Incorrect breaking of Fe-F bond in molecular complexes

[schatzsc]

According to the current "disconnection rules", metal-halogen bonds are broken in a number of clearly discrete covalent compounds such as [Mo(O)Cl3F] or [FeF(N^N)] as discussed here since the combination of fluorine with transition metals leads to electronegativity differences which are higher than the threshold value for disconnection.

While in the case of [FeF(N^N)], one could generally do an additional check whether the other directly coordinated ligand atoms are "simple" in these sense of mono- or maximum mayby diatomics while bonds to more extended (and chelating ligands) are not to be broken, this addition would not work for [Mo(O)Cl3F], as all ligands here are monoatomics.

Possibly, the threshold value for disconnection either needs to be increased instead of simply calculating it from the electronegativity values or bond disconnection needs to be restricted to specific element-element combinations (or rules out in the case of compounds which are not homoleptic (= exclusively have ligand of only one kind).

[fbaensch-beilstein]

@schatzsc Thanks for pointing it out. We are looking for rules that point in the direction you mentioned, such as non-homoleptic compounds.

[gblanke02]

@schatzsc do you have an idea how a general rule set to handle the "non-homoleptic compounds" may look like. In the case of issue #159 you have 1 F and 2 C atoms bound to the Fe center. A simple rule may be to exclude these cases from disconnection in general although the difference of electronegativity ΔE may be high enough. Here, the "ligand atoms" directly bound to the Fe differ in their ΔE The rule may be that you do not break bonds in case at least one ΔE is below the threashold of 1.7
Another rule may be not to break any bonds to metal atoms if one or more ligands consists out of multiple atoms as you described it already above. (see issue #159 as well).
Let us define tis rule-set in our next working group session.

[schatzsc]

@gblanke02

The chelating "nacnac" (= 1,3-bis(ketimine) ligand is N^N and not C^C - it's simply the atom labels missing in the left part of the figure. But that does not matter at all, since currently only the difference in eletronegativity of the individual bonds is evaluated, if I'm not mistaken.

We need rules based on information that is encoded in the molfile or easily available in table format, which definitely restricts it to either electronegativity or or "topological arguments" (don't disconnect any chelating ligands, only disconnect some small diatomics or so).

But [Mo(O)Cl3F] and related compounds would only work with the first one, since all ligands are terminal - this would suggest to first try with tweaking/optimizing the EN differences a bit further.

[gblanke02]

The alternative for „But [Mo(O)Cl3F] and related compounds would only work with the first one, since all ligands are terminal - this would suggest to first try with tweaking/optimizing the EN differences a bit further.“ may be a rule that you do not disconnect any of the metal bonds in case at least one bond is described with ΔE < 1.7.

[ebla71]

@gblanke02

My best guess would be that that this is not too far from "don't disconnect anything" if the central atom is a transition metal ;-)

[ebla71]

@gblanke02

Alternatives: don't disconnect if the ligand set is heteroleptic, regards of what the EN differences are?

This would only disconnect the simple halide salts plus aquo complexes, I guess.

[gblanke02]

@ebla71
Unfortunatley tle last rule would go to far most likely because the standard salt handling of InChI may be affected by that. As a consequence InChI of organic compounds may change which we must avoid. But we need further testing for any rue set.

[gblanke02]

After a discussion with @schatzsc the following 2 additional rules may handle the issue:

• In case one bond between a metal atom and a ligand atom must be kept according to the ΔE rule, all ligand bonds are kept and the ΔE rule is not further used.
• In case that a ligand atom links two metal atoms all bonds to the metal bonds are kept, i.e. ΔE rule is not applied.