Observation of the onset of torsion-induced, mode-specific dissipative intramolecular vibrational redistribution (IVR)

Evidence is found showing that coupling with vibration-torsion (“vibtor”) levels of both in-plane and out-of-plane vibrations is instrumental in causing dissipative intramolecular vibrational redistribution (IVR). Both zero-electron-kinetic-energy (ZEKE) and two-dimensional laser-induced fluorescence (2D-LIF) spectroscopy are employed to investigate a series of bands located ~ 1200 cm -1 above the S 1  S 0 origin in p -fluorotoluene ( p FT). Transitions in this wavenumber region have been the focus of a number of studies probing intramolecular vibrational redistribution (IVR). By recording both ZEKE and 2D-LIF spectra, a prepared S 1 population is projected onto both the ground state cation and ground state neutral energy states, respectively, giving added confidence to the assignments. The spectral region under discussion is dominated by a pair of fundamental bands, but for the first time, we present explicit evidence that this is complicated by contributions from a number of overtones and combinations, including vibtor levels. We deduce that very different extents of coupling are present across a 60 cm -1 window of the spectrum, even though the density of states (DOS) is similar; in particular, one of the fundamentals couples efficiently to the increasing bath of levels, while one does not. We explain this by the influence of serendipitous near-coincidences of same-symmetry levels.


Introduction
The ability to describe the making and breaking of chemical bonds and the flow of energy through a molecule necessitates a knowledge of the internal energy level structure; in particular, vibrations and torsions. Building on work by Parmenter and coworkers, 1,2,3,4,5,6,7 and the group of Weisshaar, 8,9,10 recent work by the Lawrance group and ourselves has identified that vibration-torsional ("vibtor") coupling is of key importance in the following para-substituted molecules that contain methyl groups: toluene, 11,12,13,14,15,16,17 pFT 18,19,20,21,22,23,24,25,26,27 and para-xylene (pXyl). 25,28,29 Some of the most recent work has employed the technique of two-dimensional laser-induced fluorescence (2D-LIF), which has been reviewed recently. 30 Following on from earlier work on pFT 31,32 , Seliskar et al. 33 and Okuyama et al. 34 presented laserinduced fluorescence (LIF) spectra under jet-cooled conditions, giving assignments of some of the vibrational bands. A number of the low-wavenumber bands have been reassigned to vibration-torsion (vibtor) levels by Zhao,5 which was confirmed in recent work by our group 21 and that of Lawrance et al. 22 An overview of the lowest 1250 cm -1 of the S1  S0 REMPI spectrum is shown in Figure 1. Recently, we have examined the bands close to 400 cm -1 (Ref. 26) and those at ~800 cm -1 (Ref. 27) using 2D-LIF, and compared the results to our earlier ZEKE work. 18,21,23 The assignments of the 2D-LIF spectra indicated that both vibrational interactions and vibration-torsional coupling were occurring.
In the present work, we examine the 1200 cm -1 region of the spectrum -see Figure 1. This region of the spectrum contains two strong fundamental transitions, and these have been assigned by Okuyama et al. 34 with LIF, and confirmed by our later ZEKE study. 18 What was remarkable about the ZEKE spectra was that the one recorded from the lower fundamental (1196 cm -1 ) was highly structured, with very limited broadness in the baseline, while the one recorded via the higher one (1232 cm -1 ) had a very large amount of complicated overlapped structure across a wide range. This was interpreted as a significant increase in IVR in the latter, despite its being only 36 cm -1 higher in wavenumber. These spectra were further discussed in Refs. 19 and 20 in tandem with additional time-resolved data from the Reid group at Nottingham. In particular, in was noted therein that there has been some ambiguity in the labelling of vibrational levels previously studied, and effects of different experimental conditions were discussed; 19 also, comparisons between observations for toluene, toluene-d3 and pFT were made. 20 Finally, more-recent work from the Reid group 35 has looked at the same fundamentals, although the focus of that work was on combination levels involving these, located higher in wavenumber.
Both a combination of an increased density of states (DOS) and symmetry-allowed vibtor coupling have been invoked to rationalize the rapid increase in interactions that occur in such molecules and which drive energy dispersal. 25 In the present work, we consider a ~60 cm -1 region close to 1200 cm -1 internal energy of the S1 state of pFT. Very different intramolecular vibrational redistribution (IVR) behaviour has been observed for two fundamental vibrations located here. It has been hypothesised that the presence of a methyl group is responsible, but over such a narrow range the DOS is not expected to change enormously, and so the reason for the difference is unclear. Here, we examine both the abovementioned fundamental levels and other levels that are located in the 1190-1250 cm -1 range. In this region we expect to see combinations involving levels on which we have recently reported, at ~400 cm -1 (Ref. 26) and ~800 cm -1 (Ref. 27); additionally, combinations involving the levels at ~400 cm -1 and those at ~845 cm -1 (Ref. 24) may be expected to lie towards the higher wavenumber end of this region -see Figure 1. We piece together evidence from the 2D-LIF and ZEKE spectra presented in the present work to assign this region more completely than has been done before. We conclude that one of the fundamentals is able to couple efficiently to the increasing number of levels, via interactions with vibtor levels, while the other is not. We discuss the rationale for this.

