First and foremost, I would like to thank my wonderful thesis partner Peter Benoit, who worked with me to complete this thesis. It is common for industrial polycrystalline alumina to contain a significant amount of liquid, but it is unclear what dictates the densification temperature. It is proposed that the viscosity of the grain boundary fluid determines the densification temperature of polycrystalline alumina.
In addition, two levels of each chemistry were evaluated to demonstrate that compaction is independent of the amount of liquid in the grain boundary. In support of the hypothesis, samples that contained a low viscosity liquid compacted at 1600°C, while samples that contained a high viscosity liquid compacted at 1700°C. These results illustrate an opportunity to tailor the grain boundary chemistry to control the densification temperature for industrial sintering systems.
INTRODUCTION
LITERATURE REVIEW
Liquid Phase Sintering
Stage I – Rearrangement and Liquid Redistribution
Depending on the volume fraction throughout the system, the low viscosity liquid will distribute itself either in the necks or the pores. If the pores have a narrow distribution of sizes, there will be a homogeneous distribution of the liquid phase. Heterogeneous packing results in the fluid filling the larger pores later in the firing process.
This causes inhomogeneous mixing resulting in a lack of driving force for fluid redistribution. After the formation of the liquid, particle rearrangement is very fast and takes place in as little as a few minutes. Initial compaction occurs during this step and determines the initial microstructure of the sintering compact.
Stage II – Solution-Precipitation
Stage III – Solid State Sintering
Glass Formation Boundary
Fragility
Otminski applied the Avramov-Milchev equation with a predicted characteristic temperature at the Littleton softening point, where η=106.6 Pa⋅s, to then calculate the brittleness parameter using the Newton-Raphson algorithm.10,11. Although there is a lack of Littleton softening point data for inverted glasses within the SciGlass database, for a glass composition having a 1:0.37 molar ratio of CaO to SiO2 at 1500°C, the viscosity is calculated to be 101.57 Pa· s. 10. 12,13 This prediction of the viscosity of inverted glasses is consistent with previous experimental observations by Moesgaard and Yue.14 Based on the structural variation of glasses, it is assumed that inverted glasses will generally exhibit at least two orders of magnitude higher low.
Sintering High Purity Alumina without a Liquid Phase
Sintering High Purity Alumina with a Liquid Phase
EXPERIMENTAL PROCEDURE
Sample Selection
Sample Preparation
- Milling
- Heterocoagulation Process for Coating Alumina with Silica
- Slip Casting
- Doping via Salt Solution Infiltration
- Chemistry Verification
As shown in Figure 9, between pH 2 and 9, alumina will have a positive surface charge, while silica will have a negative surface charge. Measured ζ-potential of micron particles of quartz and aluminum oxide as a function of pH in the aqueous medium.5. Individual suspensions of alumina and silica are prepared and adjusted to the desired pH, and then mixed together by stirring, resulting in heterocoagulation.
The amount of silica required to provide a monolayer of alumina particles can be calculated from a hexagonal close-packed array of beads. However, the use of excess silica ensures adequate particle coating and will not cause long-term suspension instability. At an ideal pH of 3.0-4.5, the negatively charged particles of colloidal silica are attracted to the positively charged surface of alumina.
Once agglomerated, the pH is then raised to 7.0 to 10 to create a uniform negative charge thereby imparting long-term stability to the new composite particles (illustration not to scale).4. The pH was adjusted with 6M nitric acid until stabilization around a pH of 6 was achieved to obtain a positive surface charge. The final alumina-coated silica suspension was then centrifuged for 30 minutes, rinsed and centrifuged again to eliminate excess silica.
This uncoated alumina suspension was then mixed with the coated suspension in ratios of 1:4 and 1:2 to produce 25%. These different levels of silica were desired to achieve different levels of liquid during firing. Of each composition, 60 pellets with a diameter of 1.27 cm were cast using a Lexan mold onto a plaster mold.
