Where the atoms go

The ring closes from the arrangement that alternates

The rotamer ceiling on the gem-dimethyl effect needs a count of the rotations a closure freezes, and three counts were in use. Building the open chain's staggered conformers in three dimensions gives a fourth that is derived rather than chosen. The conformers whose ends meet are the ones whose gauche torsions alternate in sign, not the ones of one sign, which wind into a helix; every conformer closer than the first one with an anti torsion is all gauche; and so the count is exactly n − 3. The alternating arrangement carries the syn-pentane contact the account never priced, which raises the five-ring's ceiling from 11.0 to at most 21 — still twelve times short of the measured 250.

Worth reading first: The count that was never written down · Two sweeps and one lever.

Two methyl groups on one carbon of a chain make it close to a ring faster, by 250-fold for a five-membered lactone, and the account that has survived the arithmetic in these essays is about rotations rather than angles. The angle account had the wrong sign. The rotamer account says the substituent makes the chain’s extended arrangements expensive, so the chain spends more time in arrangements that can close, and it has a ceiling that rises where the measurements fall: however hard the substituent pushes, it can at most drive every rotation gauche, so the rate ratio cannot exceed ((1+2x)/2x)r((1+2x)/2x)^r, with x=e−g/kTx = e^{-g/kT} and rr the number of rotations the closure freezes.

rr was never measured. The count that was never written down found three conventions in use — n−2n - 2, a count of four for the five-membered ring and two for the six, and n−3n - 3 — which disagree by a factor of eleven in the ceiling, and showed that no verdict moves between them. It ended by asking for a fourth count derived from the chain itself: build the conformers, find the ones that can close, and let the ceiling say how many rotations they freeze. The essays that followed swept the gauche energy and the temperature and left that request where it was.

Doing it settles the count. It also shows that the account has been naming the wrong conformer as the one that closes, and the right one carries a cost the account has never included.

Which conformers bring the ends together

A chain of nn backbone atoms has n−1n - 1 bonds, and rotating either terminal bond moves no backbone atom, so n−3n - 3 torsions decide where the ends are. Give each one of the three staggered values — anti at 180°, or gauche at +60° or −60° — and build the chain with carbon–carbon bonds of 1.53 Å and angles of 112°. The five-atom chain that closes to a five-membered ring has two torsions and nine conformers; the six-atom chain has three and twenty-seven.

The conformers whose ends meet alternate in sign; the one-sign ones wind into a helix. Every staggered conformer of a five-atom and a six-atom chain placed by the distance between its two end atoms, coloured by its torsions. 5 atoms: closest +− 2.69 Å, −+ 2.69 Å; one-sign ++ 3.65 Å, −− 3.65 Å; first with an anti torsion t+ at 4.44 Å. 6 atoms: closest +−+ 1.79 Å, −+− 1.79 Å; one-sign +++ 4.70 Å, −−− 4.70 Å; first with an anti torsion t+− at 4.05 Å.
Fig. 1 Every staggered conformer of a five-atom and a six-atom chain, placed by the distance between its two end atoms and coloured by its torsions, with the stretch below the first conformer that has an anti torsion shaded.

On the five-atom chain the closest conformers are g+g−g^+g^- and g−g+g^-g^+, with their ends 2.69 Å apart. The two of one sign, g+g+g^+g^+ and g−g−g^-g^-, are at 3.65 Å. Anything with an anti torsion is at 4.44 Å or more. On the six-atom chain the order is sharper: g+g−g+g^+g^-g^+ and its mirror at 1.79 Å, close enough that a bond of ordinary length would nearly span the gap; then the four with one sign change between a pair, at 3.10 Å; then four conformers with an anti torsion, at 4.05 Å; and only then, at 4.70 Å, the all-g+g^+ and all-g−g^- arrangements. The conformers of one sign are not the ones that close. They are among the furthest of the all-gauche states.

