ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT

Ravel, B.; Newville, M.

doi:10.1107/S0909049505012719

computer programs

JOURNAL OF
SYNCHROTRON
RADIATION

ISSN: 1600-5775

Volume 12| Part 4| July 2005| Pages 537-541

https://doi.org/10.1107/S0909049505012719

ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT

B. Ravel ^a ^* and M. Newville ^b

^aNaval Research Laboratory, Washington, DC 20375, USA, and ^bThe University of Chicago, Chicago, IL 60637, USA
^*Correspondence e-mail: ravel@phys.washington.edu

(Received 29 September 2004; accepted 21 April 2005)

A software package for the analysis of X-ray absorption spectroscopy (XAS) data is presented. This package is based on the IFEFFIT library of numerical and XAS algorithms and is written in the Perl programming language using the Perl/Tk graphics toolkit. The programs described here are: (i) ATHENA, a program for XAS data processing, (ii) ARTEMIS, a program for EXAFS data analysis using theoretical standards from FEFF and (iii) HEPHAESTUS, a collection of beamline utilities based on tables of atomic absorption data. These programs enable high-quality data analysis that is accessible to novices while still powerful enough to meet the demands of an expert practitioner. The programs run on all major computer platforms and are freely available under the terms of a free software license.

Keywords: XAS; data analysis; FEFF; IFEFFIT.

1. Introduction

Users of X-ray absorption spectroscopy (XAS) beamlines and practitioners of XAS require software that meets a broad range of analytic, programmatic and social goals. To meet these goals we have developed a collection of programs providing a rich graphical interface to the extensive analytical and numerical capabilities of the popular IFEFFIT library (Newville, 2001 ). In this report we present the main features of the three programs: (i) ATHENA, a program for XAS data processing, (ii) ARTEMIS, a program for analysis of extended X-ray absorption fine structure (EXAFS) data using theoretical standards computed by the popular FEFF program (Zabinsky et al., 1995 ; Ankudinov et al., 1998 ) and (iii) HEPHAESTUS, a collection of beamline utilities based on tables of atomic absorption data. All three programs are written in the Perl programming language (Wall et al., 1996 ), use the Perl/Tk graphical toolkit (Lidie & Walsh, 2002 ) for their user interface, and use the PGPLOT (Pearson, 1997 ) package for plotting data.

The development of this software project has been directed by a number of equally important motivations:

Versatile XAS and numerical algorithms. ATHENA and ARTEMIS use the programmers interface to IFEFFIT, as described by Newville (2001). Thus both programs have complete access to all the features of IFEFFIT, including all the capabilities of the programs AUTOBK (Newville et al., 1993 ) and FEFFIT (Newville et al., 1995 ). In this article, any analytic functionality ascribed to ATHENA or ARTEMIS is actually performed by IFEFFIT via its programming interface.

Graphical user interface. The Perl/Tk toolkit provides a full-featured forms-based environment for interacting with and processing data.

Simple for novices, powerful for experts. Those already familiar with analysis of XAS data will find ATHENA and ARTEMIS quite easy to begin using and will quickly familiarize themselves with the many features of those programs. Novices to absorption spectroscopy find that a usable graphical user interface combined with excellent analytic algorithms considerably lowers the barrier of entry into their new field of research. For the expert practitioner of the art of XAS data analysis, ATHENA and ARTEMIS provide sophisticated capabilities allowing the user to exploit deeply the information content of their measured data.

Useful at the beamline, useful at home. ATHENA and HEPHAESTUS compliment the data-acquisition software found at XAS beamlines. Since ATHENA can read most raw data formats, it provides flexible powerful data-processing capabilities right at the beamline. The utilities provided by HEPHAESTUS are useful while planning and executing XAS experiments.

Free of cost, free source code. IFEFFIT, ATHENA, ARTEMIS and HEPHAESTUS are available on the web. Because this software is free of cost, high-quality XAS analysis is available even to students and to scientists in developing countries. Because the code is licensed under very liberal terms, users are free to examine the code, understand its algorithms, modify the code to better suit their research needs, and share the codes with their colleagues.

