Determination of alpha alumina (alpha-Al2O3) content has been one of the ol
dest X-ray diffraction (XRD) methods of analysis. However, rhombohedral alu
minas may have a variety of morphological forms for which there are marked
differences in relative intensities of their major XRD reflections (the pre
ferred orientation effect). In addition, a specimen X-ray diffracting power
may vary making it difficult or impossible to employ a reference material.
Consequently, comparing intensities of selected reflections from a sample
with those of a reference material may cause a serious analytical error. A
conventional XRD and a Rietveld approach were used to analyze nine commerci
al ceramic aluminas, three NIST standards and two special grades of alumina
for alpha and beta alumina (beta-Al2O3 = Na2O . 11Al(2)O(3)) content. The
samples were also analyzed by scanning electron microscopy (SEM) to assess
morphology. Increasing the number of XRD reflections analyzed from two to e
ight did not improve the reliability of the quantification process. The mor
phological differences among samples could not have been minimized that way
. Conventional XRD (two to eight reflections) indicated alpha alumina value
s varying from 88 to 106% while the Rietveld analysis (full pattern) attrib
uted 100% for all but two. These two samples were found to contain beta alu
mina. Initial Rietveld refinements indicate that the beta phase contains on
ly a fraction of the total Na2O content. The preferred orientation was comp
uted for 14 specimens and assessed using a relative scale. The Rietveld lea
st-square refinements would not only provide relatively quick, but also rel
iable information on alpha and beta content, as well as unit cell parameter
s of alpha and beta phases. (C) 2000 Elsevier Science B.V. All rights reser
ved.