How to perform and interpret a LeBail refinement on TOPAS

Ramon13

Although it was mentioned very briefly in the technical manual I'm wondering how to do a LeBail refinement in TOPAS GUI and Launch mode. The latter seems to give an output in JEdit but no key is given for how to describe the outputs so I'm unsure how to interpret it fully.

I'm doing this for my studies to better estimate the lattice parameters of mullite solid solutions in coal fly ashes. I would appreciate any help thank you! 🙂

johnsoevans

I didn't really understand your question. There's a tutorial on Pawley fitting on the topas site.
Pawley fitting is similar to Le Bail in that it is a structure-independent model, it just uses a slightly different least squares approach. In a hkl_Is fit TOPAS defaults to Pawley fitting. You can toggle between Pawley and Le Bail by adding "lebail 1" in the hkl_Is section of the INP file.

Ramon13

johnsoevans Hi John, my question is around accurately fitting a mullite solid solution found in coal fly ashes with a le-bail fit to get the lattice parameters, then find the most appropriate mullite as it is a solid solution for the eventual Rietveld.

The question comes from another forum post of mine where, unfortunately, the original commenter hasn't been online since their reply: https://chemistry.stackexchange.com/questions/167471/what-is-a-good-strategy-for-fully-quantitative-x-ray-diffraction-on-coal-fly-ash.

Hope this helps and thank you for the reply 🙂

johnsoevans

If I understand correctly it seems relatively straightforward. You just need to do a Le Bail or Pawley fit (see the tutorials for this) then look at the cell parameters (which are listed in the INP file). You then compare these to a database to find the most likely match as a starting point for Rietveld.

Ramon13

johnsoevans thanks John, the tutorial details it for the GUI, is there a tutorial for using launch mode?

I'm also unable to download the .raw file from the link, I believe the link is broken. Do you have a copy?

I've done the 2.3.1 Pawley and LeBail fitting of PbSO4 Whole Powder Pattern Decomposition but don't see how it's relevant here seeing as I'm given very close lattice parameters as initial conditions. 2.4 seems a bit more applicable but I don't understand how the Le Bail method applies or improves the tutorial method?

johnsoevans

The tutorial I was refering to (https://topas.webspace.durham.ac.uk/tutorial_tio2pawley/) is for launch mode. I just checked the link to the data and it seems to be fine.

Ramon13

It works for me now too thanks John ! 🙂

Ramon13

Hi again,

I'm still somewhat unsure how the outputs from a LeBail and Pawley Refinement can be fed into a Rietveld refinement when doing QPA? Could I please request more assistance.

Following the link provided previously for the tutorial on JEdit I'm a bit confused on the following:

The input initial conditions for the tetragonal space group unit cell parameters from the tutorial are given but are not told to be refined. Don't I want them to be refined so that I can approximate more accurate unit cell parameters? I thought this was the whole idea? If not, am I supposed to get the d-spacing values for each peak and somehow use them all together (i.e. in the Bragg equation with the P42nm spacegroup) to get the refined unit cell parameters? Surely there's something in the software for this?
How can the information from the "load hkl_m_d_th2 I" output (as seen in the code from JEdit below) be used for QPA.

"
hkl_Is TCHZ_Peak_Type(pku, 0.0690268313,pkv, -0.08123762,pkw, 0.0228930177,!pkz, 0.0000,pkx,
.162872292,!pky, 0.0000) phase_name TiO2 Tetragonal(4.595, 2.959) lebail 1 space_group "P42nm" load hkl_m_d_th2 I { 1 1 0 4 3.24916 27.42810 45.6468304 1 0 1 8 2.48780 36.07393 33.3600771 2 0 0 4 2.29750 39.17872 5.77176564 1 1 1 8 2.18773 41.23150 21.9290494 2 1 0 8 2.05495 44.03016 9.72882227 2 1 1 16 1.68785 54.30735 118.267043 2 2 0 4 1.62458 56.60840 40.0370097 0 0 2 2 1.47950 62.75139 22.2289218 3 1 0 8 1.45307 64.02697 25.4603198 2 2 1 8 1.42406 65.49193 2.39900968 3 0 1 8 1.36024 68.98476 73.7669964 1 1 2 8 1.34648 69.79123 34.6747774 3 1 1 16 1.30429 72.39753 4.4561956 3 2 0 8 1.27442 74.37546 0.933514118 2 0 2 8 1.24390 76.52439 9.55746064 2 1 2 16 1.20068 79.81505 5.517804 3 2 1 16 1.17048 82.31103 22.4769068 4 0 0 4 1.14875 84.21955 16.2627491 4 1 0 8 1.11445 87.44883 6.33435116 2 2 2 8 1.09387 89.52956 41.4777995 }"

Apologies if my questions seem elementary, I'm still trying to piece this all together, especially with relevance to the original paper link for context where the LeBail model has been done for the mullite solid solution (with results fed into the overall Rietveld refinement I believe?) preceding the overall Rietveld refinement of other phases (e.g. quartz etc.). I understand that extra analyses need to be done for complex solid solutions, I just don't understand the practicality of the analyses I've come across so far.

Kind regards,
Ramon!

johnsoevans

You should refine cell and zero point. Point 10 of the tutorial suggests you look to see how they've shifted. The tutorial assumes you've worked through the earlier tutorial on Rietveld fitting (https://topas.webspace.durham.ac.uk/tutorial_tio2riet/) so doesn't repeat how to do every step. If you want to know about the fundamentals then you could try the video tutorials at https://www.youtube.com/channel/UCqOTanQPNxEjSOzv6pLDJqQ.

Intensities from Pawley fitting aren't normally carried through to QPA. You could carry things like cell parameters and peaks shapes forward. However, there are more specialised methods for QPA using Pawley fitting when the more common Rietveld approach isn't possible (that may be the literature you're referring to). One is the Partial or No Known Crystal Structure (PONKCS) approach (or flavours thereof) by Ian Madsen and Nikki Scarlett. They're described in, for example, "Powder Diffraction Theory and Practice" by Dinnebier/Billinge or in https://topas.awh.durham.ac.uk/doku.php?id=book .

Ramon13

Thanks for the reply John 🙂

From a lot of my reading it seems that Pawley and LeBail methods are used synonymously and I'm not sure why when they're different methods. Why is this so?

I see, so these methods are good to estimate the unit cell parameters and peak shapes when there is no coordination information but other information is known like the space group (etc.). Interesting.

johnsoevans

They are different methods in that the algorithms used differ. Details are in texts I mentioned previously. In most cases they will give equivalent fits so you can do either. There's lots of literature/on-line discussion about this going back many years.

Ramon13

No worries John, thanks for that 🙂

I've done quite a lot of Rietveld fitting after watching your great videos and doing some research from last year. One last question from me on this: the tutorial didn't mention about using scale or crystallite sizes (etc.) that are usually used for Rietveld modelling of phases, so are there any main pieces of logic when distinguishing between what parameters can and can't be refined in a LeBail rather than a Rietveld refinement? I also see that the LeBail is done in one refinement rather than a sequence of refining parameters like in a Rietveld which is interesting.

PS - Out of curiosity, will you be uploading any more videos soon? Theory, QPA or otherwise you're a great teacher and I'd be interested to learn more from your videos 🙂.