A nitrogen's two bonds cancel on an angular ring
Worth reading first: On an angular ring the carbon decides · An angular ring rescales what lies beyond it.
A heteroatom in a chain of fused benzene rings perturbs the bond orders around it, and the perturbation dies away along the chain. On an angular ring the carbon decides took apart which carbon of an angular ring the heteroatom sits on. Matched carbon for carbon against the straight chain, a heteroatom on one of an angular ring’s two CH carbons leaves its own ring almost exactly as responsive as a straight ring would be, and sends every other ring about half. The reason offered was the bond the CH carbon sits on. An angular ring’s two CH carbons are bonded to each other, and in the relaxed chain that bond is the strongest in the molecule, the 9,10 bond of phenanthrene that reacts like an alkene. A heteroatom on one end of it spends much of its effect polarising an isolated double bond, and an isolated double bond is poorly connected to the rest.
Every heteroatom in that argument was a site energy and nothing else. A pyridine-type nitrogen is more than that: it also changes the hopping to its two neighbours, and on an angular ring’s CH carbon one of those two bonds is that same strongest bond. The essay’s own lead put the question directly. Does a hopping change on that bond halve what reaches the chain as the site energy does, or undo it?
It does neither, and the reason is that the nitrogen has two bonds and they work against each other.
Two bonds of different strength
The starting point is a difference of bonds that the site energy never had to notice. On ring five of the straight chain, a CH carbon has two bonds to ring junctions, and both carry a bond order of 0.541. On the kink’s first angular ring, a CH carbon has one bond to the other CH carbon, at 0.660, and one to a junction, at 0.490 or 0.510 depending on which CH carbon it is. On a lone angular ring the same pair is 0.675 and 0.478. A heteroatom that changes its bonds therefore changes one strong bond and one weak one on the angular ring, and two middling ones on the straight ring.
The model is the one every earlier essay on this chain used: Hückel’s π system on twelve fused rings, a small stagger between the two sublattices that keeps the gap open, and couplings relaxed to the bond orders they produce, with the reference bond order taken from the plain chain so that the comparison removes no floor. The new ingredient is a scale on the heteroatom’s bonds: each scaled bond’s coupling is , so the scaled bond is relaxed against its own scale rather than fixed. At this is exactly the earlier calculation, and it reproduces its per-ring responses to .
The bonds alone send half too
The first test removes the site energy altogether and scales only the two bonds, from 0.7 to 1.3. A heteroatom that is nothing but a change of hopping is a cleaner probe of the bond explanation than one that is mostly a site energy.
With no site energy, every other ring receives between 0.57 and 0.64 of what the straight chain’s matched carbon sends under the same scaling, on all four CH carbons and at every scale. The fraction drifts down slowly as the scale rises, from about 0.63 at 0.7 to about 0.58 at 1.3, and the four carbons agree to within 0.02. That is the same deficit the site energy produced, 0.49 to 0.62 on these four carbons, from a perturbation with no site energy in it.
So whatever makes a heteroatom on this carbon send the chain half as much is not something particular to site energies. It belongs to the carbon’s place. That is already a partial answer to the lead: there is nothing for the bond change to undo, since the bond change on its own shows the same halving.
The heteroatom’s own ring keeps the other half of the earlier finding. With the bonds scaled alone its response stays between 0.93 and 1.03 of the straight ring’s, the same near-equality the site energy showed. What the angular ring changes is where the response goes, not how much of it stays at home.
Adding the bonds to the site energy
With the site energy and the bond scale together, which is what a nitrogen is in this model, the far response stays between 0.48 and 0.70 of the straight chain’s across the whole range of scales. On the carbons of one sign of the stagger it hardly moves from the site energy’s value, 0.51 to 0.55 on the kink and 0.48 to 0.53 on the lone ring. On the carbons of the other sign it falls from about 0.69 at 0.7 to 0.54 at 1.2 before turning up again at 1.3, a wobble that comes from the stagger and the scale acting on the same carbon. Nowhere does it approach one, and nowhere does it fall to a quarter.
The answer to “halve it again, or undo it?” is therefore “change it by a tenth either way”. A pyridine-type nitrogen on an angular ring’s CH carbon sends the rest of the chain about half of what it would send from a straight ring, whatever its bond parameter is in the range anyone uses.
