NNNS Chemistry blog
Prevous: Most cited part I
Next: Copper-free Sonogashira coupling

All blogs

New ruling on stability

8 August 2008 - reality check for computational chemists

In a recent essay Roald Hoffmann, Paul von Rague Schleyer and Henry F. Schaefer (HSS) invite fellow computational chemists to a reality check (DOI). They observe that too often a hypothetical molecule is declared stable (example cyclic ozone or diatomic carbon) even when it is obvious you wouldn't be able to fill a bottle with it. The authors are also annoyed with the generous use of significant figures in papers on computational chemistry to the point of being silly and they also have some serious issues with computational accuracy and precision. Fortunately they have some solutions on offer.

From now on, according to the authors, the use of stability in describing hypothetical molecules is discouraged and replaced by viable and fleeting.

A viable molecule is a molecule that meets computational criteria of persistence appropriate to ambient conditions in a typical chemical laboratory environment, namely its isolation in condensed phase, near 1 atm pressure at room temperature (perhaps in the presence of a moderately humid atmosphere), or in reasonable concentration (say 0.001m) in solution. And, with that, a half-life of a day or longer. Additional criteria stipulate that such molecule will not fragmentate or dimerize, that the molecule has well defined vibrational frequencies, a sufficiently largeHOMO/LUMO gap and that an ion has a well characterized counterion. A fleeting molecule then is one with a local minimum, with barriers, (...) preventing escape from its local basin, (...) of (...) importance in interstellar chemistry, or (...) low-temperature matrix-isolation and high-vacuum environments.

More definitions redefined: precision assesses the degree to which a particular computation approaches the exact result that should have been obtained with the specified method and basis set. Precision only means that you correctly accomplished what you claimed. In this particular expect maximum precision of around 0.6 kcal/mole. Accuracy simply means absolute theoretical accuracy and because the authors note that results are not impressive (for example for benzene accuracy is of by 6 kcal/mole with the best methods) the phrase should be used only very rarely.
With respect to significant numbers HSS rather arbitrarily recommend 1.25 and not 1.254 angstrom for distances and 40.3 and not 40.269 kcal/mole for activation energies.

The Angewandte essay is accompanied by no less than 4 referee reports (a novelty) from other chemists, three of which basically agree (Bickelhaupt, DOI), (Koch, DOI), (Reiher, DOI) and one of them definitely disagreeing. In a comment, simply called No Important Suggestions (DOI), Gernot Frenking expresses his disappointment at the intellectual level of the HSS essay feeling like a guest who was invited to a gourmet restaurant where he was only served a thin soup. According to Frenking prominent molecules relevant to interstellar space are not just fleeting but very viable, he also questions the new counterion rule reminding the reader of the existence of noncoordinating anions. Frenking also asserts that most computational chemists already behave responsibly with respect to the number of decimals with which a result is presented and that only newcomers occasionally go ballistic.

This last point about chemists behaving badly can be resolved by selecting at random some recent computational chemistry articles. This blog can only conclude it is way too easy to find offender examples. Schuquel et al. (Chemical Physics 2008 DOI) present ionization energies like 117.203292 au (clearly an offence in the new HSS regime) and interestingly when measured in eV they present values like 9.21 and 10.27 as if the precision of the method mysteriously increases beyond the value of 10. Lopez at al. (J. Phys. Org. Chem. 2008, DOI) are presenting free energies in KJ/mole with up to 4 significant figures. And finally Jia et al. (Org. Lett. 2008, DOI) present bond lengths with 4 significant figures and bond angles with 5.

Roald Hoffmann, Paul von Ragué Schleyer, Henry F. Schaefer (2008). Predicting Molecules-More Realism, Please! Angewandte Chemie International Edition, 47 (38), 7164-7167 DOI: 10.1002/anie.200801206