AVS1997 Session QS-MoA: XPS

Monday, October 20, 1997 2:00 PM in Room N
Monday Afternoon

Time Period MoA Sessions | Abstract Timeline | Topic QS Sessions | Time Periods | Topics | AVS1997 Schedule

Start Invited? Item
2:00 PM Invited QS-MoA-1 Final-State Effects in Quantitative XPS Analysis: the Ion and the Photoelectron**
C.S. Fadley (Univ. of California, Davis & Lawrence Berkeley National Lab)
We will consider the influence of various final-state effects on quantitative surface analyses with x-ray photoelectron spectroscopy (XPS) and angle-resolved XPS. These final-state effects will be broken into two categories, those involving the ion with a core hole that is left behind, and those involving the elastic scattering of the photoelectron in its exit from the sample. Several many-electron effects in the ion cause intensity to be distributed over various peaks or satellites, and this intensity has to be properly integrated over and/or calibrated for in order to derive accurate sample compositions. The origin of these many-electron effects will be reviewed, and ways for dealing with them discussed. In specimens with single-crystal character, elastic scattering can furthermore lead to strongly preferential emission along certain directions due to photoelectron diffraction effects, and such phenomena also have to be averaged over and/or calibrated for in order to carry out accurate quantitative analyses. Some recent examples of this type of analysis will also be discussed. ** Work supported by DOE, BES, Mat. Sci. Div. (Contract DOE-AC03- 76SF00098) and ONR (Contract N00014-94-1-0162).
3:00 PM Invited QS-MoA-4 Creating a Rulebase for Interpretation of XPS: A Guide for both Human-based and Computer-based Experts
J.E. Castle (University of Surrey)
The XP spectrum is an enormously rich source of information about the concentration, chemical states and distribution of elements in the near-surface region of a solid. To utilise such richness one needs the help of an expert since its interpretation is far beyond the output of a typical datasystem. Initial decisions to be made are whether the carbon signal represents contamination and whether the remainder of the spectrum is derived from an homogenous or a layered structure. Much of this can be gathered by expert eyeballing of the survey spectrum and hence can be used to gain most reward from the subsequent set of detailed scans. In setting up a rulebase we are trying to devise a preferred route to enable interpretation by human or computer in a portable manner. In writing rules we need to be highly introspective, both in making implicit decisions quite explicit and in ascribing degrees of certainty to a balance of probabilities. Above all the process cannot be left to an individual. The rules need to be written and validated by a community, since, once communicated to a computer (or a human) they tend to remain hidden and difficult to challenge. Moreover we cannot be sure that they will not be used in an unintended situation. In thinking about building a ruleset we have started from the simplest of beginnings. This talk will describe the progress which has been made and will illustrate how such introspection enables us better to teach others how to gain most from their spectra.
4:00 PM QS-MoA-7 Sharing of AES and XPS Spectral Data through Internet
K. Yoshihara, M. Yoshitake (National Research Institute for Metals, Japan)
As a part of the international collaborative reset of VAMAS, we have developed a PC-based system, Common Data Processing System(COMPRO), which integrates the spectral and physical data, peak position database, GUI query and analysis system. COMPRO Version 5.0 runs on Windows95, and is now distributed through Internet (http://sekimori.nrim.go.jp). This version can convert spectral data of ASCII format to the VAMAS Standard Data Transfer Format (ISO DIS 14976), and has an easy access interface to a Internet spectral database provided by the Surface Analysis Society of Japan(SASJ). COMPRO also has calibration systems for energy and intensity scales, and every body can easily calibrate his/her analyzer. In 1994 the Science and Technology Agency (STA) of Japanese Government launched a project to interconnect networks under various ministries and agencies. As a site of this network we are implementing a network-oriented database for surface chemical analysis such as AES and XPS spectra. SASJ is responsible to provide the spectral data, and controls its quality. Internet spectral data now has about 2,000 spectra of metals, semi-conductors and ceramics. The file structure of spectral data is based on ISO NP 14976 and ISO DIS 14975, and is fully compatible with VAMAS Standard Data Transfer Format. Because this file structure can carry the information on specimens, calibration and data-processing, we could construct a GUI searching system for Internet database. One can choose searching items (ex: host materials, in-situ preparations, keywords, techniques, instruments, operators and etc.) by GUI. The searching results are displayed on a client's PC, and a target spectrum can be downloaded to a client's PC through Internet. In future we hope all computers of the surface analysis machines can be connected to Internet system so that every surface analyst worldwide can share the spectral data and the common data processing software to identify the surface chemistry of new materials.
4:20 PM QS-MoA-8 Universal Description of Elastic Electron Scattering Effects in XPS/AES Peak Attenuation
S. Tougaard (Odense University, Denmark); A. Jablonski (Polish Academy of Sciences, Poland)
It is well known that attenuation of XPS and AES peak intensities originating from deeper layers depends not only on inelastic- but also on elastic electron scattering. For typical geometries applied in practical AES and XPS analysis (geometry near the magic angle and emission not too far from the surface normal) we show that the effect on measured peak intensities is small and amounts to less than 10 - 15 % for electrons originating from depths ≤ 1.5 IMFP (inelastic mean free path). The effect of elastic scattering is in general much larger for electrons generated at large depths (≥ 2 IMFP). The magnitude of the effect depends on the depth, the material, the peak energy, and the anisotropy of the excitation. Because of the dependency on many parameters, accurate correction has been a rather complex procedure which involves tedious Monte Carlo calculations. In an attempt to find a more practical correction procedure, we have performed Monte Carlo calculations of 360 cases where these parameters were varied systematically. From this we find a simple and universally valid expression of the peak attenuation as a function of a reduced depth. The correction formula requires knowledge of the IMFP and the transport mean free path for elastic electron scattering and is for near normal emission and geometries near the magic angle universally valid to within ca. 5% accuracy. The correction procedure is applied to a quantification by QUASES peak shape analysis 1 of Au4d spectra from a thin layer of Au-atoms situated in Ni at depths varying from 0 to 2.5 IMFP. When elastic scattering is neglected, the accuracy of the analysis (given by the RMS scatter in the determined amount of Au around the mean value) is 15.4 %. After correction for elastic scattering effects, the RMS scatter is reduced to 11 %.


