jamesrhester
commented
2 years ago I think this should work.
- I don't see any problems with the
_pd_po_mdoutline. I think it is true that all of these items are per phase, so there would be a child data name of_pd_phase.idin there too. As we are separating phase-specific information into separate data blocks, this data name would generally be invisible. - Likewise with the
_pd_po_shproposal, and definitely prefer they_i, y_j, y_pforms as they make the indices more immediately available to calculations. _pd_po_geomseems like it is not just specific to preferred orientation but something that probably belongs inPD_INSTR._pd_instr.geometryis one of those old data names that does not restrict the geometries to a specific list, so how about we add a new data name like_pd_instr.reflection_geometrywhich would have values as suggested above?- Yes, the data names will probably become
_pd_preferred_orient_March-Dollase.*and...._sphericalharmonics.*
rowlesmr
briantoby
Idea for the addition of preferred orientation parameters to pdCIF.
To my knowledge, there are two main methodologies for modelling preferred orientation (PO) in powder data: March-Dollase (MD), and spherical harmonics (SH).
In both cases, a geometrical factor must be known to properly apply the corrections. More on that at the end.
This would extend the existing
_pd_proc_ls.pref_orient_corrby providing specialisations for MD and SH, requiring any other implementations to conform with the existing specification.Note: I know that all the tag names are currently terrible.
March-Dollase
See refs 1,2,3
When applying MD, there are two required values: the amount of orientation, typically referred to with r, and the direction of the orientation (the direction is often a cleavage plane). There can be multiple PO directions, in which case, each has their own r and hkl, and there is an additional fractional contribution of each direction to the overall PO correction. In a refinement, hkl is defined, and r is refined. If there is more than one direction, the the fractional contribution of each direction is also refined.
A first suggestion:
_pd_po_md_r– the March-Dollase r factor. Float, (0, infty) , default value = 1.0_pd_po_md_h– the h index of the orientation direction. Int, (-infty, -infty)_pd_po_md_k– the k index of the orientation direction. Int, (-infty, -infty)_pd_po_md_l– the l index of the orientation direction. Int, (-infty, -infty)_pd_po_md_fract– in the case of multiple directions, the amount that is in that direction. Float, (0, infty), default value = 1.0. The sum of all values for a single structure must be 1._pd_po_md_id– in the case of multiple directions, an ID number to uniquely id a thing.To be used as:
or
or
Spherical-Harmonics
See ref 4
When applying SH, there are many required values, depending on the Laue class*. From Jarvinen [4], only even orders are used. Each value is a refinable parameter and must be reported. There are a couple of ways of giving the values: (the notation I'm using follows Jarvinen and TOPAS).
Note: SHs can also be used for anisotropic crystallite size. so setting up the tag names so that changing to crystal size would still look largely the same would probably be best.
_pd_po_sh_y_i- order of the SH; even integer, [0, infty)_pd_po_sh_y_j- terms of each order: integer, [0, i]_pd_po_sh_y_p- parity: 0, +1, -1 (can only be 0 for the 0th term of each order, otherwise, +/- 1)_pd_po_sh_cijp- the actual refined value. Float (-infty, infty)Also, as a separate item,
_pd_po_sh_texture_indexThe texture index is defined by Bunge (ref 5; eqn 4.212) as
\sum{i,j,p}\frac{1}{2i+1}c{ijp}^2)
It gives the value [1, infty), where 1 is a random powder and infty is an ideal single crystal. This is a good summary statistic to show how orientated a phase is.
To be used as:
or, you could define a tag for each. This has the downside of limiting the combinations of ijp to those specified
_pd_po_sh_y00- default value 1. If not 1, then it is suspect._pd_po_sh_y20– float (-infty, infty)_pd_po_sh_y21p– and so on…._pd_po_sh_y21m_pd_po_sh_y22p_pd_po_sh_y22m_pd_po_sh_y40_pd_po_sh_y41p_pd_po_sh_y41m_pd_po_sh_y42p_pd_po_sh_y42m_pd_po_sh_y44p_pd_po_sh_y44m_pd_po_sh_y60_pd_po_sh_y61p_pd_po_sh_y61m_pd_po_sh_y62p_pd_po_sh_y62m_pd_po_sh_y64p_pd_po_sh_y64m_pd_po_sh_y66p_pd_po_sh_y66m_pd_po_sh_y80_pd_po_sh_y81p_pd_po_sh_y81m_pd_po_sh_y82p_pd_po_sh_y82m_pd_po_sh_y84p_pd_po_sh_y84m_pd_po_sh_y86p_pd_po_sh_y86m_pd_po_sh_y88p_pd_po_sh_y88mGeometrical considerations
The exact implementation of the PO correction depends on the instrument geometry, or probably more to the point, the defaults in the program used to do the modeling of the PO. The (probable) default for (most?) programs would be symmetric reflection geometry. The other possible geometries are asymmetric reflection, symmetric transmission, asymmetric transmission, and capillary. (see [1], [2], [4])
(the sample normal is taken as the rotation axis of the sample, hence the difference between capillary and asymmetric transmission)
There are some diagrams in [1] showing these different arrangements. They are also in Table 1 [4] and the first two are discussed in [2].
It is suggested to have a tag:
_pd_po_geomwith possible values:
"refl": symmetric reflection"trans": symmetric transmission"arefl": asymmetric reflection"atrans": asymmetric transmission"cap": capillary, or Debye-ScherrerIf this tag is not supplied, then
"refl"is assumed.References
1 Rowles and Buckley 2017, §3, https://espace.curtin.edu.au/handle/20.500.11937/50104 [espace.curtin.edu.au] 2 Ida, 2013 https://www.researchgate.net/publication/271325583_Effect_of_Preferred_Orientation_in_Synchrotron_X-ray_Powder_Diffraction [researchgate.net] 3 Dollase 1986 https://doi.org/10.1107/S0021889886089458 [doi.org] 4 Jarvinen, 1993 https://doi.org/10.1107/S0021889893001219 [doi.org] 5 Bunge, https://www.researchgate.net/publication/277308168_Texture_Analysis_in_Materials_Science_H-J_Bunge [researchgate.net]