Odd looking InChI calculation for an octahedron that fulfills all calculation rules for the molecular inorganics

[gblanke02]

The attached octahedral amino-bromo-chloro-fluoro-hydroxy-methyl-molybdenum is a "handmade" 3D structure for further testing of 3D structures. During the InChI calculation the F-Mo bond is correctly broken according to our EN table while all the other ligands stay bonded within this chemical complex. The allowed valences for Mo allow the break of bonds up to a valence of 6 which is found in our Mo complex.

Although evertyhing is according to our rules the result looks odd because there is only 1 ligand of the orginally 6 ones that is broken. But from the chemical perspective I wuld have expected to keep all bonds incluing that one to the F ligand.

We should discuss the putput for this structure from a chemical perspective again.

amino-bromo-chloro-fluoro-hydroxy-methyl-molybdenum.mol.txt

[schatzsc]

@gblanke02

Simply taking the group number as the highest allowed "valence" where breaking of bonds is allowed certainly does not make any sense for the group VI elements (Cr, Mo, W), as the MX6 halides are clearly all discrete molecular compounds:

https://en.wikipedia.org/wiki/Molybdenum(VI)_chloride

[schatzsc]

@gblanke02

Even a number of five would be too high, since MoOCl4 also is molecular and should not be broken into anything (MoO + 4x Cl?!?)

https://en.wikipedia.org/wiki/Molybdenum_oxytetrachloride

[schatzsc]

@gblanke02

Another example where breaking of bonds is not correct in my opinion is MoOCl3F, which gives an InChI string of InChI=1/Cl3MoO.FH/c1-4(2,3)5;/h;1H/q+1;/p-1 even with the "Include bonds to metals option":

[schatzsc]

@gblanke02 Personally, I think that for any "valence" higher than three, no bonds should be broken at all, at least for the transition metals and f-elements - and even for a "valence" of three we need some exceptions, since there are clearly 3-coordinated molecular iron complexes for example, if you use sufficiently bulky substituents on a tridentete ligand:

A review from the Holland group back from 2008 here:

https://pubs.acs.org/doi/10.1021/ar700267b

Example with Fe-F bond here:

[schatzsc]

@gblanke02 Maybe need to restrict the breaking of bonds to homoleptic complexes, where all ligands are the same (and thus also no stereoisomers are possible)?

[schatzsc]

Molfile here:

Fe_N2X.mol
  ChemDraw10202521082D

  0  0  0     0  0              0 V3000
M  V30 BEGIN CTAB
M  V30 COUNTS 27 29 0 0 0
M  V30 BEGIN ATOM
M  V30 1 C -0.714471 0.894320 0.000000 0
M  V30 2 C -0.714471 0.069320 0.000000 0
M  V30 3 C 0.714471 0.069320 0.000000 0 CHG=-1
M  V30 4 C 0.714471 0.894320 0.000000 0
M  V30 5 C 0.000000 1.306820 0.000000 0
M  V30 6 C -1.428942 -0.343180 0.000000 0
M  V30 7 C 1.428942 -0.343180 0.000000 0
M  V30 8 C -1.428942 1.306820 0.000000 0
M  V30 9 C 1.428942 1.306820 0.000000 0
M  V30 10 C -2.143413 0.069320 0.000000 0
M  V30 11 C -2.857884 -0.343180 0.000000 0
M  V30 12 C -2.857884 -1.168180 0.000000 0
M  V30 13 C -2.143413 -1.580680 0.000000 0
M  V30 14 C -1.428942 -1.168180 0.000000 0
M  V30 15 C 1.428942 -1.168180 0.000000 0
M  V30 16 C 2.143413 -1.580680 0.000000 0
M  V30 17 C 2.857884 -1.168180 0.000000 0
M  V30 18 C 2.857884 -0.343180 0.000000 0
M  V30 19 C 2.143413 0.069320 0.000000 0
M  V30 20 C -2.555913 0.783791 0.000000 0
M  V30 21 C 2.555913 0.783791 0.000000 0
M  V30 22 C 2.769439 1.580680 0.000000 0
M  V30 23 C 3.352801 0.997317 0.000000 0
M  V30 24 C -2.769439 1.580680 0.000000 0
M  V30 25 C -3.352801 0.997317 0.000000 0
M  V30 26 Fe 0.000000 -0.343180 0.000000 0
M  V30 27 F 0.000000 -1.168180 0.000000 0
M  V30 END ATOM
M  V30 BEGIN BOND
M  V30 1 2 1 2
M  V30 2 1 3 4
M  V30 3 2 4 5
M  V30 4 1 5 1
M  V30 5 1 2 6
M  V30 6 1 3 7
M  V30 7 1 1 8
M  V30 8 1 4 9
M  V30 9 2 6 10
M  V30 10 1 10 11
M  V30 11 2 11 12
M  V30 12 1 12 13
M  V30 13 2 13 14
M  V30 14 1 14 6
M  V30 15 2 7 15
M  V30 16 1 15 16
M  V30 17 2 16 17
M  V30 18 1 17 18
M  V30 19 2 18 19
M  V30 20 1 19 7
M  V30 21 1 10 20
M  V30 22 1 19 21
M  V30 23 1 21 22
M  V30 24 1 21 23
M  V30 25 1 20 24
M  V30 26 1 20 25
M  V30 27 1 2 26
M  V30 28 1 26 27
M  V30 29 1 26 3
M  V30 END BOND
M  V30 END CTAB
M  END

[fbaensch-beilstein]

@schatzsc Thanks for the example. Would you be so kind and create a new issue for the wrong Fe-F bond disconnection stuff. And if possible provide some more examples where this happens, not only Fe-F maybe.

[schatzsc]

@fbaensch-beilstein

Happens with any M-F combinations where the electronegativitiy differences is too high.

If you don't insist on compounds which are really structurally characterized or reported in the literature, take for example:

[M(NH3)2F] for M = Fe, Ru, Os and also M = Cr and Mo but not W which gives InChI=1/FH6N2W/c1-4(2)3/h2-3H3 instead of InChI=1/FH.H6MN2/c;2-1-3/h1H;2-3H3/q;+1/p-1 for all the others.

Also M = V and Nb are disconnected but M = Ta is not

Further to the left, only M = Ti is disconnected but M = Zr and Hf are not (all again for [M(NH3)2F])

In the cobalt group, it is even less systematic, as M = Co and Ir are disconnected, but the M = Rh inbetween not

And for M = Ni, Pd, Pt all M-F bonds remain connected, as well as all other 3d to 5d elements further to the rigth

[schatzsc]

@fbaensch-beilstein

If you vary the halogenido ligand instead of the metal for [Ti(NH3)2X] for example, you get disconnection for X = F but not Cl, Br, I

For all other M-X combinations where X is Cl, Br, I also not disconnection at all

[gblanke02]

See proposal in issue #166: Ig at least one lingand bond must be kept, all bonds between the metal atom and lingands are kept. In that case all bonds in the MoOCl3F are kept because Mo-Cl is below the ΔE threashold.

[nnuk]

As per the proposed rules in GHI # 166 the issue has been resolved and can be tested on InChI Web Demo.