The reason is the one every model-builder discovers by hand. A gauche torsion turns the chain by a sixth of a turn about each bond, and doing it in the same sense repeatedly is a screw: the chain winds round an axis and advances along it, which is how a polymer helix is built. Turning in alternating senses cancels the advance and folds the chain back on itself.

Opposite signs fold a chain back on itself; one sign winds it away. Five- and six-atom chains built with every torsion gauche, drawn on their two widest axes: 5 atoms +−, ends 2.69 Å apart; 5 atoms ++, ends 3.65 Å apart; 6 atoms +−+, ends 1.79 Å apart; 6 atoms +++, ends 4.70 Å apart. The end atoms are the larger dots.
Fig. 2 Five- and six-atom chains with every torsion gauche, drawn on their two widest axes: alternating signs, and one sign, with the gap the closing bond must span dashed.

The cleanest check is a ring whose torsions are known. Cyclohexane’s chair, built the same way at an angle of 111.4°, closes — its end-to-end distance equals one bond — at torsions of ±54.9°, and when its six torsions are read back from the closed ring they alternate all the way round: −,+,−,+,−,+-, +, -, +, -, +. The chair is the alternating arrangement taken to closure. A six-membered ring formed from a chain forms from the conformer that already looks like part of a chair.

Why the published sentence was wrong and the number was not

The account as it was set out said that a chain with three internal rotations closes only from “the two in which every rotation is gauche and all of one sign”. The geometry says the closable two alternate. The ceiling, though, came out right, and the reason is worth stating because it is what makes the derived count possible. The ceiling depends only on how many of the closable conformers are all gauche and how many all-gauche conformers there are. In the limit of an infinite penalty on every anti torsion, the substituted chain lives entirely in the 2r2^r all-gauche states, so its closable fraction is the closable all-gauche states’ share of those; the plain chain’s closable fraction is their Boltzmann weight over everything. Two closable states of one sign and two closable states of alternating signs give the same ratio. The account’s arithmetic counted two patterns and was right to; it named the wrong two.

That is not a pedantic correction, for two reasons below. The alternating conformer is the one whose ends meet over a range of closure distances that can be stated, which makes the count derivable; and it is the one that carries a contact the one-sign helix does not.

The count the chain gives is n − 3

Call a conformer closable when its end atoms are within a distance dcd_c — the near-attack distance below which the ends can react — and compute the ceiling from the closable set exactly as the account defines it.

Read off the chain, the count is n − 3 for any closure distance short of the first anti conformer. The effective number of frozen rotations, ln(ceiling)/ln((1 + 2x)/2x), against the end-to-end distance below which a conformer counts as closable, for chains of four, five and six atoms. 4 atoms: 1 from 3.03 Å to 3.90 Å, then 0.00; 5 atoms: 2 from 2.70 Å to 4.44 Å, then 0.56; 6 atoms: 3 from 1.80 Å to 4.05 Å, then 1.83.
Fig. 3 The effective number of frozen rotations, the logarithm of the ceiling in units of one rotor’s factor, against the closure distance, for chains of four, five and six atoms.

For any dcd_c between the closest conformer and the first one with an anti torsion, every closable conformer is all gauche, and then the ceiling is ((1+2x)/2x)n−3((1+2x)/2x)^{n-3} exactly: the closable all-gauche weight cancels between the two fractions and what remains is the all-gauche states’ share of the whole, which is a product of one factor per torsion. The derived count is therefore n−3n - 3 for the five-atom chain anywhere from 2.69 to 4.44 Å, and for the six-atom chain anywhere from 1.79 to 4.05 Å. Within that window it does not matter which closure distance is chosen, or how many all-gauche conformers it admits. Past it, conformers with an anti torsion become closable, the substituent’s penalty works against them, and the count falls: to 0.56 on the five-atom chain beyond 4.44 Å and 1.83 on the six beyond 4.05.

