Chapter 66 Compatibility of Electrolyte and Electrode Materials with LTCC

Transcription

1 Chapter 66 Compatibility of Electrolyte and Electrode Materials with LTCC 6. Compatibility of Electrolyte and Electrode Materials with LTCC The earlier Chapters of this Thesis have given detailed description about synthesis of nano-crystalline electrolyte and electrode materials for Solid Oxide Fuel Cells (SOFC). We have seen optimization of synthesis parameters for oxygen ion conductor (Gadolinium doped Ceria), with respect to its structural, morphological and electrical characterization in Chapter 3. Synthesis and characterization of two proton ion conductors viz. Yttrium and Ytterbium doped Barium Zirconate (BZYYbO) and Yttrium doped Barium Cerate is discussed in Chapter 4. The experimental results for the preparation and characterization of anode and cathode materials, viz. Copper Zinc oxide (CuZnO) and Samarium and Strontium doped Cobaltite (SSC) is presented in Chapter 5. The present Chapter is an important part of this Thesis as it presents a detailed study on compatibility of the above mentioned materials with standard LTCC materials, leading to device fabrication. 6.1 Need for compatibility Electrolyte and electrodes are important components of solid oxide fuel cells. All these components in SOFC are made up of ceramic (oxide) materials. As we have seen before, these materials require high sintering temperature, which is usually above 1200 C. Their physical properties, such as, shrinkage, temperature Co-efficient of Expansion (TCE), thermal conductivities are dependent on constituent materials. Especially, being a ceramic material they have high TCE values and low thermal conductivity. On the hand, LTCC is a glass ceramic technology Chapter 6 Page 202

2 which has low sintering temperature, low TCE valuable 6.1 presents a comparison of important properties of SOFC and LTCC materials. Physical Properties SOFC materials LTCC materials (DuPont 951) Materials type Ceramics Glass-ceramics Sintering temperature ( o C) > Operating temperature ( o C) Temperature Co-efficient of expansion (ppm/ o >12 C) 5.8 Thermal conductivity (W/mK) Density (gms.cm -3 ) >6 3.1 Shrinkage (%) C 875 C Table 6.1: Overview of present status of physical properties of materials used in SOFC and LTCC It is clearly seen from Table 6.1 that the physical properties of materials used in SOFC are quite different from that for LTCC. Their chemical properties are also dissimilar. Electrodes of SOFC are chemically active and are involved in catalytic reactions during SOFC operation. On the contrary, LTCC materials are highly inactive in sintered state. The electrical properties of SOFC and LTCC are also very different from each other. The SOFC electrodes are mixed conductors, the electrolytes are charge conductor at high temperature, while the LTCC materials are essentially used as dielectric material having low dielectric constant and low dielectric loss. It is clearly seen that most physical, chemical and electrical properties of SOFC and LTCC materials are in conflict with each other. Nevertheless, we have undertaken the task of integrating these materials in view of enormous advantage offered by the integrated and low temperature SOFC. This Chapter presents details of our efforts to make these two different materials technologies compatible with each other. In order to overcome these dissimilarities in various properties of SOFC and LTCC materials, there is a need to first tune up the physical properties of SOFC materials by formulating new combinations of materials or by adding some additional materials as sintering aid to reduce sintering temperature, matching shrinkage and other physical Chapter 6 Page 203

3 properties. Tuning of the rest properties would follow. Note that the sintering temperature of an electrolyte is very high (1350 C) compared to anode and cathode materials ( ), Further, the density of electrolyte (must be >96%) is very important factor in ionic conductivity properties. Therefore, study of electrolyte material becomes more important than the electrode materials. Clearly there is a need of an appropriate sintering aid that can lower the sintering temperature of SOFC ceramic electrolyte to the range of LTCC firing temperatures, while simultaneously achieving the required sintering density. A carefully selected sintering aid, either in crystalline or in the form of glass, may help in lowering sintering temperature of SOFC electrolyte without compromising its ionic conductivity properties. The following Section describes the detailed study about effect of chosen sintering aids on physical, chemical and electrical properties of SOFC electrolytes GDC and BCYO. 6.2 Effect of sintering aids on Gadolinium doped Ceria To study effect of sintering aids on Gadolinium doped ceria, first Bismuth oxide (Bi 2O 3) was selected as sintering aid in GDC. This is because Bi2O3 is a well known high temperature oxygen ion conductor in its δ-fcc phase below 825 C ( C) and its crystal structure matches well with GDC. It has high polarizability in cataion network due to highly disordered anion structure. About 25% intrinsic vacancies are present in Bi 2O 3, Bi 2+ ions have good ability to accommodate asymmetry present in lattice structure. Bi 2O 3 has low melting temperature close to 850 C, and enormous densification properties due liquid phase sintering capabilities (1). Due to these favourable properties, Bi 2O 3 is good candidate as sintering aid for GDC, and was chosen for experiments (2) (3). The experimental procedure and results are described in following Sub-section Bi 2O 3 as sintering aid The experiments were carried out by adding different weight proportion of Bi2O3 (99.8% nanocrystalline Sigma Aldrich) in GDC nano Chapter 6 Page 204

