# Validation of _atom_site_frac_*: Range outside of [0,1)?

I have the following cif file

loop_
_atom_site_label
_atom_site_type_symbol
_atom_site_fract_x
_atom_site_fract_y
_atom_site_fract_z
...
Al2 Al -0.35936(1) -0.08768(9) -0.24188(8)
Al3 Al -0.27044(1) -0.35380(9) -0.13044(8)
F1 F -0.5320(2) -0.2015(2) -0.3794(2)
F2 F -0.3800(2) -0.2437(2) -0.1891(2)
F3 F -0.0350(2) 0.0944(2) 0.1528(2)
...
N3 N 1.2782(3) 0.7260(3) 0.3726(3)
...
C2 C 1.0165(4) 0.6764(4) 0.2267(3)
C3 C 1.1053(4) 0.6806(4) 0.3576(3)
...
H1 H 1.015(4) 0.603(4) 0.147(3)
...
H6 H 1.068(6) 0.773(6) 0.220(5)
H7 H 1.100(5) 0.761(5) 0.441(4)
...
H10 H 1.303(5) 0.653(4) 0.319(4)
...
H12 H 1.070(5) 0.574(5) 0.364(4)
...
H16 H 1.328(6) 0.749(6) 0.471(5)
H17 H 1.313(5) 0.801(4) 0.348(4)


Of interest is the fact that many of these "fractions" of the cell length are greater than 1, and many are negative. To my sensibilities these atoms lie outside the unit cell.

Should the IUCr definition form CIF dictionary of _atom_site_fract_x state that the permitted range is 0.0 -> 1.0, but forgot?

Should these atoms be discarded, should they be mod'd into the unit cell, or should they be interpreted in some other manner?

• I couldn't find in the reference that the range is permitted only from 0 to 1. – Martin - マーチン Dec 27 '16 at 9:06
• @Martin-マーチン: It isn't in the reference. The question is whether it should be. – user14717 Dec 27 '16 at 16:30
• Ah, thanks! I completely missed that reading it, sorry. – Martin - マーチン Dec 27 '16 at 16:46
• Everything should be ok. The first Al atom in the list (not shown above) is arbitrarily set at (0,0,0) so you could just add a constants to this (and same to all other atoms of course) if you don't like negative values etc. But if you leave things as they are and work out xyz coords using the known transformation matrices and the unit cell params given, you should have the structure. As it happens, in this structure the unit cell boundary cuts through the molecule. The unit cell is only the repeat pattern, it has no physical significance per se . – porphyrin Dec 27 '16 at 17:22
• @porphyrin: If you put this as an answer I'll mark it as accepted. As an aside, I suppose that chemists feel uncomfortable putting a slice through (say) a benzene ring and that if you want a conventional unit cell with a slice just mod it in? – user14717 Dec 28 '16 at 23:31

I completely agree with @porphyrin comment. This is an example of crystal data which formatting does not entirely meet IUCr requirements, that's all. This happens a lot, and in most cases it's no big deal. This particular set of coordinates is just not uniform within a single cell. Different crystallographic software uses different means to "reassemble" atoms "correctly" and write changes to *.cif, e.g. Olex2 uses Standardise function for that: