| AVS 55th International Symposium & Exhibition | |
| Applied Surface Science | Tuesday Sessions |
| Session AS-TuP |
| Session: | Aspects of Surface Analysis Poster Session |
| Presenter: | D.J. Surman, Kratos Analytical Ltd, UK |
| Authors: | A.J. Roberts, Kratos Analytical Ltd, UK K Macak, Kratos Analytical Ltd, UK D.J. Surman, Kratos Analytical Ltd, UK |
| Correspondent: | Click to Email |
Angle resolved XPS is a useful method for obtaining non-destructive quantification of thin (4-6nm) layers with good absolute depth resolution.1 Although acquisition of ARXPS data with modern instrumentation is easy, determining the depth distribution of elements from the data is more challenging. The maximum entropy method (MEM) is a technique frequently used for solving the inversion problem in angle resolved XPS experiments. The MEMSYS algorithm used by Livesey and Smith2 can provide an estimate of the optimum value of the regularising parameter (smoothness), of the error bars in the reconstructed profile and of the noise level in the experimental data set. While for the artificial ARXPS datasets the interpretation of these factors is straightforward, the fitting of the experimental data may be complicated by the sample structure which does not always agree with the assumptions postulated by the model used in MEM calculations : the electron transport in the sample must be determined by the inelastic scattering with the constant mean free path throughout the sample and the material density must be constant throughout the sample. The necessary condition for the consistency of experimentally measured data with the MEM model (successful fit) is that the Laplace transform of the compositional depth profile (LTCDP) calculated from the experimentally measured normalised intensity is a monotonically decreasing function for all measured elements. We have found an efficient algorithm which can estimate the LTCDP for elements with different inelastic mean free paths and thus provide an independent estimate of the noise scaling calculated by the MEMSYS algorithm and verify whether the MEM model is appropriate for the measured data. Here we present results from a number of samples and draw conclusions on the use of these methods to generate elemental and chemical-state depth profiles.
1 P.J. Cumpson, J. Electron Specrosc. Rel. Phenom. 73 (1995) 25
2 A. K. Livesey, G.C. Smith, J. of Elec. Spectroscopy 67 (1994) 439-961.