Mass-analyzed-threshold-ionization (MATI) spectroscopy of 1,2,3-substituted halogenated benzenes via different intermediate vibrational states in the S1 state

For the first time, two color resonant mass analyzed threshold ionization (MATI) spectroscopy has been applied in order to investigate the ionic properties of 1,3-dichloro-2-fluoro-benzene (1,3,2-DCFB) and 1,3-difluoro-2-chloro-benzene (1,3,2-DFCB) radical cations in their electronic ground state. The ionic ground state of the different samples has been investigated via different S1 intermediate states and compared to 1,2,3-trichlorobenzene measured in previous work. Additionally quantum chemical calculations at DFT (density functional theory) and TDDFT (time-dependent density functional theory) level of theory have been performed to support experimental findings. From the MATI spectra the adiabatic ionization energies of 1,3-dichloro-2-fluorobenzene and 1,3-fluoro-2-chlorobenzene could be determined to be 75.242 6 cm 1 and 75.627 6 cm , respectively. Several vibrational modes of both compounds have been assigned by comparison of the experimental and theoretical results.


Introduction
Halogenated aromatic molecules became a topic of great interest due to their widespread presence in industrial processes and hence presence in the environment as potentially toxic and cancerogenic pollutants. Moreover halogenated benzenes are important model substances to validate theoretical concepts, such as vibronic coupling and others. As one of the spectroscopically best-investigated molecules of all, benzene can serve as an outstanding reference system to study the inuence of tailored perturbations, such as careful choice of substitution pattern, to the (ro)vibronic structure. As an immediate effect a shi in transition energies, ionization energy, molecular geometry and vibrational frequencies can be observed. The characteristics of these effects strongly depend on the number, type and localization of the substituents. As the magnitude of these effects are rather small (from a few wavenumbers up to a hundred wavenumbers), high resolution techniques, such as mass analyzed threshold ionization (MATI) or zero kinetic energy (ZEKE) spectroscopy are ideally suited to investigate such phenomena. [1][2][3][4] Chloro-and uoro-benzenes have been subject to various studies investigating the vibronic properties of the rst electronic excited state (S 1 ) as well as cationic ground state (D 0 ). 1,[5][6][7][8][9][10][11][12][13] In this paper we will focus particularly on out-of-plane b 1 -symmetric modes. Modes of this symmetry are connected to interesting phenomena observed in a whole variety of halobenzenes. For example, diuorobenzenes exhibit large frequency lowering of b 1 (b 3u under D 2h in case of p-DFB) symmetric modes in going from S 0 to S 1 or from D 0 to S 1 , 14 respectively. The intensity gain of this symmetry forbidden mode was interpreted in terms of the pseudo-Jahn-Teller effect (PJTE). The PJTE is based on the idea that a geometrical distortion blurs the difference in symmetry between two or more electronic states of virtually equal energy. In 1,2,4,5-tetrauorobenzene (D 2h ) it was found that the b 2g -symmetric mode 11 shows an unusual strong activity in the form of a long progression (v 2n 11 ). Recent results show that the planarity of this molecule is distorted along the eigenvector of the 11 mode during excitation to 1 B 2u S 1 state. It is an interesting question to elucidate if these phenomena can be extended to further congeners and identied as a general feature. In particular in the case of the molecular geometry change it is crucial to determine the involved vibrational modes and participating electronic states. For this purpose, the experimental data are compared to theoretical calculations, which enable the possibility of assigning the observed vibrational modes. In this paper we present mass-selected two color two photon Resonance Enhanced Multi Photon Ionization (2C2P-REMPI) spectra of the S 1 ) S 0 transition and MATI spectra via different S 1 -intermediate states of 1,3-dichloro-2-uorobenzene and 1,3-dichloro-2-uorobenzene. Some assignments of 1,2,3-trichlorobenzene reported previously 1 were reconsidered and compared to the newly gained knowledge.

