The vacuum-ultraviolet photoelectron spectra of CH2F2 and CH2Cl2 revisited

• Users may freely distribute the URL that is used to identify this publication. • Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of private study or non-commercial research. • User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?) • Users may not further distribute the material nor use it for the purposes of commercial gain.

The vacuum-ultraviolet photoelectron spectra of CH 2 F 2 and CH 2  The threshold photoelectron spectrum (TPES) of difluoromethane and dichloromethane has been recorded at the Swiss Light Source with a resolution of 2 meV or 16 cm À1 . Electronic and vibronic transitions are simulated and assigned with the help of Franck-Condon (FC) calculations based on coupled cluster electronic structure calculations for the equilibrium geometries and harmonic vibrational frequencies of the neutrals, and of the ground and excited electronic states of the cations. Notwithstanding a high-resolution pulsed-field ionisation study on CH 2 F 2  in which a number of transitions to the e X þ state have been recorded with unprecedented accuracy, we report the first complete vibrationally resolved overview of the low-lying electronic states of CH 2 X 2 + , X = F or Cl. Hydrogen atom loss from CH 2 F 2 + occurs at low energy, making the ground state rather anharmonic and interpretation of the e X þ band challenging in the harmonic approximation. By Franck-Condon fits, the adiabatic ionisation energies to the e A þ 2 B 2 , e C þ 2 A 2 and e D þ 2 B 2 states have been determined as 14.3 ± 0.1, 15.57 ± 0.01 and 18.0 ± 0.1 eV, respectively. The first band in the CH 2 Cl 2 TPES is complex for a different reason, as it is the result of two overlapping ionic states, e X þ 2 B 2 and e A þ 2 B 1 , with derived ionisation energies of 11.0 ± 0.2 and 11.317 ± 0.006 eV, and dominated by an extended progression in the CCl 2 bend (in e X þ ) and a short progression in the CCl 2 symmetric stretch (in e A þ ), respectively. Furthermore, even though Koopmans' approximation holds for the vertical ionisations, the e X þ state of CH 2 Cl 2 + is stabilized by geometry relaxation and corresponds to ionisation from the (HOMOÀ1) orbital.
That is, the first two vertical ionisation energies are in the same order as the negative of the orbital energies of the highest occupied orbitals, but the adiabatic ionisation energy corresponding to electron removal from the (HOMOÀ1) is lower than the adiabatic ionisation energy corresponding to electron removal from the HOMO. The second band in the spectrum could be analysed to identify the vibrational progressions and determine adiabatic ionisation energies of 12.15 and 12.25 eV for the e B þ 2 A 1 and e C þ 2 A 2 states. A comparison of the assignment of electronic states with the literature is made difficult by the fact that the B 1 and B 2 irreducible representations in C 2v symmetry depend on the principal plane, i.e. whether the CX 2 moiety is in the xz or the yz plane, which is often undefined in older papers.

Introduction
Gas-phase vacuum ultraviolet photoelectron spectroscopy, established in the late 1960s, is now a mature technique for determining ionisation energies of molecules and studying their electronic structure. Koopmans' theorem establishes a direct connection between the ionisation energies and the orbital energies in the neutral [1]. The resolution of the technique has improved by orders of magnitude for select systems [2,3], in which pulsed field ionisation with small bandwidth lasers is possible and the spectra can be rovibrationally resolved. Threshold photoelectron detection using velocity map imaging, the approach pursued here [4], allows us to study larger molecules routinely with a few meV, i.e. vibrational resolution. In tandem with experimental improvements, transitions can now often be assigned with the aid of ab initio calculations, in particular with Franck-Condon (FC) simulations of the vibrational progressions in each cationic electronic state. These advances have led to surprises in the interpretation of the vacuum-UV photoelectron spectra of difluoromethane (CH 2 F 2 ) and dichloromethane (CH 2 Cl 2 ), especially in their ground-state bands, but for different reasons. of this paper is to track the history of these spectra since the early days of photoelectron spectroscopy. We show how improvements in both experimental techniques, primarily in the resolution of photon sources and modern electron analysers, and ab initio theoretical methods have led to a deeper understanding of these deceptively simple-looking spectra. Certainly for CH 2 F 2 , the apparent clarity of a single progression in the e X þ band, as first observed over forty years ago by several groups [5][6][7], hides a complex overlap of different vibrational modes in which the effects of anharmonicity are pronounced.
A further and unexpected complication was found to be the assignment of electronic states. Computational chemistry programs rotate molecules in a ''standard orientation''. This is a deceiving term, since the principal plane, i.e. the one spanned by the three heavy atoms in the molecule, is sometimes defined as the xz, sometimes as the yz plane in widely-used computer programs. If the molecule is rotated from one orientation to the other, electronic and vibronic states belonging to the irreducible representations B 1 and B 2 are swapped. While it is conceptually easy to find the irreducible representation of a vibrational mode with a description in any convention, molecular orbitals (MOs) are rarely described in detail. Additionally, some of the MO energies, particularly that of the highest occupied MO (HOMO) and the (HOMOÀ1) orbital in CH 2 Cl 2 , are closely spaced and their ordering can be method dependent. Thus, there is no unambiguous way to identify the orientation of the molecule unless it is explicitly given. Herein, we use yz as the principal plane, as this is how most computational works have been carried out.