Portability. ATHENA, ARTEMIS and HEPHAESTUS work on most common computer platforms including all versions of Microsoft Windows, Macintosh OSX, Linux, BSD, Solaris and other common unixes. This is possible because all of the tools upon which they rely, IFEFFIT, Perl, Perl/Tk and PGPLOT, themselves operate on all these platforms. Additionally ATHENA and ARTEMIS save data to project files which are easily transferable between users, even those working on different platforms. These project files capture the state of the program, all data and FEFF calculations, and all parameters associated with the data and FEFF calculations. Reimporting a project file thus restores the state of the data analysis project, even on a different computer.

At the time of this writing, ATHENA, ARTEMIS and HEPHAESTUS are employed by a community of several hundred users. A vigorous electronic mailing list is devoted to discussion of the use of these programs as well as to the science and practice of XAS.

2. Features of ATHENA

ATHENA is used for all the main steps in data processing including conversion of raw data to $[\mu(E)]$ spectra, background subtraction, Fourier transforming and plotting. As such it is useful for data visualization at the beamline as well as for processing of data in preparation for data analysis. The normal view of the ATHENA window is shown in Fig. 1.

Figure 1
The main window of the ATHENA program. The data sets imported into ATHENA are listed on the right. The analysis parameters for the selected data set are displayed in the main panel on the left.

Data conversion from its raw form is handled in a flexible manner, creating $[\mu(E)]$ spectra from transmission, fluorescence or electron yield experiments. The conversion utility allows for on-the-fly summation of data from a multi-element detector. Alternately, each column of multi-element data can be imported as individual data groups for later inspection and summation. ATHENA also supports powerful preprocessing of data as it is imported, including automated deglitching, truncation, alignment and constraint of analysis parameters.

Background subtraction is performed using the AUTOBK algorithm (Newville et al., 1993) which determines an empirical background spline based on a distinction between data and background in terms of Fourier components. Edge-step normalization of the data is determined by a linear pre-edge subtraction and regression of a quadratic polynomial beyond the edge. The difference between these two polynomials extrapolated to the edge energy E₀ is used as the normalization constant in the definition of $[\chi(E)]$ = $[[\mu(E)-\mu_0(E)]/\mu_0(E_0).]$ Normalized data are typically presented after subtracting the curvature of the regressed quadratic and the difference in slope between the post- and pre-edge polynomials after the edge. $[\chi(k)]$ are obtained by conversion of the abscissa of $[\chi(E)]$ by the formula k = $[(2m_{\rm{e}}/\hbar^2)(E-E_0)^{1/2}.]$ E₀ is determined automatically as the data is imported by finding the first large peak in the first derivative of the $[\mu_0(E)]$ spectrum, but can also be set interactively to the tabulated atomic value, to the position of half the edge step height or to any arbitrary value selected by the user.

ATHENA provides utilities for most common chores involved in the preparation of EXAFS data for analysis. There are tools for energy calibration of spectra, alignment of data, use of reference spectra, deglitching (i.e. the removal of spurious points from spectra) and merging of data by calculating the average and standard deviation at each point in a set of spectra. There is a utility for performing log-ratio/phase-difference analysis (Bunker, 1983 ) of back-transformed spectra. ATHENA also provides a tool for the energy calibration of dispersive XAS spectra (Hagelstein et al., 1997 ).

ATHENA provides tools for the analysis of near-edge spectra. There are utilities for calculation of difference spectra, for fitting line shapes (arctangent, Gaussian, Lorentzian) to spectra, and for fitting linear combinations of standard spectra to unknown spectra. ATHENA can also import and display the results of FEFF8 calculations (Ankudinov et al., 1998) for comparison with measured data.

The most powerful feature of ATHENA is that processing and displaying many data sets simultaneously is as simple as for a single data set. In Fig. 1, the upper row of buttons labeled with the letters E, k, R, and q are for plotting single data sets in energy, k-space, R-space or back-transformed k-space, respectively. The lower row of buttons will plot multiple data sets. It is simple to mark any subset of imported data sets for plotting in this manner. ATHENA keeps track of all changes made to the parameters of any data set and will automatically perform background removal and Fourier transforms as needed. The parameters controlling background removal and Fourier transforms can be adjusted for each data set individually or constrained in a flexible manner between data sets.