Initial ICP results showed that the desired ratios of CaO to SiO2 were not successfully achieved, so new samples were made and the doping process was re-run more precisely. The results of the second ICP analysis are presented in Table II, which verifies the dopant chemistry. -ES results confirming the correct proportions between CaO and SiO2. moles) Molar ratio (CaO:SiO2) Invert.
Sintering
Sample Characterization
Density Measurements
Microstructure
RESULTS AND DISCUSSION
The difference in alumina is due to the location of the glass formation boundary indicated in Figure 12. It was found that samples containing a low viscosity fluid experienced maximum densification at a lower temperature than samples containing a high viscosity fluid . In Figure 14 it is shown that low viscosity, inverted glass samples achieved maximum densification around 1600°C, while high viscosity, normal glass samples achieved maximum densification at 1700°C.
At 1500°C, inverted glasses are predicted to have at least two orders of magnitude lower viscosities than normal glasses.13 As previously discussed in the literature review, the controlling mechanism for densification in the second phase of liquid phase sintering is diffusion of the material by the liquid phase. It is illustrated in Figure 15 that samples at maximum density still contain some closed porosity. When comparing the samples containing a normal glass containing up to 1550°C with low and high liquid levels in Figure 16, it is clear that the overall density appears to be higher in the sample containing more liquid.
This means that the previous thickening process may be related to the amount of liquid in the system. However, the final density does not appear to be affected by the level of liquid present. It can also be seen that the maximum density for the conventional samples is reached at 1700 °C.
Contrary to the results of the normal samples, at 1600°C the density appears to be independent of the liquid level. This observation supports the idea that samples containing an inverted glass will condense at a lower temperature. The overall uniformity of the microstructures illustrates that the heterocoagulation process was a successful way to introduce glass into the alumina system.
These images show that the samples appear to increase in porosity as grain growth increases. The hypothesis appears to be correct: samples containing a low-viscosity liquid thickened at 1600 °C, while samples containing a high-viscosity liquid thickened at 1700 °C. This proves that the densification temperature depends on the viscosity of the fluid at the grain boundary.
If the viscosity of the liquid phase determines the densification temperature rather than the amount of liquid, a low liquid inverted glass will reach full densification at 1600°C (open triangle A) and a high liquid normal glass will reach full densification at 1700°C (open circle A). Chan, “Improved fault tolerance in alumina containing 1 vol% anorthite by crystallization of the intergranular glass,” J.
Chemistry
Density Measurements
Error bars are not visible because the data points are too large to allow the error bars to be seen). Densification with a liquid phase illustrating that low viscosity, inverted glasses reach maximum density around 1600°C while high viscosity, normal glasses reach maximum density around 1700°C. Liquids having a lower viscosity allow easier mass transport during densification, thus resulting in higher densification at lower temperatures.
The actual density values for full density samples are shown in Table IV and compared to the average measured densities in Figure 15.
Microstructure
- EVIDENCE OF DE-SINTERING WITH INVERT GLASSES
- CONCLUSIONS
- FUTURE WORK
- REFERENCES
- APPENDIX
De-sintering is said to occur when the rate of grain growth exceeds the rate of densification. Desintering results in a decrease in overall density as the temperature rises above the temperature at which the maximum density was reached. A t-test was performed between the two mean density values at 1600 °C and 1700 °C to determine if there was a statistical difference between the two.
This can be done by creating an inverted glass with about 5% liquid and a normal glass with about 10% liquid. Suggested future work to confirm hypothesis by low and high viscosity fluid levels in brackets (open symbols on plot). Ralph, “Grain growth in high-purity alumina ceramics sintered from mixtures of particles of different sizes,” Ceram.
ASTM Standard C20-00, “Standard Test Methods for Apparent Porosity, Water Absorption, Apparent Specific Gravity, and Bulk Density of Burned Refractory Bricks and Forms by Boiling Water” ASTM International, West Conshohocken, PA, 2010.