The geometries of neutral CH 2 F 2 and CH 2 Cl 2 are well established by microwave spectroscopy and related techniques [8,9]; equilibrium values are quoted in Tables 1 and 2 which are invariant to the isotopologue under consideration. Both molecules have nine vibrational modes. Four belong to the a 1 irreducible representation (m 1 CH 2 symmetric stretch, m 2 CH 2 bend, m 3 CX 2 symmetric stretch, m 4 CX 2 bend), one to a 2 (m 5 CH 2 twist), two to b 1 (m 6 CH 2 asymmetric stretch, m 7 CH 2 rock), and two to b 2 (m 8 CH 2 wag, m 9 CX 2 asymmetric stretch). The Mulliken convention is used: the molecule's vibrational modes of the same symmetry are labelled so that the highest-energy vibration belonging to the a 1 irreducible representation is m 1 , the second highest is m 2 , etc. Two a 1 symmetry modes in CH 2 F 2 , the CH 2 bend and CF 2 symmetric stretch, as well as the two b 2 modes, the CH 2 wag and CF 2 asymmetric stretch, have a different ordering in the e X þ electronic ground state of the cation. For consistency, and in order to have a compact way of denoting sequence and hot bands, we maintain the above labelling unchanged for CH 2 F 2 + . Thus the m 2 + mode of e X þ CH 2 F 2 + continues to describe the CH 2 bending mode, even though the m 3 + mode describing the CF 2 symmetric stretch has a higher frequency. Likewise, the eighth mode of the ground-state cation is used to describe the CH 2 wag and the ninth mode the CF 2 asymmetric stretch, even though m 8 + < m 9 + . The energetic reversal of these two pairs of vibrational modes arises essentially because of the significant change in geometry and, hence, in the force constants upon ionisation in CH 2 F 2 , and has caused confusion and misunderstandings in previous papers. Following high-resolution Fourier transform infrared (IR) measurements by Smith et al. [10,11], Deo et al. [12,13], and Cronin et al. [14] including decoupling of the multiple Coriolis interactions between m 2 and m 8 , m 3 and m 9 , m 5 and m 7 , and m 6 and 2m 2, the experimental values of the nine vibrational fundamentals in CH 2 F 2 are well established and listed in Table 1. It is, thus, not surprising that the first photoelectron band corresponding to e X þ CH 2 F 2 + is complicated on spectroscopic grounds alone because Coriolis interactions are prevalent in the neutral, making formally-forbidden IR transitions, such as m 5 (a 2 ), observable. In this context, it would be surprising if only a simple progression of totally symmetric a 1 vibrations were observed in the photoelectron spectrum. Numerous geometry optimizations and normal mode analyses have been reported for both CH 2 F 2 and its parent cation, with improving accuracy with increased basis set size and more rigorous description of electron correlation up to coupled cluster CCSD(T) theory [15][16][17][18]. However, most calculations on the neutral molecule have been made in the harmonic approximation with no interaction between the nine normal modes. The recent paper of Luckhaus et al. [19] allows for fully coupled anharmonic effects and Coriolis or Fermi interactions between close-lying vibrational levels, and the results are impressive; for the neutral molecule most calculated vibrational transitions lie within 2 cm À1 of the experimental data, with a maximum deviation of only 3 cm À1 . Data for the harmonic vibrational frequencies of the ground state of CH 2 F 2 + have been reported at different levels of theory by Takeshita and Forysinski et al. [15,18]. The most significant points to note from both studies are that the a 1 vibration describing the CF 2 symmetric stretch vibration (which we label m 3 + ) is of higher energy than the CH 2 bend (m 2 + ), and the eigenvalue for the CF 2 asymmetric stretch (m 9 + of b 2 symmetry) exceeds that of the CH 2 wag (m 8 + in our notation  [19]. Very few of the lowest-lying vibrational levels can be described as 'pure,' and the effects of anharmonicity are pronounced even at modest levels of vibrational excitation above the zero-point energy level. The situation is complicated, or perhaps partially explained, by the fact that the lowest dissociation threshold to CHF 2 + + H lies only 2740 cm À1 above the v + = 0 level [18,20], so the electronic potential of e X þ CH 2 F 2 + only supports few bound states. This energy is also close to the FC maximum of the ground state of CH 2 F 2 + , and significant anharmonic effects are therefore expected in the excitation of any of the nine vibrational fundamental modes. It is therefore predicted that attempts to analyse higher-resolution photoelectron spectra of the first band of CH 2 F 2 will encounter problems if the conventional, harmonic approach is taken. This has already been appreciated by Forysinski et al. in their attempt to analyse the pulsed field ionisation zero kinetic energy (PFI-ZEKE) photoelectron spectrum of CH 2 F 2 recorded at a superb, 1 cm À1 resolution [18]. As will be seen later in the FC simulations, anharmonic effects are by far the most pronounced in the vibrational structure of the ground state of CH 2 F 2 + . In fact, even in CH 2 F 2 + , the vibrational transitions are not defined or resolved well enough to warrant an anharmonic analysis at our spectral resolution. Instead, we analyse all vibrational fine structure in the harmonic approximation but refer the reader to Luckhaus et al. [19] for a more accurate analysis of the vibrational states in e X þ CH 2 F 2 + .
Coriolis interactions are, of course, also prevalent in the IR and Raman spectra of neutral dichloromethane [21][22][23][24][25], and it would therefore be surprising if such effects were not observed in the parent cation. However, the significant difference from difluoromethane is that the first dissociative ionisation threshold from CH 2 Cl 2 corresponds not to H-atom loss, but to Cl-atom loss and at a much higher energy above the adiabatic ionisation energy (IE). In the so far highest-resolution (ca. 0.002 eV or 16 cm À1 ) study of this process, we determined the dissociation threshold to CH 2 Cl + + Cl to be 12.108 ± 0.003 eV [20], which is 0.791 eV (or 6380 cm À1 ) above the experimental adiabatic IE of the e A þ 2 B 1 state of CH 2 Cl 2 + . Due to the large geometry relaxation, the origin of the (7b 2 ) À1 ionisation to the e X þ 2 B 2 state of CH 2 Cl 2 + lies even lower. Not only are these bond dissociation energies significantly larger than their equivalent in CH 2 F 2 + , but they also mean that the first dissociative ionisation threshold lies well above the FC envelope of e X þ CH 2 Cl 2 + . The anharmonic effects on the vibrational fine structure in the first photoelectron band of CH 2 Cl 2 are therefore less pronounced. The experimental and calculated geometries of CH 2 Cl 2 and its cation are shown in Table 2 [9,26,27]. Perhaps the most significant point to appreciate is that the origin transition in unlikely to be observed for the e X þ state because of the large decrease in ClCCl bond angle of 22°; the C-H and C-Cl bond lengths are virtually unchanged and the HCH bond angle increases by only 5°. Thus, to a first approximation, one expects the greatest FC activity in the e X þ photoelectron band of CH 2 Cl 2 to be seen in the m 4 CCl 2 bending mode, not in the m 2 CH 2 bending mode as in the corresponding e X þ 2 B 1 state of CH 2 F 2 + where the HCH bond angle is calculated to decrease significantly upon ionisation by 26°.
In this paper, we report the valence vacuum-UV threshold photoelectron spectrum of CH 2 F 2 and CH 2 Cl 2 recorded at the 3 rd generation Swiss Synchrotron Light Source, SLS, from their ionisation onset, 11-12 eV, to 20 and 16 eV, respectively. The resolution of the spectra is ca. 0.002 eV or 16 cm À1 . Photoions were also recorded in coincidence, and the results of this dynamical fragmentation study and of earlier studies using a 2 nd generation synchrotron are published elsewhere [20,28,29]. We analyse the threshold photoelectron spectra in the light of new ab initio calculations and Franck-Condon simulations, and reflect on how the interpretation of these apparently simple spectra has evolved over the years.

