Posts Tagged ‘Clar islands’

Clar islands in a π Cloud

Wednesday, December 9th, 2009

Clar islands are found not so much in an ocean, but in a type of molecule known as polycyclic aromatic hydrocarbons (PAH). One member of this class, graphene, is attracting a lot of attention recently as a potential material for use in computer chips. Clar coined the term in 1972 to explain the properties of PAHs, and the background is covered in a recent article by Fowler and co-workers (DOI: 10.1039/b604769f). The concept is illustrated by the following hydrocarbon:

Clar islands in a polybenzenoid hydrocarbon

Clar islands in a polybenzenoid hydrocarbon

The Clar islands are shown in red, and represent in effect the resonance form of this species which maximises the number of aromatic electronic sextets possible to achieve via a cyclohexatriene resonance form. It encapsulates the concept that maximum stabilization is achieved when the π-electrons in the molecule cluster together (or localize) in cyclic groups of six (rather than eg other allowed values as predicted by the 4n+2 rule of aromaticity). As a historical note, although Clar popularized the concept in the 1970s, the (C) representation had in fact been introduced almost one hundred years earlier, by Henry Armstrong (DOI: 10.1039/PL8900600095). Many demonstrations that Clar islands are reasonably based in quantum mechanical reality have been made; a very graphical and convincing one is that by Fowler and coworkers in the reference noted above, using the calculated magnetic response property known as π current densities (although this shows that the six outer islands tend merge into a single continuous outer periphery).

Current density maps showing Clar islands (taken from DOI: 10.1039/b604769f
Current density maps showing Clar islands for the molecule above (taken from DOI: 10.1039/b604769f)

Previous posts on this blog have mentioned the application of another computed quantum mechanical property known as ELF, the electron localization function introduced by Becke and Edgecombe in 1990 (DOI: 10.1063/1.458517 ) and subsequently adapted for use with DFT-based wavefunctions. ELF is normally applied to help analyze the bonding in a molecule; the value of the function is normalized to lie between 1.0 (a simple interpretation is that this is the value associated with a perfectly localized electron pair) and 0.0. ELF has no association with magnetic response (the latter being an excitation phenomenon), but since the Clar islands can also be considered a localizing property of the π electrons, it is tempting to ask whether the ELF function can also reveal their characteristics (this question was first posed in DOI: 10.1039/b810147g).

The ELF function, as isosurfaces contoured at various thresholds

The ELF function, as isosurfaces contoured at various thresholds. Click for 3D

The diagram above shows the ELF function computed for the π-electrons of the molecule above (B3LYP/6-31G(d), as isosurfaces contoured at various values. At the value of 1.0, no features are discernible, but at 0.95 features which resemble basins associated with each atom centre have appeared, in the region of the 2p-valence atomic orbital on each carbon atom we regard as contributing the π-electron to the system. As the ELF threshold is reduced, these objects start to merge into what are called valence basins associated with bonds in the molecule. The outer periphery is the first to start coalescing. By a value of 0.75 (click on the diagram above to see a 3D model) the basins have merged to form seven clear-cut rings which happen to coincide exactly with the Clar islands. This feature persists down to a threshold of 0.55. Below this value, the seven individual basins merge into a single basin contiguous across the top (and bottom) surfaces of the molecule. One can also conceptualize the journey in the other direction. At low ELF values, the function is continuous, but as the threshold increases, it starts to bifurcate into separated basins. The first clear-cut bifurcation is indeed into the Clar islands, and this persists across a relatively wide range of ELF values, which suggests it is a significant feature. What is somewhat surprising is the close apparent correspondence of this way of analysing the electronic properties of the π electrons with their magnetic response computed via current densities. But association with aromaticity has previously been made (DOI: 10.1063/1.1635799). Thus Santos and co-workers have shown that the value of the ELF function at the point where it bifurcates from a ring into discrete valence or atomic basins can be related to other metrics of aromaticity. Here, that value is around 0.75 for the Clar basins, which is also within the range of values that Santos et al associate with prominent aromaticity (benzene itself has a value around  0.95).

A C114 PAH

A C114 PAH

The ELF function for the 114-carbon unit shown above again reveals prominent Clar islands, the inner heptet being very similar to the picture painted using current densities.

Clar islands in the  ELF function for a  C114 carbon PAH

Clar islands in the ELF function for a C114 carbon PAH

The final example involves diboranyl isophlorin (DOI: 10.1002/chem.200700046), a 20 π-electron antiaromatic system. Such systems are particularly prone to forming locally aromatic Clar islands as an alternative to global antiaromaticity (DOI: 10.1039/b810147g).

A Diborinyl system.

A Diboranyl isophlorin.

The ELF function is shown for both the neutral diboranyl system and its (supposedly more aromatic) dication. Here a mystery forms. No Clar islands are seen, and instead it is the periphery that bifurcates, at ELF thresholds of 0.5 for the neutral and 0.7 for the dication. The latter value clearly is that of an aromatic species, but the former is somewhat in no-man’s land, but certainly less aromatic that the dication. One for further study I fancy!