Experimental
The 2D-LIF apparatus is the same as that employed recently. 24 The vapour above room temperature para-fluorotoluene (99% purity, Alfa Aesar) was seeded in ~5 bar of Ar and the gaseous mixture passed through a General Valve pulsed nozzle (750 μm, 10 Hz, opening time of 180-210 μs) to create a free jet expansion. This was intersected at X/D ~20 by the frequency-doubled output of a single dye laser (Sirah CobraStretch), operating with C540A. The fluorescence was collected, collimated and focused onto the entrance slits of a 1.5 m Czerny Turner spectrometer (Sciencetech 9150) operating in singlepass mode, dispersed by a 3600 groove/mm grating, and ~300 cm -1 windows of the dispersed fluorescence collected by a CCD camera (Andor iStar DH334T). At a fixed grating angle of the spectrometer, the excitation laser was scanned, and at each excitation wavenumber the camera image was accumulated for 2000 laser shots. This allowed a plot to be produced of fluorescence intensity versus both the excitation laser wavenumber and the wavenumber of the emitted and dispersed fluorescence, termed a 2D-LIF spectrum. 30 We have also recorded some separate dispersed fluorescence (DF) spectra with higher averaging to get better signal-to-noise than simply taking a vertical slice through the 2D-LIF image. These DF spectra were recorded with the same spectrometer as for the 2D-LIF spectra, and were recorded three times accumulating over 5000 shots each time, and an average taken of these.
The REMPI/ZEKE apparatus was the same as that used in earlier work. 18 The focused, frequencydoubled outputs of the two dye lasers (Sirah CobraStretch) were overlapped spatially and temporally and passed through a vacuum chamber coaxially and counterpropagating, where they intersected a free jet expansion of pFT in Ar between two biased electrical grids located in the extraction region of a time-of-flight mass spectrometer, which was employed in the REMPI experiments. These grids were also used in the ZEKE experiments by application of pulsed voltages, giving typical fields (F) of ~10 V cm -1 , after a delay of up to 2 s, where this delay was minimized while avoiding the introduction of excess noise from the prompt electron signal. The resulting ZEKE bands had widths of ~5-7 cm -1 .
Electron and ion signals were recorded on separate sets of microchannel plates.
The excitation laser operated with C503 and was pumped with the third harmonic (355 nm) of a Surelite III Nd:YAG laser. The ionization laser operated with Pyromethene 597, pumped with the second harmonic (532 nm) of a Surelite I Nd:YAG laser. The fundamental outputs produced by each dye laser were frequency doubled.

Vibrational and Torsional Labelling
Since neither Wilson 36 /Varsányi 37 nor Mulliken 38 /Herzberg 39 notations are appropriate for the vibrations of pFT, 40,41 we shall employ the Di labels from Ref. 41. In other papers, we have provided correlations with the labels used in previous work to aid the reader in referring to earlier studies. 23,26 Note that we shall refer to previously calculated values (B3LYP/aug-cc-pVTZ) of the vibrational wavenumbers in the three electronic states pertinent to the present study, presented in our previous work -those for the S1 and D0 + states are taken from Ref. 23, while those for the S0 state are from Ref.

41.
Although referred to in terms of Wilson nomenclature in earlier work, in fact a detailed analysis of the vibrational motions of various symmetric and asymmetric disubstituted benzenes 41 showed that the D5 and D6 vibrations were in-phase and out-of-phase motions of the C-X stretches, where X is the substituent in the symmetrically-substituted molecules. In the asymmetrically-substituted analogues, these vibrations evolve into localized stretches: in the halotoluenes, D5 was assigned as the C-Hal stretch, while D6 was identified as the C-CH3 stretch, where Hal represents a halogen atom. 41 Since the G12 molecular symmetry group (MSG) is appropriate for vibtor levels in pFT, we shall use these symmetry labels throughout. In addition, torsional levels will be labelled via their m quantum number. (The reader may find it useful to refer to previous work 15,16,17,21,28 if they are not familiar with these labels.) The correspondence between the C2v point group labels and the G12 MSG ones are given in Table I. To calculate the overall symmetry of a vibtor level, it is necessary to use the corresponding G12 label for the vibration, and then find the direct product with the symmetry of the torsion ( Under the free-jet expansion conditions employed here, almost all of the molecules are expected to be cooled to their zero-point vibrational level and thus essentially all S1  S0 pure vibrational excitations are expected to be from this level. In contrast, owing to nuclear-spin and rotational symmetry, the molecules can be in one of the m = 0 or m = 1 torsional levels. 28,42

