N I S T Center for Neutron Research
Accomplishments and Opportunities 2001
E X P G U I: Smoothing the Path to Powder Crystallography
Back in the dark ages of computational crystallography, crystallographers were expected to be able to modify their computer programs to suit the computation at hand. Most crystallographers could create programs to convert data between different formats, and felt comfortable reading through a program’s source code when the program documentation did not adequately explain a topic. (There is disagreement on when this ended, some say the 1950s, others say the 1980s or even the early 21st century!)
The advent of desktop computing has also introduced a new paradigm of computing — the graphical user interface (G U I). Visual analogies to physical devices (such as switches) make operating a G U I intuitive; a good G U I implementation significantly reduces learning time.
Simultaneous with the changes in computing, crystallography has evolved from a technique used exclusively by specialists to one of many tools in the arsenal of chemists, physicists and engineers. To facilitate this, the N C N R is developing G U I - based software for crystallography. There are two goals: to increase the utility of the software for the non-specialist, and to improve the productivity of experts. This allows novices to concentrate on the science and techniques of neutron powder crystallography rather than devote their efforts to learning arcane details of computer software.
The Rietveld refinement technique for crystallographic analysis was adopted within the N C N R immediately after its first publication (Refer to Reference 1). The N C N R then produced Rietveld software specifically adapted to the local instrumentation, which is still being improved (Refer to References 2 and 3). However, the General Structure Analysis System (G S A S) package of programs from Los Alamos is also widely used (Refer to Reference 4). G S A S has wide applicability. It can be used with virtually any type of neutron or x-ray diffraction data, including both powder and single crystal data. It supports a wide range of geometric and compositional restraints and constraints. While many new features have been added to G S A S on a continual basis, the original text-based user interface has not changed. Novice and occasional users often find the many levels of dialogs cumbersome and difficult to learn.
Work on a G U I for G S A S was begun several years ago, with the modest goal of providing a mechanism for accessing a small portion of the G S A S features. The project, now called E X P G U I, has grown in scope (Refer to Reference 5). Most of the commonly used powder diffraction features are now implemented, allowing a complete Rietveld refinement to be performed graphically. Features not implemented in the G U I can still be accessed via the standard G S A S user interface. E XP G U I runs identically on all platforms where G S A S is supported, including both Unix and Windows. In F Y 2001, the Advanced Technology Program recognized the value of this effort and provided funding to accelerate development.
The G U I is designed to display information and controls in a concise and intuitive fashion. An example screen, shown in Figure 1, manipulates structural parameters. This figure demonstrates how four atoms that share sites in this material can be selected together, so that their refinement options can be changed as a group. Another screen, shown in Figure 2, applies constraints that link appropriate parameters in the refinement.
Graphics Caption FIGURE 1. Sample screen from E X P G U I, showing where atomic parameters are displayed.
Graphics Caption FIGURE 2. E X P G U I screen for entering and display of atomic constraints. In this case, atoms sharing sites are constrained to have the same Uiso values and have fractional occupancies that total to unity.
Graphics Caption FIGURE 3. Graphical display of fit in E X P G U I. This plot has been “zoomed” in to show details. Several features of this plot, for example the display of reflection indices and the location of the fitted background, were not previously available in G S A S.
Recent E X P G U I development has been aimed at extending the capabilities of the G S A S package. As an example, when E X P G U I was recently given the ability to import atomic coordinates from files, support was added for C I F, the complex standard crystallographic information file format developed by the International Union of Crystallographers. Further, the import routine was written in a general fashion, so that other formats are easy to add.
Where possible, E X P G U I takes advantage of the graphical nature of modern computers to offer scientists better access to refinement tools. Figure 3 shows a graphical display of the observed and computed diffraction pattern. Through a simple set of mouse operations, the user has “zoomed in” on a portion of the data, to better see the poor agreement between observed and computed intensity values, indicating that the structure is not well modeled.
This graphical tool also allows the user to see the fitted background and to label the reflection indices — features not previously available within G S A S. Another example is a new tool for background fitting. In most cases, background fitting is nearly automatic, but for some refinements, fitting can be quite difficult. G S A S allows a scientist to define the background either through use of a function with refinable terms or as a spline drawn through a set of fixed points. A new tool in E X P G U I improves G S A S by allowing a user to define where the background curve should occur using a computer mouse. Since many experts agree that fixed background points are a poor way to treat background fitting, E X P G U I offers an augmented feature: a background function can be fit to the user’s desired background curve. The terms for this function will be sufficient to obtain an initial model with good agreement to the data. At the end, the background terms can be optimized, to obtain the best possible model. This capability was not previously available in G S A S.
E X P G U I has been widely adopted by Rietveld users. It is used in most major neutron and synchrotron facilities, as well as dozens, if not hundreds, of universities, companies and research centers around the world.
References
[1] H. M. Rietfeld, J. Appl. Cryst. 2, 65 (1965).
[2] L. W. Finger and E. Prince, National Bureau of Standards, Report No. 852 (1975).
[3] B. H. Toby, E. Prince, and J. Stalick, unpublished computer program (2000).
[4] A. C. Larson and R. B. Von Dreel, Los Alamos National Laboratory, Report LAUR 86-748, (2000).
[5] B. H. Toby, J. Appl. Cryst. 34, 210 (2001).
Authors
B. H. Toby
N I S T Center for Neutron Research
National Institute of Standards and Technology
Gaithersburg, MD 20899-8562