Thermo-Mechanical Properties and Benefits of Borosilicate Glass for SOEC Sealants

Abstract

Boro-silicate glass materials have ideal properties for sealing oxidizing and reducing environments and electrically insulating solid oxide electrolysis cells (SOEC), while maintaining a glassy phase at high temperatures (750-900°C) to minimize chemo-mechanical degradation and gas permeability 1,2. Glass seals are typically applied as a slurry; hence, they are also expected to have non-wetting properties to prevent glass infiltration into the porous electrode structures during manufacturing 3. During the first startup, the initial temperature ramp serves to evaporate any binder in the slurry and to go through the glass transition phase, which cures the glass and increases hardness to improve resistance to creep. Typical glass materials operate in visco-plastic regimes (i.e., beyond the material’s glass transition temperature) and must be chemically stable while operating in steam-rich and oxygen-rich environments. To fully understand the performance of borosilicate glass seals, optimize the manufacturing, assembly, and maintenance of SOEC, we must be able to measure and analyze the glass material properties (i.e., glass transition temperature, coefficient of thermal expansion, modulus, creep, hardness) at the same operating conditions relevant for SOEC operation. In this work, we characterize the thermo-mechanical properties of borosilicate glass powder and sintered samples at various temperatures (ranging from ambient to 850°C), and at different aging conditions (i.e., atmosphere composition, temperature, and duration). The atmosphere composition is changed between air, steam-N2 mixtures, and H2-N2 mixtures to evaluate the chemical stability of borosilicate glass relating to the possible depletion of boron and silicon via the formation of boron hydroxide and silicon vapor when exposed to steam- and hydrogen-rich environments. The aging temperature and duration is varied to evaluate the crystallization kinetics, particle size growth, and its effect on mechanical properties. We employ a combination of Simultaneous Thermal Analysis - STA (i.e., combined Thermogravimetric Analysis - TGA, Differential Scanning Calorimetry - DSC, Fourier Transform Infrared - FT-IR spectroscopy, and Quadrupole Mass Spectroscopy - QMS), and nanoindentation as characterization techniques, while surface roughness is estimated with a 3D confocal optical microscope.Surface roughness varies between 0.83-1.38 𝜇m ± 1.1-1.76 𝜇m (Sa ± Sq). Both powder and solid samples are analyzed via STA with sequential temperature ramps (constant heating rate at 10 K/min in Ar atmosphere) up to 1500°C to estimate the effect of crystallization and particle size on the glass transition onset. For a sample aged at 850°C for 340 h, the first ramp shows typical DSC exothermic peaks (mW/mg) at around 800°C, with a midpoint glass transition occurring at 900°C. However, on the second ramps, peaks lower to around 500°C and the glass transition is estimated to be around 700°C. Subsequent tests consistently mirrored the trend of the second temperature ramp, showing minimal deviation and influence of crystallization and particle growth. Nanoindentation is performed between room temperature and 600°C to characterize both sintered glass samples “as-received” and aged at temperatures between 750-950°C for 340 h. Comparing a sample aged at 750°C with an “as-received” sample, the elastic moduli equally decreased from 100 GPa at room temperature to 40 GPa at 600°C, with comparable hardness values ranging from 10 GPa at room temperature to 0.5-0.9 GPa at 600°C, respectively. A.G. Sabato, G. Cempura, D. Montinaro, A. Chrysanthou, M. Salvo, E. Bernardo, M. Secco, F. Smeacetto, Journal of Power Sources, 328, 262-270 (2016)Dilshat U. Tulyaganov, Allu Amarnath Reddy, Vladislav V. Kharton, José M.F. Ferreira, Journal of Power Sources, 242, 486-502 (2013)S.-B. Sohn, S.-Y. Choi, G.-H. Kim, H.-S. Song, G.-D. Kim, Stable sealing glass for planar solid oxide fuel cell, J. Non. Cryst. Solids. 297 (2002) 103–112. doi:10.1016/S0022-3093(01)01042-0 Figure 1