The third ring takes the long way round
Worth reading first: The zigzag is short by its gap · On an angular ring the carbon decides.
A heteroatom on one carbon of a fused ring system changes the π bond order of every bond in it, and the size of the change on each ring is the reach of a local perturbation through a delocalised system. It has been measured here in one model throughout: Hückel’s, with each bond’s coupling relaxed with its own bond order and a stagger of site energies holding the gap open. In that model the reach is the molecule’s, the same from the end and from the middle of a straight chain. An angular ring rescales what lies beyond it, on an angular ring the carbon decides how much it rescales, and the zigzag is short by its gap and by nothing else, its strong bonds taking back part of the response at one ring and giving it back at the next.
All of those are chains, where a response has one way to go. Three times the sequence named the case it could not reach: a ring fused on three sides, triphenylene’s central ring, where a response leaving one outer ring has two routes to each of the others. The question was whether those routes add or interfere at the far ring. Now that a strong bond is known to send back a response opposed to its neighbour’s, it looked possible that a loss on one route would meet a gain on the other.
The routes turn out not to be new. Phenanthrene has them already. What triphenylene’s third ring changes is which of them is open, and the response does not choose between routes at the branch; it chooses at the carbon it starts from.
Two arcs round a middle ring
Call the ring with the heteroatom A, the middle ring C and the far ring B. Phenanthrene’s middle ring touches A along one bond and B along another, and joins them twice. The short arc is the single bond across the bay, from A’s junction carbon to B’s. The long arc is the three bonds round the other side, through phenanthrene’s C9 and C10, whose bond is the most olefinic in the molecule. In the relaxed calculation its bond order is 0.676, against 0.435 for the short arc and 0.497 for each fusion bond. A response leaving A for B has both arcs available in phenanthrene, and it always has had.
Triphenylene is phenanthrene with a third ring, D, fused onto that long arc. Its C9–C10 bond becomes D’s fusion bond, and the two bonds on either side of it become single-like links between rings. The middle ring is then the ring Clar’s sextet picture calls empty: three fusion bonds at 0.508 alternating with three inter-ring bonds at 0.415, and no bond in it stronger than 0.508.
So the comparison the open question asked for is not between a molecule with one route and a molecule with two. It is between two molecules with the same two routes, one of which has had its long arc rebuilt from a double bond into a ring junction. Phenanthrene against triphenylene, heteroatom for heteroatom on the same carbon of A, asks what that rebuild does to what reaches B, and where whatever B loses goes.
Four carbons, two answers
Ring A has four CH carbons, numbered here from the bay carbon on B’s side round to the bay carbon on the far side. The heteroatom is placed on each in turn, with a site energy of +1 in units of the hopping, and each ring’s response is the mean change in bond order over its six bonds, the same measure the chain essays used, with bond orders read from the eigenvectors in the usual way and the heteroatom treated as a shifted site energy and nothing more. Both molecules are run with the published stagger, 0.6, and relaxation strength, 0.4.
The four carbons split into two pairs, and the split decides everything. From carbon 1, the bay carbon beside B’s junction, triphenylene’s ring B receives against phenanthrene’s , a ratio of 0.986, and ring D receives a ninth as much. From carbon 3, the outer carbon on the far side, B receives 1.078 of what it received in phenanthrene and D about a third of B. These two carbons send B what they always did, and the third ring hardly registers.
From carbon 2, the outer carbon on B’s side, triphenylene’s B receives only 0.641 of phenanthrene’s, and D receives nearly three times as much as B. From carbon 4, the bay carbon on the far side, B receives 0.412 and D receives . That is the same size as what B received from carbon 1, which is the mirror of carbon 4 across the axis through A and C.
The pairing is the oldest rule in the Hückel treatment of alternant hydrocarbons, and it has nothing to do with B. In an alternant π system a perturbation at one carbon is felt most strongly at carbons ortho and para to it and weakly at those meta, because the mutual atom polarisabilities Coulson and Longuet-Higgins introduced in 1947 alternate in sign and magnitude with the parity of the path. Carbons 1 and 3 are ortho and para to the junction carbon A shares with the short arc. Carbons 2 and 4 are para and ortho to the junction carbon A shares with the long arc. So each carbon of ring A addresses one arc, and the two arcs are not two routes for one response. They are two routes for two different responses.