Coupling and transitions
If an anharmonic vibration is close in wavenumber to one or more combination or overtone vibrational levels that has the same overall symmetry, then "off-diagonal" anharmonic interactions can occur, with the simplest example of two interacting states being the classic Fermi resonance (FR). 43 The noninteracting levels are termed zero-order states (ZOSs), and their interaction leads to the formation of eigenstates that are linear combinations of these, and will be at different wavenumbers to the original ZOSs. 39 For molecules that contain a hindered internal rotor, and if vibration-torsional coupling occurs, then the ZOSs can also be torsional or "vibtor" levels. The end result of such interactions is the formation of eigenstates which facilitate delocalization of energy through widespread motion of the molecule. Such couplings are only expected to be significant for small changes, v  3, of the vibrational quantum number, and also for changes, m, of 0, ±3 or ±6 in the torsional quantum number in descending order of likely strength. 16,21,28,44,45 In electronic spectroscopy, if we assume a non-coupled picture initially, then a vibrational, torsional or vibtor ZOS can be bright (i.e. it has a significant transition intensity) or dark (i.e. it has no, or a very small transition intensity); these are often termed a zero-order bright (ZOB) state and a zero-order dark (ZOD) state, respectively. Following interaction, the resulting eigenstates will be composed of mixtures of ZOB and ZOD state character and so more transitions will become observable in the spectrum as a result of the interaction, by virtue of the ZOB character.
When designating excitations, we shall generally omit the lower level, since it will be obvious from the jet-cooled conditions; similarly, for emissions, we shall omit the upper level, as that will be obvious from the excitation and context. In the usual way, vibrational transitions will be indicated by the cardinal number, i, of the Di vibration, followed by a super/subscript specifying the number of quanta in the upper/lower states, respectively; torsional transitions will be indicated by m followed by its value. Finally, vibtor transitions will be indicated by a combination of the vibrational and torsional transition labels. E.g. 5 1 m 0 implies an excitation from the torsionless (implied by symmetry) zero-point vibrational level in the S0 state to the S1 level with one quantum, of D5 excited, and no torsional excitation, while 5 1 m 1 implies an excitation from the zero-point vibrational level, but with one quantum of torsion (implied by symmetry), in the S0 state to the S1 level with one quantum, of D5 excited, and one quantum of torsion. If no m values are specified, then the transition label refers to transitions involving both m = 0 and m = 1, whose transition wavenumbers are expected to be coincident at the present resolution.
The wavenumbers of the levels will be given with respect to the relevant zero-point level in each electronic state, but noting that some excitations will originate from the m = 1 level in S0 and those transition energies are given with respect to that level, as usual. The S1  S0 origin is located at 36860.0 cm -1 (Ref. 22). The most intense transition is generally expected to be that for which no change in the vibrational, or both vibrational and torsional, quantum numbers occurs; these will be designated as v = 0 or (v, m) = 0 transitions. As has become common usage, we will generally refer to a level using the notation of a transition, with the level indicated by the specified quantum numbers, with superscripts indicating levels in the S1 state and subscripts indicating levels in the S0 state; since we will also be referring to levels in the ground state cation, D0 + , we shall indicate those levels with superscripts with a preceding superscripted + sign. Also, the eigenstates will often be referred to by the dominant contribution from one of the ZOSs, with the context implying if an admixture of other ZOSs is present. 2D-LIF band positions will be indicated by a pair of (excitation, emission) wavenumbers, and the corresponding transitions similarly.

The S1  S0 spectrum
In Figure 1, we show a REMPI spectrum of pFT up to 1250 cm -1 . Indicated are a set of bands at ~400 cm -1 that are dominated by 14 2 , 29 1 and 11 1 transitions, 18,21,22,26 but where there are also additional, including vibtor, transitions. 21,22,26 At ~800 cm -1 there are a further set of transitions, dominated by a pair of levels that largely comprise a FR between the 9 1 and 29 2 levels, 23,27,46,47 but again, there are other transitions in this region, including those involving vibtor levels (see Refs. 23 and 27). At ~845 cm -1 there is the 18 2 transition, which has been shown to be interacting strongly with several vibtor levels at this wavenumber (see Ref. 24), while at ~1015 cm -1 there is a band that is the subject of ongoing work. At ~1200 cm -1 , there are a series of bands which are dominated by two fundamental transitions, assigned as 6 1 and 5 1 ; in earlier work, these have mainly been referred to with Wilson labels 13 and 7a, respectively. 18,20,34 It is clear, however, that further bands are present and expanded views of this wavenumber region are presented in Figure 2, recorded using three different techniques: REMPI, LIF and integration of the 2D-LIF spectrum (see Section 3.2.3). Immediately apparent is the relative similarity of the LIF and integrated 2D-LIF spectra. However, the relative intensities of the two most intense bands is clearly different to those in the REMPI spectrum, and indeed the higherwavenumber feature has an even greater relative intensity in some REMPI scans. 18 This is supportive of there being some time-dependent effect on the 5 1 band, with REMPI being a far faster technique than LIF, and higher laser power densities favouring ionization over S1 population-loss mechanisms.
We shall now present the ZEKE spectra and the 2D-LIF spectrum, each recorded across the spectral region shown in Figure 2. We shall then discuss these together in deducing the detailed assignments.

ZEKE Spectra
In Figure 3, we show ZEKE spectra recorded at each of the indicated positions; the ZEKE spectrum recorded via the origin is shown for comparison purposes. The indicated transitions arise from the overall assignment to be discussed in Section 3.3. We highlight that the ZEKE spectrum via the origin has a flat baseline throughout. The spectrum recorded at 1196 cm -1 has a largely flat baseline until ~1000 cm -1 , and then a small amount of underlying complicated, overlapped structure is seen thereafter. The strong ZEKE bands in this spectrum can be assigned as + 6 1 , + 6 1 11 1 and + 6 1 9 1 . All other spectra have significant contributions from a wide range of underlying overlapped structure, although some caution is required for ZEKE spectra that were recorded via levels corresponding to weak bands in the REMPI spectrum, where the background can be relatively significant. We highlight the spectrum recorded via the level corresponding to the intense band at 1232 cm -1 , which shows this broad structure, but there are clear bands seen on top of this, notably those assigned to + 5 1 , + 5 1 11 1 and + 5 1 9 1 .
The ZEKE spectra recorded at 1196 cm -1 and 1232 cm -1 are very similar to those reported and discussed previously. 18,19,20 Other bands and indicated assignments will be discussed in Section 3.3.

