What a spectrum settles

Each band reaches three quarters

Every depolarisation probe of a distortion's direction so far sorted each split pair's two bands by frequency and took the larger departure from three quarters. Followed instead by its composition, each band of ammonia's two degenerate pairs sits at three quarters exactly — to 10⁻⁸ or better — at three directions 120° apart, where it is antisymmetric under the mirror the distortion keeps, and reaches its largest departure halfway between; its partner does the same shifted by 60°. So each band's contrast round the circle is twice the larger-of-two reading's, 2.80 × 10⁻³ against 1.42 × 10⁻³ for the stretching pair, and two old results are one: the factor of 1.98 the larger-of-two reading varied by is the envelope of two complementary curves, and the sum that was flat is their complement.

Worth reading first: The pyramid trades its pairs · One number was a direction too.

A molecule with a three-fold axis has degenerate pairs of vibrations, and a Raman band from a degenerate pair is depolarised: its depolarisation ratio is exactly three quarters. Distort the molecule a little and the pair splits, each band’s ratio moves off three quarters, and how far it moves depends on the direction of the distortion. That is the depolarisation probe this sequence has been reading since one number turned out to be a direction too: round a circle of distortions in the plane of the three bond stretches outside the totally symmetric one, the departure from three quarters varies, and the variation says which way the molecule was bent.

Every reading of the probe so far has made the same bookkeeping choice. At each direction, a split pair’s two bands are sorted by frequency and the larger of their two departures is taken as the pair’s reading. The pyramid trades its pairs used it to find that on ammonia the stretching pair carries the reading, thirty times over, where on boron trifluoride the bending pair does. Its continuation named the alternative a spectroscopist following one line would use — follow one physical band round the circle by what it is, not by where it sits in the frequency order — and named the check that labelling allows: on the pyramid, at a direction the distortion keeps a mirror, one band of each pair is antisymmetric under that mirror and its depolarisation is exactly three quarters.

Followed that way, every band goes to three quarters exactly, and two results this sequence found separately turn out to be one.

Following a band by what it is

At each direction round the circle, in steps of 2.5°, ammonia is distorted by 0.02 Å, its normal modes are found in the same valence force field, and its Raman tensors computed. Each of the four bands of the two split pairs is carried from one step to the next by the overlap of its mode shape with the shape it had one step earlier: within each pair, the assignment that keeps the shapes most nearly the same is the one taken.

The labelling is unambiguous. At no step does any band’s shape overlap its predecessor’s by less than 0.9997, on ammonia or on boron trifluoride, so there is never a choice to make. And the frequency order is never broken: for both molecules, at this amplitude, the band followed by its shape is always the lower or always the upper of its pair. The two ways of naming a band coincide here. What does not coincide is the reading.

Followed by its composition, each stretching band of ammonia reaches three quarters exactly, three times round the circle — so its contrast is twice the larger-of-two's. NH₃, the stretching pair: each of its two components followed round the circle by the continuity of its mode shape, its depolarisation's departure from three quarters against the direction of a 0.02 Å distortion; dashed, the larger of the two at each direction. Component 1: from 0 to 2.77×10⁻³, exactly three quarters at 30°, 150°, 270°; Component 2: from 0 to 2.80×10⁻³, exactly three quarters at 90°, 210°, 330°. The larger of the two runs from 1.38×10⁻³ to 2.80×10⁻³.
Fig. 1 Each of ammonia’s two stretching bands followed round the circle, its departure from three quarters against the direction of the distortion, with the larger of the two at each direction dashed.

Each stretching band’s departure swings from zero to about 2.8×10−32.8 \times 10^{-3} and back once every 120°. The lower band sits at three quarters exactly at 30°, 150° and 270°; the upper at 90°, 210° and 330° — to 10−810^{-8} or better, which is the precision of the arithmetic and of the directions themselves rather than a small number. Each band reaches its largest departure where its partner is at three quarters, and the two cross at half the peak at every multiple of 60°.

