The Main Control Window is below the Spectrum Display window and
controls most of the screen functions of DTSA.
Contents of Chapter.
The Main Control Window is below the Spectrum Display window and
controls most of the screen functions of DTSA.
The cluster of controls in the bottom left corner perform either an often needed function, like Do a Fit, or Save Work or will replace the main control window with another control window that will reconfigure the purpose of the mouse when it is in the spectrum (upper) window area or some other more specialized function.
The calibrate feature permits DTSA to adjust its gain and zero to agree with the calibration of a group of spectra being imported from another environment. This calibration is in addition to the eV/Channel value that you have entered in the EXPERIMENT HEADER and is intended only to fine tune DTSA. Coarse calibration of both eV/Channel and offset can be done when the spectra are imported.
It is necessary that at least one spectrum in the group have two peaks that are well separated in energy with good counting statistics. The Calibrate dialog has several predefined possibilities, pure Cu, Al-Fe, Si-Fe and Al-Cu and one where the user may choose any two peaks. To use these buttons, it is necessary to have in WORK the appropriate spectrum. Calibration will occur automatically when the button is pushed. A least squares procedure is used to do the centroid determination of each peak. It is important, when using these predefined buttons, that the spectrum to which the calibration is being applied, not be more than about 100 eV from the true value for any peak in the spectrum or the Least squares may fail to lock on to the peak.
When the predefined buttons are used, it is not necessary to enter values in the four boxes at the bottom, as this will be done automatically.
It is possible to use pairs of peaks other than the predefined ones. To use other pairs, select the Pick Lines button. The periodic table appears allowing the user to select any K, L, or M line. Alternatively, run SIMPLEX procedure manually for each peak in the pair before invoking this dialog, and then fill in the four boxes with the requested information. You will find the book energies in the X Ray Database under the Math Menu.
The three radio buttons on the right will select which calibration data to use and the Active Slope and Active Intercept show the values for the selection. The calibration may either be off (slope = 1, intercept = 0), or on with the data from the boxes at the left or with the values stored with the spectrum. Good calibration data may be stored in the DTSA file by clicking "Calibration values to Work". As you switch from one radio button to another, the peak markers on the screen are moved to agree with the selected calibration values.
The CALIBRATE feature was not intended to make large corrections. It will "correct" the calibration if either or both of the high and low energy peaks are no more than a quarter of a peak width away from where they should be. The closer the mechanical calibration is then the less work the calibration feature will have to do and the more meaningful will be the eV per channel value.
Multiple Linear Least Squares (MLLSQ) is a curve fitting procedure that determines the counts in families of characteristic lines in a spectrum by scaling reference distributions to the data. In the DTSA program these reference distributions or References are usually obtained from well-characterized acquired spectra with very good statistics that have been background subtracted. A Reference extends over an energy range from the low side to the high side of a complete family of characteristic lines. See the section MLLSQ References for information on making, saving and manipulating references. You will need a Reference for each family of lines you wish to measure, and for any family that has a line that overlaps a family you wish to measure. The actual fitting region for a bundle of lines is defined by the lowest and highest energies among the references required for fitting the bundle. Along with the family area and fitting standard deviation, the ratio of family peak area to 10 eV of background at the known energy of the major line in the family, and the ratio of family peak area to a chosen region of peak-free background are reported. If no chosen region of background is selected in the Set-up, the integrated background counts in the energy range 1.35 to 1. 45 keV are used in the ratio.
I. Display a spectrum from the file you wish to fit.
II. Mouse click ROI: select "PEAK WIPE" and, while holding down the mouse button, sweep the cursor over each bundle of peaks to be fit. Click "QUANT BACKGROUND" and indicate the chosen region of background in the same way as the peak ROIs. Then click "OK". ROIs may also be set automatically in 1. below.
III. Select "MLLSQ" to call up the MLLSQ Set-up dialog. When first doing a Set-up the buttons are best selected in the order presented. Once a Set-up is complete, you may choose and change Select Derivatives, Output Options or Save Set-up at anytime. EACH time you OK from Select Elements you must complete a new Set-up. "Use ML Set-up File" is useful only if you have completed a Set-up and saved the information to a file.
