Hi Lisa,
The typo in the core atom does not affect the result if run in Artemis. I think I now have a better grasp of why atoms cannot run my structure correctly. The standard i2/c setting has a unique axis a, but the structure in the input file below has a unique axis b. This appears to through atoms off. If I convert to an I2/a setting, then I get the correct results without a shift vector.
George
Hi George,
I cannot understand how you got this atoms.inp file where the core is stated to be Co1 and there is only Fe. So, I cannot confirm the crystal structure from a database.
My problem is solved. I understood that the ATOMS program implemented in Arthemis and ATOMS 2.5 have some bugs that are corrected in ATOMS 3.0 and WebATOMS. When I use those two programs I get a correct crystal structure xyz table in feff.inp.
the only reason that I can think of, why your shifting seem to work better is that the atoms.inp file was not representing the correct structure in the first place. This can happen when the situation you wnat to describe is not the situation that was measured by somebody else. Or there is a mistake in the paper.
Anyway. Thanks for sharing this idea.
Lisa
Date: Thu, 26 May 2011 11:41:08 -0400
From: George Sterbinsky <GeorgeSterbinsky@u.northwestern.edu>
To: XAFS Analysis using Ifeffit <ifeffit@millenia.cars.aps.anl.gov>Message-ID: <BANLkTi=OH=NCJjmwhkE5vLmnTEao8k3avg@mail.gmail.com>
Subject: Re: [Ifeffit] more bugs in atoms?
Content-Type: text/plain; charset="iso-8859-1"
Hi Lisa,
Let me just mention one situation I have encountered using atoms, and how I
resolved it. I am not sure if this is the result of a bug or not, but
perhaps you can try applying the approach I took to your own situation and
see if it can resolve your problem.
The issue I encountered was running atoms for a monoclinic I2/c structure,
space group 15.
Here is the atoms input file:
! This atoms input file was generated by Artemis 0.8.014
! Atoms written by and copyright (c) Bruce Ravel, 1998-2001
title = ...
space = i 2/c 1 1
a = 5.51120 b = 5.51120 c = 7.79410
alpha = 90.0 beta = 90.740 gamma = 90.0
core = Co1 edge = K rmax = 6.0
!shift 0.25000 0.25000 0.25000
atoms
! elem x y z tag occ.
Fe 0.00000 0.00000 0.00000 Fe1 1.00000
If one calculates the Fe-Fe distance between the atom at (0,0,0) and the
atom at (0, 0, 0.5), from application of the (x, -y, -z+0.5) lattice
translation, one finds a Fe-Fe distance of 3.9705. However, if one runs the
above atoms input file, this Fe-Fe distance is not found. Instead, a shift
vector of (0.25, 0.25, 0.25) is needed to get the correct Fe-Fe distance.
Note, the i2/c space group is listed with only one origin in the
international tables. I determined the necessary shift vector from trail and
error. It is still unclear to me why it was necessary to include a shift
vector. So the best suggestion I have is that you can try including
different shift vectors in your own atoms.inp file and see if you can get
agreement with the crystallography program that way. Since, as I said, it
isn't clear to me why this fixed my problem, its hard to say if this is the
same issue you are having, but it may be worth a try.
It may also be worth while to calculate some atomic distances from the
lattice positions given in the international tables, and see if atoms or the
crystallography program is giving you the same thing.
Finally, let me add on another question for the list here since it is
somewhat related. When one runs the above atoms.inp file with the (0.25,
0.25, 0.25) shift vector one finds four Fe1_1 atoms at 3.9701 and two Fe_2
atoms at 3.9705. When one then runs Feff, it combines these into a single
Fe1_1 scattering path with N=6. Is there a command can be placed in the Feff
input file to tell Feff not to combine identical paths like this?
Best,
George
_______________________________________________
Ifeffit mailing list
Ifeffit@millenia.cars.aps.anl.gov
http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit