High-resolution elastic recoil detection utilizing Bayesian probability theory

Elastic recoil detection (ERD) analysis is improved in view of depth resolu tion and the reliability of the measured spectra. Good statistics at even l ow ion fluences is obtained utilizing a large solid angle of 5 msr at the M unich Q3D magnetic spectrograph and using a 40 MeV Au-197 beam. In this way the elemental depth profiles are not essentially altered during analysis e ven if distributions with area densities below 1 x 10(14) atoms/cm(2) are m easured. As the energy spread due to the angular acceptance is fully elimin ated by ion-optical and numerical corrections, an accurate and reliable app aratus function is derived. It allows to deconvolute the measured spectra u sing the adaptive kernel method, a maximum entropy concept in the framework of Bayesian probability theory. In addition, the uncertainty of the recons tructed spectra is quantified, The concepts are demonstrated at C-13 depth profiles measured at ultra-thin films of tetrahedral amorphous carbon (ta-C ). Depth scales of those profiles are given with an accuracy of 1.4 x 10(15 ) atoms/ cm(2). (C) 2001 Elsevier Science B.V. All rights reserved.