[Jmol-developers] Re: [Open Babel] Covalent bonding radius + autobonding

[Geoff H. <ghu...@ws...>]

> I know that open babel makes other checks to determine whether or not a
> bond is appropriate. But Jmol does not currently do any other checks  
> ...
> if two atoms are close enough (but not too close) then they will bond.
...
> Q: Why are there non-zero values for the covalent radii of the noble  
> gases
> in openbabel?
>
> Q: Is there any reason why we should not set them to zero in Jmol?
...
> I am reluctant to set them to zero in Jmol unless that is also done in
> open babel. But, I am also somewhat reluctant to add more complexity to
> the bonding algorithms in Jmol. Its job is rendering, not calculating.

The checking is simply to ensure chemical consistency. It's more than  
just the noble gases! There are certainly cases where, simply from the  
XYZ coordinates, you cannot know if two atoms are bonded even though  
they are "close."

Examples include:
- certain types of hydrogen bonding (where the hydrogen is "close" to  
either partner)
- bridging hydrogen or halogen atoms between transition metal centers
- transition states

So Open Babel makes another quick pass to ensure atoms do not exceed  
their typical valency. If they do, the longest "bond" is killed.

But your question is "why are there nonzero covalent radii for noble  
gases?" Well, for one, there are noble gas compounds! Xenon is the most  
prolific, but other noble gas compounds have been discovered.

Furthermore, databases of covalent radii frequently include values for  
the noble gases:
e.g. http://www.webelements.com/webelements/elements/text/Ne/radii.html
http://www.webelements.com/webelements/properties/text/definitions/ 
covalent-radius.html

So my suggestion is that for a number of reasons, an "autobonding"  
algorithm needs to be more complex than just a nearest-neighbor search.  
But I can't imagine that this introduces a significant time delay in  
practice.

Cheers,
-Geoff