Quantifying X-Ray Fluorescence Data Using MAPS

Synchrotronbased X-ray fluorescence is an important technique for observing elemental segregation, stoichiometry relationships and clustering behavior in samples from a multitude of fields including biology, chemistry and material science. The information obtained from these studies is qualitative until proper quantification procedures are used to convert raw fluorescence counts into elemental aerial masses. This video will demonstrate how to use the quantification program created by Argonne National Laboratory to generate numerical information for two-dimensional X-ray fluorescence maps.

To use the MAPS program, it is first necessary to download the IDL software from the Internet. This can currently be done by going to the website for IDL and creating an account. Next, select My Account then Downloads, and it will show a page of all the available programs.

Scroll down and select the most recent version of IDL. Next MAPS may be downloaded from the Argonne National Laboratory website. After downloading and extracting the zip folder, there should be four files.Compound.

dat, henke. xdr, maps and xrf_library.csv. The three files other than maps should be copied and pasted in the IDL sub-folder called lib.

For Windows computers this most likely can be found within Program Files under the folder Exelis. Generally it is convenient to run the fitting from the desktop, however it is critical that the folder name and the pathway do not contain any spaces or special characters otherwise MAPS will produce an error when attempting to run the fitting. If the pathway to the desktop does contain spaces, place the folder elsewhere.

For example, directly within the C drive. For this demonstration I will place the Fitting Folder and the MAPS. sav icon on the desktop for easy access.

Within this folder place the files maps_fit_parameters_override. txt and maps_settings.txt. Examples of these files are made available in the supporting documents.

Next create a folder named mda and paste the chosen high resolution map file that will be used initially for the fitting. Files for the standard fitting are also added and should include either one or four files depending on the number of detector elements used by the sector. These files denote the standard.

If the AXO standard was used, then the file should be named axo_std. mca, otherwise if the NIST or any other standard was used, it may be named anything ending in mca as these files will be selected later on. Then for a four-quadrant detector, the standard and fit_parameter files should be named like so, ranging from mca0 to mca3 and txt0 to txt3 and including one fit_parameters file ending in txt.

Next, check the maps settings file that it is using the correct number of detector elements. In the case of this fitting, one detector element was used. With the fitting folder prepared, open MAPS and change the directory to be the folder just created on the desktop.

Then click okay and go to configuration. The configuration window has a variety of features that set the parameters for the fitting. Firstly, select the beamline that is representative of the beamline used for the measurements.

If the measurements were taken at Argonne National Laboratory, then the beamline should directly correspond. Otherwise additional information on which selection to use is included in the manuscript. Next select the mda file that will be used and then type in the incident energy used for the measurement.

Select start processing and wait until the program is finished. Once it is complete, go to file and then select the first option. Open XRF image average or single element.

A series of new folders should have been created by the program so select img and the fit files or h5 files generated should be located in this folder. Select the file corresponding to the map and then change the second drop-down menu from the left to be the normalization. In this situation, the data is normalized to the upstream ion chamber or USIC.

Selecting viewing, multi element view will produce images for the individual elemental channels. The units are now in micrograms per centimeter squared. But the values are not yet representative of the fitted quantities.

To work on the fitting, instead the data is viewed as a sum of all the spectra from each pixel of the map which can be viewed by going to viewing, plot integral spectrum. Next go to generate output, export raw integrated spectra series long to save the image. Close the window and go to tools, spectrum tool, load spectrum.

Locate the file that was just exported to the output folder. Generally sorting the output folder by date modified is the quickest way to find files as each new fitting will update the files within the folder. The exported spectrum will be named intspec followed by the beamline and scan number then txt.

Once the spectrum is opened, open the maps_fit_parameters_override file. First check that the number of detector elements is correct. Next in the elements to fit line, include all the elements expected to be in the sample.

Notice that L line elements and M line elements include the suffix _L or _M accordingly. Here copper is known to be in the sample but it will be excluded to provide an example of an incomplete fit. Scrolling down, enter the incident energy for coherent scattering energy.

Then in the subsequent two lines, enter a maximum and minimum energy range for the program to use as bounds. Generally a range of plus and minus 2 to plus and minus 5 keV is sufficient. Further down, check that the max and min energy to fit is inclusive of the elements energies of interest.

Additionally, check that the line detector element has the correct number to correspond to either germanium or silicon detector. At the bottom of the file, there is the ability to change the names for the detector channels that are used in the fitting. More information on how to change them is described in greater detail in the manuscript.

After making changes, save the document. Then select analysis, fit spectrum, and a window will pop up. At the top, the energy range for the fit may be set as well as the number of iterations used for the fitting.

After changing the range, select the third of the four buttons at the bottom and the program will run a fit. From the spec tool window, there is a series of drop-down menus which allow for the visualization of different curves. In the drop-downs, set one to fitted and the remaining selections to none.

