13 March, 06

 

Greetings All-

 

After reading over the preliminary document circulated by Arunan, Joe Dannenberg’s reaction, and so on, I thought I might chime in with my thoughts.

My first and over-arching opinion is that it is not going to be possible, or even desirable, to write down a set of strict criteria that each and every H-bond must conform to. It would seem to me much more sensible to propose a set of principles, fundamentals, ideas. Interactions could be classified as H-bonds if they fit most of these criteria, but not necessarily all. This will no doubt still leave some interactions on the borderline, with conflicting opinions as to their nature. But maybe that is best. I don’t think we want to be too rigid on this matter, and some controversy on borderline situations is a good thing for science. One example that arises is Joe’s mention of a particular system which was considered an H-bond from a density vantage point but not from a spectroscopic perspective.

As another example, for years, H-bonds were only considered so if they underwent a red shift of the X-H stretching frequency. And that made sense. At least for a while. But then along came blue-shifting H-bonds. Their violation of this simple rule led some to propose they were not H-bonds at all. On the other hand, others noted that these interactions conformed to a host of other H-bonding principles, e.g. bond strength, directionality, NMR shifts, electron density shifts, and so on. Consequently, most now classify these blue-shifters as H-bonds, even with a violation of one generally observed phenomenon.

I suspect a similar idea might apply to studies of electronic structure. I think many would concede that Bader AIM analysis of electron density is a useful tool. Certain properties and values do appear to conform nicely to other markers of H-bond presence, and even strength. But like a red shift, I don’t think this sort of model should be approached as anything more than another indicator. Deviations from normal expectations in the electron density analysis should not in and of itself disqualify an interaction as a H-bond, nor should conformity to expectations be taken as incontrovertible proof of the validity of that classification.

Another issue has to do with the means of obtaining information. Experimentalists typically do not have access to the details of the electron density, and are likely to be skeptical of its analysis in a particular way as being a sole arbiter, particularly if they must depend on a quantum calculation to extract this information. I expect they will be far more comfortable with indicators which they can easily measure, like NMR shifts or vibrational frequencies. And how would even theoreticians deal with an observation that the classification of a system as H-bond reverses if the basis set is changed? Or if the means of analysis, e.g. AIM vs natural population, provide different interpretations?

So I would propose we settle on a list of properties that would be common to most, but necessarily all H-bonds. These should probably include properties dealing with energetics, geometries, spectral properties, electron density, maybe others. Some of these ideas have been considered for years, so I concentrate on certain questions to be considered, refinements.

 

Bond Strength, i.e. Energy. How weak can an interaction be and still be considered a H-bond? 1 kcal? 2? or would any quantity suffice, no matter how small, as long as it is attractive? How about on the other end, the max side? When HF is paired with F-, one gets a very strong interaction, ion-neutral, likewise for HOH + OH3+. Can an ion pair also contain a H-bond. Let’s say OH3+ approaches OH-. Granted, there will tend to be a proton transfer forming a neutral water dimer, which will of course contain the classical H-bond. But prior to the transfer, is this a H-bonded system, or has the interaction energy maxed out, and we refer to it simply as ion-ion?

ΔE035 kcal ?50?minmax

 

Position of Proton The very strong complexes, e.g. FHF-, brings up the question of whether the proton position makes a difference. With the equilibrium proton position smack dab in the middle, is this really a H-bond? And if so, what are the two partners? We can’t really call it FH and F- since we can’t associate the H any more closely with one F than the other. The usual way of talking about such things is to think of the complex as formed from HF + F-, and after the H-bond has formed, the proton moves to the middle, possibly strengthening the interaction further.

 

Geometry Of course, not all H-bonds are free to adopt their most stable, optimal geometry. Intramolecular bonds within large molecules, or intermolecular bonds in crystals can be severely strained. Many of us have been confronted with a particular strained interaction within a crystal structure and been asked “Is that a H-bond?”

So, how much distortion is allowed before the H-bond ceases to be one? Does OH··O still constitute a H-bond when R(O··O) has been stretched to 3.5 Å? 5 Å?

And likewise for angles. Consider the carbonyl C=O as an example. Clearly H1 is within the H-bonding range. What about H2? How large can θ become before this is no longer a H-bond?

COH1H2θ

And what about motions of the proton out of the carbonyl plane that contains the lone pairs? We have all seen numerous suggestions in the literature for limits on distance and angles, but they are arbitrary to a large degree.

Another issue which relates geometry to energetics deals with linearity. I think most would consider a criterion of an H-bond as an interaction which penalizes the system energetically if

the bridging H is moved off of the internuclear axis. But by how much? Should the attractive nature of the interaction vanish entirely if the H is moved away by 30º? 45º?

 

Electron Density Referring now to the proposed definition concerning an “electron deficient” H and an “electron rich” region, how much so? By which I mean, how deficient must the H atom be? For example, if a C-H proton, the charge on the H would normally be computed as fairly small, particularly for sp3-hybridized C. Some theoretical frameworks might indicate a + charge, others -. Which method takes precedence? Or if the charge of the H is evaluated by experimental means, again which one?

With respect to the electron rich region, we can all agree on the idea of a lone pair on a particular atom, a π bond as in ethylene, or even a delocalized π-cloud as in benzene as a suitable proton acceptor. But it seems to me that using this same classification for a molecule like methane (“the plane formed by 3 H atoms”) may be stretching the definition too much. Could the same classification be extended to a σ bond, or what about any density in a σ* antibond? What about a spherically symmetric electron cloud as in Ne---HF? Can we take this idea too far, and make it so inclusive that ANY electron density at all is considered an electron rich region?

And of course, we are also confronted with the question as to how specifically to analyze the electron density. Should the AIM approach be used? Should natural populations be used to quantify density in different orbitals? Or should density shifts be examined pictorially?

In summary, I raise the latter issues as items which bear some discussion. But I emphasize that my primary opinion is that we propose a list of properties that are characteristic of H-bonds, but do not demand a system satisfy each and every one. These potential exceptions would enable us to provide some quantitative guidance in each category, without the fear that doing so would disqualify an otherwise very deserving H-bond.

Steve Scheiner