So of the three conventions, one is not a convention. n−3n - 3 is the number of torsions, and it is what the conformers give at any closure distance a near-attack geometry could plausibly mean. n−2n - 2 counts one rotation too many. The count the decided verdict used, four for the five-ring and two for the six, is not derivable at any closure distance, because no closable set on either chain gives a count above n−3n - 3 or a count that falls with ring size.

Robust to the geometry

The bond length, the angle and the gauche torsion are all rounded figures, and the identity of the closable conformer ought not to depend on them.

At every bond angle and gauche torsion tried, the alternating conformer is the closest. End-to-end distances of the alternating and the one-sign all-gauche conformers of five- and six-atom chains, at bond angles from 109.47° to 114° and gauche torsions of 55°, 60° and 65°. The alternating one is the closest conformer in every row.
Fig. 4 End-to-end distances of the alternating and the one-sign all-gauche conformers of five- and six-atom chains, at bond angles from 109.47° to 114° and gauche torsions of 55° to 65°.

It does not. Across bond angles from the tetrahedral 109.47° to 114° and gauche torsions from 55° to 65°, the alternating five-atom conformer’s ends are 2.38 to 2.97 Å apart and the one-sign conformer’s 3.35 to 3.90; on the six-atom chain, 1.33 to 2.21 Å against 4.34 to 5.02. The alternating conformer is the closest one in every row. The gap between the two kinds is about an ångström on the five-atom chain and nearly three on the six, which is the size of the effect of winding instead of folding, and no plausible change of geometry closes it.

The contact the closable conformer carries

The alternating arrangement has a familiar name in conformational analysis. Two neighbouring gauche torsions of opposite sign bring the atoms four positions apart into contact: on pentane, g+g−g^+g^- puts the two terminal carbons 2.5 to 2.7 Å apart at ordinary angles, the syn-pentane interaction, and it is why that conformer is rare. On the five-atom chain the contact is between the end atoms themselves, which is precisely the approach the ring needs; on the six-atom chain g+g−g+g^+g^-g^+ has two such contacts, between atoms one and five and between two and six. The rotamer account’s energy has a gauche cost per torsion and nothing else, so it has never priced the arrangement it depends on.

Pricing it is a sweep rather than a number, because what a contact costs when one of the atoms in it is the one about to react is not a quantity anyone has measured. Give every neighbouring pair of opposite-sign gauche torsions an extra cost ss and recompute.

Priced as a repulsion, the contact raises the ceiling but not far enough for the five-ring; as an attraction of three kilojoules it sinks the six-ring. The rotamer ceiling against the extra cost of each pair of neighbouring gauche torsions of opposite sign, negative for an attraction, for the five- and six-membered closures, with the measured accelerations as dashed lines. 5-ring: -4 kJ/mol 4.3, -3 kJ/mol 5.6, -2 kJ/mol 7.2, -1 kJ/mol 9.0, 0 kJ/mol 11.0, 1 kJ/mol 13.0, 2 kJ/mol 14.8, 3 kJ/mol 16.4, 4 kJ/mol 17.7, 6 kJ/mol 19.4, 8 kJ/mol 20.2, 10 kJ/mol 20.6, 12 kJ/mol 20.8, 15 kJ/mol 20.9; 6-ring: -4 kJ/mol 5.9, -3 kJ/mol 9.6, -2 kJ/mol 15.6, -1 kJ/mol 24.5, 0 kJ/mol 36.5, 1 kJ/mol 50.9, 2 kJ/mol 66.3, 3 kJ/mol 81.3, 4 kJ/mol 94.5, 6 kJ/mol 113.5, 8 kJ/mol 124.0, 10 kJ/mol 129.1, 12 kJ/mol 131.6, 15 kJ/mol 133.0.
Fig. 5 The rotamer ceiling against the extra cost of a neighbouring pair of opposite-sign gauche torsions, negative for an attraction, for the five- and six-membered closures, with the measured accelerations dashed.

