XAS Tutorials XAS Data Analysis with Artemis (You must first install ifeffit) Loading EXAFS Data Using either Athena or PROCESS, raw XAS data are reduced to generate EXAFS data (chi vs. k). These should be in the form of a *.chi file (generated by Athena, either directly or by conversion from a *.frm file output from PROCESS) which is the input for EXAFS data into Artemis. Start Artemis (if it detects the availability of updates, download them) and from the File menu, choose Open file, find the *.chi file and select it. As part of this tutorial, the file zrdwa.chi is available from the ifeffit_tutorial_a zip archive.
Assuming the file can be read, a Graphics Window will open immediately, displaying (by default) the Fourier transform of the EXAFS data. Also, the data filename (zrdwa.chi in this case) will appear in the middle panel of the Artemis main window (called Data & Paths). Whatever is highlighted (orange background) in this panel has detailed information in the left panel of the main Artemis window. The panel on the right side of the Artemis main window (above) controls what appears in the Graphics Window. You can change the type of plot (k = EXAFS; R = FT; q = filtered EXAFS), the k-weighting of the k and q plots, the x-range plotted, and several other parameters. Changing these parameters by clicking on (radio) buttons or checking boxes causes an immediate change to the Graphics Window. Also, the Graphics Window has a menu and perhaps the most useful is the Mode | Zoom in feature which turns your cursor into a zoom window (click and drag a rectangle around the portion of the plot you would like to expand). To return to the full display, use Mode | Back. You should become familiar with the behavior of both sets of controls. Normally, we view the EXAFS data over a truncated k region and with k3 weighting. Below is the result of clicking the 3 radio button and setting kmin, kmax to 2, 13 on zrdwa.chi.
To view the FT over the same k range, use the left panel of the Artemis main window (section called Fourier and fit parameters) and set k-range [2] to [13]. You can also change the plotting range of the FT to 0-7, which is more common for our practice (these limits are in the right panel). Then click R to see the FT.
Curve fitting Now that we have the EXAFS data loaded into Artemis, we need a structural model for fitting. For ifeffit, this is a 3d coordinate set from a feff.inp file. Feff must have been run first on the feff.inp file, generating the path files, which must be in the same folder as the feff.inp file.
Now a new entry in the middle panel (Data & Paths) describes the model (rather than the data). Initially the line FEFF0 is highlighted and this is associated with a new left panel.
The Interpretation tab of the left panel shows the header of the feff.inp file (at the top) and a set of interpreted paths below. (Clicking on the feff.inp tab, displays the entire contents of the feff.inp file (for reference).) This is a very simple case: the model is just a Zn surrounded by 4 S at 2.33 Â so there is only one single-scattering path (under Scattering Path, [+] stands for the absorbing atom, so this path is from the Zn to the S back to the Zn). Back in the Data & Paths (middle) panel, the square icon just before FEFF0 toggles the path hierarchy open or closed (it is open by default). Within this hierarchy for this simple case is only the single path labeled Path 1:[S]. Clicking on the path name brings up a new left panel with the details of that path:
Displayed in the left panel for each path is the path interpretation labeling a top section with detailed structural parameters (x,y,z coordinates, potential definitions, Reff, etc.). Below this section is a table of the path parameters written as math expressions. The three parameters most often optimized are delE0, delR, and sigma^2. (If you want to see a brief description of each parameter, just click on it and some text will appear at the bottom of the window. Describing these parameters as math expressions related to other variables (e.g., enot, dr, ss) makes the optimization very versatile in Artemis. This is one of its advantages over OPT in EXAFSPAK. In the top right corner of each path’s left panel is a checkbox labeled Include in the fit. When this is checked, the path name in the Data & Paths panel is bold. This is how you can tell quickly whether a particular path has been excluded or included in a given fit. The remaining selection in the Data & Paths (middle) panel is Guess, Def, Set (often known by its acronym GDS). This is where we can control initial guesses of parameter values as well as restraints and constraints to be used in optimization. Selecting Guess, Def, Set in the Data & Paths panel brings up another left panel in which we can define our optimization plan.
Note that the list of parameters in this GDS table are not EXAFS parameters like delr, etc., but are the variables used in the math expressions for the EXAFS parameters (compare this list to the math expressions in the path interpretation). When we highlight one of these parameter records in the left panel (the record background turns yellow), we can decide how to handle it below this panel. The following are choices:
A typical simple fit would be described as shown in the above Artemis window: amp (the variable for S02) is fixed at 0.9; enot and dr are guessed as 0 (and will be floated); ss is guessed as 0.003 (and will be floated). Before starting a fit, you should check to make sure all the settings are as desired. Selecting the data file again in the Data & Paths panel (zrdwa.chi here), brings up the data panel on the left.
Now you can choose to do phase correction for your FT displays if you like by selecting a path in the Path to use for phase corrections dropdown menu. (If you click R again in the right panel, your phase-corrected FT will appear in the Graphics Window.) You should also check the kw=3 box (and uncheck any other boxes) in the Fit k weights section of the left panel. This will do the normal k3-weighted fit. Also, it is most common for us to do curve-fitting in k space (Fitting space dropdown menu). Check the k-range again (under Fourier and fit parameters) and when you are ready, click the green button labeled Fit at the top right. Often (as in this case), you will get an Artemis palettes window with the Messages tab displayed:
Once you understand how to respond to the message, you can click the Dismiss button. In this case, we have to choose a new name for the dr parameter. Click on Path 1:[S] in the Data & Paths panel, then change dr to dr_1 (for example) in the delR: field in the left panel. You must also change its GDS settings: click on Guess, Def, Set in the Data & Paths panel, highlight the dr parameter record at top left, then change dr to dr_1 in the Edit selected parameter box at bottom left. Click the Guess radio button again to update the record above. Clicking on the green Fit button now brings up the Artemis: Information about this fit window (all problems have been cleared).
Artemis provides a way to organize a set of fits for a project. You should develop a labeling system that allows you to return to any given fit later. For example, use labels that start with the filename, then are numbered consecutively (e.g., zrdwa_001). You can use the Comment line to write something in English about the fit design.
A successful fit will generate a Graphics Window with data and fit displayed (right; note your fit label is on the graph) as well as another Artemis palettes window (below) displaying a Results tab. You can compare data and fit in k space or R space and you can display the residuals if you like using the controls in the right panel of the main Artemis window. It is instructive to expand the Artemis palettes window containing the Results tab.
There is a lot of useful information in this file and you may choose to save this results file for some of your best fits. There are also other views available. If you use the dropdown menu at the top that currently says Raw log file, and choose Column view, then scroll to the bottom, you will see a useful table:
For fits that are more complicated than this one, the table at the bottom of this view is particularly helpful to organize the results. Entire Artemis projects can be saved (as *.apj files). Use File | Save project (or Save project as …), give the file the same name as the data file (zrdwa in this case) and explicitly use the .apj extension. Then, next time Artemis is run, use File | Open file to read this project file. All data, fits, etc., will be reloaded. You can output data and fit, both as (k,chi) and (R,FT) by using the Save data as … and Save fit as … under the File menu of the Artemis main window. For each, select first chi(k), then chi(R) to save the (k,chi) and (R,FT), respectively. A suggested naming convention is
These are all (x,y,…) ASCII files with header information that describes their origin and content. They contain column headers so that you can identify the data columns. They can be opened in Igor, for example. Trust the .apj file to hold preliminary fits and save these files only for the most important final fit results. scott@chem.uga.edu |