Discussion List Archives

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Discussion #2

  • To: Chemical information in core CIF <corecifchem@iucr.org>
  • Subject: Re: Discussion #2
  • From: Howard Flack <Howard.Flack@cryst.unige.ch>
  • Date: Thu, 18 Dec 2003 17:08:56 +0100
  • Organization: University of Geneva
  • References: <3FCBAA43.6070409@mcmaster.ca>

HDF's contribution to Discussion #2

Herbert and David are both on the right track. 

Herbert suggests a very general view to the incorporation of molecular information within the core dictionary and I'm sure in the long run that suitably coded information of all these types would be of the greatest of use. However it looks like such a vast programme that it might be difficult to summon up the courage to get going. Herbert's comments fit in well with the general conceptual problems of atoms, bonds and molecules. Bader's work tries to make sense of what is meant by atoms in molecules. Solution of Schroedinger's equation naturally leads to energy levels and wavefunctions, even the electron-density distribution, however the notion of the 'bond' is not directly evident.

David's atom and connectivity information certainly has its place within the molecular cif and should definitely be incorporated.

> Peter Murray-Rust's CML defines a molecule as a composed of atoms,
> but leaves it to the author to state which atoms.

  I think it is a very wise approach in the atomistic view of molecules to allow an author, one or several free choices of which atoms constitute the molecules of interest to the author. It is important in the specification of the absolute configuration of a molecule to state which molecule, as a set of atoms, one has in mind. The same applies to specifying 'molecular' symmetry and notions such as Z' (per molecule).

  I would slightly re-express David's rendering of the PMR 'definition' of a molecule to read: One property of a molecule is that it is composed of atoms. There are several reaons for such a change.
 - Whilst a molecule, in this view, is a set of atoms, not all sets of atoms are molecules. 
 - Another property of a molecule, dear to IDB, is that the atoms are bonded together in various ways.
 - Moreover, see next paragraph, I think it is advantageous to allow CIFs to provide a means of defining set of atoms which no one would want to describe as or think of as being molecules. This recalls IDB's problem of 'dummy atoms' which no-one at all thinks of as being atoms. On this premise CIF needs a means of defining sets of atoms, and giving a label, name, type of set (e.g. molecule, tecton, building unit, etc), and various other set properties.

  I have some misgivings concerning IDB's bond graph methodology. Apart from the difficulty of being able to find a universal way of defining bond strength, I do not think that it may always be possible to make a complete description of all of the molecular entitities that one needs based on a hierachical choice limited by a cut-off value of bond strength. The bond graph methodology is important but it is not the only way to look at structures. You will find the structure of low-temperature quartz described as being formed of slightly-distorted Si-centered tetrahedra of SiO4 where the O atoms at the corners are shared by two tetrahedra. At a larger scale of description, one sees that these tetrahedra are arranged in such a way as to form various sorts of helices. These tetrahedra and helices are clearly not molecules but their 'existence' is important for the description of the structure. The bond graph method, according to the chosen cut-off in bond strength, tells only that the low-quartz is composed either of isolated Si and O atoms, or that the whole crystal is a molecule. A similar case arises for the high-temperature superconductors where the polyhedra that one sees drawn in all diagrams are not those corresponding to the shortest (strongest?) bonds.

  In short, it seems to me that CIF needs a way of defining sets of atoms of the authors choice, and giving a label, name, and properties to this set of atoms.

(PS I wrote to the person at NIST who is developing SpectraML. I have as yet not received any reply.)
coreCIFchem mailing list

[Send comment to list secretary]
[Reply to list (subscribers only)]