolinium-doped ceria films on microtubular anode supports after the shrinkage of the supports was increased by lowering the substrates' calcination temperature. They reported film densities of 55% theoretical density on almost rigid substrates as well as densities of 91% and >98% on substrates that experienced lateral shrinkage, of 9.9% and >19%, respectively, during free sintering (converted to true strain). Thus, the substrate shrinkage required was similar to the value of 13% for the free substrates obtained after 1400°C/3 h in the present work, which was needed to densify the zirconia films to >95% theoretical density (Fig. 2). Owing to the sensitivity of the coating technique used (dip coating) to the capillary activity of the support, which depends on the calcination temperature, Yamagushi et al. compared different layer thicknesses that lay in the range from 5 μm for almost rigid to 25 μm for the most porous shrinking substrates, which probably amplified the dependency observed.11

Another noticeable observation is the slow densification of the layers on top of the rigid, porous zirconia substrates (Fig. 4). Guillon et al.11 observed that the densification inside an alumina film was more retarded in the vicinity of the substrate (smooth and rigid) than elsewhere in the film, which was mainly attributed to the hindered particle rearrangement in this region. Furthermore, it is obvious that the densification of the film inside a substrate's pore is even further retarded, because not only one constraining interface exists, but up to five, which restrains the particle rearrangement and the mass flow during sintering significantly. Therefore, it can be concluded that the constraining conditions on a rough substrate are different and more effective than on a smooth substrate. Even though the anode substrate is even rougher, the substrate and the anode layer on top of it also sinter during cofiring and thus decrease the influence of the roughness. For further studies, substrates made of the same material should be used, as the materials may affect the densification of the applied layer. The alumina substrates were chosen because of their availability and superior surface quality. Diffusion can occur between the particles of the 8YSZ layer and the 8YSZ substrate, which can improve the interfacial adhesion and thus further retard the particle rearrangement. On the other hand, alumina is sometimes used as a sintering aid for 8YSZ.22 However, the huge differences between rigid and co-shrinking substrates as well as free-standing layers remain despite these considerations.

The grain size was measured on thermally etched samples for one sintering program (1400°C/5 h, Figs. 5(c) and (d)) and was found to be 2.8±0.3 μm for the free-standing layers (96.7% theoretical density), 2.4±0.4 μm for the layers sintered on the dense alumina substrates (86.6%), 1.7±0.3 μm for the layers on the porous zirconia substrates (85.5%), and 4.0±0.5 μm for the electrolyte layers (97.1%) on the coshrinking supports. This means that the grain growth was retarded for the fully constrained films and amplified for the films compressed by the coshrinking support. The pores in Figs. 5(a) and (c) are larger than the grains, whereas the opposite is found for the free and cosintering layers.

The mean apparent pore size in the cross sections of the free layers decreases slightly until approximately log Θ=−18.0, which corresponds to 95% theoretical density (1400°C/1 h), and increases somewhat again afterwards (Fig. 6). This increase could be related to pore growth and the disappearance of smaller pores.38 Whereas no clear trend is observed for the layer applied on rigid substrates, the layers applied on anode supports exhibit the largest variation in pore size evolution. Their pore size decreases continuously (with some scattering points) and becomes significantly smaller than the value of the layers applied on rigid substrates after 1400°C/3 h. Finally, it reaches the same value as the free-standing layers at 1500°C/3 h. This process apparently resembles a slight hot pressing: particle rearrangement takes place caused by creep,39 pores close, and the number of contact points increases, which in turn improves the sinter activity. This prevents the formation of thermodynamically stable pores. As irregular or connected pores are not truly represented in the cross sections, it is not possible to correlate gas tightness and the apparent pore size directly.

image

Figure 6.  Apparent mean pore size determined from cross sections during the course of sintering.

