For convenience, manuscript Figure 2 is attached below.The profile in Figure 2 was measured with the Caltech Cameca 7f instrument using conventional semiconductor-industry depth-profiling techniques (e.g., Wilson et al. 1989, manuscript Table 1). In order to minimise (1) effects of surface contamination and (2) distortion of the implant depth profile, the primary ( 16 O -) ion beam was rastered over 100 microns.(1) SIMS primary ion current densities are inhomogeneous; consequently, without rastering, even with a field aperture and electronic gating, the tails of the beam incident on a highly contaminated surface outside of the sputter crater can contribute to the measured counting rate because the Cameca ion optics, although good, are not perfect.The efficiency of these contributions is low, but all potential contamination signal levels are frequently orders of magnitude higher than those from the implant. The rastered beam cleans the surface adjacent to the analysed area greatly reducing contamination contributions. (2) After sputtering with an unrastered beam, e.g., at the peak of the implant depth profile, the high intensity core of the beam will be deeper than the lower intensity margins. For nominal depths beyond the peak, counts will still be coming from the peak due to the tails of the beam incident on the walls of the crater, causing significant distortions of the depth profile, distortions which can be avoided by rastering the beam. Normally, raster sizes less than 75 µm are required for depth profiling in insulators to avoid excessive charging; however, a somewhat larger raster appears to have