1. Click on "Select Elements" to display the periodic table dialog. Mouse Click each element that has any line in any Peak ROI you have selected. The program sorts the families that fall into each ROI. You will need a Reference for each set of lines in each ROI for each selected element. The procedure uses only that part of the reference within the energy range of the ROI. The fitting may not be successful even with derivatives if the spectrum header does not have the correct values for energy slope and energy intercept.
Click "OK" to return to the Set-up dialog.
2. Click the "Get References" button. From the menu presented, select a file of references to search. After searching the Reference file, the program will tell you which references have not been found and, if necessary, again will present the file selection menu. If all the required References are not found in five files, you will be instructed to save the required references into five or fewer files.
3. "Select Derivatives" brings up the derivative dialog. Select "Clear All Derivatives" unless you must fit a bundle of peaks with a very low count family overlapped by a very high count family and your resolution (click "1st Derivative") or energy calibration (click "2nd Derivative") is not well known. For each bundle that meets the derivatives criteria, type in the high count element atomic symbol and family name (K K, U M, PbL, etc.), click the required derivatives, and select "ACCEPT". Click "OK" to retire the dialog. You may change the derivative information at anytime.
4. "Output Options" brings up a dialog that allows you to name and save various types of files of fitting results. "Name binary DTSA file" brings up a dialog to input a name for the binary file of results that is always saved. The default binary filename is ML_AddFit or ML_BatchFit. The existing default file data is lost when a new Set-up is accepted with a default results filename. Check "Full TextReport" to save and name a MicrosoftWord ASCII text file of results likewise for spreadsheet files. "OK" to create the files or "Cancel" will discard the information..
5. You may select "Save Set-up" next or you may return to the MLLSQ dialog to do this after trying a fit. A dialog to input the Set-up file name will come up. The file will be written when you OK to exit the MLLSQ dialog.
6. Select "Cancel" to reject the new Set-up data or "OK" to activate the new Set-up data.
ROI Set-up is not necessary to use a saved Set-up file. Choose MLLSQ to bring up the Set-up dialog and simply click Use "ML Set-up File". A file selection dialog will be displayed choose the file you wish to activate. You may then change the derivative information and/or select output file names. These changes do not affect the active Set-up file.
To fit only the spectrum displayed in WORK, click "Do a Fit". The results of subtracting the fitted references from the data will be displayed. To see the fitting result before writing to the results file(s), click "See Fit" and/or click "Add Fit" to write to the file(s). Once you click "Add Fit", "See Fit" will display the entire file of fitting results at any time until "Do a Fit" is again selected.
Or to fit the entire file with one spectrum displayed in WORK, select "Analysis" from the MAIN Menu and click "Fit All Work File Spectra". As the fitting proceeds through the file, each spectrum and the results of subtracting the fitted references from the data will be displayed. Once the 'batch' fitting is complete, to see the fitting results at any time click "See Fit". If a large number of spectra were fit, it will be impossible to see them all with See Fit. This is because the window used for displaying the results can only contain 32k characters. To see them all, call up the text file in a word processor.
Once accepted, an MLLSQ Set-up will remain active until another fitting procedure is accepted. Fit any experiment file with a spectrum displayed in WORK via "Do a Fit" or "Fit All Work File Spectra". After running a Simplex fitting procedure, to re-activate the MLLSQ fitting procedure, click "MLLSQ" and then click "OK" in the MLLSQ dialog. Remember that derivatives and output options may be changed at any time.
Note the two buttons "Flow Chart" and "Help". These two buttons contain most of the important information required to run the SIMPLEX, and will not be repeated here. The following are some general comments on using the procedure.
The Simplex procedure, like the MLLSQ procedure above, requires that you Set-up a Peak ROI (use the ROI button, main control window) that straddles the unresolved bundle(s) you wish to unravel, before entering the SIMPLEX dialog (or the auto-ROI inside the Choose Elements window may be used). If you are setting up to do a biological microanalysis, then also Set-up the "Quant Background" ROI at the desired position. The default value will be a narrow window centered at the energy of aluminum Ka, which is a good place for many types of biological analysis in an analytical electron microscope.