Experimental and theoretical methods
The imaging photoelectron photoion coincidence (iPEPICO) spectrometer at the X04DB vacuum ultraviolet beamline of the SLS has been described in detail elsewhere [30,31]. Pure sample is introduced into the chamber through an effusive source at room temperature. Typical pressure in the experimental chamber was 2-4 Â 10 À6 mbar, against a background pressure of 1 Â 10 À7 mbar. The sample is ionised by monochromatic VUV radiation dispersed by a grazing incidence monochromator. Two gratings are available, interchangeable under vacuum, with maximum output at 12 eV (600 lines mm À1 ) and 20 eV (1200 lines mm À1 ). The optimum photon resolution is ca. 0.002 eV. The photon energy is calibrated against autoionisation lines of argon and neon in first and second order. Higher orders of radiation are removed using a compact gas filter with an absorption path length of 10 cm and operating at 10 mbar of Ne. Table 1 Experimental and ab initio vibrational frequencies, and geometries of CH 2 F 2 . Vibrational frequencies in cm À1 , bond lengths in Å, bond angles in degrees (°).

Mode
Description  Table 2 Experimental and ab initio vibrational frequencies, and geometries of CH 2 Cl 2 . Vibrational frequencies in cm À1 , bond lengths in Å, bond angles in degrees (°).

Mode
Description Photoelectrons are extracted by a 120 V cm À1 continuous field and velocity map imaged onto a position sensitive delay line anode (Roentdek DLD40) with a kinetic energy resolution of 1 meV at threshold. 'Hot' electrons with non-zero kinetic energy can have a velocity vector that is oriented along the flight tube axis. If so, they also arrive at the centre of the detector along with the threshold electrons, and can be removed by a subtraction technique [32]. The resulting spectra are flux normalised using a sodium salicylate coated pyrex window with a visible photomultiplier tube. We have already established that the TPES peak positions of a range of fluoroethene molecules were not Stark shifted measurably in an extraction field as high as 120 V cm À1 [33], and assume the same applies to the molecules studied herein. However, we note that peak positions may be shifted to lower energy by as much as 8 meV, in the unlikely event that this assumption is incorrect [34]. The uncertainties for both the adiabatic and vertical IEs were determined from the half width at half maximum of a Gaussian function fitted to the experimental spectrum when the corresponding origin transition was detected. When extrapolating to the onset of a progression with a 0-0 vibrational transition of negligible intensity, we estimated the uncertainty in the number of vibrational quanta at the detected transitions and have set the error bar accordingly. CH 2 F 2 (g) and CH 2 Cl 2 (l) samples were obtained from Fluorochem UK (99.9%) and Aldrich Chemical Company, respectively. The former was used without purification, the latter was subject to several freeze-pump-thaw cycles before introduction in the experimental chamber.
The geometry of the neutral ground electronic state of both CH 2 F 2 and CH 2 Cl 2 was optimised and the numerical frequency analysis was carried out employing coupled cluster theory with singles and doubles (CCSD) in the frozen core approximation with the cc-pVTZ basis set using Q-Chem 4.0.1 [35]. For the electronic ground and excited cationic states, we used the equation-ofmotion for ionisation potential (EOM-IP-)CCSD approach with the said basis set. The frequency analyses confirmed the stationary points to be minima unless noted otherwise, and the computed Hessian matrices were used in FC calculations. We note that Q-Chem uses the xz principal plane as opposed to our choice of the yz principal plane in C 2v symmetry. Irreducible representations given for electronic states and normal modes listed herein are correct when the molecule is oriented so that the CX 2 moiety is in the yz plane.
Franck-Condon simulations were performed with the program eZspectrum.OSX [36], based on the optimised geometries and force constants of the contributing states. Unscaled frequencies were used, and the Dushinsky rotation was taken into account. All spectra were simulated at room temperature and the phase space of the fit, defined by the excitation space of both neutrals and ions, was successively increased until the spectrum did not change significantly. Resulting stick spectra were subsequently convoluted with a Gaussian function to match the experimental spectrum. When the photoelectron band is comprised of excitations into two or more different electronic ion states, separate simulations were performed, which were then convoluted, weighted, and summed to fit the observed band system.

Difluoromethane, CH 2 F 2
The threshold photoelectron spectrum of CH 2 F 2 between 12.7 and 20.5 eV is shown in Fig. 1, recorded at a resolution of 0.006 eV, along with calculated adiabatic ionisation energies at the EOM-IP-CCSD/cc-pVTZ level of theory. This spectrum is a composite between two scans recorded with the lower-energy grating (up to 13.7 eV) and the higher-energy grating (above 13.7 eV). A significant overlap between the two scans ensures the validity of the relative intensities of the different photoelectron bands. A new spectrum of the first photoelectron band between 12.7 and 13.8 eV with improved resolution, 0.002 eV, and longer signal averaging is shown in Fig. 2.
