Interactions responsible for the lowest energy structure of the trimer of fluoroethanol.

Steve Bachrach on his own blog has commented on a recent article[cite]10.1002/anie.201505934[/cite] discussing the structure of the trimer of fluoroethanol. Rather than the expected triangular form with three OH—O hydrogen bonds, the lowest energy form only had two such bonds, but it matched the microwave data much better. Here I explore this a bit more.

The stability of the lowest energy form, as is evident from the title of the article, was attributed to unusual H-Bond topology and bifurcated H-bonds as teased out from bond critical points in the QTAIM analysis of the topology of the electron density. Here I add to this analysis by displaying the computed NCI (non-covalent-interaction)[cite]10.1021/jp204278k[/cite] surfaces, as you might see in the comment I posted on Steve’s blog. In essence, the QTAIM had revealed bond paths connecting an oxygen to a H-C and also a bifurcation from an F to two H-C atoms, shown with orange lines in the diagram there. What might an NCI analysis reveal? The analysis[cite]10.14469/ch/191558[/cite] is shown below, where I have added orange arrows to indicate the location of these bond paths. The arrows point to an NCI feature which corresponds to a weak dispersion-like stabilisation.

Click for 3D

Click for 3D

However, as you can spot from the diagram above (and inspect in a 3D sense if you click on the diagram above to load a 3D Jmol model), there are many more regions where NCI features appear. The most obvious are the blue-coded ones, which in fact represent the conventional O…HO hydrogen bonds, but there are plenty of others as well, including a cyan one which is not part of the published attributions. I will recapitulate my comment on Steve’s blog; the point I make here is that apart from the two regions which have been picked out in the article as responsible for stabilisation of the low energy structure, there are around 4-5 OTHER regions that also may be stabilising but for which there is no corresponding critical point in the density. So whilst the above origins are not incorrect, they may well be very incomplete!.

There is a tendency to only highlight features which can be named, and perhaps to ignore or pay less attention to those which have no name. The latter may in fact be more common than we imagine, and cumulatively they can often have a big impact.

Postscript: A structure has recently been reported[cite]10.1002/anie.201403599[/cite],[cite]10.5517/CC128NYY[/cite] illustrating an exceptionally strong OH…F interaction of 1.52Å. This is noteworthy because such hydrogen bonds are rarely strong and indeed even their very existence is controversial. The cyan NCI region mentioned above is just such an interaction (of length ~2.0Å).

Tags: , , , , , , ,

4 Responses to “Interactions responsible for the lowest energy structure of the trimer of fluoroethanol.”

  1. Henry Rzepa says:

    I noted above the surprising observation that O-H…F-C hydrogen bonds are rare and controversial. Here is a plot obtained by searching the crystal structure database for such interactions (the C is 4-coordinate). The hotspot is ~2.25Å (with an angle subtended at the H of only 100°) and the shortest (as we saw above is 1.52Å, with a small number of examples in-between these two values.


  2. Henry Rzepa says:

    I noted above (refs 4,5) that a molecule with a very short O-H…F-C bond has recently been reported. Here I provide the NCI analysis. Its shows a focused blue region corresponding to a genuinely strong hydrogen bond (blue arrow), and with an interesting green (weak) region (green arrow) reinforcing the interaction. The former does have a name, whereas the latter is nameless, but their effect should be considered as cumulative.


  3. Henry Rzepa says:

    For good measure, here is the NCI computed for the C3 symmetric form of the trimer, with three O-H…O hydrogen bonds. These appear as blue below, whilst weaker F…H-C interactions are in green.


  4. Henry Rzepa says:

    My first comment/addendum related to a search of the crystal structure database for OH…F-C instances. Here I add variations. First, OH…F-X, where X is any 4-coordinate atom.

    This shows a quite different hotspot for the H…F distance of around 1.75Å. The majority of these hits are in fact X=B. When the search is restricted to X=3A (column 3A of the periodic table), a very similar pattern emerges. So these types of hydrogen bond to F are common after all, just less so with O-H…F-C.



Leave a Reply