2D-LIF spectrum
In Figure 4, we show a composite overview of the 2D-LIF spectra recorded across the excitation range corresponding to that of Figure 2, and covering emissions to the origin up until ~2550 cm -1 of the S0 state, selected to cover the v = 0 regions of transitions that might be contributing to the spectrum in Figure 2. Integrating the 2D-LIF spectrum vertically gives an excitation spectrum similar to the REMPI and LIF spectra -see Figure 2. It is evident that there are two main vertical "stripes" of activity located around the 1196 cm -1 and 1232 cm -1 excitations, i.e. as a result of exciting 6 1 and 5 1 , respectively. It is also clear that there is a significant amount of Franck-Condon (FC) emission activity as a result of these two excitations, and that there is more intensity associated with the 6 1 excitation than with the 5 1 excitation. In Figure 5 we shown DF spectra recorded through the centres of the 6 1 and 5 1 excitations and, for comparative purposes, also through the origin. Expanded relevant sections of the 2D-LIF spectra in Figure 2 are given in Figure 6. From Figure 5, it is clear that there is much more broad underlying emission structure when exciting 5 1 than there is for 6 1 , consistent with the behaviour seen in the ZEKE spectra (see Figure 3). As indicated by the additional structure seen in Figures 2 and 4, there is additional activity at other excitation wavenumbers, located both between 6 1 and 5 1 , as well as to higher wavenumber than 5 1 , and this will be discussed and assigned in tandem with the ZEKE spectra in the following subsection.

Assignments
We commence by noting that the previous 18,19,20 and present ZEKE spectra ( Figure 3) are consistent with the main two bands in Figure 2 being 6 1 and 5 1 . The 2D-LIF spectrum is also consistent with this, with outline assignments indicated in the DF spectra in Figure 5 (further assignments are given in Figure 6). Thus, the (6 1 , 61) and (5 1 , 51) v = 0 bands are located at (1196, 1215) cm -1 and (1232, 1241) cm -1 , respectively. A series of other intense Franck-Condon (FC) active combination and overtone bands involving the main transition can be seen to higher wavenumber in each case. As well as 111 (see Figure 5), there are also other emissions, with those in the range ~840-855 cm -1 being assigned as 142, 91 and 292. (Note that there was some indication that, as well as 9 1 and 29 2 being in Fermi resonance in the S1 state, 27,47 the 91 and 292 levels may be in Fermi resonance in S0, and the same was true of + 9 1 and + 29 2 in D0 + . 27 ) We now consider the combination bands involving one of the S1 levels previously discussed 24,26,27 at ~400 cm -1 and another located at ~800 cm -1 or ~845 cm -1 (see Figure 1) that are expected to lie in the 1190-1250 cm -1 range. To aid the reader, in Figure 7 we show the expected positions of the various combinations, indicated by shifting the region of the spectrum at ~400 cm -1 to positions that indicate where the various combinations are expected. If these combinations have significant intensity, we should be able to identify activity from them in the 2D-LIF and ZEKE spectra, although we need to consider that interactions may shift levels from the expected position. In addition, it is possible for additional ZOB states to contribute to this region of the spectrum, analogous to some of those seen at similar wavenumbers in our recent studies of p-difluorobenzene (pDFB) 48 and pchlorofluorobenzene (pClFB), 49,50 but noting that the higher symmetry of pDFB means that some transitions are symmetry forbidden, notably 6 1 . Between different molecules, however, the brightness of any transition can vary, as can its wavenumber, with the latter potentially causing levels to move, affecting the possible interactions. Furthermore, the presence of the methyl group in pFT leads to vibtor levels built on each vibrational level, dramatically increasing the possibilities for coupling -see the discussion in Ref. 25 concerning comparison between the abovementioned three molecules and also pXyl.
One explanation for this would be FC activity, but it could also indicate that the 9 1 14 2 level is coincident with 6 1 . Since this coincidence would imply there was very little interaction between these two levels, and combined with the fact that we see no clear + 9 1 14 2 band in the ZEKE spectrum when exciting via 6 1 , we favour FC activity.
In Figure 8, it can be seen that, as well as the clear FC (6 1 , 142) emission, there is also a secondary 142 emission when exciting at 1198 cm -1 (see also the integrated trace in the lower half of this figure), and we suggest that this could be associated with the 9 1 14 2 v = -1 transition, with there being a ZEKE band at the correct position (Figure 3), as well as a clear + 14 2 band when exciting at this wavenumber.
(Although 14 6 and 14 4 29 1 are expected close to this excitation wavenumber neither + 14 6 nor + 14 4 29 1 bands are clearly evident, although there is a + 14 4 band at 1403 cm -1 ; in addition, the corresponding emission bands are not definitively seen in the 2D-LIF spectrum, and so evidence for activity involving these levels is weak.) We also cannot rule out a possible contribution to the ~1530 cm -1 ZEKE band from + 14 2 29 2 , which would also be consistent with FC activity of + 14 2 band. Although there is no definitive 142292 emission band seen, the 142 band could also be associated with emission from 14 2 29 2 . In summary, although there is some evidence for 9 1 14 2 and 14 2 29 2 (and possibly 14 4 29 1 and 14 6 ) activity when exciting at 1198 cm -1 it is not wholly conclusive.
We now move on to 9 1 29 1 and 29 3 , recalling that these are more correctly thought of as combinations of 29 1 with each of the (9 1 29 2 ) and (29 2 9 1 ) FR pair. The clearest evidence for this activity is from the 291 emission, which shows two significant areas of activity at excitations of 1193 cm -1 and 1203 cm -1 (see Figure 8). These regions of activity are identified as the (29 3 , 293) v = 0 band at (1203, 1275) cm -1 , with the band at (1193, 1264) cm -1 being (9 1 29 1 , 91291); further, there is some evidence for cross activity between these bands, consistent with FR between the 9 1 29 1 and 29 3 levels, as there was 27,47 between the 9 1 and 29 2 levels. In both cases the 293 emission is the strongest, consistent with the 292 band being the stronger when exciting via both of the 9 1 /29 2 FR levels. 27 The weak (6 1 , 291) activity at (1196, 424) cm -1 is assigned as Herzberg-Teller (HT) activity associated with the 6 1 transition, similar to that seen via the origin. 21,22,26 We expect the combinations of 11 1 with the 9 1 /29 2 FR pair at excitation wavenumbers of around 1205 cm -1 and 1211 cm -1 . There is clear FC activity for both 91111 (1295 cm -1 ) and 111292 (1302 cm -1 ) following 6 1 excitation, so that the emission wavenumber is secure. We thus assign the weak band at (1204, 1302) cm -1 as (9 1 11 1 , 111292), in line with the expected greater intensity of 111292 emission when exciting via both FR components, and the very weak (1204, 1296) cm -1 band to (9 1 11 1 , 91111).
Surprisingly, we only see the faintest traces of the partner (11 1 29 2 , 111292) band, expected at around (1211, 1302) cm -1 and no clear evidence for 11 1 14 4 activity was seen. The ZEKE activity (Figure 3) shows slightly more convincing evidence of the activity of both of these S1 levels, although there was no ZEKE spectrum recorded following excitation at 1211 cm -1 .
Finally, a weak band at (1198, 1528) cm -1 is a reasonably good match for (11 1 15 1 20 1 , 111151201), expected at (1203, 1524) cm -1 -see Figure 6; however, it is not possible to identify the corresponding + 11 1 15 1 20 1 ZEKE band definitively, owing to a number of transitions being expected at this wavenumber in the spectrum -see Figure 3; on the other hand, the FC-allowed + 15 1 20 1 band is seen, supporting the activity of the 11 1 15 1 20 1 level.
In summary, the main expected combinations between the ~400 cm -1 and 800 cm -1 set of lines do seem to appear in this wavenumber region, and although they were sometimes weak in the 2D-LIF spectrum, they were often more apparent in the ZEKE spectrum. More clear in the 2D-LIF spectrum is the 9 1 29 1 /29 3 FR interaction (Figure 8), while the separate bands are less easy to discern in the ZEKE spectra. Part of the explanation for these intensity variations may be different Franck-Condon factors (FCFs) for the ionization and emission processes.