4 powder synthesized using the optimized preparation condition given in Chapter 3. These weight proportions are 1, 2, 5, 8 and 10% respectively. These powders were first wet ground in agate pestle mortar for 1 hr. A 25% PVA solution prepared in DI water was then added as binder and was mixed thoroughly. Pellets of this mixture were pressed by applying 280 MPa pressure for 10 min. These pellets were fired at 850 C for 2hrs. This temperature was chosen in view of the melting temperature of Bi 2O 3. Density of fired pellets was measured by weight-volume method. Platinum paste (Heraeus CL , USA) was applied on both surface and fired again at 800 C for 2hrs. Electrical characterization was done impedance analysis using AUTOLAB PGSTAT 100 Potenstiostat/Galvanostat in the temperature range of C and in frequency perturbation of 1Hz-1MHz. Bi 2O 3-GDC pellets fired at 850 C have shown shrinkage in the range of 3-5%, which increases with Bi 2O 3 content. The density of these pellets was above 80% and it was highest (~95%) for 2% Bi 2O 3. Interestingly, the ionic conductivity of these pellets was found to be higher than that recorded for GDC at 600 C. Table 6.2 presents measured shrinkage, density and ionic conductivity of Bi 2O 3 added GDC pellets. Weight % of Shrinkage when Density (%) Ionic conductivity at Bi 2O 3 fired at 850 C 600 C ( 10-3 S.cm -1 ) Table 6.2: Effect of Bi 2O 3 added to GDC in different proportions along with the shrinkage, density and ionic conductivity, for pellets fired at 850 C Table 6.2 indicates that with increasing Bi 2O 3 content, the properties of GDC pass through a peak. The shrinkage, density and ionic conductivity properties were found to be highest at 2% Bi 2O 3 content and reduced as the Bi2O3 content increased further. Clearly, higher weight percent of Bi 2O 3 found inappropriate in all respect. The Chapter 6 Page 205

5 highest conductivity at 2% Bi 2O 3 was recorded as S.cm -1 at 600 C. In contrast, ionic conductivity of pure GDC sintered at 1350 C is reported to be 0.014S.cm -1. Decrease in shrinkage, density and ionic conductivity in case of 5-10% doped Bi 2O 3 may be attributed to longer sintering dwell times and evaporation of Bi 2O 3. Hence, it can be concluded that Bi2O3 may be useful with GDC as sintering aid and 2 wt%. Similar results 0.022S.cm -1 has been reported in literature; however, sintering temperature was 1400 C (2). Present work improved these results by achieving density and conductivity better than that reported in literature (2). Even though, Bi 2O 3 improved the results in terms of lower sintering temperature, high density and improved ionic conductivity at lower (600 C) operating temperature it is not a perfect choice as sintering aid for LTCC applications, as maximum shrinkage obtained was just 3.1%, which is far lower than 13-15% reported for LTCC. Even an increase in sintering temperature to 1000 C for GDC+2%Bi 2O 3 pellets showed shrinkage of 5.4% which was low compared to LTCC. This shrinkage mismatch would induce stresses during cooling part of the firing cycle, causing warpage and even cracks in the LTCC structures. In order to achieve LTCC equivalent shrinkage values, the sintering aid must be in the form of glass as glass addition is expected to increase shrinkage (4), (5). Low temperature melting recrystallized glasses offer higher shrinkage values due to their higher agglomeration mechanism (6). Hence, some low temperature melting glasses were prepared by conventional solid state reaction and water quenching method. Effect of different glasses on the ionic conductivity of GDC pellets is discussed in the following Sub-section Glasses as sintering aid in case of GDC Low temperature melting borate and phosphate glasses based on Bismuth oxide are prepared to use as sintering aid in GDC electrolytes. Boron oxide (B 2O 3) and Phosphorous Pentaoxide (P 2O 5) are low melting temperature oxides having melting points 450 C and 360 C Chapter 6 Page 206

6 respectively. Also it is reported that P 2O 5 glasses are very good protonic conductors (7), (8). Bi 2O 3 and V 2O 5 also have melting points around 817 C and 690 C respectively. These two oxides together form oxygen ion conducting phases commonly known as BIMEVOX. BiVO 4, Bi 4V 2O 11, Bi 3.5V 1.2O 8.25, Bi 23V 2O 44.5 and Bi 8V 2O 17 are some of the oxygen ion conducting phases of Bi2O3 and V2O5 (9). These phases are oxygen ion conductors at high temperature ( C). It is also reported that Bi 2O3-P 2O 5 glasses are exhibit phonon-assisted hopping conduction at high temperatures ( C) (10). Hence, by considering ionic conductivity and low melting temperature of these oxides, B2O3, P2O5, Bi 2O 3 and V 2O 5 were selected for synthesis of these glasses. The glasses are synthesized by solid state reaction method. Four different compositions were tried with GDC as reported in the following Synthesis of glasses Row materials in the form of oxides were selected for synthesis of glasses. Bismuth oxide (Bi 2O 3; 99.9% Sigma Aldrich), Vanadium Pentaoxide (V 2O 5; >99.6% Sigma Aldrich), Boron oxide (B 2O 3; 99.9% Sigma Aldrich) and Ammonium Phosphate (NH 4H 2PO 4; >98% Sigma Aldrich) were mixed together in different molar proportions as given in Table 6.3. Glass Name Mole % Mole % of Mole % of Mole % of (Glass code) Bi 2O 3 V 2O 5 B 2O 3 NH 4H 2PO 4 BBVP BVP BVO Table 6.3: Composition of different glasses synthesized by conventional mixing, melting and water quenching process. Acetone (HPLC grade; Merck make) was added to mixture and milled on three roll mill for 96 hrs using Zirconia balls in polythene container. This oxide mixture was then dried and loaded in a 99% Alumina crucible. This mixture was calcined at 900 C for 2hrs, at ramp rate of 5 C/min and molten mass was poured into DI water. A yellow colored glass frit was obtained in all cases. This glass frit was ground again in an agate pestle-mortar to fine powder. Pellets were pressed of a Chapter 6 Page 207