The ordering of the outer CH 2 F 2 Hartree-Fock molecular orbitals (MO) using the cc-pVTZ basis set and the yz principal plane is . . .(1b 1 ) 2 (5a 1 ) 2 (3b 2 ) 2 (1a 2 ) 2 (4b 2 ) 2 (6a 1 ) 2 (2b 1 ) 2 , where the orbital numbering includes the C and F 1s core orbitals. The 4b 2 and 6a 1 orbitals lie close in energy and are exchanged in density functional theory results, which shows that there is no unambiguous ordering of the MOs in this molecule [6,15]. Pullen et al. report a 2 B 1 ground ion state [7], whereas Brundle et al. propose it to be of 2 B 2 symmetry [6], which could indicate that they were working in different coordinate systems but identified the correct ground state. However, probably because of the limited computational possibilities at the time, they could not identify the following three states correctly. The HOMO of 2b 1 symmetry has C-H bonding and C-F antibonding character. Both vertical and adiabatic ionisation energy calculations confirm that the lowest lying cationic electronic state is indeed e X þ 2 B 1 , thus corresponding to electron  The first photoelectron band of CH 2 F 2 (ionisation to CH 2 F 2 + e X þ 2 B 1 ) recorded between 12.65 and 13.85 eV with a step size of 0.002 eV and an integration time of 90 s per point. The spectrum has been flux normalised. A Franck-Condon simulation (blue sticks) and convolution with a Gaussian function (FWHM = 18 meV) to account for unresolved rotational structure (red) are also shown. removal from the HOMO, which leads to a lengthening of the C-H and a shortening of the C-F bond (Table 1). In the harmonic approximation, Franck-Condon activity should only be observed in vibrational modes of CH 2 F 2 + (of either fundamental, overtone or combination bands) with a 1 symmetry or with even quanta in non-totally symmetric normal modes. Band positions are tabulated in Table 3, and compared with the recent PFI-ZEKE data of Forysinski et al. [18] and earlier photoelectron studies using fixed-energy lamp sources or low-resolution synchrotron studies [28,37]. At first sight, the appearance of our spectrum looks very similar to the earlier lower-resolution studies, with a progression being observed in a mode with a FC maximum around m i = 3. However, at our resolution of ca. The history of the assignment of this band when recorded with inferior resolution, ca. 0.03 eV, is interesting. From the first photoelectron spectra, Potts et al. [5] assigned the 'single progression' to the CH 2 bending mode and obtained m 2 Brundle et al. [6] agreed with this assignment and obtained m 2 + = 1010 cm À1 , but thought it likely that another overlapping and almost degenerate vibrational progression, presumably of a 1 symmetry, too, was also present, because vibrational structure was lost for CD 2 F 2 ; this indicated that the vibrational structure in CH 2 F 2 did not correspond to a single vibrational mode. Pullen et al. [7] thought the single band could be a complicated overlap of the CF 2 symmetric stretch, the CH 2 twist, the CH 2 rock and the CF 2 asymmetric stretch, although it is not clear whether they appreciated that the last three modes do not have a 1 symmetry and odd-quantum transitions are therefore formally forbidden in photoelectron spectroscopy. None of these early studies from the 1970s had the advantage of modern ab initio packages to aid in the assignment. In 1990, Takeshita first performed geometry and vibrational frequency calculations both on neutral CH 2 F 2 and the first four electronic states of CH 2 F 2 + in C 2v symmetry, i.e. its ground state and the three lowest excited states, using self-consistent field methods and the xz convention for the principal plane [15]. Calculations were made in the harmonic approximation, only a 1 progressions were therefore considered, and Franck-Condon simulations were attempted. He assigned the single progression to In 1993, Pradeep and Shirley also recorded the HeI photoelectron spectrum of cold CH 2 F 2 under supersonic beam conditions and an improved resolution of 0.013 eV, identifying the ionic ground state as 2 B 2 and therefore implying an xz principal plane [37]. They believed they could resolve three independent but overlapping progressions which they assigned to the symmet- are reported using the xz convention [28]. However, the modest resolution of the beamline, ca. 0.015 eV, meant that no new information was obtained, and the single progression observed was assigned to overlapping progressions in the CH 2 stretch, m 2 + , and the CF 2 symmetric stretch, m 3 + , with an average value of 1130 cm À1 . An extra peak between the origin band and m i = 1 was observed at ca. 12.8 eV, but was not assigned. As in the HeI studies, the maximum in the band was also observed at v i % 3.
We have performed independent FC calculations in the harmonic approximation and simulated the vibrational structure of the first photoelectron band of CH 2 F 2 , using optimised geometries and Hessian matrices obtained at the (EOM-IP-)CCSD/ cc-pVTZ level of theory. The geometries and vibrational frequencies calculated for the neutral and ground state of the cation are given in Table 1. The energies of the predicted bands, their relative FC factors and their assignments are given in Table S1 of the Supplementary Material. Fig. 2 shows our experimental spectrum (black 2 ), a stick spectrum of the vibrational bands with significant FC activity (blue), and a convolution of the predicted bands with a Gaussian function with full width at half maximum, FWHM, of 18 meV to account for the unresolved rotational structure (red). The adiabatic IE of the simulated spectrum is set to the same value as our experimental value, 12.727 eV. The assignments of the peaks are given in Table 3. The agreement with experiment is much better than that reported by Takeshita [15] in that the simulation predicts a maximum at v i = 3-4, in agreement with experiment. As expected, the effects of anharmonicity in the experimental spectrum become more pronounced as the energy increases, and there are a number of unassigned peaks near the adiabatic IE. Since these calculations are performed in the harmonic approximation, all assignments only involve vibrational modes in the cation of a 1 symmetry (m 1 + through m 4 + ) or even combinations of non-symmetric vibrations (e.g. 2m 7 + ).
A major change in understanding of this band came in 2010, when Forysinski et al. applied PFI-ZEKE spectroscopy using a vacuum-UV laser to record the spectrum at a resolution of ca. 1 cm À1 , i.e. at least two orders of magnitude better than in previous studies. They reported the band origin for ionisation to e X þ CH 2 F 2 + to be 102 636 ± 7 cm À1 (12.7252 ± 0.0009 eV) [18]. Furthermore, they observed that the 'single' peaks around 1100 and 2200 cm À1 to higher energy were in fact a triplet and, possibly, a quintet of peaks, respectively, with weak structure in between. The spectrum was not recorded between the origin band and the triplet of lines at ca. 1100 cm À1 , or at energies above the dissociative ionisation threshold to CHF 2 + + H of 105 375 ± 25 cm À1 (13.065 ± 0.003 eV).
As seen in Fig. 2, there is in fact significant threshold photoelectron signal in both these ranges. The explanation of the absence of the second set of peaks is most likely that fragmentation of the Rydberg states above this dissociative ionisation threshold is so fast that the bands are broadened considerably and the ZEKE signal falls below the detection limit. The line positions of the bands observed by Forysinski et al. are shown in Table 3. They concluded that their spectrum could not be fully assigned unless anharmonic contributions from all vibrational modes were accounted for. The results of the fully-coupled anharmonic calculations for the ground electronic states of CH 2 F 2 and CH 2 F 2 + were published later in the same year by Luckhaus et al. [19]. Whilst the calculations were relatively easy to converge for the closed-shell e X CH 2 F 2 molecule, there were more difficulties for the open-shell e X þ CH 2 F 2 + species. Nevertheless, they were able to assign most of the strong transitions observed in the PFI-ZEKE spectrum to a 1 vibrations or overtones and combination bands of a 1 symmetry, the energies and relative intensities also being shown 2 For interpretation of color in Figs. 2 and 7, the reader is referred to the web version of this article.   Ref. [18]. b Ref. [19], the numbering of the vibrational modes has been changed so that they conform to our usage. c Ref. [37]. d Ref. [28]. e Assumed value from [18].
