Dear Armand Budzianowski

Normally I or C in the monoclinic case:

https://scripts.iucr.org/cgi-bin/paper?S0108768103002829

A different centering may be used if you want to compare similar structures which belong to different crystal systems; for example with a structure in Aba2

Guido

Dear Guido,

I think checkCIF nowadays even suggests that the setting with the monoclinic angle as close as possible to 90° be used.

Best regards,

Rudi

Dear Rüdiger W. Seidel ,

thank you very much for this hint. But I still support the use of C and I. If all centering types are used, comparision and the reviewing process is unnecessarily complicated, to my point of view.

Technically it is not a problem to refine a structure in a setting, which shows a monoclinic angle, which is a few degrees apart from 90°.

Greets

Guido

Some late answer:

Previous "recommendations" were to use only C-centered cells for exactly this reason: to limit the number of "names" for identical crystal systems.

However, more recently, a view emerged that always using C-centered cells can lead to unit cells with the beta-angle further away from 90 deg than in the I-centered cell. Since the value of the beta angle is important data to judge, for example, the possibility of twinning or of phase transitions to orthorhombic, it is now recommended to choose either C- or I-centered cells based on the value of the beta-angle (actually, so that a and c are the shortest possible basis vectors, which should be more or less the same thing).

I personally agree with this view, but it is not "law". You can still use C-centered cells only. Even Acta Cryst., in its note for authors, only uses "preferable": "Note that space group settings like P21/n and I2/a are usually preferable to P21/c and C2/c, respectively, when the former lead to unit-cell β angles that are closer to 90° than the latter."

Later edit: "Re: An alternative setting for Ia space group in monoclinic crystal system are: A11a, Aa, An, B11b, Cn, Fd11, Ic, whereas space group Cc is the most popular in the organic crystal structures (CCDC)."

I agree with Guido that we should limit ourselves to C and I-centered lattices. Use of any other lattices does not add any advantage:

- A-centered lattices can be transformed into C-centered by simple a-c axis exchange. The angles stay the same.

- A B-centered lattice is normally "wrong" with b as a unique axis. It can be transformed into a smaller monoclinic primitive lattice without any loss of symmetry. The "B11b" space group uses c as the unique axis and I think we stop doing this in... 1783?

- I would need to look it up, but no way F-centered monoclinic cells are not redundant and could be reduced into a smaller cell.