The software consists of a main program that provides the user interface, and various machine input/output functions, together with at least two libraries (DLL’s) that provide the spectrum processing and ZAF calculations for spectra taken on a SEM fitted with an EDS detector.
The software runs on standard PC’s and operating systems (Windows XP, 7, etc.). Complete ZAF analysis is possible, with or without standards, using an internal database of fundamental parameters (FP) such as absorption coefficients, fluorescence yields, transition probabilities, etc. There is also a spectrum display module included.
The software also includes the acquisition of spectra using Amptek DPP hardware. The underlying methods and results file (a so-called EDX file) can be setup and re-used for routine analysis, or elements can be selected for each new spectrum analysis.
The software can analyze either bulk materials or a single-layer thin-film material. Analysis can be done without standards if the results can be normalized to 100%. When using standards, thickness can also be determined or the results do not have to be normalized.
Figure 1. EPXA Electron Probe X-Ray Analysis software interface.
The software can analyze up to 40 elements as individual elements and/or compounds. Unanalyzed elements can be specified stoichiometrically bound with an analyzed element (e.g., oxides or carbonates). Elements can be analyzed in one or more compounds within the same analysis. One compound (or element) can be analyzed by difference. Any number of compounds (or elements) can be “fixed.” For example, solutions, binders and/or hydrated crystals can be analyzed this way.
Any bulk, or single-layer (unsupported) thin-film, sample can be analyzed by either standardless or a calibration-with-standards ZAF approach.
Many detectors and windows can be fully modeled. This allows analysis without any standards, with normalization to 100% (or any specified factor). If some elements are calibrated, and some are not, the missing calibration coefficients are derived by interpolation from the existing ones.
The mass or linear thickness of the sample can either be specified or calculated. In the latter case the analysis cannot be done without a standards’ calibration. Several units are possible for thickness measurement, and the density can be calculated theoretically or specified, in the case of linear thickness calculations. Composition units may be ppm or wt%, with the additional output of atomic and mole percent.
Various detectors (e.g., Si(Li), Si-PIN, or SDD) and windows can be fully modeled. The software has provision for the user to input all the required parameters (e.g., contact material and thickness, dead layer, etc.) associated with these detectors (and their windows).
The software contains specific functions to allow the direct acquisition of spectra from Amptek detectors using either their analog or digital pulse processors (DPP).
The complete system geometry can be specified, including the sample incidence, take-off, tilt and azimuthal angles, and the sample-to-detector distance.
The ZAF calculations include full corrections for absorption and secondary fluorescence. All possible lines are considered for both excitation and fluorescence. The analysis can be performed for all elements from H through Fm, using K, L or M lines in the energy range from 0.05 keV up to 130 keV.
This module automatically analyzes a spectrum and assigns the most-likely elements and lines to each identified peak, and assembles a complete list of likely elements in the spectrum.
This module (Spectra-X) displays acquired or processed spectra. Up to 8 can be compared. KLM markers are available for peak identification and various annotation tools are available for adding text and lines to the display.
Figure 2. Spectrum Display.
Using two known peaks in the spectrum, the software can calculate the effective gain (eV/channel) and offset (zero shift) for the spectrometer. These factors can then routinely be applied to subsequent spectra prior to other spectrum processing. It is vital that peaks are located at their expected energies; otherwise the spectrum processing cannot function correctly.
This module uses iterative filtering to remove all peaks, leaving behind the spectral background. This background modeling includes details about steps in the spectrum from absorption edges in the detector and window. The background spectrum may be displayed or removed from the original spectrum. There is also an option to remove a spectrum file on a channel-by-channels basis from the sample spectrum.
Routines are available to optionally remove both detector escape and sum (pile-up) peaks.
A specified number of 1:2:1 Gaussian smooths can be applied to a spectrum.
A blank spectrum can be subtracted from the sample spectrum. This can be used to correct for artifact peaks in the spectrum from a variety of sources.
Specified element peaks may be integrated over a fixed ROI (Region Of Interest), or the complete spectrum can be fit using either synthetic Gaussians for every possible line in the regions of interest, or reference profiles that are acquired experimentally. One of six major lines (Ka, Kb, La, Lb, Lg, Ma) is selectable as the main analysis peak for intensity extraction. All relevant lines, required for deconvolution, are then automatically included by the software for full overlap correction using a least-squares fitting procedure.
All required line energies and resolutions are calculated automatically from the specified analyte line. The Gaussian peak fitting can be done with a linear or non-linear least-squares approach. The latter allows constrained changes in the peak positions, intra-series line ratios, and peak widths, from their nominal starting points. The Reference deconvolution method (separate library) is more restricted and is typically used without any adjustments of line ratios and peak positions.
In addition to calculating elemental intensities, the software automatically calculates the estimated uncertainty and background values, which allows uncertainty and Minimum Detection Limit (MDL) calculations to be performed during the ZAF analysis.
Revised March 23, 2012