in Table 3. Nearly all the observed bands had contributions from several normal mode basis functions, and no band could be described as a 'pure' transition. This is especially true near the dissociation threshold to CHF 2 + + H, 2740 cm À1 above the v + = 0 level of e X þ CH 2 F 2 + . Fundamental band transitions (nominally labelled m 5 + through m 9 + ) of a 2 , b 1 or b 2 symmetry were calculated to have very low FC intensity, and the authors commented that, far from indicating a breakdown of the FC principle when anharmonic effects are considered, the opposite is, perhaps surprisingly, the truth. As shown in Table 3, the peak positions from our threshold photoelectron spectrum recorded at a resolution of ca. 16 cm À1 and those recorded by PFI-ZEKE at a resolution of ca. 1 cm À1 are in excellent agreement, to within 1-3 meV or 8-24 cm À1 , up to the dissociation energy to CHF 2 + + H, 13.06 eV. As noted above, two low-energy bands near the adiabatic IE have not been reported in the PFI-ZEKE spectrum. The band at 12.792 eV, 65 meV or 524 cm À1 above the adiabatic IE, is significantly below the first dissociation threshold, and it therefore seems reasonable to believe that the predominant basis function of this band must be due to a vibration of a 1 symmetry. Our FC simulation shows that indeed the m 4 + = 1 transition is populated, although its vibrational frequency is significantly higher at 608 cm À1 in our harmonic calculations, compared to 597 cm À1 in the anharmonic calculations of Luckhaus et al. [19]. The weak but distinct band at 12.748 eV, 21 meV or 169 cm À1 above the adiabatic IE, is probably due to a sequence band from an excited vibrational level of e X 1 A 1 CH 2 F 2 .
The most likely assignment is m 3 = 1 ? m 3 + = 1 or the 3 1 1 transition, since the frequency of this vibrational mode is predicted to increase by 128 cm À1 upon ionisation in the harmonic calculations and will therefore appear to higher energy of the adiabatic IE. We note, however, that neither of the 3 0 1 and 4 0 1 transitions is observed in the room temperature spectrum in Fig. 2 ca. 1100 and 530 cm À1 to the red of the adiabatic IE, respectively. It is also worth noting that weak Franck-Condon activity is also present in the 4 1 1 band but this transition is lower in energy and probably blended into the 0-0 band. The assignments of bands in the photoelectron spectrum above the CHF 2 + + H dissociation threshold are even more ambiguous.
Whilst the spectrum appears to simplify and becomes a single progression of broader lines, it is now clear from the calculations of Luckhaus et al. that it is probably an oversimplification to assign these lines to a simple progression in m 2 + , m 3 + , m 4 + , or to some combination thereof [19]. For the sake of completeness, the assignments given to these higher-energy lines both by Seccombe et al. and Pradeep et al. remain in Table 2 [28,37], but they have now been shown probably to be incomplete or even incorrect.

Dichloromethane, CH 2 Cl 2
There have been surprisingly few reports on the photoelectron spectrum of CH 2 Cl 2 , either under non-resonant HeI (hm = 21.22 eV)/ HeII (hm = 40.8 eV) conditions or under threshold electron conditions. Spectra have been recorded with HeI radiation by Potts et al. [5] and Pradeep and Shirley [37], the latter in a supersonic beam with a resolution of 0.013 eV. The HeII spectrum was reported by von Niessen et al. [38]. Dixon et al. interpreted the first bands of the spectrum of CH 2 Cl 2 as ionisation from non-bonding p orbitals on the Cl atoms coupling weakly to r-bonding C-H and C-Cl orbitals [39]. To the best of our knowledge there has only been one threshold photoelectron spectrum reported over the complete valence region at a resolution of ca. 0.030 eV by Chim [29], with the ground-state band(s) also being recorded with an improved resolution of 0.011 eV. Calculated symmetries, geometries, and vibrational frequencies of the ground and valence-excited states of CH 2 Cl 2 + have been reported most recently by Takeshita and Xi et al. [26,40], with term symbols and orbital symmetries both in the yz, i.e. our, principal plane convention. Just like CH 2 F 2 , the orbital ordering in dichloromethane is also somewhat method dependent. However, it is quite similar to that of CH 2 F 2 , and the outer orbital ordering of the Hartree-Fock MOs in the cc-pVTZ basis set corresponds to . . .(2b 1 ) 2 (8a 1 ) 2 (6b 2 ) 2 (2a 2 ) 2 (9a 1 ) 2 (7b 2 ) 2 (3b 1 ) 2 , where the orbital numbering includes the C and Cl core orbitals. In the Kohn-Sham orbital ordering, 9a 1 lies below the 2a 2 orbital. The highest-lying b 2 orbital in CH 2 Cl 2 is clearly the (HOMOÀ1). However, the corresponding 2 B 2 cation state undergoes significant geometry relaxation, and becomes the ground cationic state to which the origin transition is not detected because of a negligible FC factor. It is important to note that Koopmans' approximation does hold, and the vertical transition to the 2 B 1 state lies lowest in energy at the neutral geometry (at which the separation of the first two vertical ionisation energies is small, only 0.24 eV), meaning that the first sharp peak in the spectrum corresponds to the origin transition to the e A þ state. That is, the vertical ionisation energy to the e X þ state lies above the vertical energy to the e A þ state, even though the adiabatic ionisation energy to e X þ lies below the adiabatic energy to e A þ . Consequently, in contrast with CH 2 F 2 + where they are well separated, the FC envelopes of the e X þ 2 B 2 and e A þ 2 B 1 bands overlap strongly in CH 2 Cl 2 + . The e B þ 2 A 1 and e C þ 2 A 2 photoelectron bands overlap in both CH 2 F 2 + and CH 2 Cl 2 + , but they are 3 eV higher in energy than the ground-state band in CH 2 F 2 + whilst only 1 eV higher in CH 2 Cl 2 + . When comparing the relative intensities of the bands under the three different recording conditions, the two obvious differences are that the intensity of the e X þ = e A þ overlapped band compared to that of the e B þ = e C þ band of CH 2 Cl 2 is significantly reduced under threshold photoelectron conditions [29]. Second, under HeII photon conditions, the bands corresponding to ionisation from the 6b 2 , 8a 1 and 2b 1 orbitals ( e D þ ; e E þ and e F þ ) are much stronger compared to the four lower-energy bands [38].