How large a bond change is
The fractions above compare like with like, and they hide how big each perturbation is in absolute terms. On the straight chain, a heteroatom that is a unit site energy and nothing else sends the rest of the chain a summed response of 0.0487: each ring’s mean change of bond order, added over every ring but the heteroatom’s own. Scaling its two bonds by 0.9 with no site energy sends 0.0122, a quarter as much, and scaling them by 0.7 sends 0.0395, most of the way to the site energy’s. So a pyridine-type nitrogen, with its bond scale close to one, is mostly a site energy in this model, and its bond change is a correction of a quarter or less to what the site energy does. That is why the combined fraction lies so close to the site energy’s.
On the angular ring’s CH carbons the same bond scaling sends 0.0073 to 0.0074 at 0.9, three-fifths of the straight chain’s 0.0122, and that is the halving again, now in a perturbation that is a fraction of the size. Nothing about the relative size of the two parts of a nitrogen differs between the two rings; what differs is how much of either part reaches the chain.
Relaxing the couplings matters to these numbers in the way the frame that was allowed to relax first showed it matters to any bond-order comparison: the scaled bond’s own order changes, and with it the couplings of its neighbours, so a bond scale here is not a fixed perturbation but one that the chain partly absorbs. Both rings are relaxed in the same way, and every ratio above compares a relaxed angular chain with a relaxed straight one.
Each bond on its own
The near-indifference to the scale is the surprising part, and it can be taken apart by scaling the two bonds one at a time with no site energy.
On the kink’s CH carbon at a scale of 0.9, the bond to the junction alone sends the rest of the chain a response of size 0.0205, measured as the root sum of squares of every bond-order change outside the heteroatom’s ring. The strong bond to the other CH carbon alone sends 0.0119. Both together send 0.0134 — less than the junction bond alone, and two-fifths of the sum of the two sizes. Above unit scale the pattern sharpens: at 1.2 the junction bond sends 0.044, the strong bond 0.022, and both together 0.023, barely more than the strong bond on its own.
The reason is direction. Treated as vectors over every bond outside the heteroatom’s ring, the response to the strong bond’s change and the response to the junction bond’s change point nearly opposite ways: the cosine between them is between −0.81 and −0.86 on all four carbons and at every scale. Weakening the bond to the other CH carbon pushes the chain’s bond orders one way and weakening the bond to the junction pushes them almost exactly the other, so a perturbation of both is mostly a cancellation. The sum of the two separate responses is not quite the joint one — the mismatch is 10 per cent of the joint response at 0.9 and 1.1 and 20 per cent at 0.8 and 1.2, since the relaxation is not linear — but the cancellation is the dominant thing either way.
Read with the bond orders, this is a reading of the same localisation the site energy found, from another side. The strong bond is a nearly isolated double bond: changing its hopping moves π density between its two ends and away from the junction bond. Changing the junction bond’s hopping moves density between the CH carbon and the junction, which is the route into the rest of the chain. A nitrogen does both, and the first largely undoes what the second would have sent. That is a reading, not a separate calculation, but it agrees with every number in the figures.
The whole chain, ring by ring
Ring by ring the picture is less uniform than the totals, and the exception is worth stating. With both bonds at 0.9 as well as the site energy, a heteroatom on the lone angular ring’s CH carbons leaves every other ring between 0.35 and 0.76 of the straight chain’s response, on both sides of ring five. On the kink’s (−) carbon every other ring sits between 0.43 and 0.65. The kink’s (+) carbon is different on one side: the rings before ring five receive 0.51 to 0.75, but the rings beyond the kink’s second angular ring, seven to eleven, receive 0.82 to 1.04 — nearly what the straight chain sends them. Its total far response is still 0.645 of the straight chain’s, because the nearest rings carry most of the response and they are where the deficit is.
So the halving is a statement about the near chain, where most of the response lives, and on one carbon of the kink it does not survive the second angular ring. An angular ring rescaling what lies beyond it found that each angular fusion a perturbation passes through rescales what lies beyond it, and the kink’s second angular ring evidently rescales this carbon’s perturbation upwards, towards the straight chain’s, rather than down. The same happened with the site energy alone in the earlier comparison, so it is not something the bond change introduces.
How the relaxation was carried
The chain is the twelve-ring chain of the bend that is not an end and its successors, with the kink’s angular rings at positions five and six and the lone angular ring at five, a stagger of 0.6 in units of the hopping between the two sublattices, and a coupling-to-bond-order slope . Each placement is relaxed to self-consistency from the plain chain’s couplings, with the plain chain’s mean bond order as the reference in both runs, the correction the reach essay found removes a spurious floor. A heteroatom’s site energy is one unit of hopping. Its bond scale multiplies the relaxed coupling of each scaled bond, so that the bond’s coupling and its order remain consistent with each other.
