R. E. Loehman, A. Ayala, M. Brochu, B. Gauntt, Sandia National Laboratories, Albuquerque, NM
Seals for solid oxide fuel cells (SOFC) face demanding performance requirements. They must be gas tight at temperatures to 1000°C for operating lifetimes of up to 40,000 hr. Seals are subject to stresses from thermal gradients and expansion mismatch of different stack materials, as well as adverse reactions with other fuel cell components. Mechanical design of seals is critical since performance is specific to component geometry. Materials and design choices usually require compromises among competing requirements.
This approach to SOFC seal design uses ceramic and metal filled glass composites as sealants, which provides significant design flexibility. Glass matrix composite seals provide control of properties such as glass transition temperature, viscosity, and thermal expansion coefficient by varying compositions, amounts, and microstructures of the different phases. Thermal and mechanical strains are reduced by using glass compositions with glass transition temperatures (Tgs) below the SOFC operating temperature. Varying the glass and filler compositions can minimize the amount of glass in the seal, which reduces reactivity with fuel cell materials. The choices are guided by thermochemical and composite microstructural models that allow us to target specific seal properties for a given design.
Recent work has analyzed reactivity of sealants in contact with anode and interconnect materials. The best glass composite sealants exhibit little or no reactivity with SOFC anode materials at 750-800°C or with stainless steel interconnect materials at 750°C. Preoxidizing the interconnect increases glass wetting and adhesion rates, which enlarges the processing space for making glass composite seals.
Sandia National Laboratories is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the U.S. Dept. of Energy under Contract DE-AC04-AL85000.
Summary: Developing reliable methods for sealing solid oxide fuel cell stacks presents the most challenging set of performance criteria in the entire field of ceramic joining. For SOFC applications the requirements on the sealing method include:
1. adhesion of the sealing material to fuel cell components from RT to as high as1000°C
2. need to provide a leak tight seal at the SOFC operating temperature
3. ability to maintain a seal while accommodating strains from SOFC components with different coefficients of thermal expansion (CTEs)
4. lack of adverse reaction between the sealing material(s) and the fuel cell components
5. chemical and physical stability of the sealant at temperatures up to 1000°C in oxidizing and reducing atmospheres
6. thermal shock tolerance
7. electrically insulating for some SOFC designs
All of the above properties must be maintained for SOFC operating lifetimes of up to 40,000 hours. The list is written in approximate order of decreasing stringency. That is, no matter what the SOFC design, the seal must be adherent and leak tight. On the other hand, some stack designs may require joining only similar materials and thus, a matched CTE seal may be sufficient. Note also that the requirements may be contradictory. For example, being leak tight and adherent at high temperatures suggests a refractory, stiff sealant, which may work against the requirement for thermal strain accommodation. Such situations are common and seal developers know that seal design is specific to a particular component geometry and usually requires compromises among competing requirements.
We have shown that our SOFC sealing technique can be tailored to fit a wide range of SOFC specifications. The essence of the method is to engineer ceramic-filled glass composites, metal filled glass composites, and/or ceramic-filled metal composites that can meet SOFC requirements. This approach combines extensive capabilities in composites and ceramic joining that have been developed by Sandia National Laboratory staff over the past 15 - 20 years.
The composite seals require a glass matrix that is slightly fluid at the SOFC operating temperature to reduce thermal strains. In technical terms this requires a glass transition temperature that is below the cell operating temperature. Our compositional modifications varied the glass Tgs and expansion coefficients and thus give us a wider choice of matrix materials for our composite seals. The flow properties are reflected as differences in the rate of spreading of a molten glass drop on the substrate of interest. For example, a more fluid glass at a given temperature will spread faster and assume a lower contact angle than one that is less fluid. In a recent series of tests at 750°C, ten glass compositions exhibited a range in contact angles after 5 min on 410 stainless steel (a typical SOFC alloy) from 80° (low flow, poorer wetting) to 20° (faster flow, good wetting). No one glass is necessarily better than another for all use conditions. Rather, the results demonstrate that our glasses span a wide range of physical properties, which allows us a lot of options in engineering optimized SOFC seals.
Constructing SOFC stacks may require seals to Ni - YSZ anodes. Thus, we tested different glasses for reactivity using a sessile drop test geometry in which small glass specimens were heated on anode material for different times at fixed temperature. During heating we monitored the change in glass contact angle with time. After cooling we sectioned the interface and analyzed it for evidence of reaction using electron microscopy and energy dispersive spectroscopy (EDS). The results typically show well-bonded interfaces with no evidence for Ni dissolution (the expected reaction mechanism) in the glass within the resolution of the measurement. In other experiments we mixed up to 30 vol% Ni powder into different glasses and heated the specimens for up to 50 hr at 800°C as an extreme test of glass-Ni reactivity. Our most stable glasses showed no change in coefficient of thermal expansion after heating and no microscopic evidence for Ni dissolution.
Reactivity of glasses with interconnect materials was evaluated using the sessile drop method described above. Microscopic examination of interfaces showed no evidence of Cr dissolution (the most reactive constituent of stainless steel) or reaction products with the glass within the resolution of the technique. As an extreme test we heated mixtures of glass and pure Cr powder at 750°C for increasing times. Similar to the case with stainless steels, no reaction products were observed at the interface. Since increasing temperature accelerates chemical reactions, we heated glass-Cr mixtures at 850°C. The results showed that under those more extreme conditions Cr reacts and dissolves in the glass. More work will be required to extrapolate the results for pure Cr and 850°C testing to realistic SOFC operating conditions and materials.
Preoxidizing 304 stainless steel housings used for electrical connectors has been shown to promote wetting and bonding with glass insulators. We extended those results by demonstrating that preoxidizing E-brite and 410 stainless steels is similarly beneficial for SOFC sealing. Heating the alloy at 1000°C for 5 - 30 min in Ar/1000ppm O2 produces a surface oxide layer 1-3 micrometers thick that is primarily Cr2O3 with a little SiO2. Glass melted on the preoxidized alloy wets and adheres to the alloy in as little as 2 min, whereas reaching a steady-state contact angle on the bare alloy requires heating for 5 min or more. Minimizing time at the sealing temperature reduces the extent of undesirable interface reactions so preoxidation of the interconnect may offer significant benefits in manufacturing SOFC stacks.
Conclusions
o Glass seal compositions provide excellent control over flow and thermal expansion properties.
o Seal compositions are stable in contact with anodes at 750-800°C and with stainless steel interconnect materials at 750°C.
o Preoxidizing interconnect materials offers advantages in SOFC seal processing
adhered better to preoxidized 410 stainless steel than to the as-received form