The ceiling rises. A contact that makes the closable conformer rarer in the plain chain makes it rarer in the fully substituted chain too, but the plain chain has many more places to put its population, so the ratio between them grows: on the five-membered closure from 11.0 with no contact to 14.8 at 2 kJ/mol, 19.4 at 6 and 20.9 at 15, levelling off at 21.0 however large the cost. In units of rotors that is 2.00 rising to 2.54. On the six-membered closure, with two contacts in the closable conformer, the ceiling goes from 36.5 to 133, which is 4.08 rotors.

So the contact is a real lever, and priced as a repulsion it pushes in the direction of the measurements. It is also bounded, and its bound settles the question the whole sequence started from.

Priced as an attraction it pushes the other way, and much harder for the six-membered ring than for the five. An attraction makes the closable conformer more common in the plain chain, which is where the ceiling’s denominator lives, and the six-atom chain’s closable conformer has two contacts to the five-atom chain’s one. At −1 kJ/mol a pair the six-ring’s ceiling is 24.5; at −2 it is 15.6; and at −2.9 kJ/mol it falls to the measured 10. An attraction of three kilojoules a mole, a fraction of a hydrogen bond, would refute the account on the six-membered ring as well.

The verdicts under the derived count

The derived count is the torsion count, and it refutes the five-ring by twenty-three and leaves the six-ring standing. Each measured gem-dimethyl acceleration against the rotamer ceiling under three conventions and under the count derived from the chain, with the derived ceiling's value when the syn-pentane contact is made prohibitive. γ-butyrolactone formation, gem-dimethyl, 250: derived 2 rotors, ceiling 11.0, at most 21.0 with the contact; γ-butyrolactone formation, tetramethyl, 11000: derived 2 rotors, ceiling 11.0, at most 21.0 with the contact; δ-valerolactone formation, gem-dimethyl, 10: derived 3 rotors, ceiling 36.5, at most 133.6 with the contact.
Fig. 6 Each measured gem-dimethyl acceleration against the rotamer ceiling under the three conventions, under the count derived from the chain, and under the derived count with the syn-pentane contact made prohibitive.

The derived count is the third convention, so the verdicts are the ones that row already gave. The 250-fold acceleration of γ-butyrolactone formation is above the five-ring’s ceiling of 11.0 by a factor of 22.7, and the 11,000-fold tetramethyl acceleration by a factor of a thousand: the rotamer account is refuted on the five-membered ring. The 10-fold acceleration of δ-valerolactone formation is below the six-ring’s 36.5 by a factor of 3.6: unrefuted.

The contact cannot rescue the refutation. At any cost the five-ring’s ceiling is at most 21.0, still twelve times short of 250 and 524 times short of 11,000. The only lever the derived arithmetic adds moves the ceiling in the measurements’ direction and stops a factor of twelve away. On the six-membered ring the contact makes the account more than sufficient — up to thirteen times the measured acceleration — which is a reminder that “not refuted” there was never support.

How the conformers were built and counted

Each chain is built atom by atom from the bond length, the bond angle and the torsion about each internal bond, using the standard construction that places atom ii from atoms i−3i-3, i−2i-2 and i−1i-1. A conformer’s weight in the plain chain is e−(g ng+s n±)/kTe^{-(g\,n_g + s\,n_{\pm})/kT}, with ngn_g its number of gauche torsions, n±n_{\pm} its number of neighbouring opposite-sign gauche pairs, gg = 3.8 kJ/mol (butane’s gauche energy, quoted) and TT = 298.15 K. The ceiling is the closable all-gauche weight over the all-gauche weight, divided by the closable weight over the total; the effective count is its logarithm divided by ln⁡((1+2x)/2x)\ln((1+2x)/2x).

