The most widely used raw material in the technical ceramic industry is alumina, most of which is manufactured using an output from the Bayer process. While most aluminas are purified and calcined in downstream operations, there are practical limits to the ultimate purity levels that can be achieved.

Impurities such as calcia (CaO), silica (SiO₂), and soda (Na₂O) are nearly ubiquitous in these aluminas, and typical purity levels are in the 99.0-99.8% range. In order to produce high-quality microstructures in real technical ceramics, magnesia (MgO) is often added to reactive alumina powders to control the level of abnormal grain growth.
 

Investigating Abnormal Grain Growth

The presence and influence of impurities on grain growth behavior in alumina systems has been well documented by the research community. However, the vast majority is based on ultra-high-purity systems. These systems form model platforms to study specific fundamental cases, but often fail to capture the essence or form a suitable link to production-scale powders. While calcia is known to strongly segregate to grain boundaries and cause anisotropic abnormal grain growth, magnesia can prevent abnormal grain growth and is added to alumina to act as a sintering aid.^1-4

The presence of magnesia is reported to reduce anisotropic segregation, of calcia as well as other impurities, which promotes homogeneous grain growth.^5-7 Several studies have found that the relative amounts of different impurities play an important role in tracking microstructure characteristics in various types of aluminas. Key among these findings is that there must be at least as much MgO as CaO to prevent elongated abnormal grain growth.^4,7-9

This work aims to evaluate the microstructural evolution of two commercially available reactive alumina powders with similar overall purity levels. Each powder was considered before and after the addition of specific impurities in order to test the impact of particular relationships reported in the literature on grain growth abnormalities. Both powders have an overall purity of 99.81 wt%, though they exhibit different grain growth behaviors.

The composition of the powders can be seen in Table 1. Powder A contains 440 ppm MgO, 280 ppm CaO, as well as 820 ppm Na₂O and 140 ppm SiO₂. Powder B contains 1290 ppm MgO, 70 ppm CaO, 330 ppm Na₂O and 270 ppm SiO₂. While the two powders have the same nominal impurity level, the actual impurity chemistries differ significantly.

The powders were pressed into pellets using spark plasma sintering (Thermal Technologies, LLC) at 1,300°C with 50 MPa applied pressure, and then annealed at 1,650°C for two hours (M60, Centorr Vacuum Industries, Inc.) to allow for further microstructure development. The samples were polished and prepared for observation by scanning electron microscopy (Hitachi 4300 FEG SEM, Hitachi High America) and electron backscatter diffraction (EBSD). Grain sizes were measured and calculated using the EBSD software (TSL OIM).

Experimental Results

An EBSD map and histogram of the grain size distribution for each of the samples can be seen in Figures 1(a) and 1(b); the colors represent the orientation of each grain. The grain size data for all samples can be seen in Table 2. The average grain size for powder A was 1.68 µm with a maximum of 12.26 µm. The average grain size in sample B was 1.78 µm and the maximum was 8.98 µm.

The ratio of the maximum grain size to average grain size (max/avg) can be used to describe the amount of abnormal grain growth in the sample by giving a measure of the difference between the highest extreme and the bulk of the data. The max/avg for sample A was 7.3, and 5.0 for sample B. While A had a smaller average grain size, it also had a larger maximum. The lower max/avg for sample B indicates a more continuous grain size distribution with fewer outliers, meaning sample B had less abnormal grain growth.

MgO was added to powder A in an attempt to make the microstructure more homogeneous, such as powder B. Following the idea that both CaO and SiO₂ induce anisotropic segregation and abnormal grain growth while MgO prevents abnormal grain growth, MgO was added in a sufficient amount to make the ratio of (SiO₂+CaO):MgO equal to that in powder B. The addition increased the MgO content to 1,450 ppm.

This new powder is referred to as A+MgO, and the composition changes have been highlighted in Table 1. The microstructure for A+MgO can be seen in Figure 1(c). While the max/avg grain size ratio decreased 25% from the original powder, the average grain size more than doubled. While the decrease in the max/avg ratio is the same as powder B, the average grain size increased significantly.

In addition, the level of CaO in powder B was elevated to match that of powder A; this new powder is called B+CaO. The purpose of this experiment was to test if powder B could be made to behave similarly to powder A—effectively the reverse of the MgO addition experiment. The microstructure of sample B+CaO can be seen in Figure 1(d). An escalation in grain growth was observed for this sample. The average grain size increased 84% and the max/avg ratio increased 38%, indicating an increase in both normal and abnormal grain growth.

Experimental Discussion

By considering the CaO:MgO ratio, as well as the absolute amounts of CaO and MgO for each sample, it becomes apparent that both play a role in the character of the microstructure. Sample A had the largest CaO:MgO ratio and the largest max/avg grain size ratio, as well as the smallest total amount of MgO. Sample B had the smallest CaO:MgO ratio and the smallest max/avg ratio, but not the largest absolute amount of MgO.

When the relative amounts of CaO and MgO were nearly the same for samples A+MgO and B+CaO, the microstructures were similar. By altering the MgO content in sample A, the microstructure improved, but not drastically. However, by adding CaO to sample B, the microstructure was essentially degraded to the same condition with an increase in both grain size measures.