Download figure to PowerPoint

The gas tightness of a layer does not only depend on its density but also on the configuration of the remaining porosity. In accordance with published data,10,11,26Figs. 7 and 8 illustrate that elongated pores develop perpendicular toward the film plane in the constrained-sintered layers, which oppose gas tightness. Again, the coshrinking anode support alters this undesired behavior to properties that are similar to free-standing layers. The difference between rigid and shrinking substrates becomes apparent after 1400°C/0.25 h. Bordia and Scherer40 suggested that compressed particle necks grow faster than necks sufffering no stresses or even tensile stresses. One could also argue that pore channels that collapse due to the sintering stress and particle rearrangement respond differently on superposing stresses depending on their orientation. If, for example, the layer is constrained and lateral tensile stresses superpose the compressive sintering stress, channels perpendicular to the film plane cannot collapse, whereas channels parallel to it still do, which sustains perpedicular pores on rigid substrates. If the substrate compresses the layer, vertical pore channels can be closed faster as horizontal channels because of their smaller lateral dimension; thus an isotropic or a slightly horizontal pore configuration is recovered or formed. This continuum mechanical explanation should be appropriate, if pores are not small in comparision with the grain size (which holds for the layers in Figs. 5(a) and (c)). The fluctuation in the pore orientation of the free-standing layers is possibly an effect of friction with the support (log Θ<−19) and afterwards probably a consequence of elongated and aligned particles in the green body,10,30 which seem to be eliminated after a very long sintering (1500°C/3 h).

image

Figure 7.  Pore orientation factor as a function of log Θ; positive values correspond to a alignment along the thickness of the layer.

Download figure to PowerPoint

image

Figure 8.  Pore elongation factor versus relative density. A value of one corresponds to a spherical shape.

Download figure to PowerPoint

From these microstructural investigations, the electrolyte can be considered as dense after sintering at 1400°C/3 h. It was shown25 that the remaining leakages originate from singular defects due to the defects in the substrate surface or defects introduced by the processing that can only be removed slowly by compressing the layer during cosintering. For future studies, the analysis of the top surfaces of the layers may yield additional information, especially if local defects are to be tracked. In order to develop a correlation between gas tightness and the investigated microstructural quantities, a large number of different leak-test samples would be required to assure statistical relevance.

(3) Camber Development

Besides the gas tightness, the flatness of the produced cells is of importance for later assembling and operation. First, the curvature during the course of sintering was measured on small test samples (46 mm × 23 mm × 1.1 mm, anode substrate/electrolyte) that were placed vertically and horizontally. The temperature program +3 K/min, 1400°C/5 h, −5 K/min was followed and terminated at different times with a cooling rate of −10 K/min. Additionally, one measurement with a prolonged dwelling time of 8 h was carried out. The data plotted in Fig. 9 represent the average of two similar samples; they deviated 10% at maximum. The bending behavior of the free-hanging substrates can easily be related to the free-sintering rates of Fig. 2. Initially, the densification of the electrolyte causes a concave bending (from electrolyte side) that is reduced shortly after the start of the dwelling period as the substrate's sintering exceeds the electrolyte. After a certain time (105 min, approximately −12.1% true lateral strain), the samples became flat again as reported for other cosintering structures.15,17 From this point, the bending increases steeply with further shrinkage as the electrolyte is already approximately 95% dense and its viscosity becomes large in comparison with the substrate,31 rendering further densification ineffective. Whereas Steinbrech et al.17 used substrates presintered at 1300°C/3 h with vacuum slip-cast electrolytes of a presumably lower green density and observed almost no free bending after final sintering at 1400°C/5 h (at the cost of electrolyte densification), the curvature using the cells of this work exceeds −6 m–1 at that point. The difference of the two values at 1400°C/5 h is due to the slightly different densities of the presintered substrates (47.1% and 45.6% theoretical density; the lower density corresponds to higher bending, which is the main trend).

image

Figure 9.  Curvature of horizontally placed and vertically hung up electrolyte/substrate two-layer compounds after different times of sintering (heating, +3 K/min; cooling, −10 K/min, for 1400°C/5 h; and 8 h, −5 K/min).