1 )
Higher in excitation wavenumber, at ~ 1215 cm -1 , we expect to find the combinations of the excitation bands that appear at ~818 cm -1 , 12 1 14 1 and 11 2 , with each of 29 1 and 14 2 , at excitation wavenumbers of ~1215 cm -1 . However, no clear evidence of bands corresponding to these levels could be found.
Slightly higher than this, at 1226 cm -1 , we would expect 11 1 12 1 14 1 and 11 3 . We see an unexpectedly bright (compared to the corresponding FC-active band via 6 1 ) emission at (1230, 1816) cm -1 , i.e. essentially coincident with excitation at 5 1 , which we assign to 111121141 -see Figure 6. Because of this unexpected brightness, we assign this band as the (11 1 12 1 14 1 , 111121141) v = 0 band, which is coincident, but apparently not interacting, with 5 1 . Note that in the cation, the expected wavenumber of + 5 1 11 1 (1772 cm -1 ) is almost identical to + 11 1 12 1 14 1 (1775 cm -1 ) and so both could be contributing to the ZEKE band seen at 1781 cm -1 , when exciting at 1232 cm -1 -see Figure 3. Weak 113 activity is also seen when exciting close to 5 1 , but such activity is also seen via 6 1 , and so these may simply be attributed to FC activity.

Other v = 0 bands
In Figure 9 we show another expanded view of the 2D-LIF spectrum where other v = 0 bands are located. Quite an intense band at (1208, 1654) cm -1 is assigned as (14 1 16 1 29 1 , 141161291), which is at the expected wavenumber in both S0 and S1; we would thus expect a ZEKE band at 1608 cm -1 when exciting at 1208 cm -1 , and indeed a band is seen at 1607 cm -1 -see Figure 3. We are comfortable with this assignment, since the wavenumbers of this combination are consistent across three electronic states and we observe that its three components are all active in the spectrum in different guises.
The assignments above yield a wavenumber for D16 in the S1 state of ~610 cm -1 . Although this value is somewhat different to the calculated value 23  To higher excitation wavenumber in Figure 9, there is a band at (1245, 1900) cm -1 , which is consistent with being the (12 2 , 122) band; the 12 2 band was also active in pDFB. 48 We also see in the ZEKE spectrum in Figure 3, that when exciting at 1248 cm -1 , a strong band was seen at 1977 cm -1 assignable to + 12 2 . Interestingly, there is significant, overlapped underlying structure here, suggesting widespread coupling, which will be commented on in subsection 3.4. It is also interesting to see the + 12 1 band in this ZEKE spectrum, which is reminiscent of the activity of + 14 1 in other ZEKE spectra, despite being non-totally symmetric. Activity in such bands was discussed in relation to pDFB in Ref. 48.
We also saw very weak 2D-LIF bands (not shown), plausibly assignable as combinations of 182 with 111 and 291, suggesting that the activity of the corresponding S1 levels is low.