7 mixture of GDC added with 10 wt% of each glass. These pellets were fired at 1000 C for 40 min. The shrinkage was calculated by comparing dimensions before and after firing and its density was measured by weight-volume method. Platinum paste (Heraeus CL , USA) was applied on both surface and pellets were fired again at 850 C. These pellets were used to measure ionic conductivity using impedance analysis (AUTOLAB PGSTAT 100 otentiostat/galvanostat) by employing frequency perturbation in 1Hz to 1 MHz frequency range and at different temperatures between 400 C-600 C. The selection of best glass composition was done on the basis of sintering properties and ionic conductivity measurements Impedance measurement of Glass doped GDC pellets The fired pellet showed a shrinkage of the order of 7-10% when fired at 1000 C. The density of the pellets was found to be in the range of 60-70%. The impedance spectra for these pellets are presented in the form of Nyquist plots in Figure 6.1. Figure 6.1: Nyquist plot measured at 600 C for different glasses doped at 10weight% in GDC The Nyquist plots, plotted for 10 wt% doped glass-gdc pellets shows increase in impedance of the pellet compared to un-doped GDC. The impedance is seen to be increase in order of un-doped GDC < BVP < BBVP < BVO. It is seen that bulk conductivity is affected by glass doping, this may be due to the distortion in lattice planes of host Chapter 6 Page 208

8 material due to crystallization of glass. The addition of glass at 10wt% is not only located at grain boundaries but also can diffuses in grains of GDC and deteriorates the lattice structure of GDC. Grain boundary impedance is also seen increased indicating poor grain boundary structure due to presence of glasses. The grain boundary impedance is found to be relatively lower for the pellets with BVP glass. Comparing with the impedance of BVO glass it can be concluded that P 2O 5 in the glass structure may be aiding ionic conduction across grain boundaries (10). Higher impedance in case of BBVP glass is due to presence of higher wt% of B2O3 in glass, addition of B2O3 above 0.4wt% and sintering temperature above 850 C lower ionic conductivity is reported in literature (11). The effect of glass doping on GDC in terms of shrinkage, density and ionic conductivity is tabulated in Table 6.4. Glass added Shrinkage at Total ionic conductivity at Density (%) in GDC 1000 C (%) 600 C (x10-3 S.cm -1 ) BBVP BVP BVO Table 6.4: Shrinkage, density and total ionic conductivity measured for 10 wt% glass doped GDC pellets fired at 1000 C From the results presented in Table 6.4 one can conclude that using glass as sintering aid is much helpful with respect to the shrinkage, especially in comparison with Bi2O3. However, the ionic conductivity deteriorates. Clearly, the presently used glasses would not be useful, and a different glass composition may be required. One of the noticeable outcomes of the above experiments conducted with glasses was the observation of lower grain boundary impedance for BVP glass, although this glass composition showed lowest shrinkage, density and ionic conductivity. It was, therefore, decided to use this glass with the addition of some alkali oxide, since alkali oxides are known ionic conductors. For example, Lithium and Chapter 6 Page 209

9 Sodium are well know alkali oxide used in batteries (12). Potassium oxide (K 2O) is an alkali oxide having anti-fluorite crystal structure, large atomic radii of Potassium comparative with Phosphorous and Vanadium. Hence it was decided to add K 2O in Bi 2O 3-V 2O 5-P 2O 5 glass. A glass composition used to study was x(bi 2O 3-K 2O)-y(V 2O 5-P 2O 5), wherein both x and y vary from 0 to x(bi 2O 3-K 2O)-y(V 2O 5-P 2O 5) (abbreviated as BKVP) was synthesized using conventional solid state reaction, melting and water quenching method. Initially 10 wt% glass was added in GDC and pellets were pressed and fired at 1000 C for 40 min. The platinum paste was applied on both surfaces and fired again at 850 C for 40min. The ionic was conductivity measured using impedance analysis as stated above. The sintered pellet showed shrinkage of 12.78% and density measured was found to be 6.49gms.cm -1, which is 92% of the theoretical density. The increase in shrinkage and density was attributed to lowered glass transition temperature of BKVP glass compared to BVP glass. The glass transition temperature of BKVP was found to be 650 C compared to BVP glass 730 C. This improvement also brought about improvement in the electrical properties. The ionic conductivity results indicated improvement in the conductivity due to addition of K 2O in glass. Figure 6.2 presents comparison of Nyquist plots measured for all the glasses added to GDC at 10 wt%, the 2 wt% Bi 2O 3 in GDC and undoped GDC measured at 600 C. 1 Exact glass composition is under patenting process. Chapter 6 Page 210

10 Figure 6.2: Comparison of Nyquist plots measured for all glasses with 10wt% addition to GDC, 2% Bi 2O 3 in GDC and un-doped GDC Nyquist plot presented in Figure 6.2 clearly indicates that, addition of the K 2O in BVP glass improve its bulk conductivity, although this is less than the 2 wt% Bi2O3 added GDC and un-doped GDC. The ionic conductivity of the S.cm -1 was observed at 600 C. This conductivity is highest amongst the glass-ceramic composite electrolytes. The earlier reported highest glass-ceramic conductivity is S.cm -1. This increase in bulk conductivity is due to addition of alkali oxide (13). Addition of alkali oxides in the glass was reported to be favorable in producing homogeneous melts at low temperature. Also, addition of alkali oxides in glasses reduces dielectric losses of samples, which found advantageous for ionic conductivity in glass ceramic composite electrolytes (13). Due to the improvement observed with 10 wt% addition of BKVP glass, the BKVP doping percentage in GDC was increased up to 40 wt% on steps of 5 wt% in the next set of experiments. The ionic conductivity found to have increased enormously with increasing BKVP glass content in GDC. Figure 6.3 presents effect of BKVP glass addition on ionic conductivity of GDC based glass-ceramic electrolytes through Nyquist plots and dependence of shrinkage and ionic conductivity of electrolyte. Chapter 6 Page 211

