atpolonsky
Ok, so I'm able to follow the simple rietveld refinement examples via jEdit from Professor Evans' website, so I believe I have everything for jEdit and TOPAS installed correctly. I am having some trouble getting my actual refinement to work in jEdit, however.
I'm trying to refine the lattice parameter of a cubic spinel that has been slightly doped, so the lattice parameter is minutely different from that of the stoichiometric material. Basically these are the steps I follow to perform my refinement in TOPAS (version 4.2, not academic), which has given good results:
1. Load .raw scan file and change order of Chubychev polynomial to order 5
2. Load my instrument settings for the instrument (right now they're saved as a .inp file) because the geometry is different from that which shows up normally in TOPAS (i.e. goniometer radius = 141.5 mm, not 217.5 mm, among other differences)
3. Select the option to correct for sample height under the Corrections tab
4. Change the LP factor in the Corrections tab from 90 to 0 (another instrument geometry thing)
5. Load an .str file of Silicon, which I add to my sample to act as an internal standard which I use to correct for sample height offset (exact peak position is critical to determining an accurate lattice parameter value)
6. Run a refinement
7. Uncheck the "use phase" option for Silicon.
8. In the Corrections tab, change sample height option from refine to fix (now that the sample height offset has been determined)
9. Load the .inp file for the stoichiometric version of my material (a cubic spinel), which already has information about atoms positions, space group, etc., as well as a starting point for the lattice parameter and crystallite size that is set to be refined.
10. Run a refinement and obtain the refined value of the lattice parameter (which is luckily only one value as the structure is cubic).
This analysis works great but is pretty tedious to teach to others working with me. I have been trying to replicate these steps in jEdit in an attempt to fully or partially automate the analysis, but am having difficulty figuring out whether or not selections I make from the Xinsert menus are actually having any effect on the scan file. I constantly get error messages along the lines of "cannot locate pk_to_Lam from riet_app_3 in data structures" after running the modified input file from jEdit in TOPAS. I have not had any insight as to what these errors mean.
Any guidance at all to help me with my refinement via jEdit would be greatly appreciated.
ahhill
I would suggest get a working refinement in TOPAS, then save it to an INP file. You can then use that as a quick template for future refinements. Just change your structural phase to one you want.
Adrian
mfisch
If you're only interested in lattice parameters, I would suggest a LeBail or Pawley refinement anyway. There are many examples on the web and in the examples folder.
I don't think your concern about correlation of sample displacement with lattice parameters is true for a cubic phase either, there are plenty of constraints due to the cubic space group.
Martin
johnsoevans
The errors you're getting sound like they're due to instructions being put in the wrong part of the input file (their scope). Certain keywords only make sense in certain places in the file. For example information about the diffractometer wavelength only makes sense in the part of the .inp file that deals with data file.
The tutorial examples and jEdit menus are set up so you get this correct for simple examples. Another way of doing it is to use one of the "template" files provided. The other good way in practice is (as suggested by others) start from an inp file from one of the examples or created via the gui (though the layout this way isn't always as clear).
atpolonsky
Thanks for all the input everyone. It seems like using one of my working refinements as a template (as suggested by Professor Evans and Adrian Hill) is the easiest way for me to go. Now that I figured out how to do my refinement through jEdit, I should be able to easily automate the entire process via macros. You've all been a great help!
-Andrew
arazbat
Dear All,
I have the same question in TOPAS. As our machine is not calibrated I should always check displacement or Zero error in my refinement. But as I my aim is finding lattice parameter I am finding I should have always fix values to that error to relay on lattice parameter results.
I read atpolonsky procedure, it is really sounds good for me to fix error(either displacement or Zero error) by reference. Our bruker machine has corundum reference. Then I wonder if I refine it by TOPAS and calculate errors and replace scan with my sample and uncheck errors, values for lattice parameters are gonna be reliable?
Another question is that is it important to refine values you want one by one? for example should I refine lattice parameter first and run it second refine crystal L, run it and so one or I can run all at the same time?
thanks for your help
johnsoevans
Arazbat,
The procedure you suggest sounds dangerous to me. Zero point error and sample displacement are highly correlated, especially if you have a narrow 2-theta range. A rule of thumb is to refine one or the other but not both. If your diffractometer is well aligned it would often make sense to refine displacement as this will vary according to the way you mount samples.
If you really want high accuracy in cell parameters then the best approach is to use an internal standard (ideally something like NIST Si). You can then effectively calibrate the diffractometer from the measurement itself.
In general you shouldn't refine parameters sequentially as you suggest. Their optimum values may change during the course of refinement. If you have uncorrelated, well-determined parameters in your model you ought to be able to refine everything in the final Rietveld cycles.
John
arazbat
Dear John,
Thanks for your informative answer.
I try to dope MgB2 with C. In my case more doping will shift peaks more and then lattice parameters are changing more. As zero point and sample displacement refinement dealing with peak shifting and they are correcting peak shifts, I wonder if I check them can my doping effect got mistaken by errors?
For sure internal calibration is best way, just I wanted to know if I trust on this procedure which I have mentioned am I going far from reality or not?
Best
Araz
johnsoevans
My view is as above. If you want accurate (not just precise) cell parameters you have two choices.
1. You make sure that your instrument is well aligned and that your sample mounting procedure puts the sample in exactly the right position in the diffractometer each time. You check calibration with an external standard and ensure that there are e.g. no absorption differences between the standard and sample that could effect peak positions.
2. Use an internal standard.
I would personally opt for choice 2.
arazbat
John, Thanks for answer
I think will go for Si internal standard as well
Thanks again