Experimental setup
The experimental setup consists of a homemade time-of-ight (TOF) mass spectrometer as described in detail previously. 1,15,16 Briey, the spectrometer consists of a standard second order corrected reectron time-of-ight mass spectrometer equipped with single stage ion source. The laser system used for excitation and ionization consists of two dye lasers (Laser Analytical Systems LDL 205, Lambda Physics FL 3002). Each dye laser is pumped by a dedicated Nd:YAG laser (Lumonics HY-1200, Continuum Surelite II). A Quantum Composers 9600+ delay pulse generator fed by an external clock operated at 10 Hz controls ash lamp and Q-switch delays. The output of each dye laser is frequency-doubled by a BBO-I crystal yielding tunable ranges from 245 to 285 nm. Wavelength calibration of both dye lasers is performed by recording an optogalvanic spectrum with a neon hollow cathode lamp yielding accuracy better than 2 cm À1 . Initially a supersonic molecular beam of sample molecules and seed gas (argon) with a backing pressure of approximately 2 bar is expanded via a pulsed jet valve (General Valves Series 9) into the ion source. To obtain a sufficient vapor pressure the sample is heated to approximately 80 C. Excitation or ionization is accomplished by multi photon absorption under eld-free conditions. In the MATI modus, promptly generated ions are discriminated against Rydberg neutrals through the application of a weak electrical eld of 1-2 V cm À1 by a subsequent delay ($100 ns) to multi-photon excitation. Finally, a high voltage pulse (890 V cm À1 ) switched by a Behlke HS56-01 fast thyristor ionizes the Rydberg neutrals and accelerates them into the mass spectrometer. In REMPI modus, the generated ions are accelerated directly into the TOF without the retarding eld. The ion signal is detected by a conventional dual micro-channel-plate detector and transferred to a LeCroy 534M digital oscilloscope. A computer, linked to the oscilloscope by a GPIB connection performs the data acquisition and processing. The 123-DCFB was purchased from Aldrich, 126-DFCB from ABCR and were used without further purication.

Quantum-chemical calculations
Quantum chemical calculations were performed in order to assign the observed vibrational bands and to support the experimental ndings. The soware package Turbomole [17][18][19][20] was used for all quantum chemical calculations. Geometry optimizations and subsequent frequency analyses for molecules in the electronic ground state (S 0 ) and the cationic ground state (D 0 ) were conducted at the density functional theory (DFT) level of theory with both the gradient corrected functional BP86 21a and the hybrid functional B3LYP. 21b The triple basis set TZVPP 21c has been applied to all DFT calculations. Calculations for the S 1 were done analogously using the time dependent density functional theory (TDDFT). For comparison reasons, geometry calculations and frequency analyses were also treated at the coupled cluster (CC2) level of theory for the S 0 -and S 1state using the Ahlrichs basis set cc-pVTZ. 21d The nomenclature used for assignment of vibrational bands is according to Varsanyi and Szoke,22 which is derived from Wilson's notation of the Benzene modes. 23 We refrained from the use of scaling factors to t the calculated frequencies.

Experimental results
4.1. 1,3-Dichloro-2-uorobenzene (1,3,2-DCFB) 4.1.1. REMPI spectrum. According to electric dipole selection rules the transition to rst excited singlet state 1 B 2 S 1 (p* ) p) in 1,3,2-DCFB is allowed (y-polarized). For benzene, the corresponding transition A 1 B 2u ) X 1 A 1g is dipole forbidden under D 6h symmetry. The reduced symmetry in 1,3,2-DCFB (C 2v ) leads to an electronically as well as vibronically allowed transition. Starting from the premise that the molecules are sufficiently cooled with regards to their degrees of freedom, only total symmetric (a 1 ) vibrations are allowed according to the Franck-Condon (FC) principle. Due to vibronic coupling to intensive, nearby states, vibrations of a 2 and b 2 symmetry can gain intensity as well. The REMPI spectrum shown in Fig. 1 in units of internal energy could be recorded in the range up to 1300 cm À1 . The excitation energy of the rst excited state S 1 could be determined for the rst time to be 36 460 AE 2 cm À1 .
