Skip to main content

Analytical Chemistry

The Return of Kekulene

Kekulene! This is one of those molecules that someone who’s learning organic chemistry might sketch out on a whiteboard, wondering if it really exists. It does, but it’s not like we have a lot of recent information about it. There was a preparation of it in 1978 (from the Staab group at the Max-Planck Institute for Medical Research in Heidelberg), but that one-off is it for experimental data. As you would imagine, the synthesis was a taxing and low-yielding one, so there’s not exactly a big jar of the stuff sitting on a shelf somewhere in Germany. If you want to have a look at it by more up-to-date instrumental techniques, you’re going to have to make it yourself.

That’s what this group (a team from Spain, Portugal, and the IBM center in Zürich) did, and many of you will have guessed that the IBM folks are in there in order to provide a look via high-end single molecule atomic force microscopy. That’s just the sort of thing you’d want for this molecule, because for many years there was a debate about what sort of structure it has. You can draw it two ways, as shown above: as six aryl rings connected by bridging groups or as a fully delocalized collection of conjugated double bonds, and those two options are going to give you very different patterns of bond lengths (aromatic rings alternating with nonaromatic ones versus pretty much everything the same). I note that Wikipedia has it drawn in the latter form. (Edit: actually, it’s correct).

The original synthesis is quite an achievement; this is not the kind of molecule that’s just going to zip itself together. It has eleven steps, and as the current authors note, the last four go in about 20% yield, and the middle three are 23% overall, but the first four. . .well, that intermediate (2 in the scheme at right) is only produced in 2.8% yield, which is a pretty fearsome beating to start off your synthesis with. I realize that it’s probably better than taking a hammering in the last four steps, after you’ve already put the work in, but still. These authors have done a lot of work in the polyaromatic field in recent years, and they have a much-improved aryne Diels-Alder route that’s one step from commercially available starting materials. It gives you a mixture of regioisomers, true, but that can be separated by SFC or by repeated crystallization, either of which are a lot more enjoyable than the previous route. From there, the authors mention trying to explore some newer methods to get to kekulene, but ended up just pushing through the seven steps of the 1978 route, which worked exactly as described.

When examined by AFM at various tip heights above a single molecule, the structure becomes clear (see at right; the lower image is the upper one after Laplace filtration, which highlights edges and other sudden contrast changes in an original image). What you can see is that alternating rings have different character (shape, size, and AFM response), and the tip seems to be picking up the isolated bridging alkenes as higher-contrast features. That argues strongly for structure 1a in the scheme above, and it’s in agreement with the X-ray bond lengths determined by the Staab group. So the delocalized structure just doesn’t reflect reality; kekulene is indeed a sextet of benzene rings stitched together by non-aromatic bonds, and we now have experimental proof by two different methods.

I should note the experimental difficulties in obtaining any of these data. In the current work, the group had to sublime kekulene onto a copper surface by rapid heating under high vacuum, at high enough temperatures that most of the molecules broke up – scanning the AFM tip around showed mostly “small and often mobile” molecules on the surface, which are surely shattered pieces of the original. But they tracked down a few intact molecules by searching around on the copper-atom plain, whose snowflake form immediately stood out. Meanwhile, back in 1978-79, Staab and co-workers crystallized the stuff from pyrene, of all things, heating the mixture up to 450C in a sealed tube and gradually cooling it to 350 to get yellow needles of kekulene crystals to form at that relatively chilly temperature. They then removed the pyrene by careful sublimation to leave the pure crystals behind, and you can be sure that this was not exactly the first attempt that they made at getting X-ray quality crystals. It would take a while for that method to occur to you.

19 comments on “The Return of Kekulene”

  1. Project Osprey says:

    The wiki image was already correct. Would the 1b form not have major issues with cross-conjugation?

    1. RM says:

      Yeah, the wiki form is a bit odd with which side of the line the double bonds are drawn on, but it’s definitely 1a and not 1b.

      The main distinguishing feature is whether or not the “radial” bonds have double bonded character. Form 1a has double bonds in the “radial” position, in form 1b they’re all drawn as single bonds. The version on Wikipedia does have double bonds on the “radial” bonds, though not necessarily drawn in a way which makes the six aryl rings immediately obvious.

  2. NC says:

    Well, with 48 electrons, the molecule doesn’t follow Hückel’s rule (4n + 2 pi electrons), so although interesting this result shouldn’t really surprise anyone.

    1. Hap says:

      But the key cycles aren’t 48 (which would be antiaromatic) but 30 (outside) and 18 (inside) which could be aromatic (since they’re held in a plane). You have a choice of six separate aromatic rings or two big ones, and apparently Nature prefers six little ones (the aromatic stabilization per carbon is higher for them than for the two big rings, even with the twelve nonaromatic carbons left over).