ATHENA comes with extensive documentation accessible within the program and a number of examples demonstrating its many features. Features which are planned for implementation in ATHENA include principle component analysis, self-absorption corrections for fluorescence data, and dead-time corrections for data measured with solid-state detectors.

3. Features of ARTEMIS

ARTEMIS works within the framework of FEFF's multiple-scattering path expansion (Zabinsky et al., 1995; Rehr & Albers, 1990 ). This means that the data are described as a summation of one or more scattering paths as computed by FEFF. For each path, the XAFS equation must be evaluated. This means that values of the parametric terms in the EXAFS equation, N, S₀², E₀, $[\Delta{R}]$ and $[\sigma^2]$ , must be specified for each path included in the fit. The path expansion and how it is parameterized and used in a fit is described elsewhere (Newville et al., 1995; Ravel, 2000 ). As discussed in those references, EXAFS analysis can be quite subtle and complicated. By providing a graphical forms-based mechanism for handling the many bookkeeping chores required by FEFF and IFEFFIT, ARTEMIS makes basic analysis chores simple and sophisticated analysis problems tractable.

Shown in Fig. 2 is the ARTEMIS window in its data view. The data-processing parameters, including the range of the Fourier transform from k-space and the fitting range in R-space, are set in the main window. The far right panel is used to control how plots are displayed. The middle panel is a list of all data sets, FEFF calculations and paths used in the fit. This list is integral to the generation of plots and is used to determine which view is displayed in the main window. Other views (not shown) allow the user to set fitting parameters, to examine log files from previous fits, to manage FEFF calculations or to manage individual FEFF paths and the mathematical expressions used to evaluate the terms in the EXAFS equation.

Figure 2
The main window of the ARTEMIS program. Shown in the main panel is the view for setting the Fourier transform and fit range parameters. The middle panel lists all data sets and FEFF paths used in the fit and provides a history of all fits performed within the project. The panel on the right contains plotting controls.

ARTEMIS includes access to the functionality of the ATOMS (Ravel, 2001 ) program, which converts crystallographic data into a format suitable for FEFF. This interface can read crystallographic data either in the form of an ATOMS input file or a crystallographic information file (CIF) (Hall et al., 1991 ). These data are then converted into a FEFF input file or can be converted into a form suitable for export to a molecule viewer. For materials that are not crystalline or do not have a crystalline analogue, FEFF input files prepared in some other way can be imported directly into ARTEMIS.

A simple interface to FEFF is included in ARTEMIS. This is little more than a simple text editor for altering the contents of the input file as needed and a button for launching an external instance of FEFF. Fortunately, this simple interface is quite adequate for the needs of EXAFS data analysis. After the FEFF run, ARTEMIS displays a page containing a concise interpretation of the paths computed by FEFF including the path distances, degeneracy, atomic species in the path, whether the path is single- or multiple-scattering, and estimated importance of the path. ARTEMIS has tools for organizing and plotting the results of the calculation, including plotting the individual contributions of each path in both k- and R-space.

ARTEMIS has several features which facilitate the creation and examination of fitting models. One such is the automated generation of parameters appropriate for a first-shell fit. When a FEFF calculation is made or imported, a set of fitting parameters are automatically generated. In many cases, a fit can be run immediately which will give a physically reasonable result. Because all fitting problems are unique, this fully automated fit is usually incomplete, but it provides a sensible starting point for almost any fitting problem from which the user can begin to refine the fitting model. ARTEMIS provides convenient plotting in k-, R- or back-transformed k-space of the data, the fit and any number of the individual paths used in the fit. Numerous tools are available to check the consistency and sensibility of the parameters used in the fit and to evaluate the results of the fit. For instance, if one of the parametric terms in the EXAFS equation for any path evaluates to a suspicious value (e.g. a negative $[\sigma^2]$ , a very large energy correction E₀ or a very large change in path length), ARTEMIS will issue a warning message explaining the possible problem to the user. Warnings about large changes in path length or E₀ are particularly important as they might indicate that an inappropriate structural model was used in the FEFF calculation or that the fit has settled upon an unphysical solution (Michalowicz & Vlaic, 1998 ).