Normal coordinate analysis for 1,3,2-DCFB yields following distribution of fundamental modes under C 2v symmetry: G vib ¼ 11 Â a 1 , 10 Â b 2 , 3 Â a 2 , 6 Â b 1 . Clearly, the spectrum is dominated by the total symmetric modes 1 1 (565 cm À1 ) and 18a (977 cm À1 ). Moreover the a 1 -symmetric modes 9a 1 , 6a 1 , 7a 1 , 12 1 (186 cm À1 , 349 cm À1 , 792 cm À1 , 1104 cm À1 ) could be identied with lower intensity. With 16a 1 (360 cm À1 ) also an a 2 -symmetric mode could be assigned. As expected, no modes of b 1 -symmetry could be assigned to the spectrum in accordance with selection rules. Particular attention should be paid to the low frequency band at 120 cm À1 . A band with such a low frequency can with certainty be assigned to an out-of-plain mode. Based on correlation with MATI spectra we assigned the rst even overtone of the b 1 -symmetric mode 17b to this band. The frequency of 52 cm À1 calculated with the coupled cluster method is in reasonable agreement with the experimental value for the half of the overtone for 17b 2 . The bands at 247 cm À1 (15 1 ), 337 cm À1 (6b 1 ), 500 cm À1 (3 1 ) and 1236 cm À1 (18b 1 ) were assigned to b 2symmetric modes. The latter two experimental values are in excellent agreement with both calculated values (TDDFT, CC2) for the rst excited singlet state of 1,2,3-DCFB. Comparing the calculated values for the 15 1 and 6b 1 mode a major deviation becomes apparent, just like in the case of 17b. All performed calculations show good agreement with the experiment for the modes 9a 1 , 6a 1 , 1 1 , 7a 1 , 18a 1 , 12 1 , 16a 1 , whereas the best results were obtained with the CC2 level of theory (see Table 1).
4.1.2. MATI spectra. The MATI spectra via the S 1 intermediate states 0 0 , 17b 1 , 9a 1 , 1 1 are shown in Fig. 2. The origin of the D 0 ( 2 B 1 ) state and with that the adiabatic ionization energy was found to be 75.242 AE 6 cm À1 (9.3288 AE 0.0007 eV). This value is in good accordance with the previously by conventional photoelectron spectroscopy determined value of 9.32 AE 0.02 eV. 25  Table 1 Comparison experimental and calculated normal modes observed in the S 0 , S 1 and D 0 state of 1,3,2 DCFB Besides the 0 0 transition, the MATI spectrum obtained via the electronic origin exhibits several more additional resonances. The most prominent peak in the spectrum is the band assigned to the 17b mode, which is constituting a violation of the Dv ¼ 0 propensity rule. Additionally, the mode 17b 1 at 95 cm À1 appears as a combination band (17b 1 9a 1 ) at 288 cm À1 . Moreover the spectrum is conspicuously rich in b 1 -symmetric modes. In addition to 17b 1 we assigned the 10b 1 (264 cm À1 ), 16b 1 (452 cm À1 ), 11 1 (678 cm À1 ) and 4 1 (732 cm À1 ). Another outof-plane vibration (10a 1 ) with a 2 -symmetry and low intensity has been assigned to the band at 174 cm À1 . The bands at 361 cm À1 , 578 cm À1 have been assigned to the total symmetric modes 6a 1 and 1 1 , respectively. b 2 symmetric modes were identied at 242 cm À1 (15 1 ), 331 cm À1 (6b 1 ), 471 cm À1 (3 1 ) and 827 cm À1 (7b 1 ). The band at 657 cm À1 ts the overtone 6b 1 .
The MATI spectrum obtained via the S 1 17b 2 shows a short, three-membered regular progression with transitions corresponding to the excitation of one, two and three quanta. In contradiction to the Dv ¼ 0 propensity rule, the band labeled 17b 1 is the most prominent peak in the spectrum. It is notable that the 17b mode also appears in the combination bands 17b 1 9a 1 (288 cm À1 ) and 16b 1 6b 1 (423 cm À1 ). Also apparent is the richness in b 1 -symmetric modes, rst and foremost the intense 16b 1 (448 cm À1 ), but also the 10b 1 (264 cm À1 ), 11 1 (680 cm À1 ), 4 1 (728 cm À1 ). The MATI spectrum obtained via the S 1 9a 1 mode also shows a breakdown of the Dv ¼ 0 propensity rule. Not the vertical transition into the D 0 9a 1 state is the dominating one, but the band at 242 cm À1 (15 1 ). In addition the 17b mode appears again, also as a combination in 17b 1 9a 1 . A further band at 331 cm À1 could be identied with 6b 1 . The measured values are in good accordance with the calculated values (195 and 189 cm À1 for the 9a 1 , 233 and 253 cm À1 for the15 1 and also 300 and 331 cm À1 for the 6b 1 ).
The MATI spectrum via S 1 1 1 continues the series of MATI spectra characterized by a breakdown of the Dv ¼ 0 propensity rule in favor for the vibronic transition into the D 0 17b 1 state. Further active modes that could be assigned in the recorded range up to 750 cm À1 are the 15 1 (241 cm À1 ), 6b 1 (331 cm À1 ) and 1 1 (578 cm À1 ). The experimentally determined frequency for the 1 1 mode is in good accordance with the calculated values of 578 cm À1 or 603 cm À1 respectively. It should be noticed that, owing to the heavy substituents, the displacement pattern of the 'ringbreathing-mode' shows a striking deviation from the original benzene pattern. It resembles clearly the pattern of mode 6a found for benzene. The assignment of the band at 578 cm À1 to the mode 6a is excluded since it was already doubtlessly assigned to the band at 361 cm À1 .
