Background Fitting in the Low-Loss Region of Electron Energy Loss Spectra

Author(s):  
C P Scott ◽  
A J Craven ◽  
C J Gilmore ◽  
A W Bowen

The normal method of background subtraction in quantitative EELS analysis involves fitting an expression of the form I=AE-r to an energy window preceding the edge of interest; E is energy loss, A and r are fitting parameters. The calculated fit is then extrapolated under the edge, allowing the required signal to be extracted. In the case where the characteristic energy loss is small (E < 100eV), the background does not approximate to this simple form. One cause of this is multiple scattering. Even if the effects of multiple scattering are removed by deconvolution, it is not clear that the background from the recovered single scattering distribution follows this simple form, and, in any case, deconvolution can introduce artefacts.The above difficulties are particularly severe in the case of Al-Li alloys, where the Li K edge at ~52eV overlaps the Al L2,3 edge at ~72eV, and sharp plasmon peaks occur at intervals of ~15eV in the low loss region. An alternative background fitting technique, based on the work of Zanchi et al, has been tested on spectra taken from pure Al films, with a view to extending the analysis to Al-Li alloys.

2014 ◽  
Vol 20 (3) ◽  
pp. 798-806 ◽  
Author(s):  
Raman D. Narayan ◽  
J. K. Weiss ◽  
Peter Rez

AbstractA model-based fitting algorithm for electron energy-loss spectroscopy spectra is introduced, along with an intuitive user-interface. As with Verbeeck & Van Aert, the measured spectrum, rather than the single scattering distribution, is fit over a wide range. An approximation is developed that allows for accurate modeling while maintaining linearity in the parameters that represent elemental composition. Also, a method is given for generating a model for the low-loss background that incorporates plural scattering. Operation of the user-interface is described to demonstrate the ease of use that allows even nonexpert users to quickly obtain elemental analysis results.


2002 ◽  
Vol 14 (48) ◽  
pp. 12793-12800 ◽  
Author(s):  
R Jones ◽  
C J Fall ◽  
A Guti rrez-Sosa ◽  
U Bangert ◽  
M I Heggie ◽  
...  

Author(s):  
Xudong Weng

A method using fourier transform and logarithm for the retrieval of the single scattering electron energy loss spectrum from the experimental multiple data has been described and used previously. However, the uniqueness of this method is proved only under an additional assumption that no energy gain processes occur as the incident electron pass through the specimen. This proof is adopted in the later review papers. In this paper we prove the uniqueness without such an assumption, although in comparison with the energy loss processes, the energy gain processes usually have negligible differential cross sections, provided the relaxation times for the excitation inside the specimen are short enough with respect to the amount of energy loss. Also we shall compare deconvolution methods for the low loss region.


Author(s):  
M. Sarikaya ◽  
P. Rez

One factor limiting energy loss analysis is the effect of multiple scattering on core loss edge shapes. Multiple scattering distorts fine structures, leads to incorrect quantitative analyses and even affects analysis of extended fine structure (EXELFS). Two procedures for extracting the single scattering spectrum from a spectrum showing the effects of multiple scattering have been proposed. Johnson and Spence derive the single scattering profile by taking the logarithm of the fourier transformed spectrum. If all scattering angles are accepted by the spectrometer this is an exact procedure. Leapman and Swyt have had some success assuming that multiple scattering imposes low loss structure on the high loss part of the spectrum. It is of interest to know how stable these procedures are with thickness and whether the logarithmic deconvolution can be used in thicker specimens than the method described by Leapman and Swyt.


Author(s):  
R.D. Leapman ◽  
C.R. Swyt

Core edges in electron energy loss spectra are generally complicated by thickness effects involving plural inelastic scattering. The fine structure can be modified and errors can be caused in quantitation based on measured edge intensities. Sometimes plural scattering can confuse even identification of elements. In this paper we describe a practical method for eliminating these difficulties.We derive the single scattering distribution, assuming valence electron excitation is small in the core loss region, a reasonable approximation for edges above 100 or 200 eV and for thicknesses of a few hundred Å. We can then separate the spectrum into a “high loss” region H(E) consisting of core edges (less background) and a “low loss” region L(E) containing the zero loss peak, plasmons and one-electron excitations.


2011 ◽  
Vol 84 (24) ◽  
Author(s):  
J. Palisaitis ◽  
C.-L. Hsiao ◽  
M. Junaid ◽  
J. Birch ◽  
L. Hultman ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document