We acknowledge that there is a potential limitation of using FEFF calculations with an assumed distance in the analysis. That distance cannot be refined too far from its starting value (typically, less than 0.1 Å^-1). Outside that range the scattering amplitude and phase from the FEFF calculation should be recalculated with the new distance. In practice, we find that this is not a serious limitation, but can cause some slight errors especially at the beginning of an analysis. This is one of the reasons that ARTEMIS issues a warning about a distance that is too different from that used in the FEFF calculation. To overcome this limitation, the analysis package would have to recalculate the scattering contributions whenever the path distance changes too much. This is an area for future development.

After a fit finishes, ARTEMIS displays a log file which includes a statistical analysis of the fit. The major fitting statistics, including the reduced $[\chi^2]$ fitting metric and the $[\cal{R}]$ -factor are displayed along with uncertainties in and correlations between all fitting parameters. Furthermore, the parameteric terms of the EXAFS equation are evaluated for each path used in the fit. This information along with easy graphical display of the paths can be used to evaluate the importance of each path to the fitting result.

While ARTEMIS is certainly well suited to simple first-shell analysis of single data sets, it is a very powerful tool applicable to the most challenging fitting models. Among its advanced features are:

(i) Simultaneous refinement of multiple data sets.

(ii) Refinement using multiple k-weightings for any data set.

(iii) Use of multiple FEFF calculations in a single fit.

(iv) Co-refinement of a background spline to reveal correlations between the spline and the fitting parameters.

(v) Arbitrary constraints between and restraints on parameters.

4. Features of HEPHAESTUS

HEPHAESTUS is a container application for several small utilities based on the periodic table of elements and on a database of absorption cross sections and other data for the elements. All calculations made by HEPHAESTUS use one of several published compilations of atomic cross sections. The default is that of Elam et al. (2002 ). The other sources of absorption data are by McMaster et al. (1969 ), Chantler (1995 ), Cromer & Liberman (1970 ) and Henke et al. (1993 ). It is easy to switch between the data resources to compare their results for common calculations.

Currently there are seven utilities in HEPHAESTUS. In the same order as the icons in the button bar on the left side of Fig. 3, they are:

Figure 3
The HEPHAESTUS program, showing the absorption and fluorescence line energy utility. The various utilities provided by HEPHAESTUS are accessed by the button bar on the left.

(i) Display the absorption and fluorescence line energies of the elements using a periodic table interface.

(ii) Compute the absorption length and quantity of sample required for measurement given the chemical formula and density of a material.

(iii) Display chemical data such as melting point and electronegativity of the elements using a periodic table interface.

(iv) Compute the ideal gas mixture for ion chambers of various lengths and internal pressures.

(v) Display a chart showing the electronic transitions associated with the various absorption lines.

(vi) Find all absorption edges in the vicinity of a given energy or of the harmonics of that energy.

(vii) Find all fluorescence lines in the vicinity of a given energy.

(viii) Plot the real and imaginary parts of the complex scattering factors of the elements using a periodic table interface.

HEPHAESTUS includes extensive built-in documentation and a tool for converting between energies and wavelengths.

5. ATHENA and ARTEMIS as teaching tools

As discussed in §1, one of the motivations for these programs is to provide an entry into data analysis for students and non-experts in XAS. In our own teaching experience we have found these programs to be good pedagogical tools with relatively shallow learning curves. With plotting and data handling tightly integrated into a graphical user environment, users are freed from such minutiae as deciphering raw data formats, constructing $[\mu(E)]$ spectra or performing Fourier transforms. The user can arrive quickly at the point of considering the scientific content of her data.