The zeros are the check the earlier essay asked for, and they are symmetry, not coincidence. A distortion along one of those directions leaves the pyramid with a mirror plane, and of the two bands a split pair becomes, one is symmetric under that mirror and one antisymmetric. A Raman tensor antisymmetric under a mirror has no trace and a fixed shape of anisotropy, and its depolarisation ratio is three quarters for any amplitude. The other band carries everything the distortion did. Turn the distortion by 60° and the mirror lies the other way: the roles swap within the pair.

The pair is larger of two; each band is the whole range

The larger-of-two reading is the dashed curve above, and it is the upper envelope of the two followed bands. It never falls to zero, because at every direction one of the two bands is far from three quarters; its lowest point is where the two cross, at half the peak. So the larger-of-two reading of ammonia’s stretching pair runs from 1.38×10−31.38 \times 10^{-3} to 2.80×10−32.80 \times 10^{-3}, a contrast round the circle of 1.42×10−31.42 \times 10^{-3}, while each followed band runs from zero to its peak, a contrast of 2.772.77 and 2.80×10−32.80 \times 10^{-3}. Following one band doubles the contrast a measurement has to resolve.

The bending pair does the same at a thirtieth of the size: its two bands alternate, each exactly at three quarters where the other is largest. NH₃, the bending pair: each of its two components followed round the circle by the continuity of its mode shape, its depolarisation's departure from three quarters against the direction of a 0.02 Å distortion; dashed, the larger of the two at each direction. Component 1: from 1.11×10⁻¹⁶ to 8.90×10⁻⁵, exactly three quarters at 30°, 150°, 270°; Component 2: from 1.11×10⁻¹⁶ to 9.71×10⁻⁵, exactly three quarters at 90°, 210°, 330°. The larger of the two runs from 4.70×10⁻⁵ to 9.71×10⁻⁵.
Fig. 2 Ammonia’s two bending bands followed round the circle, with the larger of the two dashed.

The bending pair does the same at a thirtieth of the size: its lower band at three quarters at 30°, 150° and 270°, the upper at 90°, 210° and 330°, each reaching about 9×10−59 \times 10^{-5} where the other is at three quarters. The larger of the two runs from 4.70×10−54.70 \times 10^{-5} to 9.71×10−59.71 \times 10^{-5}, again almost exactly a factor of two.

The factor of two was the envelope

That factor has been in the argument since its beginning. One number was a direction too found boron trifluoride’s reading, the largest departure over its four bands, varying by a factor of 1.98 round the circle, smallest for one bond against another and largest for two bonds against one. It was read as the probe’s anisotropy — the distortion’s effect on the spectrum being twice as large in some directions as in others.

Followed, the bands say something different. Each band’s departure has the shape of a squared sine of the direction, going to zero at its own mirror directions; its partner’s is the complementary squared cosine. The larger of sin⁡2\sin^2 and cos⁡2\cos^2 never falls below one half and never rises above one, so the larger-of-two reading of any pair whose bands behave this way varies by a factor of two whatever the molecule. Measured on the fine circle: 1.978 for boron trifluoride’s bending pair, 2.012 for its stretching pair, 2.065 and 2.027 for ammonia’s bending and stretching pairs. The departures from exactly two are how far each band’s curve is from a pure squared sine, which at an amplitude of 0.02 Å is a few per cent.

On the planar molecule the bending bands alternate the same way, with no exact zero to check against. BF₃, the bending pair: each of its two components followed round the circle by the continuity of its mode shape, its depolarisation's departure from three quarters against the direction of a 0.02 Å distortion; dashed, the larger of the two at each direction. Component 1: from 0 to 1.62×10⁻⁵; Component 2: from 0 to 1.60×10⁻⁵. The larger of the two runs from 8.18×10⁻⁶ to 1.62×10⁻⁵.
Fig. 3 Boron trifluoride’s two bending bands followed round the circle, with the larger of the two dashed.