Vibtor levels
In examining the emission when exciting from 5 1 , a band is seen at 518 cm -1 , which is the same band seen in our earlier study 24 of the S1 levels at ~845 cm -1 and assigned to 181m2, i.e. a vibtor level of e symmetry -see Figure 10. This band was a signature that combinations of the 18 1 m 2 vibtor level were coupled to 18 2 m 1 in the S1 state, also of e symmetry. In the present case, 9 1 18 1 m 2 would be expected at ~1238 cm -1 and so very close to 5 1 ; we thus conclude that 9 1 18 1 m 2 is coupled to 5 1 m 1 , both being of e symmetry. There is a weak band that is tentatively assigned to 91181m2 at around 1358 cm -1 ( Figure   10), with the weakness being explained as being due to this level being a doorway state and so coupled to many other levels, explaining the rich emission coming from a range of different so-formed eigenstates: the 181m2 emission band is the resultant sum of FC-active emissions from all of these.
There is also clear weak background emission when exciting at 1246 cm -1 , and some clue as to its origin comes from the observation of a 2D-LIF band at (1246, 476) cm -1 -see Figure 10. The emission can be assigned to 291m3(+) and so it is then straightforward to deduce that the coupled level in the S1 state is likely to be 9 1 29 1 m 3(+) , which is expected to be close in wavenumber to 12 2 and so may couple to its m = 0 level. Indeed, the extent of the 122 emission band suggests coupling between these levels. A weak band at (1245, 1325) cm -1 (not shown) is the likely 9 1 29 1 m 3(+) (v, m) = 0 band.
In discussing the levels at ~400 cm -1 , 26 we identified 20 2 m 6(+) vibtor activity, and the 9 1 20 2 m 6(+) level would lie very close in energy to 6 1 ; this would couple to only the m = 0 level. Further evidence for this is provided by some weak emission bands observed when exciting at wavenumbers corresponding to 6 1 : 204 cm -1 (m6(+)), 350 cm -1 (201m6(+)) and 492 cm -1 (202m6(+)) -see Figure 10. The (v, m) = 0 band would be expected at around 1332 cm -1 and there are bands close to this value, and although it was not possible to associate one definitively with this transition, significant coupling would explain this being weak. We also see weak 291m3(+) and m3(+) bands, which may be associated with another vibtor level coupling -a plausible candidate is 17 1  We also did look for activity associated with combination bands associated with emission or ionization from combination bands involving 14 1 m 6(-) , since this transition is observed at ~400 cm -1 (Ref. 26), but could not unambiguously identify any such.
In the following section we shall discuss the above vibtor levels further, and their role in promoting IVR.