On the bottom left, selecting add element allows the user to search through the spectrum for missing peaks. Using the plus sign and clicking through, it appears as though the peak missing from the fit is the copper K alpha one peak. For certain peaks, particularly towards the left of the image, the fit appears to be including the correct elements but the lines are still very far off in the proper intensity from the spectrum's intensity.

This can be improved by increasing the number of iterations. Usually at least 50 is enough to make a noticeable difference. Now returning to the fit_parameters file, adding in copper, saving, and then rerunning the fit shows that the peak is now being well fit.

After searching for all remaining missing elements, the fit looks good. In some cases, there are still some peaks that do not have the lines matched up perfectly. For example, the two peaks are on four keV which correspond to the indium Lg1 through Lg4 lines appear to have the correct element being fit, but the fitting is valuing the peak intensities higher than what was actually produced from the measurement.

This situation happens most frequently for L line elements. As K line elements have tabulated peak intensity ratios in literature, while instead the ratios of peak heights for L lines are much more dependent on the incident energy. To improve the fitting of these lines, first a line must be made in the fit_parameters file for the branching family adjustment.

These numbers denote the relative intensities compared to the literature for the L1, L2, and L3 families which are shown as the yellow, pink, and blue lines in the spec tool. Often these numbers can remain as ones or equal to the literature values. Instead the ratio for each individual line will be altered.

Preceding to the branching ratio adjustment for indium, notice that the branching ratios for the L gamma lines are all set to one. By looking at the integral spectrum, it is clear that the literature value is too high. Estimating the percentage difference between the green and white lines for each energy, then changing the branching ratio, saving and rerunning the fitting, there is an observable improvement in the fit of the green line to the white spectrum line.

Often times this process will take a few tries but it is necessary to ensure the accuracy of the fitting. After identifying the fit_parameters that produce the best possible fit, run the fitting once more at 10 or 50 K iterations. This is done because every fit updates the average resulting maps_fit_parameters_override file which will be the file that is actually implemented for the fitting.

Once the final fit is completed, close the spec tool window. Then add _input to the maps_fit_parameters file and rename the average resulting file to read maps_fit_parameters_override.txt. With this completed, return to the configuration window and select the beamline.

Then check off use fitting and copy and paste all of the mda files to be fit in the mda folder. Using select mda files, go through and highlight all of the files to be fit. The incident energy will be already entered from the fitting process.

To the right of the window, using the plus and minus symbols, click through and check off the boxes for the elements that are included in the fit_parameters file. Some elements are not included in this box. For instance, indium is not.

To include indium, mark off a box for any of the other elements that are not being fit. Then in the ROI name category, change the name to that of the element needed. Next using any fluorescence database for example the application Hephaestus, find the energy for the main energy line.

In this case, the indium L alpha one. Continue scrolling through the elements until the end also selecting S_I, S_E, S_A, TFY and background. In the top left, select right settings to config file to save the fitting settings for future use.

At this time, if the NIST standard is to be used for the fitting, select the button corresponding to the NIST standard number either NBS 1832 or 1833. And then select the file name for the standard from the parent folder. After this, the fitting is ready.

So select start processing to begin. Once the fittings are complete, they may be visualized as before by going to file, open XRF image, average or single element. And then to viewing, multi element view.

Using select elements detectors on the bottom right, it is possible to change the channels being analyzed. From this, numerical values in micrograms per centimeters square on the order of what is expected for the sample are displayed. The calculation used to estimate anticipated values is subscribed in the manuscript.

For example, shown here is the quantified data for the majority elements of a sig solar cell, copper, indium and gallium. Due to the incident energy used, the measurement was not sensitive to or able to detect the selenium peak. So it has been excluded.

From this data, it is now possible to relate the distribution of various elements within the sample to each other drawing conclusions on how the various cations of a sig solar cell distribute within a device and the degree of inhomogeneity they exhibit. The fit spectrum for each of the fit maps can also be viewed again by going to viewing, plot integral spectrum. Here one should be able to see the spectrum of the data in white and the fit in color.

This can be used to check the fit for all of the data files just to be certain that the process was applied properly to each map. Finally to export the data, go to generate output and select export, make combined ASCII files of maps. This will create an Excel file that contains the quantified fluorescence data for all of the elements being displayed.

To change or add elements, use the option select elements detectors. The data can then be found in the output folder. This video has explained step-by-step how to use the fitting software MAPS created by Argonne National Laboratory for the quantification of X-ray fluorescence data.

While the procedure is very useful for a variety of situations, there are many special case scenarios and challenges that require additional consideration. These are described in greater detail below and continual improvements are being made to further advance the accuracy of fitting X-ray fluorescence spectra. However, the programs ability to transform qualitative, high-resolution 2D fluorescence maps into quantitative spatially resolved elemental quantities provides a significant increase in the information obtainable from these measurements.

We hope this demonstration has been helpful for better understanding the process of quantifying X-ray fluorescence microscopy data. Thanks for watching.