[RMCProfile-users] G(r) from X-ray diffraction data

Fri May 18 09:58:58 BST 2018

Dear Clemens

The simplest way to do this is to "pretend" that the X-ray G(r) is a neutron G(r), whilst explicitly specifying the scattering coefficients based on the number of electrons (rather than letting RMCProfile calculate them for you).  This isn't rigorously correct but is what most people do when analysing their X-ray PDF data - what Yuanpeng is working on will be much better (even if slower...).  The other alternative is to fit the X-ray data in reciprocal space which is correctly incorporated into RMCProfile.

Here is an example from a material that I was studying:

MATERIAL :: Ca3Al2(O4D4)3
ATOMS :: Ca Al O D

NEUTRON_REAL_SPACE_DATA :: 1
  > FILENAME :: Xray_PDFdata.gofr
  > DATA_TYPE :: G(r)
  > FIT_TYPE :: G(r)
  > START_POINT :: 160
  > END_POINT :: 1500
  > NEUTRON_COEFFICIENTS :: 0.053554 0.046413 0.171372 0.021422 0.010056 0.074261 0.009283 0.137098 0.034274 0.002142  
  > NO_FITTED_SCALE
  > CONSTANT_OFFSET ::   0.0000
  > WEIGHT ::   0.010

This is for a G(r) function which has a low-r value defined by Equation(15) of my paper (J. Appl. Cryst. 34, 172 (2001)) and tends to zero at high-r

The coefficients are calculated based on Equation(64) in the same paper - i.e. the separate pre-factors for each [g_ij(r) - 1] in the summation - and by using the number of electrons for each atom you are effectively ignoring the variations in the Q-dependent nature of the form-factors.

Hopefully this will allow you to get your refinements working - and you can cross-check what you are doing with the details for Ca3Al2(O4D4)3 above.  If my numbers aren't right then I probably don't want to know - because the paper on this work is already published! [Phys Chem Miner 45, 333 (2018) doi:10.1007/s00269-017-0923-0 if you are interested.]

Best wishes
Dave

On 16/05/2018, 16:13, "rmcprofile-users on behalf of Clemens Prescher" <rmcprofile-users-bounces at shadow.nd.rl.ac.uk on behalf of clemens.prescher at gmail.com> wrote:

    Hi there,

    i have problems figuring out how X-ray derived PDF should be correctly
    inputted into RMCProfile.

    On page 157 in the RMCProfile manual it is written that the X-ray G(r)
    needs to be normalised so that it is scaled from 0 to 1. This is exactly
    the format i have. However, if I init my calculation and plot the "PDFs of
    rmc fit" using rmcplot it has an offset (please see the attached figure "
    gr_without_offset.ps")

    It is clear, that the RMC fit is actually normalized to 0 when going into
    infinity.

    I thought, this might just a result by subtracting 1 from the previously 0
    to 1 normalized G(r). But after several iterations using an offset of -1 it
    became clear that low part is not normlized to -1. (see the
    "gr_with_offset" figure attached).

    In the manual it is also written that G(r) goes from -(sum c_j b_j)² to 0.
    But for X-ray this does not make sense due to q-dependent form factors...

    Can you please explain me how i should transform my data?

    Best wishes,
    Clemens
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20180515/f421cd8d/attachment.html>
    -------------- next part --------------
    A non-text attachment was scrubbed...
    Name: gr_without_offset.ps
    Type: application/postscript
    Size: 21694 bytes
    Desc: not available
    URL: <http://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20180515/f421cd8d/attachment.ps>
    -------------- next part --------------
    A non-text attachment was scrubbed...
    Name: gr_with_offset.ps
    Type: application/postscript
    Size: 19422 bytes
    Desc: not available
    URL: <http://lists.rmcprofile.org/pipermail/rmcprofile-users/attachments/20180515/f421cd8d/attachment-0001.ps>
    _______________________________________________
    rmcprofile-users mailing list
    rmcprofile-users at rmcprofile.org
    http://lists.rmcprofile.org/mailman/listinfo/rmcprofile-users