Discussion
Referring to Figure 1, in previous work on the S1 state, evidence for Fermi resonances and coupling to vibtor levels has been found in the low wavenumber region (< 400 cm -1 ), 21,22 the bands at ~ 400 cm -1 , 21 the bands at ~800 cm -1 , 23,47 and the bands at ~845 cm -1 . 24 However, it is the bands at ~1200 cm -1 , as studied in the present work, where evidence for widespread coupling is first apparent. 18,19,20,35 The evidence from previous work is somewhat confusing, with uncertainties and apparent contradictions in the deductions. We thus briefly discuss that work, while rationalizing it in terms of the present data.
We first note that, as commented on in Refs. 19 and 20, some of the time-domain experiments by Parmenter's group 1,2,6 have some inconsistencies as to the cited wavenumber employed in those studies, and also the designation of the vibration being excited. We also recall that D6 is predominantly the C-CH3 stretch, while D5 is the C-F stretch. 41 As such, with the large amplitude motion (the CH3 torsion) being the likely perpetrator of any acceleration, the expectation would be 51,52 that the D6 mode would be subject to more-widespread coupling than D5, and this can clearly be seen not to be the case from the spectra and assignments discussed above. Furthermore, the ZEKE (see Figure 3 6 , 14 5 m 6(-) , 9 1 14 1 m 6(-) and 14 5 m 6(-) ; in addition, in the SEVI spectrum, unresolved from the main + 6 1 band are likely to be + 9 1 29 1 , + 9 1 11 1 , + 29 3 and a number of others. Thus, the timedependent SEVI signal that was monitored may well comprise a number of different contributions.
That said, the fact that the intensity dropped to 50% after a few hundred ps, would be consistent with Ref. 20 is consistent with a ZOB state coupling to a wide range of levels, and similar broad underlying structure is also evident in the ZEKE spectrum in Figure 3.
We have noted above that the 9 1 20 2 m 6(+) and 17 1 19 1 29 1 m 3(+) levels would lie very close in energy to 6 1 , and that these would couple to the m = 0 level only. Also, once such vibration-torsional coupling has occurred, further interactions with other levels open up and this, together with couplings involving other ZOSs in this region, would explain the underlying structure seen in the ZEKE and DF spectra in Figures 3 and 5. That said, it does not seem that the ZEKE and 2D-LIF/DF spectra recorded when exciting 6 1 are consistent with a 50% loss in intensity, suggesting that there may be another mechanism causing the time-dependent SEVI signal 20 to decay. One such would be rotational dephasing 53 occurring between the 6 1 m 0 rotational levels, caused by an m-and J,K-dependent vibtor interaction. 27,47 Such dephasing in pFT was put forward with regards to the two main levels at ~800 cm -1 by Davies and Reid, 47 and commented on further in our recent study of the same levels. 27 As noted above, the suggested vibtor interactions would be with just the 6 1 m 0 level, and so could cause the rotation-dependent coupling required for dephasing. If the dephasing were complete in a few 100 ps (as was in the case of the levels at ~800 cm -1 ) 47 then this would be an explanation for the decay in the 6 1 signal in Refs. 20 and 35. As discussed in Ref. 27, the rotational dephasing does not lead to a population change, and so in the frequency-resolved experiments no effect is seen from this; it only arises from the simultaneous excitation of the perturbed rotational levels. This explanation would then explain the very small unstructured background seen in the present work, and the very significant (~50%) drop in the " + 6 1 " SEVI signal. Given that the amount of unstructured background appears to be comparable in those SEVI spectra 20,35 as in the present ZEKE and 2D-LIF/DF spectra, we conclude the two sets of results are consistent within this explanation.
For the 5 1 level, the situation is somewhat different, with only ~10% of the + 5 1 SEVI signal remaining after 100 ps. 35 We have seen that we observe the 181m2 emission band when exciting at the wavenumber of 5 1 (Figure 10) and from this we conclude that the 9 1 18 1 m 2 level is coupled to 5 1 m 1 in a similar way that 11 1 18 1 m 2 and 14 2 18 1 m 2 are coupled to 18 2 m 1 (Ref. 24). Assuming the coupling to 9 1 18 1 m 2 opens up efficient routes for second-order coupling of 5 1 m 1 to a significant set of bath states, then this could be a rationale for loss of 50% of the + 5 1 SEVI signal in Ref. 35. Although there are a wealth of bands seen in the 2D-LIF/DF spectra at the wavenumber of 5 1 (for example, see Figures 5,6 and 10), it is difficult to assign them all uniquely, but there are various emission bands that can be assigned as analogous FC-active bands, but now in combination with 181m2, seen via 9 1 , but now active close to the 5 1 position (Figure 10). In contrast to the ZEKE spectra in Ref. 24, we did not observe the + 18 1 m 2 ZEKE band in the present work, but the UV intensity of the ionization laser was poor in the spectral region where this band is expected; for the same reason, other torsional levels were also not always seen in the ZEKE spectra that were seen here in emission. Overall, though, on the basis of earlier work and the 2D-LIF results, we are confident that the 5 1 m 1 level is strongly coupled to a background bath of states, with 9 1 18 1 m 2 acting as a doorway state. With this in mind, looking back at the ZEKE and 2D-LIF/DF spectra recorded via 5 1 we see that there is still significant intensity in the + 5 1 and 51 bands, respectively; indeed, the amount is consistent with the majority of the 5 1 m 0 intensity remaining. We thus conclude that, similar to the 6 1 m 1 case above, in the time-resolved experiments, 35 rotational dephasing causes the loss of essentially all of the remaining + 5 1 m 0 SEVI signal, with a possible interacting level being 11 1 12 1 m 6(-) . In our ZEKE study of the ~800 cm -1 bands 23 we assigned a ZEKE band to + 12 1 m 6(-) , but reassigned this on the basis of our recently-reported 2D-LIF spectrum. 27 A large aspect of this reassignment was that we would expect the 12 1 m 6(-) and 12 1 14 1 levels to be close in wavenumber, in the same way that the 14 2 and 14 1 m 6(-) levels are close in wavenumber, 21,22,26 as are 14 1 and m 6(-) . 22 Hence, here we would expect the 11 1 12 1 14 1 and 11 1 12 1 m 6(-) levels to be close together, with the 11 1 12 1 14 1 level already having been deduced to be almost coincident with 5 1 (subsection 3.3.2). What then of the remaining SEVI signal seen when exciting at ~1230 cm -1 ? Again, we note that the resolution of the SEVI experiment is not sufficient to resolve all contributions, and so we suspect that there are underlying FC-active contributions to this SEVI signal that are independent of + 5 1 , with + 11 3 and + 11 1 12 1 14 1 being possibilities at this wavenumber. Hence, we conclude that the residual signal of the decay of the " + 5 1 " SEVI signal (~10% remaining) may be assigned to activity arising from uncoupled levels.
In summary, the 5 1 SEVI signal is lost by two mechanisms, with the + 5 1 m 1 intensity being rapidly lost mainly via coupling to a bath of levels (dissipative IVR) -seen in both frequency-and time-resolved studies, while the + 5 1 m 0 intensity is lost to rotational dephasing in the time-resolved experiments (the + 5 1 m 1 signal could also be partially affected by rotational dephasing). The IVR causes the unstructured background in the SEVI, ZEKE and 2D-LIF/DF spectra, which arises from activity originating from a large number of bath states; the dephasing only occurs as a time-dependent interference effect, and consequently means that half the intensity is still present, as the + 5 1 m 0 or 51m0 ZEKE or 2D-LIF/DF signals, respectively. The residual signal in the time-resolved SEVI spectrum when exciting via 5 1 is concluded to arise from FC activity from overlapped, but uncoupled, S1 levels.
Consistent with comments above, we ruled out an assignment of the ZEKE band at 1606 cm -1 , seen when exciting at 1208 cm -1 , to + 11 1 12 1 m 6(-) , since the excitation position is 25 cm -1 away from the 11 1 12 1 14 1 transition, which has been deduced to be almost coincident with 5 1 (subsection 3.3.2).
Instead the 1606 cm -1 band has been assigned to + 14 1 16 1 29 1 on the basis that this fits the observed bands in the S0, S1 and D0 + states. Close to the 14 1 16 1 29 1 band is the 16 2 band (Figure 9), whose assignment is secure (despite the lack of a ZEKE spectrum) on the basis of the excellent agreement with the expected position of the 162 band; furthermore, this is also in agreement with the observation of the 16 2 band in pDFB. 48 We can see a weak band that corresponds to (14 1 16 1 29 1 , 162), but do not see the partner (16 2 , 141161291) band, from which we conclude that it is unlikely these levels are in FR, but that 162 is FC active via 14 1 16 1 29 1 . We note that there is some FC activity of these when exciting 5 1 and, most clearly, 6 1 .
Lastly, we discuss the excitation region close to ~1250 cm -1 where the 12 2 band is located. We have noted the observation of the 291m3(+) band, and have associated this with 9 1 29 1 m 3(+) . We conclude this level is likely interacting with the 12 2 m 0 level, since 12 2 appears to be the only bright level in this region (and is a transition also seen for pDFB). 48 Although we do not see the 181m2 band, it is likely that there are some interactions with vibtor combinations involving this level, as we see evidence of weak emissions from 18 2 29 1 and 11 1 18 2 levels (not shown), and our earlier work 24 has indicated interactions between 18 2 m 1 levels and levels such as 11 1 18 1 m 2 . However, the interaction with the 29 1 m 3(+) combination levels is the most likely explanation of the unstructured background that is present in the ZEKE ( Figure 3) and 2D-LIF/DF (Figures 2 and 5) spectra when exciting at 1248 cm -1 . In Figures 3 and 8, we see that there are still substantial + 12 2 and 122 ZEKE and 2D-LIF signals in the respective spectra, attributable to the 12 2 m 1 level.
In Table 2, we present the number of levels of different types that have the correct symmetry and lie close in wavenumber to the 6 1 and 5 1 levels, while in Figure 11 we show the rise in the number of S1 vibrational and vibtor levels with increasing internal energy. The stark difference between the coupling of 6 1 and 5 1 cannot be straightforwardly related to the density of vibrational states since, as Table 2 and Figure 11 indicate, there are only a handful of extra totally-symmetric levels for the latter.
There are more vibtor levels, with ~25% more levels that have the correct symmetry (a and e) that could interact, but this is unlikely to explain the marked difference in behaviour between these two close-lying fundamentals. As such, and as discussed in Ref. 25, there is a certain amount of serendipity required, in that not only must there be levels of the correct symmetry to interact, but the coupling must also be efficient -with at least one doorway state.
We concluded in Ref. 25 that such coupling is aided by torsional motion, which allows out-of-plane (and so non-totally-symmetric) vibrations to become coupled to totally-symmetric vibrations, which will be FC-active. This idea was explored in Ref. 35 54 , by modelling the van der Waals interaction between the methyl group and the closest phenyl hydrogens. It was found that, on the basis of the calculated vibrational motion, the D5 vibration produced a larger interaction than did D6. This is in apparent contrast to the localized motions, which are C-F stretch for D5 and C-CH3 stretch for D6, with the latter then being expected to cause more interaction. 55,56 The explanation lies in the relative motions of the in-plane wags of the two closest phenyl C-H bonds, and the CH3 stretch. Referring to Figure 11, these may be seen to be in-phase in the case of D6, so that the methyl group moves away from the phenyl ring, in sync with the two neighbouring C-H bonds, and vice versa, so that the interaction between the sets of hydrogen atoms remains very similar; in contrast, since the C-CH3 bond remains largely stationary in the D5 motion ( Figure 11), then the same C-H wagging motions cause a greater interaction than in D6. This shows that there is vibrational mode specificity in determining the efficiency of IVR, as concluded by Davies et al. 35 The analysis here emphasises that the whole vibrational motion should be considered when examining this, not just looking at localized motions. Also, Figure 11 shows that the D18 mode has out-of-plane "wagging" of the two C-H bonds closest to the methyl group, suggestive of their being able to "waft" the methyl group (see also Ref. 24). Also, the motion of D29 indicates significant methyl motion, causing it to come into closer contact with a phenyl C-H bond, causing interaction. Hence, both vibrations may reasonably be expected to cause motion of the methyl group.