11 Figure 6.3: Nyquist plots at 600 C for different wt% of BKVP glass added to GDC The Nyquist plots presented in Figure 6.3 indicate that the grain and grain boundary impedance decreases with increasing glass content in GDC. It is seen that grain and grain boundary impedance curves are distinguishable up to 20 wt% addition of glass. Further increase in glass doping caused very low grain boundary impedance. Low frequency curves started vanishing from Nyquist plots with every 5 wt% increase in glass content (25 wt% onwards) and the high frequency loop was relatively dominant. This change is attributed to reduction in traps at the grain boundaries due to glass diffusion in the grains and probably into the lattice. Nyquist plots at the high frequency show presence of Warburg in intra-grain regions, implying increase in ionic conductivity due to increase in anion vacancies in lattice. Glass Fitting parameter values wt % R1 R2 R3/L W CPE1 (Ω) (Ω) (Ω)/H (Ω) (F) n CPE2 (F) n Error (%) Table 6.5: Values of fitting parameter for each Nyquist plot presented in Figure 6.3, for samples with wt% of glass in GDC Chapter 6 Page 212

12 It is seen from the Table that, grain and grain boundary contribution to ionic conductivity increases with increase in glass weight addition. The fitting values of R2 lowered with increase in glass addition. The R3 resistance is corresponding to grain boundary resistance replaced by inductance. This implies very high ionic diffusion across grain boundaries. Figure 6.4 presents effect of glass wt% on shrinkage and ionic conductivity. These pellets were fired at 1000 C and ionic conductivity was measure at 600 C. It is observed that, increase in glass wt% decreases shrinkage due to higher percentage of glass in molten state cannot agglomerate GDC particles. Further increase in glass above 50% shows negative shrinkage for the pellets (e.g. 60 wt% glass in GDC showed shrinkage of -0.2%). Figure 6.4: Effect of glass wt% on shrinkage and ionic conductivity of glass-gdc composite It can be seen from the figure that the ionic conductivity increases continuously with the glass content and reaches 0.1 S.cm -1 at 40 wt% addition of BKVP glass. This is a significant increase in ionic conductivity for the un-doped GDC powders showed S.cm -1 ionic conductivity at 600 C. The figure also indicates that shrinkage decreased with glass content simultaneously. The two major requirements for LTCC compatibility, therefore, are found to be moving in opposite direction as the glass content increased. This implies that there is a need to select a golden mean or an optimum values for the both properties. Based upon the shrinkage requirements the 25 wt% Chapter 6 Page 213

13 glass content in GDC was chosen as an optimized condition for the time being, where shrinkage was close to 9% and ionic conductivity 0.04 S.cm -1 at 600 C. It may be noted that this ionic conductivity is 10 times higher than un-doped GDC at 600 C and is equivalent to the ionic conductivity observed at 700 C operating temperature. This comparison of results is presented by comparing the Nyquist plots in Figure 6.5. Thus, it can be conclude from above discussion that using BKVP glass is found to be useful in increasing both, the shrinkage and ionic conductivity values of GDC. Figure 6.5: (a) Comparison of Nyquist plots measured at 600 C for 25 wt% BKVP glass in GDC and un-doped glass, (b) Nyquist plot comparison of un-doped GDC and Glass doped GDC indicating similarities at 100 C lower operating temperature using glass To confirm that measured conductivity of glass ceramic composite electrolyte is pure ionic conductivity, ionic transference number was measured at 600 C for all samples. This ionic transference number found close to 1 for all samples having glass content from wt%. Table 6.5 presents density, shrinkage, ionic conductivity and ionic transference number for 10-40wt% glass added to GDC. Chapter 6 Page 214

14 Weight% glass added in GDC Shrinkage (%) Density of pellets (gm.cm -3 ) Ionic conductivity at 600 C ( 10-3 S.cm -1 ) Ionic transference number Table 6.6: Effect of different glass weight % in GDC on shrinkage, Density, ionic conductivity and ionic transeference number measured at 600 C It is seen from Table 6.5 that ionic transference number for all samples is This confirms that this glass-ceramic composite electrolyte has only ionic conductivity and the electronic component is absent. Thus, it can be concluded that these novel glass-ceramics composites show better performance as electrolytes for SOFC, and at the same time these electrolyte materials shows better physical compatibility with LTCC materials. The following Sub-section of this Chapter provides detailed analysis of structural and morphological studies of GDC-glass composite electrolyte Materials Characterization of GDC-glass composite electrolyte GDC-glass composite electrolyte powders were characterized by X-ray diffraction, Raman and TEM analysis. The surface morphology was observed using FE-SEM (Field Emission-SEM) images. The results are discussed in the following Sub-sections XRD analysis GDC-glass composite powders were synthesized by adding different proportions of glasses in GDC and milling together for 2 hr and fired at 1000 C for 40 min. XRD patterns of these pellets are presented in Figure 6.6. Chapter 6 Page 215

15 Figure 6.6: XRD patterns of pellets with glass in various weight contents from wt% in GDC and fired at 1000 C, along with pattern of pure GDC powder XRD patterns indicate that adding of glasses in GDC shifts position of peaks towards higher 2θ values compared to pure GDC. This implies decrease in the d-spacing of the structure. In case of 10 wt% glass doping, glass gets accommodated in GDC structures and no additional peaks was observed in XRD patterns compared to GDC. An effect of glass doping becomes significant at 15 wt% glass content and above. A small peak is appeared at 2θ of ~30.4 for glass content between wt%. This peak corresponds to the phase Bi 10V 2O 20. Further addition of glass in GDC increases impurity peaks around <111> plane and peak at 30.4 split into two peaks. These peaks correspond to Bi 2O 3, Bi 2O 4 and KO 3. Peak intensities of all peaks are also found diminishing with further increase in glass content in GDC, attributed to deviation from pure phase of FCC structure to mixed phase due to addition of recrystallized glass. All peaks corresponding to FCC crystal structure of GDC remained unaffected by added impurities. From these results it can be conclude that addition of glasses up to 10 wt% in GDC affects only the d-spacings in lattice; increase in glass content above 10 wt% causes re-crystallization and some such peaks are appear in XRD patterns without deteriorating GDC the matrix. It is clear from above discussion that increase in glass content does affect the lattice structure of GDC and the glass not only improves the grain Chapter 6 Page 216