Download figure to PowerPoint

Gravity reduced the bending significantly; up to an isothermal sintering time of 3 h, the absolute value of the average curvature is <1.6 m–1, whereas free-hanging samples exhibit curvatures up to −5 m–1. In contrast to the hanging samples, the horizontally placed specimens exhibited local curvature deviations, and so their maximum curvature was added to the plot. The bending moments and shear forces due to gravity diminish near the samples' edges when they lie horizontally and bend concavely. Therefore, their maximum curvature equals the curvature of the hanging samples (until 1400°C/∼1 h). Once the samples bend convexly, the shear forces do not diminish at the samples' edges in contact with the support and, therefore, the whole sample is affected by gravity. For this reason, the maximum magnitude of the curvature of the horizontally placed samples is lower than for the hanging specimens, and the bending is more uniform. Considering these results, a dwelling time of approximately 3 h is recommended.

The influence of heating and cooling rates of the sintering program, +3 K/min, 1400°C/5 h, –5 K/min, is shown in Fig. 10. The cooling rate changes the curvature very little; a very slow rate yields additional shrinkage, and thus increases the bending. However, a very slow heating rate decreases the curvature significantly: approximately 25% for the free-hanging samples and 50% for the horizontally placed samples. Kanters et al.12 observed a slightly opposite behavior as their substrate preceded the layer in sintering. However, the differences in curvature they obtained were not considered significant. The reason for this decrease in curvature is still not exactly known. It may be that gravity also affects the bending of the vertically hung samples.

image

Figure 10.  Curvature versus temperature rates for heating up and cooling down (1400°C/5 h).

Download figure to PowerPoint

(4) Application to Half-Cells

The observed camber behavior was reproduced with larger half-cells (75 mm × 75 mm × 1.1 mm), including the anode layer. Two similar specimens were tested and averaged. Whereas the default sintering program (+3 K/min, 1400°C 5 h, −5 K/min) yielded a deflection of 1.5 mm, it was reduced by 20% to 1.2 mm using a heating rate of +1 K/min. A heating rate of +3 K/min and the suggested shorter dwelling time of 3 h yielded a deflection of 1.0 mm (−30%), which was further reduced to 0.60 mm (−60%) after both the parameters were adjusted at once in the sintering regime +5 K/min, 1200°C, +1 K/min, 1400°C/2.5 h, −5 K/min. The corresponding leak rates increased slightly; values of 3.4 × 10−6, 1.1 × 10−5, 1.6 × 10−5, and 1.7 × 10−5 (hPa·dm3)/(s·cm2) were observed in the order of the above-mentioned sintering programs. All these values pass the quality control threshold. It should be considered that the leak rate is a logarithmical quantity and typically subjected to large scattering. The deflections achieved still need to be reduced further, as acceptable values for this cell size lie in the range of a few 100 μm. Currently, the half-cells are still flattened by an automated application of a load in a specially designed furnace directly after sintering. We hope that further studies, utilizing thinner electrolytes, will render this step unnecessary, because low-cost manufacturing requires not to use any special flattening aids.

IV. Summary and Conclusions

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

The green density of electrolyte films was significantly enhanced after calcination and milling of delivered powder grades. Additionally, using the pretreated powder resulted in a reduction of the sintering mismatch between the anode support and the electrolyte layer, and lower leak rates were observed for the end-fired electrolyte. Whereas no gas-tight layers were obtained on rigid substrates, utilizing coshrinking anode supports yields the same or higher final electrolyte densities compared with free-standing layers. Moreover, disadvantageous elongated pores oriented perpendicular to the film plane disappeared during co-firing on shrinking substrates. For this reason, the use of coshrinking substrates is recommended for many applications where dense layers are of high importance, including the SOFC and e.g., gas-separation membranes, especially if the coating is otherwise difficult to densify. From its microstructure, the studied layer can be considered dense after sintering at 1400°C/3 h and further densification was accompanied by significant camber. The bending of the layered structure was reduced by 20% by utilizing a slower heating rate and by 60% on combining this with a shorter dwelling time.

