Introduction to the powder refinement in Jana2000

The objective when implementing powders into Jana2000 was creation of a unique interface for single crystals and powders. We used tools known from the single crystal refinement wherever it was possible. The most different part is the initial stage, i.e. reading of data and the profile refinement. In this document it is explained using two examples taken from the following publication:
Dusek, M., Petricek, V., Wunschel, M., Dinnebier, R.E. and Smaalen, S. van (2001), J.Appl.Cryst.34,398-404,
Refinement of modulated structures against X-ray powder difraction data with Jana2000.
The data of the examples can be downloaded and used for testing of Jana2000. For other purposes the mentioned publication should be properly cited.

Example 1: quartz

(download powder data for Windows)
(download powder data for Unix)

This is a powder measurement of natural quartz (alpha-SiO2).
Cell parameters: a=4.914, b=4.914, c=5.405, alpha=90, beta=90, gamma=120
Space group: P3221 (No. 154)
Data type: GSAS measurement
Wavelength: CuKalpha1

Instructions:

  1. Enter cell parameters, space group, radiation type, wavelength and atom form factors through EditM50.
    Cell parameters, space group and radiation is necessary for proper generation of Bragg positions. An indexing program for powder data is not available.
    Enter also atom form factors through Editm50. This information is not necessary for profile refinement but Jana cannot work without having this basic information in m50.
  2. Reading of the profile data:
    File ->Import reflection file ->Powder diffraction data ->GSAS
    Note: reading of profile data is not possible before entering the information from the point 1.
  3. The experimental profile can be visualized by
    Tools ->Powder ->Plot powder profile
  4. Setting of basic powder parameters:
    Check the wavelength in Parameters ->Powder ->Basic. Warning: the values given in this tool have only informative character. The program uses the wavelength and radiation type set in EditM50 (see point 1) and these values cannot be changed here although they are listed in editable textboxes.
    We recommend enabling of Apply weight in le Bail decomposition.
  5. Setting of background refinement parameters:
    Parameters ->Powder ->Corrections ->Legendre polynoms is used for setting the background correction type. Number of terms can be something between 5 and 12 or more. In this example we are using 10 terms.
    Press Edit background and Refine all to set refinement keys of background parameters. The refinement can start from zero values.
    After closing the form the file m41 with profile parameters is created. Refine works with this file similarly like with m40 in case of structure parameters. The previous version of m41 can be recovered by Tools ->Recover files.
  6. Setting of Refinement commands:
    In SetCommands ->Refine ->Basic commands we recommend setting of Frequency of le Bail decomposition from 3 to 5 and ticking off the Berar's correction. The damping factor 1 can be used. Profile refinement without having a structure in m40 may be rather unstable in the cycle following the le Bail decomposition. Therefore number of cycles must be much larger than the frequency of the le Bail decomposition. If the automating checking for convergence is used the number of consecutive cycles must also be larger than the frequency number.
  7. Run refinement of background parameters. The resulting Rp will be about 50%.
  8. Use Tools ->Powder ->Plot powder profile to estimate quality of the background correction. As a refined profile already exists it is automatically displayed like a polyline superimposed with crosses representing the measured points. The measured points can be removed from the plot by button Pnts OFF and replaced by a polyline (possibly in a different color) through Options.
    Warning: The calculated profile is not available if refinement has been breaked by Cancel button.
  9. In Parameters ->Powder ->Profile choose Gaussian peak shape function and select refinement of GU and GW. Refinement of GU can start from zero value, GW starts from default value 5. Refinement will converge at Rp about 48%.
  10. In Parameters ->Powder ->Cell select refinement of a and c. Refinement will converge at Rp about 31%.
    The refined cell parameters will appear not only in m41 but also in m50.
  11. In Parameters ->Powder ->Corrections select refinement of zero shift. Refinement will converge at Rp about 18%.
  12. In Parameters ->Powder ->Profile choose Pseudo-Voigt peak shape function and select refinement of LY. Refinement of LY can start from zero value. It will converge at 13%.
The plotting tool will show good fit between calculated and observed profiles. See the article cited above for finer details of this refinement and for comparison with GSAS refinement.

Structure refinement:
The refinement program generates file m80 containing observed and calculated structure factors belonging to generated Bragg intensities. This is used as an input to Fourier program so that all tools related with Fourier maps are available both to single crystal and powder data. The initial structure is set by the same way like for single crystals: from Fourier maps orf by import from SHELX or CIF or other sources. The structure and profile parameters should be refined together.

Example 2: NbTe4

(download powder data for Windows)
(download powder data for Unix)

This is a powder measurement of modulated structure of NbTe4.
Cell parameters: a=9.193, b=9.193, c=6.836, alpha=90, beta=90, gamma=90
q-vector (0,0,0.691)
Space group: P4/mcc(0,0,gamma)
Data type: GSAS measurement
Wavelength: CuKalpha1

Instructions:

  1. Enter cell parameters, number of dimensions (4), q-vector, superspace group, radiation type, wavelength and atom form factors through EditM50.
  2. Reading of the profile data: like in the first example.
  3. Visualization of the experimental profile: like in the first example.
  4. Setting of basic powder parameters: like in the first example.
    In Parameters ->Powder ->Cell set maximal satellite index to 0 in order to use for the beginning only main reflections.
  5. Setting of background refinement parameters: like in the first example, 10 terms of the Legendre polynom.
  6. Setting of Refinement commands: like in the first example.
  7. Run refinement of background parameters. The resulting Rp will be about 26%.
  8. Use Tools ->Powder ->Plot powder profile to estimate quality of the background correction.
  9. In Parameters ->Powder ->Profile choose Gaussian peak shape function. Set GU to 0 and GP to 5. Select refinement of GP, other gaussian parameters will not be used. Refinement will converge at Rp about 23%.
    Note: refinement of gaussian parameters cannot run properly if all initial values are zero. In previous refinements GW was fixed to 5. Now we need GW = 0, so that GP has to be set to non-zero initial value.
    Note: the same results can be achieved by refinement of GW and GU.
  10. In Parameters ->Powder ->Profile choose Pseudo-Voigt peak shape function and select refinement of LX and LY. Refinement will converge at Rp about 20%.
  11. In Parameters ->Powder ->Cell select refinement of a and c. In Parameters ->Powder ->Corrections select refinement of zero shift. Refinement will converge at Rp about 19%.
    In the plot small satellite peaks are clearly visible.
  12. In Parameters ->Powder ->Cell set maximal satellite index to 1 and select refinement of q3 component. Refinement will converge with Rp about 16%.
See the article cited above for finer details of this refinement and for comparison with single crystal data results.

Structure refinement: like in the first example.

The course of the refinement and exact Rp values may depend on the used computer. Also the order of steps given in the examples is not a strict receipt but only a hint of solution.