16 boundary regions but may also be diffusing into lattice of GDC and increase vacancies. This effect is already seen in the Nyquist plot where intra-grain ionic conductivity also showed significant enhancement Raman spectroscopy The GDC-glass composite powders were also characterized by Raman analysis in energy range of cm -1, to find effect of glass addition in GDC on bond lengths and symmetries of FCC crystal structure. Figure 6.7 presents Raman spectra of GDC-glass composite materials with different glass contents. Figure 6.7: Raman spectra of pure GDC and with glass added in different wt%; three significant effects are highlighted using arrows and rectangle It is clearly seen from Figure that addition of glasses has significant effect in Raman spectra of bond vibrations in the glass-gdc composite. The comparison of Raman spectra for doped and undoped glasses shows three major differences. These three major changes are pointed out using arrows in red and blue color and a rectangle of red color. A signature peak of F 2g symmetry is observed in all cases, showing that the symmetry of basic GDC lattice is unaffected by glass doping. The peaks at 563cm -1 and 630cm -1 correspond to oxygen vacancies. These peaks are seen shifting towards higher energy and shoulder at higher energy becomes increasingly intense with increasing glass content. This implies increase in oxygen vacancies with addition of glasses in GDC. Raman active mode at lower energy of ~100cm -1 show Chapter 6 Page 217

17 vibrational interactions between glass and GDC lattice due high glass content. Intensity of F2g symmetry peak decreases due to polarizability and lowering in concentration of Raman active groups. Peak at 120cm -1 to shifted at higher Raman frequency at 150cm -1 in case of 40 wt% glass added in GDC shows incorporation of glass in GDC lattice structure, and hence changing lattice parameters. This incorporation is clearly seen in TEM images in next section. It is clear from the Raman spectra that addition of glass in GDC sets-in close lattice interactions between them. The increase in glass doping does not affect symmetry of GDC, but it increases oxygen vacancies in lattice TEM analysis FE-TEM of GDC-glass composite powder was studied to study the effect of glass content on lattice planes and crystallite size of composites. Figure 6.8 presents FE-TEM images of GDC-glass composite electrolyte powder with glass 40 wt%. (a) (b) Figure 6.8: FE-TEM images of GDC-glass composite powder showing lattice planes and distortion of lattice plane due to glass re-crystallization It can be very easily observed from the figure that addition of glass in GDC distorts the lattice planes. The glass is seen partially crystallized and amorphous. The highlighted area by a red oval shape magnified in image (b). The crystallization of the glass is clearly Chapter 6 Page 218

18 Ph.D Thesis 2014 observed in this image. During the crystallization it distorted <111> lattice plane of GDC. The area heighted by square indicated area where glass is in amorphous form. Figure 6.9 presents TEM image and corresponding SAED patterns. Presence of this amorphous phase is observed in the Selected Area Electron Diffraction (SAED) pattern as a broad and diffusive rings corresponding to amorphous glass could be identified. Glass may not have fully crystallizedd due to lower sintering dwell time. Increased bright spots compared to FCC crystal structure on the periphery of each polycrystalline circle shows diffractions from planes of the re-crystallized glass. Figure 6.9: FEG-TEM image showing nanocrystallites compostite of GDC-glass and corresponding of SAED pattern confirming mixed polycrystalline and partial amarphous phases The above structural analysis of the GDC-glass composites concludes that, glass doped in GDC partially crystallizes and the composite becomes a mixture of polycrystalline and amorphous phases. Crystalline glass distorts the <111> planes and creates its own planes as confirmed as impurity peaks in the XRD SEM analysis of pellets The surface microstructure of the pellets was observed under SEM to understand the surface morphology of the pellets. Figure 6.10 presents, surface morphology of the pellets doped with 25 wt% BKVP glass and sintered at 1000 C for 40 min and 3hr. Chapter 6 Page 219

19 (a) (b) (c) Figure 6.10: SEM micrographs of the surface of pellets prepared by 25 wt% glass added in GDC and fired at 1000 C (a) and (b) 40min and (c) 3 hr dwell time It is seen from the micrographs of 40min dwell time that some rod and bush-like structure is developed over most of the pellet surface. The EDAX of the pellet in this area shows all the elements content in the glass as well as GDC are present in these structures, but those coming from the glass elements were dominant. This implies these rods and bushes are grown because of high glass content in the pellet. Glass added in the pellet flows through voids present between the grains along the grain boundaries, and the excess of glass may be forming such structures on the pellet surface. Figure 6.10 (b) presents the pellet surface sintered with dwell time of 3 hr. The surface of this pellet shows void free surface throughout the pellet, and rods or bush like structures are not found on the surface. This implies that with longer sintering dwell time glass gets completely settled down on surface and between the grains and there is complete grain growth. Such grain growth is not observed in earlier case (Figure 6.10 (b)) where small pits were observed on the surface. Hence, it can concluded from the SEM micrographs that excess glass in pellets flow over the surface of the pellet and form grain structure dependent on sintering dwell times. However, this grain growth and change in surface morphology does not affect ionic conductivity. Figure 6.11 presents ionic conductivity measurement for pellets fired at different dwell times. Chapter 6 Page 220