Analysis of EXAFS data using theoretical standards from FEFF can be a daunting task. Along with processing the measured data, the user must run the FEFF program, organize its output and construct a fitting model using the contributions from the various scattering paths generated by FEFF. In some cases, tens or even hundreds of these paths might be required for adequate data analysis. ARTEMIS organizes all these tasks for the users. It integrates wrappers for ATOMS and FEFF and helps organize the paths from the FEFF calculation. The time for learning FEFF-based data analysis for new users is shortened and complex analysis is facilitated for experienced users. As an example, ARTEMIS makes it simple to plot the data along with the contributions from any number of scattering paths in k- or R-space, allowing the user to quickly evaluate which parts of the FEFF calculation are relevant to the data analysis. An extensive log file is presented upon completion of the fit, allowing the user to evaluate the quality of the fit, the reasonableness of the fitting parameters and the effects of the fitting parameters on the FEFF paths used in the fit.

6. Free software and the scientific method

Full disclosure of experimental procedures and results are integral aspects of the scientific method. Because scientific work should be independently reproducible, sufficient detail of sample preparation, experimental procedure, theoretical background and interpretation of the data are normally required for most scientific publications. For data such as XAS that are interpreted with the aid of complex computer programs such as those described here, we believe that the algorithms used in the analysis be also disclosed fully, both for the purposes of review and improvement. The source code for the programs described here are freely available and distributed under a free software license (The Open Source Initiative, 2004 ) allowing use, modification and re-distribution of the source code with very few restrictions. All users are allowed to study, modify and improve these programs to meet their own scientific needs and to disseminate these changes to others. Because of this license, we believe that these programs make a suitable basis for study and development of new EXAFS algorithms.

7. Resources

There are many examples in the XAS literature demonstrating the use of the software and analysis principles described in this paper. Some articles that demonstrate excellent analysis and good practices for reporting both the parameters used in the analysis and the results of the analysis include these references: Kelly et al. (2002 ) and Kolobov et al. (2004 ).

A project page for IFEFFIT exists at SourceForge.net, https://sourceforge.net/projects/ifeffit/. Information about IFEFFIT can be found at https://cars.uchicago.edu/ifeffit. A link to the IFEFFIT mailing list can also be found there. ATHENA, ARTEMIS and HEPHAESTUS have their own web pages, https://FEFF.phys.washington.edu/~ravel/software/exafs/. ATHENA, ARTEMIS and HEPHAESTUS as well as IFEFFIT are being actively developed.

The source codes for ATHENA, ARTEMIS and HEPHAESTUS are available at the web sites mentioned above. The codes are also packaged in ways that facilitate their installation on the various computer platforms. Automated installer packages are available for Windows, Macintosh OSX and Debian Linux. The source code packages include build scripts which make them easy to install on any other Unix-like platform.

Acknowledgements

We would like to thank S. Kelly and everyone on the IFEFFIT mailing list for their enthusiasm and support for this project. D. Barton contributed code for importing CIF files. Versions 0.8.038 of ATHENA, 0.8.0 of ARTEMIS and 0.08 of HEPHAESTUS are shown in this paper. Newer versions will, no doubt, be available by the time of publication. The screen shots in this article were made on a GNU/Linux computer with a KDE desktop. The programs may be different in appearance on other systems.

