Symmetry-Mode (ISODISTORT) Rietveld Refinement - WO3


Files needed: wo3_pm3m.cif, wo3_p21n_80056.cif, d8_03901_030c_traditional.inp, d8_03901_030c.xy

Learning Outcomes:  This example is based on lab x-ray diffraction data from the distorted structure of room-temperature WO3, which has space group P21/n (#14) and a 2 x 2 x 2 supercell relative to the cubic ReO3 parent structure.  Because each of the atoms sit on general Wyckoff sites, there are a total of 24 displacive degrees of freedom.  We will begin by decomposing the structure reported by Woodward et al., J. Phys. Chem. Solids 56, 1305-1315 (1995) into symmetry-motivated distortion modes relative to the cubic parent.  We will then modify a working TOPAS input file to use symmetry-mode amplitudes as the refinable structural parameters.  On this basis, the redetermination of the structure is fairly straightforward.

Authors: Branton and John.

Note: depending on when you installed jedit menus "distortion mode" may be used instead of "symmetry mode" and ISODISPLACE for ISODISTORT.

1. Download the files listed above to the working directory on your local computer.  The TOPAS input file provided allows us to start with working refinement of the P21/n structure of WO3 using traditional xyz atomic coordinates -- the profile shape, cell parameters and other experimental details have already been worked out.

2. Go to the  ISODISTORT website and follow the "upload parent structure from a CIF file" link. Browse to find the "wo3_pm3m.cif" file in your working directory, click the "Upload" button, and then click "OK" to proceed to the "search" page.

3. Under the heading "Method 4: Mode decomposition of a distorted structure", upload the distorted WO3 structure file from "wo3_p21n_80056.cif" and click "OK" to proceed to the "distorted structure (basis)" page.  Here, select (2,0,0),(0,2,0),(0,0,2) from the drop-down menu of candidate basis and keep all other default settings.  Click "OK" to perform the mode decomposition.

4. Having arrived at the "distortion" page, use the "View distortion" (default) option to open the interactive applet and visually explore the symmetry modes available.  Close the applet window when finished.  Back on the "distortion page", use the "TOPAS.STR" option to save this superstructure to a file called "wo3_RTmono.str" in the working directory.

5.  Copy the traditional input file ("d8_03901_030c_traditional.inp") to a new filename ("d8_03901_030c_symmodes.inp") and open the new file in jedit.  Run a single-convergence refinement in TOPAS to verify the quality of the fit.  Rwp should be approximately 9.0%.

6. Convert the input file ("d8_03901_030c_symmodes.inp") to use symmetry modes as follows.

      (a) Use "/*" and "*/" to comment out the entire str section, including the five lines at the bottom from "scale" to "Phase Density_g_on_cm3"

      (b) Create a few blank lines above the old str section and position the cursor in this space in preparation for creating a new one.

      (c)  Within jedit's "Symmetry Mode Rietveld Refinement" menu, choose "Read ISODISTORT.str" option to import the symmetry-mode information from "wo3_RTmono.str".

      (d) Copy the cell parameters from the old str section and use them to replace the cell parameters in the new str section.

      (e) Copy the five lines from "scale" to "Phase Density_g_on_cm3" near the bottom of the old str section, and copy them up to the bottom of the new str section.

7. Turn on the isotropic thermal parameter of each atom by adding "@" immediately after each "beq" keyword, and perform a single-convergence refinement of the thermal parameters.  Then turn them off again (delete the "@" symbols).  Observe that the mode amplitudes are already pretty good based on the decomposition of the published structure.

8. Set all mode amplitudes to zero and do a single-convergence refinement of the scale factor by itself.  The result should be pretty ugly.  Then enable only the five most important modes (a2, a3, a7, a8, a19) that we identified from the decomposition, and switch to simulated-annealing mode by uncommenting the "continue_after_convergence" and "randomize_on_errors" lines.  Run several convergence cycles, which should bring Rwp down below 10%.  Much of the work is done now.

9. Disable all of the mode amplitudes without changing their values.  Switch back to single-convergence mode and perform a single refinement.  The structure should converge rapidly to an Rwp near 9.0%.  Note that turning on all 24 symmetry modes at once would not have helped much, even in simulated annealing mode.  Knowing which modes are important is more than half the battle.  How might one figure out which modes are important without prior knowledge of the distorted structure?

10. To bypass the workshop exercise and go straight to the result, download this input file: d8_03901_030c_symmodes.inp.


[Modified 01-Oct-2017 by John S.O. Evans. Pages checked for Google Chrome.]