The photoelectron spectrum of CH 2 Cl 2 between 11.0 and 15.7 eV recorded with a resolution of ca. 0.003 eV is shown in Fig. 3 along with calculated adiabatic ionisation energies at the EOM-IP-CCSD/ cc-pVTZ level of theory. An expansion of the ground-state band between 11.1 and 12.0 eV along with the FC simulation is shown in Fig. 4. This is the highest-resolution spectrum of this molecule reported in the literature so far. Our harmonic spectral simulations make it clear that there are two overlapping electronic band systems of the parent cation in this energy range. This is in agreement with Takeshita, who calculated the difference in vertical IE of the e X þ 2 B 2 and e A þ 2 B 1 states of CH 2 Cl 2 + to be only 0.04 eV [26]. However, the difference in adiabatic IE is calculated to be much greater, 0.41 eV, because ionisation to the e X þ 2 B 2 ground state involves a significant change in the ClCCl bond angle, whereas ionisation to e A þ 2 B 1 does not entail such a large change in any parameter. As stated in Section 1, harmonic calculations are more appropriate for CH 2 Cl 2 + than for CH 2 F 2 + because the lowest dissociation energy of the former cation corresponds not to H-atom but to Cl-atom loss, and occurs at 12.108 ± 0.003 eV which is ca. 1.3 eV (0.9 eV) above the predicted adiabatic IE of e X þ 2 B 2 ( e A þ 2 B 1 ); the corresponding energy difference in CH 2 F 2 + is only 0.34 eV.
Our Franck-Condon calculations (details in Table S2  In attempting a complete spectral assignment of all the bands, it becomes quickly apparent that whilst the three strongest peaks at 11.317, 11.401 and 11.483 eV can easily be assigned to have their strongest component as e A þ 2 B 1 0 0 0 , 3 1 0 and 3 2 0 respectively, the picture for the remaining peaks is less clear. First, as is well known with long Franck-Condon progressions arising from a significant geometrical change between the two electronic states involved, the establishment of the absolute vibrational numbering in the unique progression observed (in this case, the e X þ 4 v 0 and corresponding 4 v+1 1 hot band) can be difficult to determine. Second, it is not clear how accurate harmonic Franck-Condon calculations are for ionisation of CH 2 Cl 2 to either the e X þ 2 B 2 or e A þ 2 B 1 states of the cation. In order to settle these issues, a PFI-ZEKE measurement of CH 2 Cl 2 would be needed, similar to that reported for CH 2 F 2 by Forysinski et al. [18], at a resolution an order of magnitude better than the TPES reported herein. With accompanying calculations such as reported in [19], this should then determine anharmonic effects on the vibrational energies and FC factors for ionisation of CH 2 Cl 2 to ionic states which lie ca. 1.3 ( e X þ 2 B 2 ) and 0.9 ( e A þ 2 B 1 ) eV below the dissociation threshold to CH 2 Cl + + Cl + e À .
With these caveats stated, Fig. 4 shows a composite of the experimental threshold photoelectron spectrum of CH 2 Cl 2 from 11.1 to 12.0 eV (black), simulations to the e X þ 2 B 2 (blue) and e A þ 2 B 1 (green) states of CH 2 Cl 2 + , and simulation to the sum of both states (red). The agreement of experimental and theoretical spectrum is good. However, the simulation predicts a monotonously increasing intensity in the e X þ progression for four quanta longer than actually observed, probably due to anharmonic effects in the TPES. The calculated energies and intensities of the individual components and comparison with literature assignments are given in Table 4. Whilst most of the peaks with energies below 11.401 eV, the e A þ 3 1 0 FC maximum, have a unique vibrational assignment involving one level of either e X þ or e A þ , all peaks above this energy have a number of components, often from both ionic states, contributing to the blended assignment. This is a consequence of CH 2 Cl 2 being a relatively heavy polyatomic molecule with several overlapping vibrational bands all allowed by the selection rules in photoelectron spectroscopy, quite apart from the additional problem of two ionic states being approximately degenerate. A comparison of our assignments with those of Pradeep and Shirley [37], Chim [29] and the early HeI study of Potts et al. [5] shows that we believe all of these earlier assignments are in parts incorrect; either it was not appreciated that there were two near-degenerate overlapping ionic states of CH 2 Cl 2 + [29], or that ionisation to e X þ 2 B 2 involved such a large geometry change in the ClCCl bond angle that the m 4 + progression would not start to gain significant FC factor until levels of v well above zero were accessed [5]. Pradeep and Shirley assign our e A þ progression to 2 B 2 and the e X þ progression to the 2 B 1 state, which implies their use of the xz convention just as they did in CH 2 F 2 . However, they refer to Takeshita's results, who used the yz convention, and cite a large ClCCl bond angle change in the 2 B 2 state, which must be a misunderstanding. They also label the higher energy electronic states in the 15-17 eV range according to the yz convention also used herein. It is presumably because of this confusion that they could not appreciate Takeshita's prediction about the offset origin of the e X state and assume to see origin transitions for both lower lying states. Takeshita may also have contributed to this misunderstanding by using two different conventions on his treatise of CH 2 F 2 and CH 2 Cl 2 within a time span of a few months [15,26]. However, putting misunderstandings and absence of coordinate systems aside, an assignment of any spectrum can only relate to the resolution at which it is recorded, and the story of the last forty five years for the ground photoelectron band(s) of both CH 2 F 2 and CH 2 Cl 2 bears witness to this statement. We have no doubt that if and when this spectrum of CH 2 Cl 2 is re-recorded with yet better resolution, new features will emerge, as happened with CH 2 F 2 [18,19], and improved assignments will result.

3.3.
Higher energy peaks in the valence photoelectron spectra of CH 2 F 2 and CH 2 Cl 2 3.3.1. Higher valence electronic states of CH 2 F 2 + The complete valence threshold photoelectron spectrum of CH 2 F 2 at a resolution of ca. 0.006 eV is shown in Fig. 1. There are two major bands lying at higher energy to the ground-state band   Fig. 3 between 11.1 and 12.0 eV. The first photoelectron band of CH 2 Cl 2 consists of ionising transitions to the two overlapping states of CH 2 Cl 2 + e X þ 2 B 2 and e A þ 2 B 1 . The vibrational structure based on a Franck-Condon simulation was calculated for both states. The sum of both simulations is also shown. already discussed in Section 3.1. They are (i) a broad peak from 14.5 to 16.5 eV with possible vibrational structure on the rising shoulder at lower energy and definite structure on the falling shoulder at high energy; and (ii) a peak from 18.5 to 20.0 eV with resolved structure in the low-energy rising edge. In addition, there appears to be a very weak and broad peak between 17 and 18 eV, not observed before, which may be due to autoionisation effects which are known to be prevalent in threshold photoelectron spectroscopy. We consider these two bands in turn.