A straight chain makes the same point by contrast. In anthracene the far ring sits opposite ring A, and either of A’s junctions leads to it through the middle ring. So there every carbon of A reaches B, and the four carbons send it , , and , all within a factor of about two of each other. Phenanthrene’s angular middle ring is what gives the two junctions different destinations, and that was already the substance of a bend is not an end: an angular fusion is not a weaker linear one but a different set of paths.
The middle ring’s own response follows the same split. From carbons 1 and 3 triphenylene’s ring C changes by 0.98 and 1.01 times as much as phenanthrene’s, since those carbons address the short arc and the short arc is unchanged. From carbons 2 and 4 it changes by 0.77 and 0.69 times as much. The long arc in phenanthrene ran through C9=C10, two of the middle ring’s own bonds, and a response travelling along it counted toward C’s mean. In triphenylene the corresponding bonds are D’s fusion bond and the two single-like links, and the response passes through them into D rather than lingering on them.
Neither adding nor cancelling
The bond map for carbon 4 shows where the long arc’s response goes. In phenanthrene it leaves A at the long-arc junction, runs round through C9 and C10, and arrives at B from the far side, so that B’s bonds change by an amount comparable to what carbon 1 delivers across the bay. In triphenylene the same response leaves A at the same junction and finds ring D in the way. The long arc’s first bond is now an inter-ring link of order 0.415. Beyond it lies a fusion bond that belongs to a six-membered ring with its own sextet. The response spreads round D, and ring D’s mean change is nearly seven times B’s.
If the two routes were interfering at B, B’s pattern of bond changes would turn as well as shrink, since a second contribution arriving by another path would add a different pattern. It does not turn. The cosine between triphenylene’s and phenanthrene’s signed changes on B’s six bonds is 0.996 and 0.995 for carbons 1 and 3, and 0.927 and 0.959 for carbons 2 and 4. Even where B loses most of its response, what remains has nearly the shape it had. The short-arc part is untouched and the long-arc part is removed. On the reading of the next section, what B still receives from carbons 2 and 4 is the small share they send across the short arc as its meta neighbours.
And the third ring is a destination, not a drain. Added together, B and D receive between 1.10 and 3.21 times what phenanthrene’s B received from the same carbon. The response that phenanthrene’s olefinic C9=C10 carried on to B does not disappear in triphenylene. It stops one ring earlier, in a ring that did not exist before, and it is larger there than it ever was at B, because D’s bonds stand on the long arc itself, where B’s lie beyond its end.
The branch at every gap
The split between the two pairs of carbons is not a property of one setting. Across staggers from 0.3 to 1.2, which take triphenylene’s relaxed gap from 1.51 to 2.77, and relaxation strengths of 0.2, 0.4 and 0.6, carbons 1 and 3 keep B’s response between 0.89 and 1.14 of phenanthrene’s. Carbons 2 and 4 lose half or more of it where the gap is narrow, and the loss closes as the gap opens: at the widest gap carbon 2’s ratio is 0.97 and carbon 4’s 0.82 to 0.85. A wider gap makes every response more local, and the long arc’s longer path through C9 and C10 is the first thing a local response stops using. Phenanthrene then sends less round the long arc for triphenylene to divert.
Relaxation matters least where the loss is largest. At the narrowest gap carbon 4’s ratio is 0.43 to 0.45 at all three strengths. At the intermediate gaps it spreads from 0.38 to 0.44 and is lowest at the strongest relaxation, the setting that makes triphenylene’s single-like inter-ring links weaker still.
Where the expectation came from
The expectation of interference came from the chains. In the nitrogen essay two contributions to one far response, from a heteroatom’s two changed bonds, pointed almost opposite ways and largely cancelled, and in the zigzag a strong bond’s loss was met by the next ring’s gain. Both were cancellations between two pieces of one response travelling one path. A branch looked like the place where two pieces could meet from different paths.
What the four carbons show is that in an alternant system the pieces were never travelling together. A carbon of ring A is coupled strongly to the junction carbons ortho and para to it and hardly at all to the one meta to it, and no CH carbon of ring A is ortho or para to both of its junctions. So the decision about which arc carries a response is made at the heteroatom’s own carbon, two bonds from the branch, before anything reaches the middle ring. The middle ring then passes on whatever arrives on each arc separately. The carbon decides was the same lesson read on a chain, where the two junctions led on to the same next ring. Here they lead to different rings, and the lesson becomes visible as a split.
