Dear All,

I also had a similar issue in my PhD thesis when using the that-time GSAS: wanting to determing d-spacings of unindexed peaks in half-interpreted diffraction pattern (it was magnetic superstructure peaks for which I did not have a model).

I may add a small trick to the technique described by Thierry Roisnel, which is helpful, if you do not access to the individual peaks of the phase (step 4).  You may have a tetragonal cell, e.g. a = b = 0.1 Å and c = d, with d to be refined and using an appropriate profile function. If you then put an atom with occupancy 1 at 0 0 0 and one of the same kind with occupancy -1 at 0 0 0.5, then you get a finite structure factor for all 00l with odd l and a zero structure factor for all even l. By this method you have fewer higher order reflections of that artificial phase, which might interfere with the rest of the pattern. The first higher order peak is 003, if I get it correct at the moment. I am not sure whether this worked really like this (it was GSAS at the end of the 1990s), because some programs do not allow negative occupancy. It is also possible that I did the thing by considering antiparallel magnetic moments at the mentioned fractional coordinates.

I, however, definitely enjoy using software where you can combine Rietveld refinement with individually fitted peaks without relying to such tricks.

Best regards

Andreas Leineweber

Am 25.04.2022 um 17:36 schrieb Magnus Sørby:

Dear Thierry and Alan,

@Thierry: Thank you very much! This is a great solution.

@Alan: I agree that we should strive to use as few and as meaningful refinable parameters as reasonably possible, not just in Rietveld refinements but in all kinds of modelling. The task I’m engaged in now is to estimate crystallite sizes in Si particles that are produced along with a lot of non-crystalline stuff in a semi-industrial process.  Taking more time to produce “clean samples” is therefore not an option (although it would surely be the best advice for a pure structure determination/refinement study). Since there are a lot of sample and time is an issue, I don’t have the luxury to spend a lot of efforts on the other features in each dataset; at least not for the moment. I’m therefore more than happy to take some shortcuts in the description of the non-Bragg parts of the data.

Best regards,


*From:* <> *On Behalf Of *Alan W Hewat
*Sent:* 25 April 2022 15:50
*Subject:* Re: Introduction of non-structural peaks in Fullprof

Dear Thierry and Magnus.

Introducing an extra pseudo-phase, and then editing out all the peaks you don't want, is ingenious, but is it really less work than creating a background file as Rietveld advocated?

The whole idea of Rietveld was to refine only physically meaningful parameters. The more parameters you introduce, the less confidence you can have in any of them being physically meaningful, especially if you then have to constrain some of them.


Dr Alan Hewat, NeutronOptics
Grenoble, FRANCE (from phone)
+33.476984168 VAT:FR79499450856 <>

On Mon, 25 Apr 2022, 14:53 Thierry Roisnel, <> wrote:

        Dear Magnus,

        Here is a way to fit independent peaks in FullProf :

    1. edit .pcr file and add for every independent peak an artificial
    phase to treat in Profile Matching refinement mode (Jbt=2 and
    Irf=0). You can use P m m m space group for example and adjust the
    a parameter to fit with the diffraction peak 2theta position, b
    and c parameters can have arbitrary values, smaller than a. So,
    the special reflection will have 1 0 0 indices.

    2. add refinement of one special reflection (Nsp_Ref = 1)

    3. Run FullProf and update .pcr file, to provide Irf = 2

    4. edit the created xxn.hkl file ( n for phase #n) and remove all
    reflections excepted the first one (1 0 0)

    5. edit .pcr file and refine peak position and profile parameters
    for the special reflection you want to fit.