Acknowledgments

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References

We would like to sincerely thank Dr. D. Sebold for acquiring the numerous SEM pictures and CeramTec AG, Marktredwitz, Germany, for the generous support with Rubalit(R) alumina substrates.

References

  1. Top of page
  2. Abstract
  3. I. Introduction
  4. II. Experimental Procedure
  5. III. Results and Discussion
  6. IV. Summary and Conclusions
  7. Acknowledgments
  8. References
  • 1
    T. Ishihara, N. M. Sammes, and O. Yamanoto, “Electrolytes”; pp. 83117 in High Temperature Solid Oxide Fuel Cells: Fundamentals, Design and Applications, Edited by S. C.Singhal, and K.Kendall. Elsevier, Oxford, 2003.
  • 2
    A. Mai, V. A. C. Haanappel, S. Uhlenbruck, F. Tietz, and D. Stöver, “Ferrite-Based Perovskites as Cathode Materials for Anode-Supported Solid Oxide Fuel Cells. Part I. Variation of Composition,” Solid State Ionics, 176 [15–16] 134150 (2005).
  • 3
    L. Blum, W. A. Meulenberg, H. Nabielek, and R. Steinberger-Wilckens, “Worldwide SOFC Technology Overview and Benchmark,” Int. J. Appl. Ceram. Technol., 2, 48292 (2005).
  • 4
    D. Stöver, H. P. Buchkremer, and J. P. P. Hujsmans, “MEA/Cell Preparation Methods: Europe/USA”; pp. 101531 in Handbook of Fuel Cells—Fundamentals, Technology and Applications, Vol. 4, Edited by W.Vielstich, H. A.Gasteiger, and A.Lamm. John Wiley & Sons, UK, 2003.
  • 5
    A. A. E. Hassan, N. H. Menzler, G. Blass, H. P. Buchkremer, and D. Stöver, “Development of an Optimized Anode Functional Layer for Solid Oxide Fuel Cell Applications,” Adv. Eng. Mater., 4 [3] 1259 (2002).
  • 6
    K. J. Yoon, W. Huang, G. Ye, S. Gopalan, U. B. Pal, and D. A. Seccombe Jr., “Electrochemical Performance of Solid Oxide Fuel Cells Manufactured by Single Step Co-Firing Process,” J. Electrochem. Soc., 154 [4] B38995 (2007).
  • 7
    K. J. Yoon, S. Gopalan, and U. B. Pal, “One-Step Cofiring of Anode-Supported Planar SOFCs”; Eighth European SOFC Forum, June 30–July 4, 2008, Lucerne, Switzerland, Paper No. A0525.
  • 8
    N. Jordan, W. Assenmacher, S. Uhlenbruck, V. A. C. Haanappel, H. P. Buchkremer, D. Stöver, and W. Mader, “Ce0.8Gd0.2O2−δ Protecting Layers Manufactured by Physical Vapor Deposition for IT-SOFC,” Solid State Ionics, 179 [21-26] 91923 (2008).
  • 9
    R. K. Bordia and R. Raj, “Sintering Behavior of Ceramic Films Constrained by a Rigid Substrate,” J. Am. Ceram. Soc., 68 [6] 28792 (1985).
  • 10
    O. Guillon, L. Weiler, and J. Rödel, “Anisotropic Microstructural Development During the Constrained Sintering of Dip-Coated Alumina Thin Films,” J. Am. Ceram. Soc., 90 [5] 1394400 (2007).
  • 11
    O. Guillon, S. Krauss, and J. Rödel, “Influence of Thickness on the Constrained Sintering of Alumina Films,” J. Eur. Ceram. Soc., 27, 26237 (2007).
  • 12
    J. Kanters, U. Eisele, and J. Rödel, “Cosintering Simulation and Experimentation: Case Study of Nanocrystalline Zirconia,” J. Am. Ceram. Soc., 84 [12] 275763 (2001).
  • 13