3.3.1.1. 14.5-16.5 eV. Previous calculations predict that this peak encompasses the photoelectron bands arising from the overlap of electron removal from the (HOMOÀn), n = 1, 2, and 3, orbitals, i.e. e A þ 2 B 2 , e B þ 2 A 1 and e C þ 2 A 2 [15,17,18]. Our spectrum has comparable signal to noise and resolution to that reported by Pradeep and Shirley [37], and our assignments and conclusions are quite similar. Within Koopmans' approximation these states arise from one-electron removal with no extensive configuration interaction [1]. Takeshita has calculated the geometries and reported harmonic vibrational frequencies of the a 1 symmetry normal modes for these three electronically excited states [15]. According to our calculations, for ionisation into e A þ 2 B 2 state the most significant change in geometry is a reduction in the FCF angle from 108.6°t o 83.8°, a similar effect to that predicted and observed for the ClCCl angle in the CH 2 Cl 2 + e X þ 2 B 2 band (Section 3.2). Our spectrum in Fig. 5 shows partially-resolved steps in the rising edge of this broad band peaking at 15.34 eV. Pradeep and Shirley resolved these peaks slightly better, and reported an average vibrational spacing of 583 cm À1 , which they assigned to the CF 2 bending mode m 4 + [37]. It seems likely that the use of a supersonic beam and hence a low rotational temperature and narrow rotational envelope in the Pradeep and Shirley study means that the vibrational fine structure of the band is better resolved than in the room temperature spectrum in Fig. 5, even if our spectral resolution is better. The increase in the m 4 frequency upon ionisation, the value in neutral CH 2 F 2 being 528 cm À1 [11], is consistent with a reduction in FCF For electron removal from the 6a 1 and 1a 2 molecular orbitals producing CH 2 F 2 + e B þ 2 A 1 and e C þ 2 A 2 , we calculated again significant changes in the FCF bond angle; +9.3°in e B þ 2 A 1 , and À11.2°in e C þ 2 A 2 , in good agreement with Takeshita's optimised geometries [15]. In addition, a 13.3°increase in HCH bond angle is predicted in e B þ 2 A 1 . While all other geometric changes are relatively small, the optimised C 2v symmetry e B þ 2 A 1 structure was nevertheless found to be a transition state. Upon breaking symmetry, the ion relaxed into a C S [CH 2 F +. . . F] A 0 structure with three quasi-degenerate states corresponding to the three possible p-holes on the fluorine atom. Thus, the potential energy surface is unlikely to support bound C 2v e B þ 2 A 1 states, and we do not expect any vibrational fine structure in the FC envelope of this state. On the high-energy part of this broad band, however, we observe two overlapping vibrational progressions ( Fig. 1 16.007 and 16.115 eV has a similar average spacing, and the peaks are displaced to higher energy by an average of 450 cm À1 . Pradeep and Shirley also observe these peaks and assign them to a m 3 + progression (CF 2 symmetric stretch, with some weak contribution from FCF bend), but they do not specify in which electronic state this progression is being observed [37].
The Franck-Condon simulation of this overlapped broad band supports the experimental findings. For the e C þ 2 A 2 state, we determine an energy-minimum structure of r e (C-H) = 1.083 Å, r e Ref. [37]. The state symmetries are given as in the paper, the e X þ and e A þ progressions were misidentified and in fact correspond to the first excited and the ground electronic state of the cation, respectively. The convention used for the term symbols is unclear. b Ref. [29]. (CF 2 asymmetric stretch) is expected because of the 0.07 Å increase in C-F bond length and 11.2°decrease in FCF bond angle upon ionisation to this state. The combined simulation for the e A þ and e C þ states of CH 2 F 2 + is shown in Fig. 5, and details are given in Table S3 of the Supplementary Material. The simulation matches the experimental spectrum surprisingly well, notwithstanding any contributions from the repulsive e B þ 2 A 1 state. When comparing our results with those of Takeshita, we note that there is no mention of the e B þ state being unbound as he carried out his calculations in C 2v symmetry. He also reported significantly higher values for m 3 + and m 4 + in the e C þ 2 A 2 state of CH 2 F 2 + of 1122 and 544 cm À1 , respectively [15]. Xi and Huang, on the other hand, reported a higher lying transition state (see later), but found that the e B þ state was a minimum in C 2v [17]. By matching the simulated FC profiles to the experimental spectrum, we can now determine the adiabatic ionisation energies of the e A þ and e C þ states of CH 2 F 2 to be 14.3 ± 0.1 and 15.57 ± 0.01 eV. While the e C þ state can be clearly identified from the spectrum, the origin of e A þ can only be extrapolated by shifting the whole FC envelope to fit the band, resulting in a larger error bar of at least one vibrational quantum in m 4 + .
In summary, we agree with the assignment for this band of Pradeep and Shirley [37], but believe that the discrete structure on the high-energy shoulder is due to intensity in m 3 ) and e F þ (727 cm À1 ) states [37], but the signal-to-noise ratio of their published spectrum hardly justifies such an optimistic assignment. However, all studies agree that the significant reduction in the vibrational energy of this m 3 mode from its value in neutral CH 2 F 2 , 1112 cm À1 [10], is consistent with electron removal from C-F bonding orbitals. We succeeded in optimising the geometries of these states at the EOM-IP-CCSD/cc-pVTZ level of theory. However, only the e D þ state is a true minimum on the potential energy surface. The e E þ and e F þ states possess imaginary frequencies in the CF asymmetric stretch and CH 2 twist mode, respectively, and it was not possible to locate further minima other than the e D þ state by breaking the symmetry and letting the geometry relax. This is at odds with the results of Xi and Huang, who found e D þ 2 B 2 to be a transition state and the 2 A 1 state to be a minimum [17]. A FC simulation predicts the vibrational progression on the low-energy side of the band very well, with dominant contributions from the CF 2 stretch (m 3 + ) and FCF bend (m 4 + ) modes of the e D þ state and their combination bands. Based on the simulation, details for which are given in Table S4 of the Supplementary Material, we can determine the adiabatic ionisation energy to the e D þ 2 B 2 state to be 18.0 ± 0.1 eV.

