When investigating ion-beam induced structural changes of single crystals by in-situ x-ray diffraction, we usually measured only one intense diffraction peak in order to keep a reasonable balance between irradiation time and data collection time. However, this strategy prevents insight into the causes of a large peak shift as it has been observed with the 112 0 peak of -alumina. An alternative approach is to use polycrystalline samples and to measure many diffraction peaks. Polycrystalline

more »

... ered -alumina (BCE Special Ceramics GmbH, Mannheim, 99.7% purity, density 3.85 g/cm3, average crystallite size 5 m) were cut into pieces of 10102.6 mm 3 in size and annealed at 1250°C for 2 h under flowing argon gas in order to remove microcracks and residual strain introduced by the cutting procedure. Subsequently, the pieces were glued with epoxy resin onto copper plates which fit to the sample holder of the x-ray diffraction apparatus in-situ at the UNILAC M2-beamline. The ion beam ( 197 Au, 4.8 MeV/u, ion flux  210 9 ions/cm 2 s) was carefully scanned across the samples at normal beam incidence. The specimen temperature during ion bombardment was (25  5)°C. At certain fluences  t the irradiation was interrupted and x-ray diffraction patterns were measured in symmetric Bragg geometry for 24°  2  70° using CuK radiation and comprising 8 intense diffraction peaks of alumina. The Bragg angles, 2 B of the unirradiated sample agreed within  0.01° with those of a calibration reference sample (-alumina from NIST, SRM 1976a). No foreign phases could be detected. Virgin alumina exhibited bright luminescence radiation which decreased rapidly during bombardment. At fluences 510 12 Au/cm 2 tiny bright spots of luminescence radiation were seen indicating spall-off, which limited the fluence range in which x-ray diffraction yielded reliable results. The 2 B position of the 8 diffraction peaks versus fluence is shown in fig. 1 . No additional peaks appeared, i.e. the formation of a significant volume fraction of a new (non-equilibrium) alumina phase can be excluded. 0.0 0.2 0.4 0.6 0.8 1.0 25.4 25.5 25.6 34.9 35.0 35.1 35.2 37.5 37.6 37.7 37.8 43.1 43.2 43.3 43.4 52.2 52.4 52.6 57.2 57.4 57.6 66.2 66.4 66.6 67.8 68.0 68.2 fluence t (10 13 Au/cm 2 ) 01-1 2 10-1 4 11-2 0 11-2 3 2(degrees) 02-2 4 11-2 6 21-3 4 30-3 0 Fig. 1: Variation of the 2 B peak position versus ion fluence for eight diffraction peaks of -alumina. The Miller-Bravais indices are indicated on the right. Hence, the observed line shifts have to be attributed to in-plane ion-beam induced stresses, which result from an expansion of defective alumina. The lines displayed in fig. 1 are the results of fitting the expression 2 B =2 B  2δ B exp(A t ) to the experimental data using a single value A = 6.810 -13 cm 2 for all reflections. 2 B denotes an apparent saturation value. The "saturation" elastic strain is given by  zz = δ B /tan B and is  zz = (5.2  1)10 -3 . Neglecting creep, the elastic in-plane stress is  = E /(1-)  zz . With a Young's modulus E = 416 GPa and a Poisson number  = 0.23, one obtains    2.8 GPa, which matches with the compressive strength of sintered alumina. Thus, spalling limits and ion-beam induced stresses give rise for the apparent saturation of stress.