As with any non-linear curve fitting procedure, it is very important to start as close as is reasonable to the true answer set. We have automated this tedious aspect by providing automatically the amplitudes and energies. The width estimates are calculated from your entry of detector resolution in the header of WORK. Experiment with different values of the resolution of the detector and watch the effects on speed of convergence and accuracy of fit as revealed by the residuals. The energy estimates we provide are, of course, the known energies of the x-ray lines. The Simplex procedure is therefore, captive to how well you have calibrated your spectrometer system. By this we mainly mean, how well do you know the eV/channel. The spectrum can be relatively uncalibrated in terms of "shift", i.e., the "zero" can be somewhat wrong. The "gain", i.e., eV/channel, however, must be known reasonably well, and that known value is "valid" in the SPECTRUM HEADER.
If you use the constraint option in the simplex dialog, the closer you are to the true resolution the tighter you can make the width constraints.
If peaks are separated, fit them in separate windows. This point can not be over emphasized. Separated peaks are mathematically independent and it is mathematically naive to add degrees of freedom to any fitting procedure when it is unnecessary. An exception to this exists for some L-peaks. If the L-beta peaks are overlapped so that the largest peak is a wide and distorted composite peak, then these peaks must be included with the L-alpha peaks, otherwise, the simplex will try to make this distorted peak into one peak with disastrous results.
Use the "Fit family peaks using linked areas" mainly for K lines. This feature locks the peak areas together by the known relative transition probabilities. It makes sense to do this when there are few matrix effects such as for a thin specimen in an AEM, or a bulk specimen in an SEM if there are no absorption edges from other elements under the peaks being unraveled. When we get a good set of L and M relative transition probabilities, then this feature will work well for those cases too.
When possible, include a number (a few) of channels of background at either end of the fitting window. This will obviously improve the relationship between the true background distribution and the weighted fit to the residuals.
_. (apple period) can be used to terminate a SIMPLEX fit and force display of its current position in fitting space; however, all the fitting results will be lost. Apple comma will also stop the fit but preserve the data as far as the fit proceeded.
During the fitting process, the up-down arrows on the extended keyboard will change the scale of the display. The response is a little sluggish since a lot of arithmetic is going on at the time.
The SIMPLEX procedure may also be used to determine accurate peak areas for WDS applications. Certain (but not all) cases of peak overlap can be handled. Fortunately, peak overlap is rare in WDS. In later releases of DTSA, the WDS capability of the SIMPLEX will be expanded considerably. In this version only a few basic capabilities are provided, such as fitting with a combined Gaussian-Lorentzian peak function.
Putting too small a number in for Tolerance may cause the simplex to run for a very long time. Sometimes certain spectra will also not converge. Additionally, the simplex may be halted upon reaching a maximum number of iterations (positive integer < 32768).
This button will execute the curve fit operation chosen in one of the two buttons above it. Either the MLLSQ (Multiple Linear Least Squares) procedure or the SIMPLEX procedure must have been Set up. Both procedures will place the fit residuals (the result of subtracting the fitted peaks from the spectrum) in RESULTS and a weighted polynomial fit to the residuals in Spectrum 1. The polynomial is between order 1-5 depending on the best c2 fit. The polynomial is used to calculate the peak to local background ratio. The continuum used is always 10 eV at the peak centroid.
This button will add the fit to any fit results file the user chose when setting up the curve fit procedure.
This button permits the user to view the results of a fit, either before or after an Add Fit. A scrollable text window will appear with all results from previous Add Fit operations. A known problem in this version of DTSA is that this window will only hold 32,000 characters. If this value is exceeded, then only the first entries will be seen as the rest will be off the display buffer. The chosen results files on disk, however, do not have a limit other than disk space. This 32000 character limit will be removed in a later release of the program.
The Calculator dialog permits vector (spectrum) arithmetic to be performed. There are two columns, A and B that are the calculator operands. A third column, C, is the result. By use of a radio button in each of the three columns, it is possible to choose one of the ten DTSA spectra for each column. The mathematical operator is chosen from the scrollable list. A single click on the operator will provide more information. A double click will perform the operation. Not all operations require two operands in which case column B is not used. Constants are entered into one or more of the text boxes.
