Date

2015

Document Type

Dissertation

Degree

Doctor of Philosophy

Department

Materials Science and Engineering

First Adviser

Harmer, Martin P.

Other advisers/committee members

Chan, Helen M.; Rickman, Jeffrey M.; Rollett, Anthony D.

Abstract

Grain growth in alumina is strongly affected by the impurities present in the material. Certain impurity elements are known to have characteristic effects on abnormal grain growth in alumina. Specialty alumina powders contain multiple impurity species including MgO, CaO, SiO2, and Na2O. In this work, sintered samples made from alumina powders containing various amounts of the impurities in question were characterized by their grain size and aspect ratio distributions. Multiple quantitative methods were used to characterize and classify samples with varying microstructures. The grain size distributions were used to partition the grain size population into subpopulations depending on the observed deviation from normal behavior. Using both grain size and aspect ratio a new visual representation for a microstructure was introduced called a morphology frequency map that gives a fingerprint for the material. The number of subpopulations within a sample and the shape of the distribution on the morphology map provided the basis for a classification scheme for different types of microstructures.Also using the two parameters a series of five metrics were calculated that describe the character of the abnormal grains in the sample, these were called abnormal character values. The abnormal character values describe the fraction of grains that are considered abnormal, the average magnitude of abnormality (including both grain size and aspect ratio), the average size, and variance in size. The final metric is the correlation between grain size and aspect ratio for the entire population of grains. The abnormal character values give a sense of how different from “normal” the sample is, given the assumption that a normal sample has a lognormal distribution of grain size and a Gaussian distribution of aspect ratios. In the second part of the work the quantified measures of abnormality were correlated with processing parameters such as composition and heat treatment conditions. A multivariate statistical tool called canonical correlation analysis was adopted to seek out relationships between a set of input variables and the abnormal character values. The input variables include the MgO, CaO, Na2O, and SiO2 contents, the ratio of MgO:(CaO+SiO2), and the annealing time and temperature. The analysis was applied to 33 different samples and showed that the composition ratio and MgO content were the strongest processing variables. These variables are most closely related to the correlation between grain size and aspect ratio, the average magnitude of abnormality, and the variance in grain size. The physical implications of these relationships are explored for a number of samples with different abnormal grain growth behaviors. Several of the samples contained a β”-alumina phase that is shown to have a dampening effect on abnormal grain growth. TEM investigation provides evidence that there is a grain boundary complexion with a different composition and structure than the second phase. A series of samples are compared after annealing for different times and are shown to have very different behaviors as a result of the second phase competing with complexions for control over the microstructure.

Share

COinS