20 Figure 6.11: (a) Nyquist plot of pellets sintered at different dwell times at 1000 C for 40, 120 and 180 min, (b) effect of sintering temperature on the Nyquist plot when fired at 875 C, 950 C and 1000 C The Nyquist plots of pellets fired at different dwell times at 1000 C shows that the impedance of the pellets is almost independent of the sintering dwell times. There is very small change in ionic conductivity observed for these pellets. However, small change in sintering temperature made a huge difference in the ionic conductivity and consequently the Nyquist plot of the GDC-glass doped pellet. Figure 6.12 (b) presents effect of sintering temperature on impedance of the pellets. The ionic conductivity of the sample decreased with decrease in sintering temperature from 1000 C to 875 C. The ionic conductivity as measured at 600 C for samples sintered at 850 C, 950 C sintering temperature was found to be S.cm -1, S.cm -1 respectively and S.cm -1 for 1000 C sintering temperature. However, unlike 1000 o C sintering temperature, the Nyquist plot in case of 875 C and 950 o C sintering temperature showed two semicircles, implying contribution from the bulk of the grain and grain boundaries and the grain conductivity is found to be less than grain boundary conductivity. This implies that the sintering temperature was insufficient for glass diffusion in grain interior. It may be noted though, that this ionic conductivity at 900 C is almost equal to the 1350 o C fired pure GDC pellet at 600 C. Thus, this study concludes that addition of Chapter 6 Page 221

21 BKVP glass in GDC not only improves the sintering of GDC, but also imparts LTCC compatibility without deteriorating the ionic conductivity properties of electrolyte, instead it improved the conductivity. Table 6.6 presents modified physical properties of SOFC ceramic electrolyte after addition of BKVP glass. Property Un-doped GDC Glass doped GDC LTCC DuPont Remarks 951 Firing temperature ( o C) Close to LTCC firing temperature Ionic conductivity Ionic conductivity o C (S.cm -1 ) less than GDC TCE (0-300 o C) (ppm/ o C) > Close to DuPont 951 LTCC tapes Density (gms/cm 3 ) Almost 90% of pure GDC achieved at 400 o C lower temp Shrinkage (%) Shrinkage is less than DuPont 951 tapes but close to pellet prepared by LTCC powder Table 6.7: Significant properties improvements in SOFC electrolyte and its comparison with LTCC properties Chemical compatibility with LTCC To study chemical compatibility of these novel electrolyte GDCglass composite with LTCC, a screen printable paste of 25 wt% BKVP glass doped in GDC was prepared by adding different binders, plasticizers and dispersant organic / inorganic liquids and solids 3 were mixed together ground for 2 hrs in an agate pestle-mortar until paste became viscous and flowing paste. This paste was printed on commercially available LTCC green tapes (DuPont 951) with thicknesses 160 and 250µm (single tapes) and stacks of thickness 350 and 500µm, both with X-Y dimensions 15 15mm. A screen having a square pattern of 10 10mm was prepared and used for printing. This pattern was printed and dried in normal atmosphere. The bottom surface of tape 2 Ionic conductivity obtained for GDC sample measured during synthesis parameters study. This powder is used to prepare GDC-glass composite electrolyte, GDC powder optimized in Taguchi study showed ionic conductivity of the order of 0.11S.cm -1 at 600 C. If this powder is used for glass-ceramic composite, much better results can be expected. 3 This composition is being patented. Chapter 6 Page 222

22 was observed removing mylar to verify penetration of organic or inorganic component of paste through tape. Such evidence was not observed confirming that neither organics nor inorganic materials from paste reacted with LTCC at room temperature and normal humidity conditions. These printed tapes were kept under observation for two days to verify inertness of the paste material with LTCC tapes. Later, these tapes were fired using standard firing cycle recommended by the manufacturer. The fired single tapes showed slight warpage only for 100µm thickness tape, while all other tapes and stacks showed good flatness, confirming matching of shrinkage of paste and tape during firing process. Figure 6.12 shows the line scans measured using inducting gauge of thickness profiler (TALLYSURF CLI 2000, Talyor Hobson). Thick film (a) Thick film Figure 6.12: Line scans recorded by inductive gauge showing warpage and thickness of screen printed thick film printed on (a) 160µm thick and (b) 250µm thick commercial LTCC tapes (DuPont 951) It is observed from Figure 6.6 (a) that screen printed thick films on 160µm tape shows slight warpage at the edges of thick film. However, this may also be the result of thickness variation across Chapter 6 Page 223

23 printing direction. The average fired thickness of thick film was 15µm. Printing direction in this both cases was from left to right of the figure. Figure 6.6 (b) shows surface scan of 250µm tape, the warpage on this tape is slightly reduced than that for 160µm thick tape. The thickness of thick film was 12µm and average warpage is observed in the range ±3µm across the ~8mm width. These warpage and thickness results are comparable with the commercially available DuPont 6142D screen printable thick film LTCC compatible paste (warpage ±1.8 µm over 8mm width). The above description concludes that the prepared GDC-glass composite thick film paste is physically compatible with LTCC. In order to confirm chemical compatibility with LTCC, the film surface and cross section of tape was first observed under microscope. There was no unusual change in the appearance, indicating less possibility of any chemical interaction. Thus, prima facie it was concluded that LTCC tape and GDC-glass composite paste both are compatible with each other. However, final confirmation of chemical compatibility can be made only after analyzing the interface and the whole sample using analytical techniques, such as, X-ray diffraction and Raman spectroscopy etc. and through microstructural observations. Presently, this analysis has not been completed. Similar to GDC, different sintering aids and some proton conducting glasses were tried for achieving low sintering temperature for the proton ion conducting electrolyte, viz. Yttrium doped Barium Cerate (BCYO). Following Sub-section report results of glass-bcyo composites Effect of sintering aids on proton conducting BCYO Bi 2O 3 was used as sintering aid for sol gel synthesized Yttrium doped Barium Cerate. The pellets turned to black color after firing at 850 C. The electrical measurement by impedance brought out occurrence of very high impedance for these pellets. The ionic conductivity measured for these samples shows ionic conductivity in the range of 10-6 to 10-7 S.cm -1 at 600 C. Such high impedance is possibly a result of formation of insulating phase between Barium oxide Chapter 6 Page 224