When a mathematical operation requires two operands, one must be careful which of the two spectrum headers is assigned to the result spectrum. The first three items assist in making the right assignment.
This button will save the spectrum in WORK to the active file of DTSA Spectra. If a file is not open for saving then the user will be asked if one should be created. See also Save DTSA Spectrum to Disk.
These labels will use Greek notation if the Symbol fonts are installed in your system. The labels can be printed with the spectrum. The features are mostly self explanatory. By clicking as close as possible in the horizontal (energy) direction to the main a peak of a family, and to its height, a set of markers will be placed on the display in the approximately correct positions. When using this feature, it is important that the spectrum be calibrated for both gain and shift.
After the labels appear, it is possible to move them by placing the mouse cursor over any one. As the cursor passes over the marker it will turn into a "hand" cursor. If you hold the mouse button down you can move the marker up and down.
The "Energy Lock the Labels" check box, when unchecked, will permit the labels to be moved left and right over a limited distance.
The labels will follow the spectrum as it is expanded or contracted horizontally but will not move with vertical expansions and contractions of any kind.
If you wish to change the size of the labels, use the "Label Font Size" from the File Menu.
The 1 Label/Family radio button will mark only the main a peak. This is useful for elements such as silicon with only one observable peak.
The Some Labels/Family is useful for most other peaks.
The All Labels/Family will label all peaks in a family and is primarily intended to be used for WDS spectra as it will put up too many labels for most L and M peaks.
The abbreviations "esc" and "dbl", which will appear on the display, stand for escape peak and double energy peak, respectively.
"Auto Peak Options" allows the user to select an eV range for identifying peaks and a sensitivity factor (how many times the standard deviation of the background do you want a peak to be in order to be called a peak). The " Qual a File of Spectra" in the Analysis Menu will do a whole file of spectra and write the results to a spreadsheet file. The intensity number reported is the peak value from the digitally filtered spectrum. The digitally filtered spectrum may be examine by looking in Display 8. This is not the same digital filter that is used for background subtraction elsewhere in the program, but a version that enhances the ability to locate peaks.
The ROI window permits marking a group of adjacent channels in a spectrum with an internal label that will permit DTSA to perform certain specialized operations.
The labeling procedure is accomplished by "wiping" the mouse cursor, from left to right, across the desired part of the spectrum. A warning will occur if the user attempts to wipe from right to left.
After a given "wipe", the chosen region will be marked and the left and right channel numbers, and the integral counts in this region, will be added to the display stack in the middle of the ROI window.
There are four types of ROI labels.
The Scale ROI is used by the spectrum Calculator to scale one spectrum to another. See item 23 in the scrollable list in the calculator.
The "Peak Wipe" is used by the curve fitting operations SIMPLEX and Multiple Linear Least Squares as an integral part of their respective Set-up operations. See the appropriate sections for details.
The "Quant Background" (Quantitative Analysis Background ROI) is currently used only by the Hall Bio Quant quantitative analysis procedure of DTSA. See HALL Bio Analysis for details.
The "Background Wipe" is used by the Background Subtract item under the Math Menu. See Background Subtract for more details.
The display stack in the middle of the ROI window is a "sorting" stack. As new items are added, they will move to the appropriate location in the list. Clearly, P stands for Peak and B stands for Background.
The "Normalize (per Channel) " check box will transform all the integral counts in the window to a counts per one channel basis. This feature is useful for comparison purposes. Only this display is affected.
The "Auto PEAK ROIs" button brings up a periodic table window that allows the user to set peak ROIs by simply clicking on the element (after determining if a K, L or M peak ROI is to be selected). Peak ROIs may also be set at any time from the keyboard. On an extended keyboard, holding down the keypad "del" key and typing the element symbol will set an ROI for that elements K peaks. Likewise, holding down the "end" key while typing the element symbol will set an ROI for the L peaks, and holding down the "page down" key while typing the element will set an ROI for the M peaks. To clear all peak ROIs, hold down the "clear" key on the number keypad and type the "=" key next to it.