        The .pcr file will look as follows :

    Phase #2 : fit  a single peak
    !Nat Dis Ang Pr1 Pr2 Pr3 *Jbt Irf* Isy Str Furth       ATZ    Nvk
    Npr More
       0   0   0 0.0 0.0 1.0 *2   2*   0   0 0      32516.641   0   7   1
    !Jvi Jdi Hel Sol Mom Ter  Brind   RMua RMub    RMuc   Jtyp
    *Nsp_Ref* Ph_Shift N_Domains
       0   0   0   0   0   0  1.0000  0.0000 0.0000  0.0000    0     
    1      0      0
    *P m m m* <--Space group symbol
    !-------> Profile Parameters for Pattern #   1  ----> Phase #   2
    !  Scale          Shape1      Bov Str1      Str2      Str3  
     0.1252800E-04   0.00000   0.00000   0.00000 0.00000   0.00000       0
           0.00000     0.000     0.000 0.000     0.000     0.000
    !       U         V          W X          Y        GauSiz   LorSiz
       0.005115  -0.007545   0.020850   0.012780 0.020016   0.000000  
    0.000000    0
          0.000      0.000      0.000 0.000      0.000      0.000     
    !     a          b         c        alpha beta       gamma     
    #Cell Info
    *3.400000 0.500000   0.500000  90.000000  90.000000 90.000000**
        0.00000    0.00000    0.00000 0.00000    0.00000    0.00000*
    !  Pref1    Pref2      Asy1     Asy2 Asy3     Asy4      S_L      D_L
      0.00000  0.00000  0.00000  0.00000  0.00000 0.00000  0.01849 
         0.00     0.00     0.00     0.00 0.00     0.00     0.00     0.00
    *! Special reflections:**
    !  h   k   l  nvk       D-HG^2 Cod_D-HG^2      D-HL      Cod_D-HL
    Shift    Cod_Shift    Del_U/sig1 Code     Del_Y/gam2     Code
       1   0   0    0   0.0000 0.000     0.00000       0.000
    0.00000       0.000     0.00000 0.000     0.00000       0.000*

        I hope you will success in what you want to do.


        Thierry Roisnel

    Le 25/04/2022 à 11:18, Magnus Sørby a écrit :


        I’m playing with Fullprof again for the first time in many
        years since I’ve changed employer and lost my access to Topas.

        Is it possible to introduce independent peaks (i.e. peaks
        whose positions are not determined by a unit cell) with
        refinable positions, intensities and shapes in Fullprof? I’ve
        found this possibility very useful in Topas for dealing with
        features in the datasets that are not of interest and not
        possible to possible to describe with a crystalline phase e.g.
        diffuse scattering from an amorphous component.

        I’m aware of the option to include a self-defined background
        (.bac file) with refinable scale, but this is time-consuming

        Best regards,



        Magnus H. Sørby, PhD

        Senior Scientist  - Materials Science and Analytical Chemistry

        Cenate AS



        Please do NOT attach files to the whole list<>  

        Send commands to<>  <>  eg: HELP 
as the subject with no body text

        The Rietveld_L list archive is 



    * Thierry Roisnel*

    Centre de Diffractométrie X (CDIFX)
    Institut des Sciences Chimiques de Rennes
    UMR6226 CNRS - Université de Rennes 1
    Bât. 10B, p. 153
    Campus de Beaulieu
    Avenue du Général Leclerc
    35042 Rennes cedex, France

    Tél: 02 23 23 59 02
    Email :


    Centre de Diffractométrie X, ISCR UMR6226

    ISCR UMR6226 <>



    Membre de ЯÉCIPROCS


    Réseau ЯÉCIPROCS <>

    Please do NOT attach files to the whole list
    Send commands to <> eg: HELP as the subject with no
    body text
    The Rietveld_L list archive is on

Please do NOT attach files to the whole list<>
Send commands to<>  eg: HELP as the subject with no body text
The Rietveld_L list archive is on

Prof. Dr. rer. nat. habil. Andreas Leineweber
Institute of Materials Science
TU Bergakademie Freiberg
Gustav-Zeuner-Str. 5
D-09599 Freiberg
Phone: +49-3731-39-2622
Fax:   +49-3731-39-2604
Please do NOT attach files to the whole list <>
Send commands to <> eg: HELP as the subject with no body text
The Rietveld_L list archive is on

Reply via email to