In the preliminary part of a Rietveld refinement, fitting
the profile can be the most difficult operation. This may
not be easy if the initial starting structure is quite far
off resulting in a poor match in peak intensities.
One possible way around this is to temporarily do a Le Bail "structureless" whole profile fit, fix the profile and background parameters, then recall the structure to continue the refinement focussing on the structure.
In theory, the Le Bail fit is the best possible fit to the powder pattern you can get with that particular spacegroup; adding in a structure then complicates matters.
This is quite easy to do in LHPM-Rietica and an example of this follows.
In the following case, the starting structure is a solution obtained from the EXPO software from the Sirware group.
Assuming you have the input file already setup, enter the Model,
Phases option. On the top half of the structure information,
right click on the mouse and select Add to database.
From the Calculation Method, click on the drop box and select Le Bail. Also make sure the scale factor is not being refined. Don't worry if the structure information is still there. It will disappear on doing some rounds of least squares refinement.
Now fit the profile using the Le Bail method. Don't be too worried if the first
few cycles look very ugly while the initial starting HKL
intensities sort themselves out; let it go for 10 cycles or so.
If you have a nasty background, you can also manually define and set
the background using the LHPM-Rietica GUI.
Initially, it may be safe to only refine on the background. Also, to save hassles, doing a separate (quick) unit-cell refinement with a specialised unit cell refinement program (such as celref)could save problematic refinements due to the calculated HKLs being to far away from the observed peak positions. Refer webpage on: Available Software for Unit Cell Refinement of Powder Diffraction Data
Refining on the cell and zero offset. Slowly releasing the relevant width, shape and assymetry parameters. It can be useful to do this with an "automatic Marquartd" refinement option under Model, General.
Normally, after Rietica generates the initial HKL file, the "recycle Le Bail" will be automatically switched on. If things do go very wonky with the HKL intensities, deleting the HKL file is a guaranteed way for Rietica to regenerate the HKL file from scratch. If doing this, check that the background is reasonably set.
Following is the starting Le Bail fit when refining only on the background.
Zero offset and unit cell now refining.
After slowly releasing width, shape and asymmetry parameters.
Now that the whole profile fitting is completed, turn off all the
background, cell and profile refinement flags (locking them into place).
Enter the Model, Phases option. On the top half of the structure information, right click on the mouse and select Select from database.
Now select the saved structure.
As the structure was saved before refining the unit-cell. It would be wise to let the unit cell refine while also trying to refine on the atomic positions. Also, make sure Calculation Method is set to Normal (not Le Bail) plus that the Scale Factor is again set to refine.
Now you can continue on with the structure refinement.