How the comparison was made
Each molecule is built from ring centres a distance apart in multiples of sixty degrees, with vertices at unit radius and deduplicated by position. That is the same construction as the chains, so phenanthrene built this way has the same bonds and the same Hückel levels as phenanthrene built as a chain, and triphenylene is the same rule with the third ring placed from the middle one. The stagger alternates site energies by sublattice, which every catacondensed benzenoid has. Couplings relax as to a fixed point, as in the frame that was allowed to relax, and both the plain and the substituted molecule use the plain molecule’s reference bond order, the correction that the floor was in the bookkeeping found necessary. Ring responses are means of over each ring’s six bonds. The cosines compare signed changes on B’s six bonds, matched by position.
The checks are of the construction and of the comparison. Triphenylene with no heteroatom has three outer rings with identical bond orders, to . Without the stagger, a heteroatom on carbon 1 sends ring B exactly what one on carbon 4 sends ring D, and carbon 2 to B equals carbon 3 to D, to , which is the mirror through A and C. That check is made without the stagger because the stagger colours the two mirrored carbons oppositely. The refusal is the one that would make the comparison meaningless. What reaches B must depend on which carbon of A carries the heteroatom, by more than a tenth of its mean, or the four-carbon table would be measuring the junction and not the routes. It does, by a factor of nearly seven between carbons 1 and 4.
What a Hückel branch cannot say
One electron at a time. Every number here is a Hückel number with a relaxation rule bolted on. The ortho–para alternation that drives the result is exact in Hückel theory for an alternant system and approximate in any theory with electron repulsion, where a heteroatom’s charge also polarises the σ frame and the response has a part that does not alternate.
A site energy is not a nitrogen. The heteroatom here changes only its own carbon’s site energy. A real aza-triphenylene changes the couplings to its two neighbours as well, which the nitrogen essay found sends the far chain responses pointing almost opposite ways. Whether that changes which arc a carbon addresses is not computed.
Mean bond-order changes, not observables. The response on a ring is an average of over its bonds, chosen because it is defined for every ring of every molecule here. Bond lengths, chemical shifts or ring currents would each weight the bonds differently. The split between carbons that address different arcs is a property of the polarisabilities and should survive any of them, but the ratios quoted are this average’s.
Two molecules’ worth of geometry. Triphenylene is the smallest ring fused on three sides. Larger branched systems put a branch further from the heteroatom, where the ortho–para decision has already been made and the middle ring may pass the two arcs differently.
A branch that was decided upstream
The interesting thing about triphenylene’s central ring, for a response, is what it does not do. It is the ring Clar’s picture leaves empty, and its bonds are the weakest in the molecule, so it was natural to expect it to be where two paths meet and either reinforce or spoil each other. Instead each path has been chosen before the response reaches it, by the parity of the heteroatom’s carbon relative to ring A’s two junctions, and the central ring passes the two separately. What the third ring changes is the destination of one of them. A phenanthrene that sent its long-arc response through a double bond to the far ring becomes a triphenylene that delivers the same response to the ring now standing on that double bond’s site. The total the far rings receive goes up, B’s share goes down, and the shape of what B still receives hardly changes.
Still open: a branch further out, and the nitrogen’s own bonds
The obvious open question is distance. Here the branch is one ring from the heteroatom, so the arc each carbon addresses is decided in the heteroatom’s own ring. In a larger branched system — a fourth ring fused to triphenylene’s D, or the heteroatom on a ring two fusions away from the branch — the response would reach the branch as a mixture already spread over both of the previous ring’s junctions, and whether the branch then splits it by arc or divides it by some other rule is a different calculation.
The nearer question is a heteroatom that changes its bonds. A pyridine-type nitrogen alters the couplings to its two neighbours as well as its own site energy. For a carbon ortho to one junction and meta to the other, one of those bonds leads toward each arc, so the bond changes might send part of the response down the arc the site energy alone does not address. If they do, the clean split by carbon seen here becomes a split by carbon and bond, and triphenylene’s ring D would receive a share from every position.
Shares its objects with
Essays naming at least two of the same things, that neither author linked.
- An anomaly that is not the first of a series — both name bond order, delocalisation, hückel theory, reference state
- Six electrons in a ring that is not all carbon — both name bond order, delocalisation, hückel theory, reference state
- The same ring, three charges — both name bond order, delocalisation, hückel theory, reference state
- Which numbers carry a frame — both name bond order, delocalisation, hückel theory, reference state
- A reach that has no length — both name band gap, delocalisation, hückel theory
- A stabilisation is measured from somewhere — both name delocalisation, hückel theory, reference state
Named objects
A dashed tag is an object no other essay names yet.
Band gapBond orderDelocalisationHückel theoryPolarisabilityReference state