These results indicate that the microstructure cannot be controlled by magnesia content alone. The results imply a diminishing return on improvement in the microstructure with increasing MgO. Such a cap on microstructure improvement with magnesia addition was also reported by Park and Yoon.¹⁰ Powder B had the least amount of abnormal grain growth and also the lowest CaO content, which suggests that reducing calcia content is an alternative—and likely more effective—way to improve the microstructure and deserves further investigation.

Past studies have had a similar outcome. As mentioned previously, Bae and Baik determined that in order to prevent elongated abnormal grain growth the magnesia content must at least match the CaO content in the system.⁷ Goswami et al. looked at the ratio of MgO to the total liquid forming impurities, including CaO and Na₂O, in liquid-phase-sintered alumina of 91-94% purity. Their results were in agreement with Bae and Baik, as well as Song and Coble in finding that at least an equal amount of MgO as CaO was necessary to avoid elongated grains.^{4, 7-8}

Yoo et al. investigated impurity levels in terms of the “equivalent silica concentration” (ESC) in which the total amounts of glass-forming impurities above the overall solubility in alumina were considered. The grain size at the onset of abnormal grain growth was correlated with the ESC value for a given system.⁹

Consistent with findings in the literature, each of the four powders studied here had at least as much MgO as CaO, and accordingly none had the large, anisotropic elongated grains known to be characteristic of CaO-driven abnormal grain growth. In a connected study, related powders of different chemistries with purities ranging from 99.90-99.97% were investigated in the same manner as the samples here. Six of the eight powders considered contained more MgO than CaO. For two powders in which the MgO content was reduced to below the level of CaO, there was substantial elongated abnormal grain growth, with the maximum grain sizes reaching 10 times the average grain size. Interestingly, these two powders had the highest overall purities, yet the most significant abnormal grain growth. These observations add strength to the conclusion that both the absolute and relative impurity levels are important to maintaining a high-quality, homogeneous microstructure.

Research Implications

These results indicate that simply adding more MgO to a sample will not solve the problem of abnormal grain growth. In fact, this work shows that calcia has a dominant role in driving microstructure development, as evidenced by the sample with the lowest CaO content having the best microstructure. With lower calcia content, less MgO will be necessary to homogenize the microstructure.

While continually adding MgO will reduce the severity of abnormal grain growth, increasing impurity contents can have undesired secondary effects, such as precipitation or stabilization of a second phase in the microstructure. Examples of second-phase formation could be the formation of glassy phases, nucleation of spinels or stabilization of beta-alumina. Therefore, continuing to add more of a single improving element is not an adequate solution. Starting with the lowest possible calcia content, and adding only the appropriate relative level of magnesia, will lead to a better microstructure than adding additional magnesia alone. 

References

1. Baik, S., White, C.L., “Anisotropic Calcium Segregation to the Surface of Al₂O₃,” J. Am. Ceram. Soc., 1987, 70 [9] 682-88.

2. Bae, S.I., Baik, S., “Determination of Critical Concentrations of Silica and/or Calcia for Abnormal Grain Growth in Alumina,” J. Am. Ceram. Soc., 1993, 76 [4] 1065-67.

3. Bennison, S.J., Harmer, M.P., “Effect of MgO Solute on the Kinetics of Grain Growth in Al₂O₃,” J. Am. Ceram. Soc., 1983, 66 C90-92.

4. Song, H., Coble, R.L., “Origin and Growth Kinetics of Platelike Abnormal Grains in Liquid Phase Sintered Alumina,” J. Am. Ceram. Soc., 1990, 73 [7] 2077-85.

5. Gavrilov, K.L., Bennison, S.J., Mikeska, K.R., Chabala, J.M., LeviSetti, R. “Silica and Magnesia Dopand Distributions in Alumina by High-Resolution Scanning Secondary Ion Mass Spectrometry,” J. Am. Ceram. Soc., 1999, 82 [4] 1001-08.

6. Baik, S., Moon, J.H., “Effects of Magnesium Oxide on Grain-Boundary Segregation of Calcium During Sintering of Alumina,” J. Am. Ceram. Soc., 1991, 74 [4] 819-22.

7. Bae, S.I., Baik, S., “Critical Concentration of MgO for the Prevention of Abnormal Grain Growth in Alumina,”J. Am. Ceram. Soc., 1994, 77 [10] 2499-04.

8. Yoo, J.H., Nam, J.C., Baik, S., “Quantitative Evaluation of Glass-Forming Impurities in Alumina: Equivalent Silica Concentration (ESC),” J. Am. Ceram. Soc., 1999, 82 [8] 2233-38.

9. Goswami, A.P., Roy, S., Mitra, M.K., Das, G.C., “Impurity-Dependent Morphology and Grain Growth in Liquid-Phase-Sintered Alumina,” J. Am. Ceram. Soc., 2001, 84 [7] 1630-26.

 10. Park, C.W., Yoon, D.Y., “Abnormal Grain Growth in Alumina with Anorthite Liquid and the Effect of MgO Addition,” J. Am. Ceram. Soc., 2002, 85 [6].