      1. Anonymous says:

        My first “sight” was to see the separate inner and outer cycles, too. Thinking back to the SIMPLE (Orgo 101) MOs of benzene (bonding: psi1 2e psi2 2e psi2′ 2e | antibonding psi3 psi3′ psi4 empty – filled bonding orbitals, aromatic), COT (psi1 2e psi2 2e psi2′ 2e psi3 1e psi3′ 1e | psi4 psi4′ psi5 empty – half-filled psi3 and psi3′ non-aromatic) but COT(-2) has filled bonding orbitals and is aromatic.

        When I try to draw the simple (Orgo 101) MOs of kekulene, I get either three degenerate psi1s, four psi2s, … which would be aromatic. OR, I get two psi1s, four psi2s, etc. … which would have only 2e’s in four degenerate psi5s and would be anti-aromatic.

        (With a drawing applet, I’d draw them out here. Hint, hint.)

        If we add in 2e via reduction, the four psi5s are half filled with 4e (spin paired? stable?). If we add in 6e, there are 8e in four psi5s, filling that level = stable and aromatic? 48e + 6e = 54e = 4n+2 = 4×13+2 = 52+2 = 54.

        Reduce kekulene with K and see if you get aromatic (kekulene)(-6). Then make ferrocene analogs (kekulenocene?) using high oxidation state metals.

  3. Mad Chemist says:

    Dang, Wikipedia is fast. They already have the paper you’re talking about included as a reference.

  4. Anton Smith says:

    Thank you for the throwback to the ‘Strychnine!’ intro. And a clever post to sprinkle it in for sure.

  5. Another Guy says:

    Does kekulene react easily with bromine? I was wondering if anyone had explored its ease of reactivity and stereochemistry with electrophiles. Addition of bromine on every other ring would have been a give-away as to kekulene’s predominant structure.

  6. Andre St. Amant says:

    I thought my compounds had poor solubility! I’ve never had to recrystallize them in triphenylene or pyrene

  7. Scott says:

    Gah, sounds like the old alchemist’s gig of ‘turning lead into gold’ (ie, refining the gold out of lead ore).

    *Excruciatingly* tedious.

    1. anon says:

      Yeah. This is just like total synthesis. All that tedious work for something we already know.

  8. gippgig says:

    These polycyclic aromatics can be viewed as a resonant blend of all possible structures with localized bonds. Just adding up all the possible structures & averaging normally gives the correct result. For structure 1a there are 64 possible localized structures (2 to the 6th power) since each benzene ring can be drawn 2 ways while for 1b there are 4 possible structures (2 alternatives for the outer ring & 2 for the inner). That would suggest that kekulene is 16/17 1a & 1/17 1b.

  9. Barry says:

    What AFM can’t tell us is whether this thing would be planar in the vapor phase. Anything you find stuck to a copper surface is likely to be flat (when you see it)

  10. Kelvin says:

    “kekulene is indeed a sextet of benzene rings stitched together by non-aromatic bonds, and we now have experimental proof by two different methods.”

    Could this also be confirmed by susceptibility of the non-aromatic alkenes to oxidation, etc.?

  11. Henry Rzepa says:

    Kekulene has an interesting Kekulé vibration (which in benzene is the vibration that oscillates between two cyclohexatriene distortions). It could take the form of an extended vibration across the larger rings (of 18 or 30 atoms), but in fact it instead oscillates across the six Clar rings, each individual ring oscillation being in perfect synchrony with the others. Since this aspect of the molecule appears not to have been mentioned anywhere, I have added detail here at Thus the Kekulé vibration agrees with the current interpretation of this molecule and adds a further property of such rings that can be both calculated and measured.

  12. Henry Rzepa says:

    Re: “whether this thing would be planar in the vapor phase”, it seems extremely unlikely. The non-bonded H…H distance in the inner periphery is ~1.84A, and in all probability it distorts to a non-planar system with D3d symmetry , thus alleviating the H…H repulsions.

    1. Barry says:

      If the enthalpy of sticking to the Copper surface is enough to force the inward-facing C-Hs of Kekulene into the plane, you’d better believe that that same effect can distort bound proteins away from their functional conformation, too. AFM can teach us what proteins look like when stuck to the surface, but not how that might differ from their folding in solution.

  13. wildyr says:

    “It would take a while for that method to occur to you.”

    This molecule was probably insoluble in everything in the solvent cabinet at reflux, and insoluble also in things that are solids at RT, so superheating it in the most structurally similar molecule on hand was the only shot they had of dissolving it enough to recrystallize. Assuredly after months of agony.

    1. Derek Lowe says:

      Yeah, I’d have been a “Hot toluene!” guy, and when that didn’t work, I’d have been “Hot xylene!”, and would have moved on to several weeks’ worth of similarly imaginative choices. . .

Comments are closed.