    T. Yamaguchi, T. Suzuki, S. Shimizu, Y. Fujishiro, and M. Awano, “Examination of Wet Coating and Co-Sintering Technologies for Micro-SOFCs Fabrication,” J. Membr. Sci., 300, 4550 (2007).
  • 14
    P. Z. Cai, D. J. Green, and G. L. Messing, “Constrained Densification of Alumina/Zirconia Hybrid Laminates, I: Experimental Observations of Processing Defects,” J. Am. Ceram. Soc., 80 [8] 192939 (1997).
  • 15
    G. Q. Ku, R. C. Sutterlin, and T. K. Gupta, “Effect of Mismatched Sintering Kinetics on Camber in a Low-Temperature Cofired Ceramic Package,” J. Am. Ceram. Soc., 76 [8] 190714 (1993).
  • 16
    D. J. Green, O. Guillon, and J. Rödel, “Constrained Sintering: A Delicate Balance of Scales,” J. Eur. Ceram. Soc., 28, 145166 (2008).
  • 17
    R. W. Steinbrech, A. Caron, G. Blass, and F. Dias, “Influence of Sintering Characteristics on Component Curvature of Electrolyte-Coated Anode Substrates”; pp. 72736 in Proceedings of the 5th International Symposium on Solid Oxide Fuel Cells, June 2–5, 1997 in Aachen, Electrochem. Soc. Proc. PV97-18, Pennington, Edited by U.Stimming, S. C.Singhal, H.Tagawa, and W.Lehnert. The Electrochemical Society, Pennington, NJ, 1997.
  • 18
    J.-H. Jean and C.-R. Chang, “Effect of Densification Mismatch on Camber Development during Cofiring of Nickel-Based Multilayer Ceramic Capacitors,” J. Am. Ceram. Soc., 80 [9] 24016 (1997).
  • 19
    P. Z. Cai, D. J. Green, and G. L. Messing, “Constrained Densification of Alumina/Zirconia Hybrid Laminates, II: Viscoelastic Stress Computation,” J. Am. Ceram. Soc., 80 [8] 19408 (1997).
  • 20
    S.-H. Lee, G. L. Messing, and M. Awano, “Sintering Arches for Cosintering Camber-Free SOFC Multilayers,” J. Am. Ceram. Soc., 91 [2] 4217 (2008).
  • 21
    R. Förthmann, G. Blass, and H. P. Buchkremer, “Deposition of Thin Anode and Electrolyte Layers for Medium Operation Temperatures of Solid Oxide Fuel Cells”; pp. 10038 in Proceedings of the 5th International Symposium on Solid Oxide Fuel Cells, 2–5, June, 1997 in Aachen, Electrochem. Soc. Proc. PV97-18, Pennington, Edited by U.Stimming, S. C.Singhal, H.Tagawa, and W.Lehnert. The Electrochemical Society, Pennington, NJ, 1997.
  • 22
    A. A. E. Hassan, N. H. Menzler, G. Blass, M. E. Ali, H. P. Buchkremer, and D. Stöver, “Influence of Alumina Dopant on the Properties of Yttria-Stabilized Zirconia for SOFC Applications,” J. Mater. Sci., 37 [16] 346775 (2002).
  • 23
    R. Mücke, N. H. Menzler, F. Kemnitzer, and K. Blöchl, “Application of Laser Topography in SOFC Development”; Eighth European SOFC Forum, June 30–July 4, 2008, Lucerne, Switzerland, Paper No. A0301.
  • 24
    D. Simwonis, H. Thülen, F. J. Dias, A. Naoumidis, and D. Stöver, “Properties of Ni/YSZ Porous Cermets for SOFC Anode Substrates Prepared by Tape Casting and Coat-Mix® Process,” J. Mater. Process. Technol., 92–93, 10711 (1999).
  • 25
    R. Mücke, N. H. Menzler, H. P. Buchkremer, and D. Stöver, “Sintering of Thin Zirconia SOFC Electrolyte Films”; pp. 217585 in Proceedings of the 10th International Symposium on Solid Oxide Fuel Cells, ECS Transactions, 7,1, Edited by K.Eguchi, S. C.Singhal, H.Yokokawa, and J.Mizusaki. The Electrochemical Society, Pennington, NJ, 2007.