Higher valence electronic states of CH 2 Cl 2 +
The second photoelectron band of CH 2 Cl 2 between 12.0 and 12.5 eV corresponds to ionisation from the 9a 1 and the 2a 2 molecular orbitals, which are essentially associated with the chlorine p lone pair orbitals perpendicular (9a 1 , 2pp bonding character) and parallel to the principal Cl-C-Cl plane (2a 2 , 2pp non-bonding character) [39]. Removing an electron from the 9a 1 orbital with p bonding character between the chlorine atoms leads to an increase in the Cl-C-Cl bond angle of 8.5°(  (Fig. 3) the e B þ 2 A 1 and e C þ 2 A 2 states are almost degenerate, as was also reported by Takeshita [26]. Ionisation from the 2a 2 orbital, having antibonding character between the chlorine ligands, results in a decrease of the Cl-C-Cl angle of 7.6° (Table 2) and thus activity in m 4 + also upon ionisation (Fig. 7, blue). Minor contributions from the m 3 + mode are also visible in the stick spectrum but are blended out in the convolution with a Gaussian function with a FWHM of 27 meV. At 12.2 eV, a hot band in m 4 contributes to the simulated spectrum, which is followed by the 0-0 transition at 12.25 eV. The sum of both simulated spectra (details for which are given in Table S5 of the Supplementary Material) reproduces the overall shape of the second photoelectron band of CH 2 Cl 2 extremely well. Based on the features of the experimental spectrum and the simulations, we determine the adiabatic ionisation energies for these two ion states to be 12.15 and 12.25 eV.
The spectrum of Pradeep and Shirley [37] is very similar in signal-to-noise ratio and resolution to ours of Fig. 7. They assign the partially resolved peaks to vibrational progressions in m 4 + in both of these ionic states, in agreement with our findings. Our calculations for the e C þ 2 A 2 state also show that Franck-Condon intensity in m 3 + is comparable to that in m 4 + at around 12.33 eV, leading to significant broadening of the spectral features. As noted earlier, a PFI-ZEKE spectrum of a rovibrationally cold sample recorded at ca. 1 cm À1 or better resolution would be needed to take this analysis further. The next band with an onset at ca. 15 eV (Fig. 3) corresponds to ionisation to three overlapping states of CH 2 Cl 2 + ; e D þ 2 B 2 , e E þ 2 A 1 and e F þ 2 B 1 . This is the equivalent band to that in CH 2 F 2 between 18.5 and 20.0 eV (Section 3.3.1), but here we were not able to resolve vibrational structure.

Conclusions
Using the imaging photoelectron photoion coincidence spectrometer at the VUV beamline of the 3 rd generation Swiss Light Source, we have recorded the threshold photoelectron spectrum of CH 2 F 2 and CH 2 Cl 2 from threshold up to ca. 20 eV at a resolution of up to 2 meV or 16 cm À1 . With the help of (EOM-IP-)CCSD calculations and Franck-Condon simulations, we were able to model the observed vibrational fine structure of the ground and excited ion states. For CH 2 F 2 we determine an adiabatic (vertical) ionisation energy of 12.727 ± 0.004 (13.158 ± 0.004) eV. FC modelling only delivered a qualitative but not a quantitative fit to the barely-bound ground state of the CH 2 F 2 + cation, pointing to the importance of the effects of anharmonicity close to dissociation barriers, as discussed in detail by Luckhaus et al. [19]. The CH 2 F 2 photoelectron band between 14.5 and 16.5 eV encompasses ionisation to three overlapping states of the cation, e A þ 2 B 2 , e B þ 2 A 1 and e C þ 2 A 2 . Applying EOM-IP-CCSD calculations, we were able to find C 2v equilibrium structures of two of the three states, to carry out a Franck-Condon simulation of the observed vibrational progressions, and derive adiabatic (vertical) ionisation energies of 14.3 ± 0.1 (15.18 ± 0.07) and 15.57 ± 0.01 (15.73 ± 0.02) eV for the e A þ 2 B 2 and e C þ 2 A 2 states, respectively. The e B þ 2 A 1 state was found to relax to a C S symmetry [CH 2 F +. . . F] structure and is considered to be repulsive in its FC envelope. In a third band, we were able to fit a single vibrational progression to derive an adiabatic (vertical) ionisation energy to the e D þ 2 B 2 state of 18.0 ± 0.1 (18.58 ± 0.05) eV. The first photoelectron band of CH 2 Cl 2 has two overlapping components, namely ionisation to the e X þ 2 B 2 and e A þ 2 B 1 states.
The first ionic state shows an extended progression in m 4 + (CCl 2 bend) arising from a large decrease in ClCCl bond angle upon ionisation, while the first excited state has a much shorter progression in m 3 + (CCl 2 symmetric stretch) with intensity condensed into v = 0-2. A simultaneous Franck-Condon modelling of the two states yielded adiabatic (vertical) ionisation energies of 11.0 ± 0.2 (11.64 ± 0.08) and 11.317 ± 0.006 (11.404 ± 0.006) eV for the e X þ and e A þ states, respectively. The second photoelectron band between 12.0 and 12.5 eV is also comprised of two states exhibiting vibrational structure, e B þ 2 A 1 and e C þ 2 A 2 , with derived adiabatic (vertical) ionisation energies of 12.15 ± 0.02 (12.221 ± 0.008) and 12.25 ± 0.02 (12.284 ± 0.008) eV, respectively.
When we compared our spectra and assignments with the literature, it became evident how crucial it is to provide the auxiliary information necessary to interpret term symbols and symmetries. There is no single accepted way of orienting C 2v symmetry molecules in space, which means that the B 1 and B 2 representations are only defined if the principal plane of the molecule is specified. In some papers, this was done. Sometimes, it was possible to deduce the convention used based on, for example, orbital ordering even when the principal plane was not given. However, in the more complicated case of CH 2 Cl 2 in which the first observed transition belongs to the first excited e A þ state which is a result of ionisation from the HOMO, and not the (HOMOÀ1) orbital, deducing the convention(s) used can prove to be an insurmountable obstacle to even the most intrepid molecular spectroscopist. In this work, we chose yz as the principal plane.
As both experimental and theoretical methods have improved over the last four decades, so has our understanding of the photoelectron spectrum of these two medium-sized molecules.
The spectrum of the first band of CH 2 F 2 , ionisation to e X þ 2 B 1 , has now been recorded at sub-wavenumber resolution by PFI-ZEKE spectroscopy [18], and further work on the overlapping e X þ 2 B 2 and e A þ 2 B 1 bands in the spectrum of CH 2 Cl 2 can only progress with the improvement in resolution that this laser-based technique brings. However, as shown here, even smaller improvements in spectral resolution can yield deeper insights into the assignment of the transitions, particularly in conjunction with modelling the spectra with computational chemistry methods.   7. The experimental spectrum of the third photoelectron band of CH 2 Cl 2 (black), ionisation to e B þ 2 A 1 (green) and e C þ 2 A 2 (blue), and simulations of the Franck Condon vibrational components, calculated in the harmonic approximation (see Section 3.3.2). The red curve, the sum of both calculations, is in good agreement with the experimental spectrum.