ELF Function for diboranyl molecules (red=neutral, green=dication). Click for 3D

ELF Function for diboranyl molecules (red=neutral, green=dication). Click for 3D

Does the ELF function have any possible advantage over the use of current density methods for analysing aromaticity? Well, the latter is normally applied to flat systems with planes of symmetry defining the π-system, and with respect to which an applied magnetic field is oriented. How to orient this magnetic field is not so obvious for prominently non-planar or helical molecules. Since the ELF function does not depend on the orientation of an applied magnetic field, it may be a useful adjunct for studying the properties of π-electrons in non-planar systems.

A molecule with an identity crisis: Aromatic or anti-aromatic?

Monday, April 13th, 2009

In 1988, Wilke (DOI: 10.1002/anie.198801851) reported molecule 1

A [24] annulene. Click on image for model.

A 24-annulene. Click for 3D.


It was a highly unexpected outcome of a nickel-catalyzed reaction and was described as a 24-annulene with an unusual 3D shape. Little attention has been paid to this molecule since its original report, but the focus has now returned! The reason is that a 24- annulene belongs formally to a class of molecule with 4n (n=6) π-electrons, and which makes it antiaromatic according to the (extended) Hückel rule. This is a select class of molecule, of which the first two members are cyclobutadiene and cyclo-octatetraene. The first of these is exceptionally reactive and unstable and is the archetypal anti-aromatic molecule. The second is not actually unstable, but it is reactive and conventional wisdom has it that it avoids the undesirable antiaromaticity by adopting a highly non-planar tub shape and hence instead adopts reactive non-aromaticity. Both these examples have localized double bonds, a great contrast with the molecule which sandwiches them, cyclo-hexatriene (i.e. benzene). The reason for the resurgent interest is that a number of crystalline, apparently stable, antiaromatic molecules have recently been discovered, and ostensibly, molecule 1 belongs to this select class!

So is 1 actually anti-aromatic? Let us look at some of the ways in which this might be estimated.

  1. One can inspect the bond lengths, measured from X-ray analysis. The longest is 1.463Å, labelled a above, and it corresponds to a single bond (value from the crystal structure).
  2. If the molecule had a bond alternating structure, the adjacent bonds would be expected to be much shorter, in the region of 1.32Å. In fact, they are rather longer, at 1.37Å. Indeed, in the cycloheptatriene part of the molecule, the alternation is much less than one might expect of an anti-aromatic molecule, oscillating between 1.37 and 1.43Å.
  3. One can also inspect aromaticity via a variety of magnetic indices. The simplest of these is the NICS probe. Placed at a ring centroid, a negative value of this index of around -10ppm indicates aromaticity (this is the value for benzene), whilst a strongly positive value (of up to +20 ppm) indicates anti-aromaticity. Molecule 1 has two potential centroids, one placed at the absolute centre of the system, and one placed at the centroid of the ~6-membered ring completed using bond b (in reality, the centroids were computed from the positions of ring critical points obtained from an AIM analysis). The NICS values at these two positions are both ~-4.4 ppm (See DOI: 10042/to-2156 for details of the calculation). These values does not indicate antiaromaticity! They could even be described as mildly aromatic. So what is going on?
  4. What about the chemical shifts of the other protons? All the hydrogens attached to sp2 carbons are predicted to resonate at around 6.7ppm (unfortunately Wilke does not report the experimental spectrum), which is typical of an aromatic system (anti-aromatic systems have high upfield shifts for such protons, at around 2ppm or even -2 ppm, see DOI: 10.1021/ol703129z for examples). The two protons of the methylene bridges are also quite different; 2.9 and 0.8 ppm. The latter is the proton endo to the cycloheptatrienyl ring, and is typical of a proton placed in the anisotropic magnetic shielding region of e.g. benzene. Thus the cycloheptatrienyl ring is itself behaving as if it were aromatic, whereas the overarching 24-annulene ring is certainly not behaving as if it were antiaromatic.

One possible explanation involves a concept known as homoaromaticity. The bond marked as b could be regarded as completing the 6π-electron local aromaticity of that ring (it would be formally considered as a 1π-electron bond, with no underlying σ-framework, see 10.1021/ct8001915 for further detail). Well characterised examples of such neutral homoaromatics are in fact very rare indeed; the phenomenon is thought to manifest mostly through cationic homoaromatics (i.e. the homotropylium cation, see 10.1021/jo801022b for discussion). So has the case been made for 1 being the first clear cut example of a neutral homoaromatic molecule, containing no less than four rings exhibiting this type of aromaticity?

There is one further concept that can be introduced. Clar (for a discussion, see DOI: 10.1021/cr0300946) proposed that benzenoid 6π-electron local aromaticity is preferred to less local or more extended cyclic conjugations, if the two compete. Many examples in a type of compound known as polybenzenoid aromatics are known where the most favourable resonance structure is that which maximises the number of Clar rings. More recently, quite a few ostensibly antiaromatic molecules have been shown to attenuate this unfavourable effect by forming instead groups of aromatic Clar islands containing delocalized benzene like rings (discussion of this point can be found at DOI: 10.1039/b810147g). In molecule 1, we could have a new phenomenon; a homoClar ring, formed to avoid antiaromaticity.

For further discussion, see the comment posted to Steve Bachrach’s blog.