The checks run where the figures are drawn. Cyclohexane’s chair is rebuilt and closed by bisection on the torsion, and all six of its torsions, read back from the closed ring with a separate measuring function, alternate in sign — which fixes the sign convention against a structure whose torsions are known rather than trusting the builder. At three closure distances inside each chain’s window, for chains of four to eight atoms, the derived count equals n−3n - 3 to 10−910^{-9}, and a finite enumeration with a penalty of 400 kJ/mol on every anti torsion reproduces the ceiling to 10−910^{-9}. The alternating conformer is confirmed the closest at all fifteen combinations of angle and gauche torsion. And the refusal: build the six-atom chain with every other torsion mirrored, the error a sign convention invites, and the closest conformer must become the one-sign arrangement. It does. If it had not, “alternating” would have been a label the geometry could not tell from its opposite.

What a staggered chain cannot say

Lactones are not alkanes. The measured closures form an ester: an oxygen attacks a carbonyl carbon, the bond lengths and angles at both are not 1.53 Å and 112°, and the ester group prefers its own planar arrangement. The chain built here is the carbon chain the account was written for. The robustness scan says the identity of the closable conformer survives a few degrees and tenths of an ångström; it says nothing about a carbonyl’s planarity.

Staggered torsions only. Every conformer here sits at a torsional minimum. A real chain closes through arrangements a little off them, and a closure distance is a stand-in for a reaction coordinate. That the derived count is the same anywhere in a window several tenths of an ångström wide is what makes the stand-in harmless.

The contact is swept, not known. The syn-pentane cost between two carbons is a well-studied number; between a hydroxyl oxygen and a carbonyl carbon in the arrangement that precedes their bonding, it is not, and it may well be attractive rather than repulsive. A negative cost lowers both ceilings — the five-ring’s to 5.6 at −3 kJ/mol and the six-ring’s below its measurement at −2.9. The sweep shows the most the contact could do in one direction, which is the direction that could have rescued the account; the other direction makes the refutation stronger.

A count that was a property of the chain all along

The count that was never written down treated rr as a parameter the account needed and nobody supplied, and was right to test every value in use. Built from the chain, it is not a parameter. It is the number of torsions, it holds for every reasonable closure distance, and the convention that disagreed with it in direction was not counting anything. The lever that was supposed to be smaller and the curve between two rows moved the ceiling by changing an energy; the count was the one input that could be computed instead of chosen, and computing it found the conformer the account had misnamed.

The same thing has happened more than once in the measurements of strain: an account is right about a number and wrong about the arrangement that produces it, and the arrangement turns out to carry a term the number never had. Cyclopentane’s missing strain was torsional; here the closable conformer’s missing cost is a contact. Both were invisible as long as nobody built the molecule.

Still open: the ester’s own geometry, and a contact with a sign

The obvious open question is the lactone itself. Building the chain with the ester’s bond lengths, the carbonyl’s planarity and the hydroxyl’s two lone-pair directions would say whether the alternating arrangement is still the closable one when one end is an oxygen and the other a trigonal carbon, and whether the count stays n−3n - 3 when the torsion next to the carbonyl prefers to be planar rather than staggered. The window measured here is wide, which suggests it will; a carbonyl that pins one torsion would reduce the count by one, and that would lower every ceiling by a factor of 3.3.

The nearer question is the sign of the contact. The sweep priced it as a repulsion, the way syn-pentane is priced between carbons. For a hydroxyl approaching a carbonyl it is the first stage of a bond, and a modest attraction there would make the closable conformer more common in the plain chain and lower the ceiling. The sweep above already says how attractive it would have to be to change the six-membered verdict, the one verdict in this sequence that every lever so far has left standing: 2.9 kJ/mol a pair. What it cannot say is whether the real contact is that attractive. An energy of the hydroxyl oxygen against the carbonyl carbon at the alternating conformer’s 1.8 Å-scale separation, from any calculation that treats the two groups as what they are, would turn that threshold into a verdict.

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.

Named objects

A dashed tag is an object no other essay names yet.

ClosureConformerConventionRing strainRotamerTorsion