24 and Bismuth oxide due to chemical reaction (14). Due to repeated, high impedance observations for such composite, this aspect of interaction was not investigated further. It was concluded that Bi2O3 is an improper choice as sintering aid for BCYO electrolyte. Some other low melting temperature oxides, such as, Vanadium oxide (V 2O 5), Antimony oxide (Sb2O5), Thallium oxide (Tl2O5) were also used as sintering aids for BCYO. However, there were no significant improvement in the results obtained using these sintering aids. Table 6.7 listed sintering aids and corresponding shrinkage, density and ionic conductivity comparison with BCYO pellet sintered at 1350 C. Sintering aid in Shrinkage when Density (%) Ionic conductivity at BCYO fired at 1000 C 600 C(x10-3 S.cm -1 ) BCYO % V 2O % Sb2O % Tl 2O Table 6.8: Effect of different sintering aids on shrinkage, density and ionic conductivity of BCYO The results presented in this table indicate that low temperature melting oxides including Bi 2O 3 are not useful as sintering aid for BCYO. The ionic conductivity lowered due to addition of theses oxides due to disorder produced in grain boundaries and produced insulating phases with BCYO by these oxides. Therefore, instead of inorganic oxides as sintering aid, known proton ion conducting glasses were tried as sintering aid in BCYO. Synthesis of glasses and their effect on shrinkage and ionic conductivity are presented the following Subsection Effect of glasses as sintering aid This work was carried out selecting two different glasses based on Silica, Zirconia and P2O5. These components were selected because glasses of these oxides are reported to be good protonic ion conductors at C operating temperature (15). These glasses are well known electrolytes for Proton Electrolyte Membrane Fuel Cell (PEMFC) (16). Chapter 6 Page 225

25 Two different compositions were selected based on literature survey. These glasses are synthesized by sol-gel method, obtained membranes were fired at temperature range C. A high proton conductivity of S cm -1 was reported for the 83SiO 2 5P 2O 5 2TiO 2-10ZrO 2 (mol%) composite at 80 C under 90% relative humidity (RH), while for the 83SiO 2 5P 2O 5 2ZrO 2-10TiO 2 (mol%) composite the conductivity was only S.cm -1 under the same conditions (16). Two compositions of glasses were selected for this work, viz. 83SiO 2 5P2O 5 2ZrO 2-10TiO 2 (SZPT) and 48SiO 2-2ZrO 2-25P 2O 5-25V 2O 5 (SZPV) (15), (16), (17). SZPT glass was synthesized by sol-gel method and SZPV glass was synthesized by standard solid state route. For the synthesis of SZPT glass by sol gel method, the raw materials used were tetraethyl orthosilicate (Si(OC2H5)4, TEOS, 99% Merck), trimethyl phosphate (PO(OCH 3) 3), 99% Merck), titanium-isopropoxide (Ti(OC 4H 9) 4, 99% Merck), Zirconium-iso-propoxide (Zr(OC 4H 9) 4, 99% Merck) and N-N- Dimethyl formamide (HCON(CH 3) 2, 99% Merck). The experimental procedure was followed as given in literature (16). These synthesis steps are also presented in Figure The SZPT glass synthesized following the steps given in flow chart was calcined at 600 C. A black colored glassy powder was obtained. This powder was crushed to obtain fine powder. 10 wt% of this glass was added to BCYO and its pellets were sintered at 1000 C. However, it was seen that these pellets show an increase in dimensions up to 1%. This would not be suitable for LTCC applications. Further, this glass also caused increase in impedance of the glass-bcyo composite when tested with 10 wt% glass content (Figure 6.14). Due to these results, it was decided not to proceed with further experiments. Chapter 6 Page 226

26 Figure 6.13: Flow chart for synthesis of SZPT glass by sol gel method. Figure taken from reference (16) The other glass used for this investigation was having composition 48SiO 2-2ZrO 2-25P 2O 5-25V 2O 5 (SZPV). In this glass TiO 2 was replace by V 2O 5 considering its lower melting point, also molar % of Silica was reduced to 48mole%. SZPV glass was synthesized by solid state reaction using the raw materials as Silicon oxide (SiO2; 99.9% Sigma Aldrich), Zirconium oxide (ZrO 2; 99% Sigma Aldrich), Vanadium Pentaoxide (V 2O 5; >99.6% Sigma Aldrich) and Ammonium Phosphate (NH4H2PO4; >98% Sigma Aldrich). These raw materials were mixed together in given molar percentage and wet-milled for 96 hr using Acetone (HPLC grade; Merck) solvent. The milled powder was dried and kept in furnace and fired at 800 C for 2hrs and molten mass was transferred to water to obtain quenched glass. This glass was again ball milled in polyurethane container to obtain a fine powder. 10 wt% of this glass was added to BCYO and pellets were pressed. These pellets were fired at 1000 C for 2hr. The sintered pellets show about 4.1 % shrinkage. Platinum paste was applied on both the surfaces and was fired again at 1000 C for 40 min. Impedance analysis these pellets was carried out at 700 C in frequency perturbation of 1Hz to 1MHz. The Chapter 6 Page 227

27 impedance results for both glasses added to BCYO were compared with the best results of pure BCYO. Figure 6.14 presents the Nyquist plots for pure BCYO and for composites with 10wt% SZPT an SZPV glasses. Figure 6.14: Nyquist plots for composites of BCYO with 10 wt% SZPT and SZPV glasses measured at 700 C and its comparison with pure BCYO The Nyquist plots presented in Figure 6.14 indicate that doping of glass is affecting the performance of BCYO resulting in high impedance across pellet thickness. SZPT glass showed very high impedance attributed to lower density and distortion in grain boundaries of the pellets. The pellets with SZPV glass shows lower impedance compared to SZPT glass, but its this impedance is still as high as 400 times higher than pure BCYO. From the above results, it can concluded that the SZPV as well as SZPT glasses are not compatible with BCYO. To make BCYO electrolyte compatible with LTCC, more options of glass compositions need to be explored. Due to lack of time this work could not be undertaken for this Thesis. Apart from the electrolytes, the electrode materials also need to be made compatible with LTCC. The following Sub-section of this Chapter presents details of compatibility studies for anode and cathode materials. 6.3 Effect of sintering aids on electrode materials Synthesis of electrode materials for SOFC was described in Chapter 5. These electrode materials must be compatible with GDC- Chapter 6 Page 228

