Atomic Positions in Rietveld refinement

originlab1

My question is regarding choosing atomic positions while performing Rietveld refinement. I will give an example. Suppose I have I vs 2th XRD data for PbSO4 ( I take this as an example as the data is available in Durham website). Now, I performed Pawley refinement as discussed in TOPAS user manual and obtained the information for space group and the I(h,k,l) values. Next, I want to do Rietveld refinement for which I will need to provide atomic positions of the Pb, S, O atoms in the unit cell (in addition to the space group and lattice parameters I have determined already). And, please pardon my ignorance here, I really do not know how these are assigned. The TOPAS user manual gives the atomic coordinates but how one arrives at those numbers is not known to me. Durham website starts from completely arbitrary initial values. Does not it matter where you start assuming TOPAS is smart enough to find the correct solution no matter where you start? My question is how one decides on the starting values of the atomic positions while doing Rietveld refinement? Is there a simple way to decide on the atomic positions, e.g., when will an atom be placed on a special position and how?

I have been randomly reading stuff online and in a few other places, but could not arrive at a quick conclusion (I admit that I am looking for a quick answer and not going through the resources properly as I am so pressed on time). I have been reading about space groups, symmetry operations, Wyckoff positions, special positions etc. but it's as if I am becoming a "Jack of "few" trades but master of none". Please pardon my ignorance, but is there a easy way to determine the initial positions for the atoms while starting Rietveld refinement? Most likely I am wrong, but I was of the impression that if you know the space group then it will be easy to figure out the initial position? Am I right? If so, how one could accomplish that? Is there any software that can help? I have been looking at Bilbao Crystallographic Server(http://www.cryst.ehu.es) for GENPOS (Generators and General Positions of Space Groups) and WYCKPOS (Wyckoff Positions of Space Groups) but I am too ignorant to do anything with all these valuable information. AM really at a loss what to do. Please help.

I am looking forward to a "black box" like thing to which input will be say, the space group, lattice parameters (if needed), chemical formula (e.g., PbSO4) and the output will be the fractional atomic coordinates (along with the totoal number of atoms to be kept in the unit cell) which I can then take and put in the Rietveld refinement code (or inp file). Am I asking too much?

I do not even know or understand how many atoms will be there in the unit cell and how one calculates that number? Is there any place where I can quickly educate myself on these? Going through tons of papers and books is not helping me much. Could any of you kindly help in this regard? Thanks in advance.

originlab1

BTW, I am following the following strategy as outlined in "http://pd.chem.ucl.ac.uk/pdnn/solve1/strategy.htm""

1. Collect very High-Resolution and High-Quality Powder Data Fit for the Purpose
↓
2. Visually Inspect Data & Determine Peak Positions
↓
3. Index Peaks to Get Refined Lattice Parameters and Bravais Lattice
↓
4. Deduce Possible Space Groups
↓
5. Extract Integrated Peak Intensities I(hkl) by Whole Pattern Fitting
↓
6. Solve Structure using I(hkl)
↓
7. Refine Initial Model Structure
↓
8. Find any Missing Atoms by Fourier Difference Methods or Chemical Logic
↓
9. Complete Structure Refinement by Rietveld Method
↓
10. Generate Tables and Pretty Pictures of the Structure for Publication
-----------------------------------------------------------------------------------------
And I would say, I am stuck between 5 and 6. Hope one of you could help. Thanks in advance.

johnsoevans

I'm afraid the only answer here is to read one of the many introductory texts on crystallography.

mfisch

There is no general answer for your question. The PbSO4 example is one of the nice examples where simulated annealing brings a fast and correct solution, meaning you start from the (best case) correct (!) number of atoms and you "shake" until the best solution is found. In most of the cases this does not work so easily from powder data and thus single crystal diffraction data is preferred. In neither case, solving and refining a structure is just a "press the button" thing and requires some experience.

You can try whether you find a structure that is similar to yours in crystal structure databases and then think and change the chemical composition, refine the atomic positions until you arrive at a decent solution. For PbSO4, a possible isotype to start with could be BaSO4. Draw the structures and then you will get the point.