Every carbon is compared with the straight chain’s CH carbon on the same ring with the same sign of stagger, under exactly the same perturbation, as the earlier essay required. A single bond of the angular ring has no counterpart there — the straight ring’s CH carbon has no bond to another CH carbon — so the single-bond scalings are read as sizes and directions on the angular chain alone rather than as ratios.
What is checked where the figures are drawn: at unit scale the relaxation reproduces the earlier site-energy calculation’s per-ring responses to ; every relaxation converges to in its couplings; the bond to the other CH carbon is the stronger of each carbon’s two by more than 0.1; and the two single-bond far responses point against each other, with a cosine below −0.7, at every scale tried. The refusal is the straight chain against itself: every placement compared with its own matched carbon must give exactly one at every scale, and does.
A model nitrogen
A site energy and a bond scale are a caricature of a nitrogen. A ring that is not all carbon is still counted as six π electrons, and real aza-aromatics differ from their hydrocarbons in bond lengths, in σ framework and in the lone pair, and the Hückel parameters for pyridine-type nitrogen — a site energy around half a hopping and a bond scale near one — are fitted to particular molecules. The range swept here, 0.7 to 1.3, brackets every value in use, and the finding is that within it the answer does not depend on the value.
One electron, relaxed. The bond orders are those of a one-electron model with a linear relaxation of the couplings, and the localisation of phenanthrene’s 9,10 bond is stronger in this model than a correlated calculation would make it. Hückel with a heteroatom laid out what the model can and cannot be asked; the cancellation between two bonds is a statement about the model’s linear response, and what survives a better model is its direction rather than its size.
Two angular rings. The kink and the lone angular ring agree, and they are the only two angular environments tried. A fully angular chain, where every interior ring has the strong bond, could behave differently because the far response would have to pass through further strong bonds on its way.
What a bond change cannot reach
The earlier finding put the halving down to the strong bond: a heteroatom there polarises an isolated double bond. What a bond scale adds is that the strong bond, when its own hopping is changed, does exactly the kind of thing one would expect from an isolated double bond — it moves density along itself — and that this motion cancels most of what the other bond would have sent into the chain. The isolation that halves a site energy’s reach also stops a bond change from restoring it.
That gives a small rule a chemist could use. On an angular ring’s CH carbon, how far a heteroatom’s influence reaches along the chain is set by where the carbon is, not by what kind of heteroatom it is. A substituent that changes only its site, one that changes only its bonds and one that changes both send the chain about half of what they would send from a straight ring, and the strongest bond in the molecule is the reason for all three.
Still open: a branch, and a ring with three strong bonds
The obvious open question is the one this sequence has deferred twice: a ring fused on three sides, triphenylene’s central ring, where a perturbation leaving one outer ring has two routes to the other two. The outer rings of triphenylene each carry one strong bond on their CH carbons, as the angular rings here do, and whether the two routes add or interfere is now a sharper question, because a perturbation entering at a CH carbon already arrives at the central ring as the difference of two nearly opposite parts.
The nearer question is a chain that is angular everywhere. In the zigzag every interior ring has its own strong bond between its CH carbons, and a heteroatom’s far response would have to cross each of them. If each strong bond takes back part of what reaches it in the way the heteroatom’s own strong bond does here, the zigzag’s reach should fall off faster than the kink’s for a reason that has nothing to do with its gap — which is the opposite of what a reach that has no length found for the acenes, where the gap was everything.
What links here
Computed from the collection rather than written here: the essays that point at this one.
Reads more easily once this is understood
Essays that name this one as worth reading first.
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, model limit, reference state
- The floor was in the bookkeeping — both name bond order, delocalisation, hückel theory, model limit, reference state
- Which numbers carry a frame — both name bond order, delocalisation, hückel theory, model limit, reference state
- A stabilisation is measured from somewhere — both name delocalisation, hückel theory, model limit, reference state
- An end effect with two signs — both name delocalisation, hückel theory, model limit, reference state
- One integer, and everything it changes — both name delocalisation, hückel theory, model limit, reference state
Named objects
A dashed tag is an object no other essay names yet.
Bond orderDelocalisationHückel theoryModel limitReference state