1http://www.spo.dk/quases

4:40 PM QS-MoA-9 Loss Functions of Various Materials Obtained from XPS Spectra
M. Jo (Electrotechnical Laboratory, Japan)
X-ray photoelectron spectrum usually shows large and continuous backgroud produced by the inelastic scattering in the solid. The shape of the inelastic background is related to the energy-loss probabilty of escaping electron via Tougaard's formula1. Recently I proposed a new algorithm for solving the formula self-consistently using optimization technique2. The algorithm is mainly based on two assumptions, (1) peak intensity ratio after background removal is proportional to the photoelectron excitation probability, (2) there is only background contribution enough far from the peak. The obtained probability (loss function) shows many peaks, some of which coincide with the results of measured energy loss structure3 . The loss function's shape is also dependent on the ranges when the assumptions are set4. In the present report, analysis results of various materials will be discussed together with the relation of range setting noted above and the loss function's shape.


1S. Tougaard, J. Electron Spectrosc. Relat. Phenom. 52, 243 (1990), and references therein
2M. Jo, Surf. Sci. 320, 191 (1994)
3M. Jo, J. Surf. Anal. 1, 220 (1995)
4M. Jo and A. Tanaka, "European Conference on Applications of Surface and Interface Analysis (ECASIA 95)", ed. by H.J. Mathieu et al., (John Wiley & Sons, 1996, Chichester), p.433

5:00 PM QS-MoA-10 Calculation of Electron Inelastic Mean Free Paths in Germanium, Gadolinium, and Dysprosium
S. Tanuma (Japan Energy ARC Co. Ltd.); C.J. Powell, D.R. Penn (National Institute of Standards & Technology)
We report calculations of electron inelastic mean free paths (IMFPs) of 50-10,000 eV electrons in Ge, Gd, and Dy. These calculations were made to evaluate different possible choices for the parameter Nv (number of valence electrons per atom or molecule) in our IMFP predictive formula TPP-2M for elements having shallow core levels (those with binding energies less than about 30 eV). In such cases, the energy-loss spectrum associated with excitations of these core levels overlaps that due to excitations of the valence electrons, and it has not been clear whether to calculate Nv from the total of the number of valence electrons or from the number of valence and shallow core-level electrons. IMFPs for Ge, Gd, and Dy were calculated using the Penn algorithm while IMFPs for Bi have been published. For Ge, the choices of Nv = 4 or 14 lead to IMFPs from TPP-2M that differ by less than 15%; the calculated IMFPs are intermediate between the TPP-2M values. For Gd and Dy, the choices Nv = 3 and 9 correspond to the number of valence electrons and the number of valence plus 5p electrons, respectively. These choices lead to larger differences in the IMFPs from TPP-2M than for Ge, but the IMFPs for the larger values of Nv agree very well with the directly calculated values. For Bi, the calculated IMFPs were intermediate between values computed from TPP-2M with Nv = 5 and Nv = 15 (for which the IMFPs from TPP-2M differ by up to about 40%). These results indicate that, particularly for the rare-earth elements, it is appropriate to include the number of shallow core electrons (together with the number of valence electrons) in the parameter Nv when calculating IMFPs from TPP-2M.
Time Period MoA Sessions | Abstract Timeline | Topic QS Sessions | Time Periods | Topics | AVS1997 Schedule