On the planar molecule the bands alternate the same way. There is no exact zero to check against at this precision — its departures are a hundred times smaller than ammonia’s and its antisymmetric band’s departure is zero only to 10−710^{-7} over several neighbouring steps — but the shape is the same: two complementary curves and an envelope at half to one.

And the sum was flat all along is the same statement from the other side. That essay found that summing the bands instead of taking the largest gave a reading constant round the circle to four parts in ten thousand, and concluded the anisotropy belonged to the maximum rather than to the physics. It did, and now the reason can be seen: sin⁡2+cos⁡2\sin^2 + \cos^2 is one. The followed bands’ departures summed over each pair are constant to between 1 and 9 per cent — the summed intensities of the earlier essay were flatter still, because intensity weighting makes the complementarity closer — and the factor of two and the flat sum are two readings of one pair of complementary curves.

Why exactly a factor of two

The squared-sine shape is not an accident of these two molecules, and the reason is the one the shape was the group found for the larger-of-two reading’s period. A band’s departure round the circle must repeat every 120°, because a three-fold rotation carries one direction of the distortion to another and the molecule to itself. The simplest function with that period is a constant plus a single third harmonic, A+Bcos⁡3(θ−θ0)A + B\cos 3(\theta - \theta_0). The zero at the band’s own mirror direction then fixes A=BA = B, since the departure cannot be negative and must reach zero there. So to the lowest order in the harmonics, each band’s departure is A (1−cos⁡3(θ−θ0))A\,(1 - \cos 3(\theta - \theta_0)), its partner’s is A (1+cos⁡3(θ−θ0))A\,(1 + \cos 3(\theta - \theta_0)), the larger of the two runs from AA to 2A2A, and their sum is 2A2A everywhere.

That makes the factor of two a theorem about the lowest harmonic, and the departures from it — 1.978 for one pair, 2.065 for another — measures of the higher harmonics: terms in cos⁡6θ\cos 6\theta that a 0.02 Å distortion produces at the level of a few per cent. Two integers made one exponent split the flat sum’s small residual into an even part growing as the square of the amplitude and an odd part growing linearly; the higher harmonics here are plausibly the same kind of correction read band by band, and whether they vanish as the distortion shrinks, as that residual’s leading part does, is a sweep over the amplitude that was not run here.

The roles belong to the pairs

The essay on the pyramid found that ammonia’s stretching pair carries its reading and boron trifluoride’s bending pair carries the plane’s, and asked whether that division belongs to the pairs or to particular bands within them.

Read band by band, the roles still belong to the pairs: ammonia's stretching bands carry thirty times its bending bands, and the plane's the other way round. The contrast round the circle — largest departure from three quarters less the smallest — of each followed component, and of the larger-of-two reading of each pair, at zero line width: NH₃ bend 1, 8.90×10⁻⁵; NH₃ bend 2, 9.71×10⁻⁵; NH₃ bends, larger of two, 5.01×10⁻⁵; NH₃ stretch 1, 2.77×10⁻³; NH₃ stretch 2, 2.80×10⁻³; NH₃ stretches, larger of two, 1.42×10⁻³; BF₃ bend 1, 1.62×10⁻⁵; BF₃ bend 2, 1.60×10⁻⁵; BF₃ bends, larger of two, 8.00×10⁻⁶; BF₃ stretch 1, 4.59×10⁻⁶; BF₃ stretch 2, 4.76×10⁻⁶; BF₃ stretches, larger of two, 2.39×10⁻⁶.
Fig. 4 The contrast round the circle of every followed band of both molecules, and of each pair’s larger-of-two reading, at zero line width.

It belongs to the pairs. Within each pair the two followed bands have contrasts equal to within ten per cent — ammonia’s stretching bands 2.772.77 and 2.80×10−32.80 \times 10^{-3}, its bending bands 8.908.90 and 9.71×10−59.71 \times 10^{-5}; boron trifluoride’s bending bands 1.621.62 and 1.60×10−51.60 \times 10^{-5}, its stretching bands 4.594.59 and 4.76×10−64.76 \times 10^{-6}. Between pairs they differ by the factors the larger-of-two reading found: ammonia’s stretching bands thirty times its bending bands, the plane’s bending bands three and a half times its stretching ones. No single band carries the role while its partner does not; both members of a pair carry it, each over the half of the circle where the other is near three quarters.