28 glass composite electrolyte with respect to their physical properties such as shrinkage, sintering temperature, Temperature co-efficient of Expansion (TCE) and operating temperature. Amongst these matching of shrinkage during sintering is an important parameter which must tackled first. As we have seen in earlier, Copper Zinc oxide (CuZnO) was selected as anode material and Samarium and Strontium doped Cobaltite (SSC) was selected as cathode material. CuZnO showed 14% shrinkage when fired at 900 C, however, SSC showed such shrinkage only when it was fired at 1100 C. Thus, these two electrode material have higher shrinkage compared to GDC-glass composite electrolyte and their firing temperatures are found varied from 900 to 1100 C. The target of this study was to match their shrinkages when fired at 1000 C. Initially, efforts were taken to make cathode material SSC compatible with GDC-glass composite electrolyte material. This was done by adding different weight proportion of BKVP glass in SSC and fired at 1000 C for 40min. Shrinkage and density of the pellets were measured after firing and found that shrinkage of SSC-glass composite electrode was close to that of the composite electrolyte when 40% glass was added to SSC. Table 6.8 presents results of adding glass with different weight percentages to SSC with respect to shrinkage, density and measured ionic conductivity at 600 C. Weight % of BKVP glass added in SSC Shrinkage (%) when fired at 1000 C Density (%) Area specific resistance (ASR) (Ω.cm 2 ) at 600 C Table 6.9: Effect of different glass weight% added in SSC on shrinkage, density and ASR The results indicate that increase in glass content in SSC increased the shrinkage of pellets as well as lowered Area Specific resistance, which is a good sign for any cathode material. However, increase in glass content increased the density of the pellets, reducing Chapter 6 Page 229

29 their porosity. This may reduce catalytic activity of the cathode. It is clear that ASR is lowered in presence of glass due to increase in ionic conductivity of the pellet. TCE measured for 40 wt% glass added SSC was measured at 600 C, which was found to be 7.2ppm/ C, which is close to composite electrolyte. Initial results on LTCC tape by making paste of SSC shows no evidence of chemical reaction, however, in case of 250µm tape thickness diffusion was observed from backside of tape. These results are better when a stack of 500µm is used as co-firing substrate. Hence, it can be conclude that 40 wt% BKVP glass addition in to SSC has matching shrinkage properties with the GDC-glass composite electrolyte and its ASR value also improves. However, increase in density may reduce the catalytic performance. This aspect remains to be verified. The anode material selected for this work was Copper Zinc oxide which shows 14% shrinkage at 900 C without any addition of sintering aid. This oxide has melting point somewhere in between C and hence its pellet melts when fired at 1000 C. Due to time limitation, this anode material was applied after firing (post-fire process) for cell fabrication discussed in the next Chapter. 6.4 Co-firing of SOFC material with LTCC In this Chapter we have reported that, BKVP glass was useful for making SOFC electrode and electrolyte materials compatible with the LTCC materials. It is seen that the ionic conductivity results are exceptionally good when sintered at 1000 C, when sintering temperature lowered its ionic conductivity found lowered. On the other hand, at 1000 C sintering temperature, anode material Copper Zinc Oxide (CuZnO) melts and cathode material shows very high density. Therefore, lower firing temperature is favorable for these electrode materials. Further, LTCC firing temperature is around C considering silver metal is used as via filling conductor and for conducting tracks. Increase in firing temperature in the range of 1000 C for LTCC will require replacement of silver by gold and platinum. These Chapter 6 Page 230

30 materials are costlier and will ultimately increase the cost of fabrication. Clearly, the present LTCC co-firing study of SOFC electrolyte materials bring up three different sintering or co-firing options where one has compromise either properties of electrolyte or co-firing of the fuel cell. Table 6.10 presents details of firing temperature options and compromises forced for selecting any one of the. Firing temp. Electrolyte (GDC-glass composite) 1000 C Good ionic conductivity properties 950 C Ionic conductivity slightly lowered 875 C Ionic conductivity lowered to 35% of that at 1000 o C Anode (Copper Zinc oxide) Cannot used as have low melting temperature Cannot used as have low melting temperature Can be used at optimum catalytic activity Cathode (SSC-glass composite) Porosity lowered, poor catalytic performance Porosity increases shrinkage matches with LTCC and electrolyte Can be used with optimum porosity and catalytic activity LTCC (DuPont 951 tapes) Silver cannot be use, costlier gold and platinum materials needed Silver cannot be use, costlier gold and platinum materials needed Silver can be used as conducting metal, all components can be cofired Table 6.10: Comparison of different sintering temperature possible in co-firing of LTCC and SOFC and their effects on electrode, electrolyte and LTCC material systems It is clear from the table that, three different temperatures are possible for co-firing of SOFC in LTCC depending on compromise chosen. At 1000 C sintering/firing temperature electrolyte properties are optimum, however, this temperature not suitable for electrode and silver based LTCC packaging. At 950 C temperature, electrolyte performance lowered in terms of conductivity, even though this temperature is high for the electrodes and silver. Sintering temperature 875 C is suitable for electrode porosity, CuZnO can be use as anode and silver can be used in LTCC structures, however, ionic conductivity Chapter 6 Page 231