Contents of Chapter.
Make File of n Noise Spectra. This feature enables creation of a file of n Spectra each containing different Poissón noise (the counting noise observed in EDS spectra). It allows a Monte Carlo simulation of n spectra, each taken under similar conditions to determine detectability levels and variance. 'Generate' the desired spectrum, bulk or thin, put the resulting spectrum in 'Work' and then do this dialog. When the calculations finish (which can take several minutes for 50 spectra), 'curve fit' the spectra in auto mode and send the results to a spread sheet file for statistical analysis. See Appendix IV for a detailed discussion of this feature.
An X-ray spectrum is generated from first principles for either a thin target (e.g., a specimen in a transmission electron microscope) or a thick target (e.g., a specimen in a scanning electron microscope thick enough to be opaque to the impinging electron beam). ALL instrumental and physical parameters must be meaningful and specified.
Enter the elemental concentration for all elements present. Enter the concentrations designating format from the options allowed to the top-right of the list. The program will convert to weight fraction before leaving the dialog. If you enter oxides using the "Ox. Wt. Fr." format, (e.g. Al2O3), the program automatically enters the compositions of Al and O as well as their valences. Above the bottom are three lines that show the current sum, 1-sum, and the average atomic number of the hypothetical specimen. Before converting either way between atomic or weight concentration, the sum must be essentially 1 or conversion errors can result.
Enter the kilovoltage, the time and the Faraday current. If the specimen is "thin" then the thickness and the density boxes become active. Obviously, we need to know how many atoms the beam interacts with, so both density and thickness are required for thin targets. Neither of these quantities are required for thick targets because the mass range is accurately predicted by an analytical formula.
Database controls are provided. If more than a few elements and their concentrations are entered, it is possible to place these into a database to prevent having to type them in again. "Load a Database" will allow you to open a previously created composition database and "Make a DataBase" opens a file for you to add compositions to (compositions may be added or removed from files opened either way). Before entering a composition manually in the columns at the left, it is a good idea to click the "Zero All" button to clear out the previous data. Data may become confused if some numbers are changed and the totals become ridiculous (even temporarily).
This dialog permits use of any or all of the ten spectral displays to plot out most of the physical and mathematical quantities used by DTSA. The choices are in a scrollable window and clicking once on any item will cause a brief description of the choice to appear at the top of the dialog window. The most important item is the final composite spectrum without counting noise, item 100. This is the spectrum with no counting noise present and will always be written into WORK. To add the correct amount of noise to this spectrum go to the MATH menu and "Add Poissón Noise". This will add the correct noise for the Faraday current, time, detector area etc. that you specified in the previous menu (putting the result of this operation into RESULTS).
Some of the options are plotted versus atomic number - these options have nothing to do with the composition entered in the "Generate" window. There are also some options that are a function of the beam overvoltage (U) - these are valid for only ONE element.
Choose from a variety of x-ray cross sections, all from the current literature or cited in current literature. In general, the user should not use other than the default settings.
However, for those users who desire to obtain the most accurate answers using the "standardless" analysis ZAF procedure, it is possible to "fine tune" the cross section multiplier Scaling Factors for a particular SEM/Microprobe configuration. The K multiplier can probably be left at 1.00, but the values for L and M can be changed to produce better standardless answers. If it is required to deviate significantly from 1.00 then something else is probably at fault.
To experiment with these numbers, place actual spectra, taken under known conditions and containing various mixes of K, L and M x-ray families, in the scratch spectra (1-8), and then try to generate the same spectra. The default continuum cross sections for both the "bulk" and "thin" cases are quite accurate. If the magnitude of the generated continuum is different than that from the real spectra, then one or more instrument parameters are probably not what you think they are. The most likely candidates are the distance from the specimen to the detector, or the Faraday current.
For Detector Parameters, See Experiment Header in Headers menu.
There are two radio buttons, EDS and WDS that will control the shape and width of the generated x-ray lines. For a discussion of the underlying theory see Appendix V.
Last Generated into work Spectrum brings the last spectrum generated and puts it into WORK. This feature is useful when a series of mathematical operations have been performed and the original generated spectrum was inadvertently erased.