References

Ankudinov, A. L., Ravel, B., Rehr, J. J. & Conradson, S. D. (1998). Phys. Rev. B, 58, 7565–7576. Web of Science CrossRef CAS Google Scholar
Bunker, G. (1983). Nucl. Instrum. Methods, 207, 437–444. CrossRef CAS Web of Science Google Scholar
Chantler, C. T. (1995). J. Phys. Chem. Ref. Data, 24, 71. CrossRef Google Scholar
Cromer, D. T. & Liberman, D. (1970). J. Chem. Phys. 53, 1891–1898. CrossRef CAS Web of Science Google Scholar
Elam, W., Ravel, B. & Sieber, J. (2002). Radiat. Phys. Chem. 63, 121–128. Web of Science CrossRef CAS Google Scholar
Hagelstein, M., San Miguel, A., Fontaine, A. & Goulon, J. (1997). J. Phys. IV (Colloque), 7(C2), 303–308. Google Scholar
Hall, S., Allen, F. & Brown, I. (1991). Acta Cryst. A47, 655–685. CrossRef CAS Web of Science IUCr Journals Google Scholar
Henke, B. L., Gullikson, E. M. & Davis, J. C. (1993). Atom. Data Nucl. Data Tables, 54, 181–342. CrossRef CAS Web of Science Google Scholar
Kelly, S., Kemner, K., Fowle, D., Boyanov, M., Bunker, B. & Yee, N. (2002). Geochim. Cosmochim. Acta, 66, 3855–3871. Web of Science CrossRef CAS Google Scholar
Kolobov, A., Fons, P., Frenkel, A., Ankudinov, A., Tominaga, J. & Uruga, T. (2004). Nature Mater. 3, 703–708. Web of Science CrossRef CAS Google Scholar
Lidie, S. & Walsh, N. (2002). Mastering Perl/Tk, 1st ed. Cambridge: O'Reilly. Google Scholar
McMaster, W. H., Grande, N. K.-D., Mallett, J. H. & Hubbell, J. H. (1969). Compilation of X-ray Cross Sections. Lawrence Radiation Laboratory Report UCRL-50174. Springfield, VA: National Bureau of Standards. Google Scholar
Michalowicz, A. & Vlaic, G. (1998). J. Synchrotron Rad. 5, 1317–1320. Web of Science CrossRef CAS IUCr Journals Google Scholar
Newville, M. (2001). J. Synchrotron Rad. 8, 322–324. Web of Science CrossRef CAS IUCr Journals Google Scholar
Newville, M., Līvinš, P., Yacoby, Y., Rehr, J. J. & Stern, E. A. (1993). Phys. Rev. B, 47, 14126–14131. CrossRef CAS Web of Science Google Scholar
Newville, M., Ravel, B., Haskel, D., Rehr, J. J., Stern, E. A. & Yacoby, Y. (1995). Physica B, 208/209, 154–156. CrossRef Web of Science Google Scholar
Pearson, J. (1997). PGPLOT Graphic Library version 5.2.0, https://www.astro.caltech.edu/~tjp/pgplot/. Google Scholar
Ravel, B. (2000). EXAFS Analysis Using FEFF and FEFFIT, https://feff.phys.washington.edu/~ravel/course/. Google Scholar
Ravel, B. (2001). J. Synchrotron Rad. 8, 314–316. Web of Science CrossRef CAS IUCr Journals Google Scholar
Rehr, J. J. & Albers, R. C. (1990). Phys. Rev. B, 41, 8139–8149. CrossRef Web of Science Google Scholar
The Open Source Initiative (2004). The Open Source Initiative, https://www.opensource.org/. Google Scholar
Wall, L., Christiansen, T. & Schwartz, R. L. (1996). Programming Perl, 2nd ed. Cambridge: O'Reilly. Google Scholar
Zabinsky, S. I., Rehr, J. J., Ankudinov, A., Albers, R. C. & Eller, M. J. (1995). Phys. Rev. B, 52, 2995–3009. CrossRef CAS Web of Science Google Scholar

© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.

JOURNAL OF
SYNCHROTRON
RADIATION

ISSN: 1600-5775

Volume 12| Part 4| July 2005| Pages 537-541

https://doi.org/10.1107/S0909049505012719

Format		BIBTeX
		EndNote
		RefMan
		Refer
		Medline
		CIF
		SGML
		Plain Text
		Text

Format		BIBTeX
		EndNote
		RefMan
		Refer
		Medline
		CIF
		SGML
		Plain Text
		Text

Search IUCr Journals		doi		Advanced search
Author		volume	page

computer programs\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT

1. Introduction

2. Features of ATHENA

3. Features of ARTEMIS

4. Features of HEPHAESTUS

5. ATHENA and ARTEMIS as teaching tools

6. Free software and the scientific method

7. Resources

Acknowledgements

References

computer programs