Putting the Mouse cursor on Work -> and holding the button down allows the user to select which of the displays the numbers to the right refer to. If several spectra are being compared, you may put them in different Displays and switch from one to another with this function.
These copy operations will move both the spectrum and its header.
DTSA has ten Spectra available to the user. Each spectrum is 8192 channels and each channel is a single precision (4 byte) floating point number.
This button is used to copy a spectrum from any one of the ten possibilities to any other. The ten spectra are called WORK, RESULTS and 1 through 8. The numbered spectra are mainly convenience (scratch) spectra.
The Copy Rslt button is similar in effect to the preceding feature.
This group of radio buttons is located in the center of the Main Control Window. Each button will determine the action of the mouse cursor when it is clicked in the Main Display Window.
If either Expand Horiz, or, Contract Horiz are on, then the display will expand or contract horizontally about the click point by factors of two for each click.
If either Expand Vert, or, Contract Vert are on, then the display will expand or contract vertically about the click point by factors of two for each click. In this case, when it is desired to remove the differential vertical expansion, it will be necessary to push either the Vertical Scale Increase or the Vertical Scale Decrease Arrow icons to the right of the above controls. See Spectrum Scaling Controls. The vertical expansion feature is primarily useful for energy loss spectra (EELS) or certain spectra which can be "generated". See Thin Target Spectrum and Bulk Target Spectrum, subsection Output Options.
This button will rotate the contents of Work and Results. It is important to remember that this operation will not rotate the headers of the spectra.
When a DTSA file of spectra or a file of Reference Spectra is opened for reading by DTSA, the Select Spectrum cluster of buttons becomes active.
This cluster will be found just right of center in the Main Control Window. When the cluster is active, the arrows will go from black to blue. The arrows increment either direction through the spectra stored in the file. If there is a large number of spectra in the file the use of the arrows may become tedious. The first spectrum can be instantly accessed by pushing the Fst button and the last spectrum by pushing the Lst button. To access intermediate Spectra, push the Scan File button. A scrollable list is loaded by culling the spectrum headers of all spectra in the file and loading selected information in the list. Single click on the items for more information. Double click to read the spectrum into Work.
The KLM markers are manipulated by the following cluster of controls and by the four "arrows" on the keyboard.
Clicking the Off button will turn the markers off and change the Off button to an On button and vice-versa. This feature is useful when photographing the screen, or printing.
By clicking on the Peaks box, the following "popup" menu appears:
For the atomic number markers, the arrows on an extended keyboard are functional. That includes the up and down arrows which will move the markers up and down. Hold the option key down and the up and down arrows will change the height of the markers. Markers can be positioned by typing the atomic symbol or number. For single letter or digit atomic numbers, use the space bar either before or after to pad to two spaces.
The relative heights of the markers are as realistic as is possible. This is particularly true of the escape peak markers which are correctly scaled to the height of the "parent" marker. If the escape peak is less than one pixel high, then it will not be observable. We have not made the escape peaks artificially high.
The double energy option is a simple doubling of the energy of the main alpha peak in each family. The height of the double energy peak is arbitrary and has no meaning.
Note that the marker heights will change to the appropriate scale if "Log", or "Sqr Root" are chosen for the display in the following section.
These controls are located in the bottom right of the main control window and are responsible for both vertical and horizontal scaling of the main display window. All of these options affect only the display and not the data in memory.
If the Auto Scale is clicked twice, it will set that vertical scale expansion as the default. The vertical elevator bar will scale the spectrum between the factor of two available with the vertical arrow icons. The hatched part of the control has ten times the effect as the arrow part of the control. The horizontal controls are the group of buttons and horizontal elevator bar at the very bottom of the display. If the scale is differentially expanded horizontally then this expanded scale can be moved left and right with the elevator bar. The hatched part of the control has ten times the effect as the arrow part of the control.
Clicking on the scale button which is highlighted (1K in the figure) will automatically restore any differential expansion to the full range display for that scale.
If Auto Scale is turned off, the vertical scale will be adjusted to a round number. Clicking the arrows up or down will change the scale by a factor of two.