At a finite line width

At zero line width the reading of a band is its own depolarisation. In a spectrum with lines of width ww, a band’s own depolarisation is not what a measurement sees: the pair’s two lines overlap, and the depolarisation measured at any frequency is that of the envelope there. The essays that brought in line widths read the envelope at its peaks, which is what a measurement of a band’s depolarisation usually does, and found ammonia’s direction going at a width of 2.6 cm⁻¹, because the stretching pair is split by only 2.1 cm⁻¹ and its two peaks merge into one.

A followed band can instead be read off the envelope at its own frequency, which a spectrometer can do whenever it knows where the band is — the depolarisation ratio measured as a function of frequency across an unresolved band, at the positions the resolved spectrum would have its lines.

Read at its own frequency, a followed stretching band keeps half its contrast to 3.0 cm⁻¹; the envelope's peaks lose it as the two lines merge. Ammonia's stretching pair: contrast round the circle against line width, on logarithmic axes, for each followed band read off the envelope at its own frequency, for the larger of the two read at the two frequencies, and for the larger of the two read at the envelope's peaks as the published probe reads it. Half the resolved contrast is reached at 2.99 and 3.01 cm⁻¹ for the followed bands, 3.00 for the larger of two at the lines.
Fig. 5 Ammonia’s stretching pair: the contrast round the circle against line width, for each followed band read off the envelope at its own frequency, for the larger of the two read at the same frequencies, and for the larger of two read at the envelope’s peaks.

Read that way, a followed stretching band keeps half its contrast to a line width of 3.0 cm⁻¹ and still holds a twelfth of it at 10 — a contrast of 2.3×10−42.3 \times 10^{-4}, where the reading at the envelope’s peaks has fallen to 6×10−66 \times 10^{-6}. The peak reading collapses at about 4 cm⁻¹, where the two lines merge into one peak. The reason is the flat sum: a merged peak’s depolarisation is that of the two lines added together, and the two complementary bands added together barely vary round the circle. Reading at the lines does not suffer that cancellation, though at widths well beyond the splitting each line’s reading is mostly its partner’s too and the contrast falls as the overlap grows.

The bending bands, split by 6.5 cm⁻¹, keep half their contrast to 8.5 and 9.7 cm⁻¹ read the same way. On boron trifluoride the bending bands halve at about 4.0 cm⁻¹ and the stretching bands, split by less, at 0.75.

Every followed band of both molecules. For each component of each degenerate pair, followed round the circle: its frequency at the reference direction, its largest departure from three quarters, its contrast round the circle at zero width, and the line width at which that contrast has halved. NH₃ bend 1: 1676.6 cm⁻¹, 8.90×10⁻⁵, 8.90×10⁻⁵, 8.53 cm⁻¹; NH₃ bend 2: 1683.1 cm⁻¹, 9.71×10⁻⁵, 9.71×10⁻⁵, 9.71 cm⁻¹; NH₃ stretch 1: 3565.7 cm⁻¹, 2.77×10⁻³, 2.77×10⁻³, 2.99 cm⁻¹; NH₃ stretch 2: 3567.8 cm⁻¹, 2.80×10⁻³, 2.80×10⁻³, 3.01 cm⁻¹; BF₃ bend 1: 479.0 cm⁻¹, 1.62×10⁻⁵, 1.62×10⁻⁵, 3.93 cm⁻¹; BF₃ bend 2: 481.8 cm⁻¹, 1.60×10⁻⁵, 1.60×10⁻⁵, 4.04 cm⁻¹; BF₃ stretch 1: 1453.6 cm⁻¹, 4.59×10⁻⁶, 4.59×10⁻⁶, 0.75 cm⁻¹; BF₃ stretch 2: 1454.2 cm⁻¹, 4.76×10⁻⁶, 4.76×10⁻⁶, 0.74 cm⁻¹.
Fig. 6 Every followed band of both molecules: its frequency, its largest departure, its contrast round the circle and the line width at which that contrast halves.

