Radial distribution study of anodic Al2O3

Abstract

Diffuse X-ray scattering curves were obtained for three different anodic Al2O3 films. The three anodic films were produced in electrolytic solutions of sulphuric acid, chromic acid, and oxalic acid. Point counting techniques were employed using Ross matched filters (Zr,Y) to produce monochromatic molybdenum radiation. The intensity was measured over a range of 3 degrees to 110 degrees (2 theta). Scattering curves were obtained also for four dehydration products of gamma-Al2O3. These samples were dehydrated at 425 C, 600 C, 700 and 800 C. Fourier analysis was performed to produce five types of radial distributions. These are: (1) the atomic radial density, (2) the atomic radial density, with thermal factor, (3) the electronic radial density, (4) the reduced atomic radial density and (5) the reduced differential atomic radial density functions. Within the limits of present accuracy, the radial distribution functions obtained for the three anodic oxides are quite similar. They differ appreciably from the radial distribution curves of the dehydration products. Isolated peaks for the three anodic oxides occur approximately at r = 1.81A and r = 3.10A plus some unresolved peaks at greater r. The tetrahedral aluminum - oxygen distance in the aluminosilicates has been reported to be 1.78 +- .02A while octahedral Al-O in corundum has been found to be 1.90A. The area under the first peak calculated on the assumption of tetrahedral and octahedral oxygen around aluminum. The calculated values are: tetrahedral = 1630, octahedral= 2450. The experimental values averaged 1450. The ratio of the second interatomic distance to the first is approximately 1.72 while for regular tetrahedral coordination this is expected to be 1.63. For octahedral coordination the expected ratio is 1.41. The second peak was identified as both a O-O and Al-Al peak from the position and area. The positions and areas of the peak indicate a tetrahedral coordination. An error analysis was performed by varying various parameters by amounts larger than the estimates of the possible error. None of these variations could change the radial distribution curves to that characteristic of octahedral coordination. In the process of computing the radial distributions the effect of Compton scattering is subtracted from the observed intensities. Since Compton scattering cannot be measured with our present technique, the effect must be calculated theoretically. The adequacy of this procedure, has been questioned and this remains as perhaps the greatest source of error in this experimental procedure. Calculations in which the Compton correction is varied by +- 30 percent cause major shifts in peak positions and areas. Even these shifted peaks, however, lend no major support to an octahedral coordination.