  • 26
    J.-B. Ollagnier, O. Guillon, and J. Rödel, “Effect of Anisotropic Mictrostructure on the Viscous Properties of an LTCC Material,” J. Am. Ceram. Soc., 90 [12] 384651 (2007).
  • 27
    P. Perona and J. Malik, “Scale-Space and Edge Detection Using Anisotropic Diffusion,” IEEE Trans. Pattern Anal. Mach. Intell., 12 [7] 62939 (1990).
  • 28
    H. Su and L. Johnson, “Master Sintering Curve: A Practical Approach to Sintering,” J. Am. Ceram. Soc., 79 [12] 32117 (1996).
  • 29
    S. J. Park, S. H. Chung, D. Blaine, P. Suri, and R. M. German, “Master Sintering Curve Construction Software and its Applications”; pp. 1324 in Advances in Powder Metallurgy and Particulate Materials 2004. Proceedings of the 2004 International Conference on Powder Metallurgy & Particulate Materials, 13—17, June, 2004, Chicago, Edited by R. A.Chernenkoff, and W. B.James. Metal Powder Industries Fed., Princeton, 2004.
  • 30
    R. Mücke, “Sinterung von Zirkoniumdioxid-Elektrolyten im Mehrlagenverbund der oxidkeramischen Brennstoffzelle (SOFC). (Sintering of Zirconia Electrolytes in the Mutli-Layer Compound of Solid Oxide Fuel Cells)”; Ph.D. Thesis, University of Bochum, 2007, Section 4.2.6.
  • 31
    R. Mücke, “Sinterung von Zirkoniumdioxid-Elektrolyten im Mehrlagenverbund der oxidkeramischen Brennstoffzelle (SOFC). (Sintering of Zirconia Electrolytes in the Mutli-Layer Compound of Solid Oxide Fuel Cells)”; Ph.D. Thesis, University of Bochum, 2007, Section 4.5.3
  • 32
    R. K. Bordia and A. Jagota, “Crack Growth and Damage in Constrained Sintering Flims,” J. Am. Ceram. Soc., 76 [10] 247485 (1993).
  • 33
    R. Zuo and J. Rödel, “Temperature Dependence of Constitutive Behaviour for Solid-state Sintering of Alumina,” Acta Mater., 52, 305967 (2004).
  • 34
    H. Moon, S. D. Kim, S. H. Hyun, and H. S. Kim, “Development of IT-SOFC Unit Cells with Anode-Supported Thin Electrolytes Via Tape Casting and Co-Firing,” Int. J. Hydrogen Energy, 33, 175868 (2008).
  • 35
    T. J. Garino and H. K. Bowen, “Kinetics of Constrained-Film Sintering,” J. Am. Ceram. Soc., 73 [2] 2517 (1990).
  • 36
    J. N. Calata, G.-Q. Lu, and T.-J. Chuang, “Constrained Sintering of Glass, Glass–Ceramic and Ceramic Coatings on Metal Substrates,” Surf. Interface Anal., 31, 67381 (2001).
  • 37
    R. K. Bordia, R. Zuo, O. Guillon, S. M. Salamone, and J. Rödel, “Anistropic Constitutive Laws for Sintering Bodies,” Acta Mater., 54, 1118 (2006).
  • 38
    J. Zhao and M. P. Harmer, “Effect of Pore Distribution on Microstructure Development: II, First- and Second-Generation Pores,” J. Am. Ceram. Soc., 71 [7] 5309 (1988).
  • 39
    B. Sudhir and A. H. Chokshi, “Compression Creep Characteristics of 8-Mol%-Yttria-Stabilized Cubic Zirconia,” J. Am. Ceram. Soc., 84 [11] 262532 (2001).
  • 40
    R. K. Bordia and G. W. Scherer, “On Constrained Sintering–III. Rigid Inclusions,” Acta Metall., 36 [9] 24116 (1988).