An interesting little discussion on the meaning of “protobranching” appears in a comment1 and reply2 in J. Phys. Chem. A. Fishtik1 calls out the concept of protobranching on three counts:
- It is inconsistent to count a single protobranch for propane, but then not have three protobranches in cyclopropane
- It is inappropriate to utilize methane as a reference species.
- Group additivities work well.
I tend to side more with Schleyer2 in his rebuttal of these charges, and so will present from this perspective. First off, Schleyer argues that he can define protobranch anyway he wants! (He in fact cites a quote of Humpty Dumpty from Lewis Carroll to support this stance!) Schleyer is of course correct. Fishtik should really have argued “Does Schleyer’s definition of protobranch add to our understanding of strain?” So Fishtik claims that there is an internal inconsistency in Schleyer’s definition – taking the view point that the C-(C)2(H)2 group is identical to the protobranch. Schleyer counters that no, the protobranch is this group along with the caveat that the two terminal carbons are not connected, like they are in cyclopropane. I really prefer Gronert’s approach here – where he argues for just what are the implications of Schleyer’s definition (see this post).
Fishtik refuses to use methane as a reference since it is a unique molecule. Again, if one takes the group-centric view, then methane possesses a group that no other compound has. But Schleyer counters that one is free to choose whatever reference one thinks is appropriate, just be sure to understand what properties are conserved or not conserved when using that reference selection. To me, this is really the key for the entire discussion: choose one’s references in such a way as to minimize differences between your reference compound(s) and the molecule(s) you are trying to explore to just the property of interest. So, if one is interested in quantifying ring strain, the reference compounds should be not only be strain-free but they should differ in no other way from the cyclic molecule other than the presence of the ring! Unfortunately, there is no unique or non-arbitrary way to do this! Schleyer’s approach and Fishtik’s approach differ in just what properties they believe are important to conserve and which properties they are going to lump into the concept “ring strain”.
Fishtik shows a whole slew of reactions that demonstrate the consistency of group additivity methods. Schleyer correctly points out that these examples are really intimately related and represent only one type of definition. Again, there is really no unique set of references, and many, many different models have been developed, all of which can match experimental data quite well – like for example heats of formation. The key is what these models say in terms of interpreting, say, these heats of formation. Can one rationalize trends and make predictions with the model? If so, then it has utility. If not, then the model should be discarded. Ultimately, Fishtik’s argument is that the protobranching model does not assist us in understanding strain – Schleyer would obviously beg to differ!
References
(1) Fishtik, I., "Comment on "The Concept of Protobranching and Its Many Paradigm Shifting Implications for Energy Evaluations"," J. Phys. Chem. A, 2010, ASAP, DOI: 10.1021/jp908894q
(2) Schleyer, P. v. R.; McKee, W. C., "Reply to the "Comment on ‘The Concept of Protobranching and Its Many Paradigm Shifting Implications for Energy Evaluations’"," J. Phys. Chem. A, 2010, ASAP, DOI: 10.1021/jp909910f
Henry Rzepa responded on 16 Apr 2010 at 4:19 am #
I realize that I should really read all the articles on this theme from start to end before asking the following, but perhaps someone familiar with all the tiny details can enlighten me here?
Recently having had to write a new course on conformational analysis (an honour, since the first ever version of this course was presented by one of the originators of the concept himself, Derek Barton, in the same institute), I decided to check as many assertions as I could using a proper long-range dispersion corrected method (which in fact Steve alerted me to); ωB97XD. In an alkanes, the contributions to the dispersion term mostly originate from H…H contacts in the range 2.2-2.4Å, and these are very approximately stabilizing by about 0.3 kcal/mol per contact at this sort of distance. The effect accumulates, and by about C50 it is thought to add up to ~12 kcal/mol.
My question is the following: in all this discussion about proto-branching, are the effects of all the close dispersion H…H contacts factored out, or are they in fact an essential part of the analysis?
By the way, the dispersion terms are computed purely on distance (not orientation), and its interesting that the van der Waals radius of H is nowadays taken as 1.2Å when the contact is to another H, but is reduced to 1.1Å when the contact is to a non-hydrogen. I do wonder how many molecular visualisation programs make this distinction, and indeed whether it is fully absorbed into all these dispersion corrected DFT methods?
Computational Organic Chemistry » A Protobranching model? responded on 15 Jun 2010 at 8:04 am #
[…] and co-workers have added to the protobranching debate (see these earlier posts i, ii, iii) with a proposal for how branching can be stabilizing.1 A normal chemical bond can be […]
Computational Organic Chemistry » Phenylhydroxycarbene responded on 01 Feb 2011 at 1:00 pm #
[…] Allen reported on their third study of hydroxyl carbenes (see these posts on dihydroxymethylene and hydroxymethylene), this time examining phenylhydroxycarbene.1 As I covered in my book, there is a lot of work on […]