Nuclear Instruments and Methods in Physics Research B 85 (1994) 499-502 North-Holland zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Beam Interactions with Materials B Atoms Reconstruction of Ar depth profiles from PIXE measurements T. Osipowicz *, S.C. Liew, K.K. Lob, I. Orlic, S.M. Tang zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQ Physics Department, National University of Singapore, 10 Kent Ridge Crescent, Singapore 0511 Th. Weber II. Physikalisches Institut, University of G&tingen, Germany We report on the application of an iterative maximum likelihood algorithm [ll to the reconstruction of depth profiles from PIXE measurements. PIXE spectra of 450 and 800 keV Ar implanted Al samples were taken at 1.5 and 1.3 MeV He+ energy and at angles ranging from 18 to 83”. The measured Ar yields are in good agreement with those calculated by a PIXE simulation program. The reconstructed depth profiles are compared to those predicted by an ion implantation simulation code (TRIM). 1. Introduction Several ion-beam analysis techniques (e.g., RBS, NRA, SIMS) allow the determination of depth profiles in the near-surface region. While quantitative results can be obtained from RBS and NRA, the application ranges of these two methods are somewhat limited: RBS analysis allows one to measure the profiles of heavy elements distributed in a matrix of lower atomic mass. In the case of a matrix of higher mass, however, a strong overlap of the signals from the different com- ponents usually occurs in RBS spectra. NRA, on the other hand, which can detect light isotopes, is re- stricted to the detection of certain isotopes. The PIXE method allows the detection of X-rays from all elements with Z > 10. The fact that the signals from different elements can be deconvoluted makes it an advantageous technique for depth profiling. Since a single PIXE measurement only gives a single integral yield-value for each of the different elemental compo- nents, measurements at different energies and/or an- gles are necessary to gain information on depth distri- butions. Since the ionization cross-section decreases steeply as the ion is slowing down, the deeper regions of a sample contribute only a minor amount to the total yield. As the yield data always contains random noise, the reconstruction of a depth distribution is not as straightforward as with RBS or NRA. Several strate- gies for the reconstruction of depth profiles from PIXE * Corresponding author, phone + 65 772 2619, fax + 65 777 6126. data have been reported, many of them give spurious oscillations [2-61. Recently, the use of an iterative maximum likelihood algorithm has been proposed [7]. This algorithm has been successfully applied to com- puter tomography. Poisson noise added yield data ob- tained using a PIXE simulation program [8] was em- ployed to demonstrate that this algorithm would recon- struct the depth profiles of trace elements which are distributed within a few km from the surface. This paper presents our work on the detection of the depth profile of Ar in Al using the PIXE technique and the maximum likelihood algorithm. Ar+ (A?‘) ions were implanted into two Al samples at 450 and 800 keV energy and fluences of 1 x 10” and 0.5 x 10” ions/cm’. PIXE spectra were taken at incidence an- gles 6 (target normal with respect to the beam) be- tween 18” and 83”. The PIXE measurements were carried out with a 4He+ beam at energies of 1.3 and 1.5 MeV. The Ar depth profiles were then recon- structed from the X-ray yields and compared to results from the ion implantation simulation code TRIM [9]. 2. Experimental The Ar implantations were performed at the 530 kV ion implanter IONAS [lo] at the University of GGttingen. The irradiated area was 8 x 8 mm2, the lateral homogeneity of the implantation profile over this area was secured by means of an electrostatic X-Y sweep system. A water cooling system was used to keep the samples at room temperature during the implanta- tions. 0168-583X/94/$07.00 0 1994 - Elsevier Science B.V. All rights reserved SSDI 0168-583X(93)E0650-6 VIII. LATTICE SITES