How the bands were followed

The force fields, the distortion and the Raman tensors are those of the earlier essays: a valence force field fitted to each molecule’s fundamentals, a distortion of 0.02 Å along a direction in the two-dimensional plane of bond-stretch combinations outside the totally symmetric one, normal modes from the distorted geometry with the force constants held, and each mode’s polarisability derivative computed as in those essays. The depolarisation ratio of a band is 3γ2/(45αˉ2+4γ2)3\gamma^2/(45\bar\alpha^2 + 4\gamma^2) in terms of its tensor’s mean and anisotropy. Directions are taken every 2.5° round the full circle.

Following is by the absolute overlap of each band’s Cartesian mode shape with the four shapes one step earlier, the assignment within each pair being the one with the larger total overlap, and the sign of each shape carried so that successive shapes point the same way. At finite width each band is a Lorentzian of full width ww in both polarisations, and a followed band’s reading is the envelope’s depolarisation at that band’s own frequency. The larger-of-two reading at zero width reproduces the published stretching-pair contrast on the published nine directions to within a tenth, and every followed stretching band’s contrast exceeds it.

What a followed band cannot be in a real spectrum

Following needs the bands’ compositions. A spectroscopist does not see mode shapes. What can be followed in practice is a band’s position, and here the frequency order and the composition coincide, so a band followed by frequency is the band followed by shape. At larger distortions or near an accidental crossing of a pair’s two bands that would fail, and the reading at the lines would be reading the wrong line on part of the circle.

Three quarters exactly is a symmetry statement about a model. The zeros hold for any Raman tensor antisymmetric under a mirror, so they survive any change of the polarisability model that respects the symmetry. They do not survive anharmonicity that mixes the pair with other modes, nor a distortion with components outside the plane the circle is drawn in.

Ammonia’s bands are tunnelling doublets. The inversion splits every level of ammonia into two, and the essays on the pyramid computed a harmonic field without it. A stretching pair split by 2.1 cm⁻¹ by the distortion sits beside inversion splittings of the order of a wavenumber in the stretching fundamentals; which of the two splittings a followed band would follow is not decided here.

A pair is two complementary bands

Reading the larger of two was a reasonable choice for a probe that had to report one number per pair, and it is the reading a frequency-sorted table invites. It hid the simplest thing about each band: that at the directions where the distortion keeps a mirror, half of each pair is not distorted at all, in the one sense a depolarisation can see. The factor of two that looked like an anisotropy of the distortion, and the flatness of the sum that looked like a separate finding, were both that fact read through two different bookkeeping choices.

Still open: the inversion doublets, and a distortion that leaves the plane

The obvious open question is the one the harmonic field hides. Each of ammonia’s bands is a tunnelling doublet, and the distortion and the inversion compete for the same levels: at a small distortion the tunnelling splitting is the larger and the degenerate pair is split within each tunnelling component, at a larger one the distortion dominates. A followed band reaches three quarters exactly only where its own mirror survives, and the inversion’s two components have opposite parity under the pyramid’s own symmetry. Whether the zeros survive in both tunnelling components, or one of them is pinned at three quarters at directions where the other is not, needs the inversion coordinate added to the same force field.

The nearer question is a distortion with a totally symmetric part. The circle here lies in the plane outside the totally symmetric species, where the probe is exactly blind to the symmetric direction. A real distortion has both, and adding a symmetric part changes the frequencies of both bands of a pair together and leaves the mirror at the same directions. Whether the followed bands’ zeros stay exactly where they are when the distortion leaves the plane — they should, by the same symmetry — and how their peaks move, is a check that needs only a third coordinate on the circle.

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.

DegeneracyDepolarisation ratioPolarisabilityRaman spectroscopySymmetry breaking