I would like to perform DFT calculations and extract atom coordinates for calculations in Gaussian. I am using Mercury and have been able to isolate the atoms that I am interested in simulating. However, when I save the CIF files and open them again i see something very strange.

Atoms that have fractional occupancy from the XRD are shown to be fully occupied, this gives me nonphysical molecules that contain to many atoms. How can I get rid of these fractional occupied atom positions?

I am attaching an image to illustrate what I mean.

This is supposed to be a regular hydro-carbon branch, but the connectivity is wrong

When i retrieve the data from CCDC's webpage, the fractional occupied positions are marked with crosses, is there a way to clean the file?:

Before editing and trying to save the file, i have crosses to represent these fractionally occupied states

  • $\begingroup$ Technically, you are not allowed to "clean" the file since these positions are also occupied by atoms. And it's not "strange", most of structures contain atoms with fractional occupancies. Unless you are doing geometry optimization, you need to account for this. Alternatively, pick atoms from the same dominant site. I don't know about Mercury, but in Olex2 you can use showp <site #s> to hide minor occupancies. $\endgroup$
    – andselisk
    May 25, 2021 at 19:23
  • 2
    $\begingroup$ @andselisk In principle one may use CCDC Mercury to edit an orientationally disordered .cif into two (or more than two) ordered .cif (Mercury handbook, section 12.1 / Manually Editing a Structure). Compared to Olex2, it demands more attention. Depending on the later intended computation, it might not be needed, though (e.g., CrystalExplorer lets you choose how to deal with disorder in .cif when e.g., computing Hirshfeld surfaces.) $\endgroup$
    – Buttonwood
    May 25, 2021 at 19:46
  • $\begingroup$ @HanDen You state that the model model was published in a journal or / and in the CCDC database. If you add a reference to the primary publication, the CCDC deposition number (sometimes in the experimental section or mentioned in the SI of a paper) or the CCDC structure code (six Latin characters, on occasion two additional digits), one could have a look at this special entry, because of CCDC's special site. $\endgroup$
    – Buttonwood
    May 25, 2021 at 20:00

1 Answer 1


In general, OpenBabel may convert some of the information of a .cif into a Gaussian .gjf. If your are fine with using the CLI, the command would follow the syntax pattern of

obabel -icif input.cif -ogjf -O output.gjf

The parameters -icif and -ogjf hereby instruct OpenBabel about the expected input / output format of the files, while -O actually triggers the program to write the output into a permanent record. Further information may be obtained here.

Depending on the computation intended, the crosses Mercury displays may be very valuable information. It is Mercury's way to tell the user that the structure is disordered, i.e. there is more than one orientation of a molecule in the unit cell. For example, that a conformational flexible $\ce{-CH3}$ or $\ce{-CF3}$ group has been refined into more than one position (think staggered / eclipsed conformation of butane, there are more than one energetic minima, and one is the absolute one.) But this orientational disorder may extend over larger portions and complete molecules, too (e.g., small solvent molecules). It would be better if you could display more of the structure, and tell a bit in addition about the intended computation.

Do you know to read a .cif file about site occupancy factors? E.g., for molecules with an up / down or ($\uparrow$ / $\downarrow$) orientational disorder and only two realizations, these often are refined to add up to $p(\uparrow{}) + p(\downarrow) = 1.00$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.