Concluding Remarks
In this work, we have presented further explicit evidence that vibration-torsion interactions are a key factor in facilitating IVR in molecules that contain a methyl group. Further, we have shown that there is a certain amount of serendipity involved, particularly at moderately low internal energy, in doorway states being present close to a ZOB state and which can enable efficient coupling to the background bath states, whose density is increasing rapidly with increasing internal energy. The efficiency of a doorway state in effecting IVR is dependent on having the correct symmetry, being energetically close to the ZOB state, and being well-coupled both to the ZOB state and a significant number of the bath states -this is likely to be linked to its motion and how this interacts with other motions in the molecule, particularly the methyl group. In this regard, we have suggested that the whole vibrational motion should be considered, and not simply local motions.
The above conclusions have been obtained from both fluorescence and ZEKE spectroscopies, with consistent activity in both producing the most reliable results. The 2D-LIF spectra, in particular, have aided in the identification of overlapped activity and particularly allowing the extraction of the vibtor activity. We have been further aided by earlier analyses 26,27 of the lower wavenumber regions that underpin much of the energy level structure in the present wavenumber region. The detailed knowledge of the wavenumbers of the vibrational and vibtor levels in the S0, S1 and D0 + states (present work, Ref. 23 and work cited therein) provides great insight into the couplings that occur. Notably, the greater detail of the present study has allowed further insight into the results of time-resolved studies.
It is clear that many of these ideas will be more generally applicable, with coupling between chromophore-localized vibrations and wide-amplitude motions that will provide routes for IVR in a range of molecules, with one key set being biomolecules which often contain NH2 and CH3 groups, as well as other flexible side chains.   All vibtors 426 585 a Numbers of S1 levels calculated within a ±20 cm -1 window of each fundamental, based upon calculated harmonic vibrational wavenumbers 23 and approximate energies of the torsional levels. It is clear that both anharmonicity and vibration-torsional coupling will produce minor changes in these numbers, but are not expected to be a large.
b Symmetry-allowed coupling to m = 0 levels of totally-symmetric vibrations.
c Symmetry-allowed coupling to m = 1 levels of totally-symmetric vibrations.      Selected assignments are given, with a number of these discussed in the text. Figure 7: Indications of the expected S1  S0 activity arising from combinations of features at ~400 cm -1 and ~800 cm -1 . The former region of the spectrum is shown offset by amounts corresponding to the three main bands of the latter region. The approximate positions of the various vibrational combination bands is given, although not all are observed; further movement of levels as a result of interactions is expected to occur -see text for further discussion.    windows is plotted as a function of the S1 state internal energy. The positions of 6 1 and 5 1 are indicated. Below this, the mode diagrams of the D6 and D5 fundamentals are shown, as are those for D18 and D29. In the latter two cases, it can be seen how the motion can cause the vibrations to interact with the torsional motion of the methyl group.