4.2.1. REMPI spectrum. 1,3,2-DFCB belongs to C 2v point group and has a dipole allowed transition to the 1 A 1 S 1 (p* ) p) rst excited state. According to FC-principle only transitions in total symmetric (a 1 ) modes are allowed, b 1 and b 2 can gain intensity due to vibronic coupling mechanism, a 2 modes are symmetry forbidden. The REMPI spectrum shown in Fig. 3 in units of internal energy could be recorded in the range up to 800 cm À1 . The excitation energy of the rst excited state S 1 could be determined for the rst time to be 37 449 AE 2 cm À1 .
The electronic origin and the total symmetric mode 9a 1 at 228 cm À1 dominates clearly the measured REMPI spectrum. It should be noted that the remaining bands at 381 cm À1 and 607 cm À1 that were both assigned to the total symmetric modes 6 1 and 1 1 , respectively. Quite unusual for total symmetric modes we found poor consistency between calculated and observed bands: calculation substantially underestimate the 6a 1 and 1 1 with 188 cm À1 to 202 cm À1 and 488 cm À1 to 517 cm À1 , respectively. The 9a 1 has been overestimated with 364 cm À1 to 374 cm À1 . Nevertheless the correlation with MATI spectra backups the assignment of the total symmetric modes. Moreover, the calculations suggest the mode 7a 1 , that cannot be seen in the spectrum, to appear in the spectrum between 710 cm À1 and 750 cm À1 . As expected, no modes of a 2 symmetry could be assigned to the spectrum in accordance with selection rules. In contrast to 1,2,3-TCB and 1,3,2-DCFB, transitions to b 1symmetric modes are allowed in 1,3,2-DFCB. The bands at 25, 98 and 404 cm À1 could be identied with the b 1 -symmetric modes 17b 1 , 10b 1 and 16b 1 . In this case, the calculated frequencies are in good agreement with the observed ones. The    Fig. 4. The origin of the D 0 ( 2 B 1 ) state and with that the adiabatic ionization energy was found to be 75.627 AE 6 cm À1 (9.3765 AE 0.0007 eV). This value is in good accordance the previously by photoelectron spectroscopy determined value of 9.37 AE 0.02 eV. 25 In accordance with the Dv ¼ 0 propensity rule, the MATI spectrum obtained via the electronic origin is dominated by the 0 0 -band. Furthermore the spectrum is characterized by strong activity of the two total symmetric modes 9a 1 and 6a 1 . Within the recorded range of 1000 cm À1 , the spectrum exhibits a progression in 9a composed of the rst three overtones 9a 1 (293 cm À1 ), 9a 2 (588 cm À1 ) and 9a 3 (882 cm À1 ). The 6a appears as ground vibration 6a 1 (418 cm À1 ), rst overtone 6a 2 (839 cm À1 ) as well as combination vibration 9a 1 6a 1 (706 cm À1 ).
Remarkably, the spectrum exhibits a strong activity in 6a modes: the bands assigned to the modes 6a 1 and 9a 1 6a 1 are among the three most intense bands in the spectrum. The 6a 1 vibration shows progression activity (6a 1 and 6a 2 in the recorded range). In addition, the 6a 1 mode contributes to overtone bands as already discussed above. The weak features at 97, 193 and 213 cm À1 have been assigned on the basis of our previous

1,3,2-DCFB
Compared to electronic and cationic ground states the 17b exhibits a drastically decreased frequency in the rst electronic excited state (see Table 1). Such a frequency lowering suggests a geometrical distortion along the eigenvector of 17b in going from S 0 to S 1 or from D 0 to S 1 , respectively. As considered earlier by Tsuchiya et al. 14 for diuorobenzene it is highly suggested that this phenomenon is the result of strong vibronic coupling between the S 1 and nearby states. But the most important indication for a such a distortion shows up in the MATI spectra via the electronic origin and, rst and foremost, via the 17b 2 . The latter exhibits a three-membered progression of the mode 17b. Not just that this progression shows a shi of the Franck-Condon maximum between S 1 and D 0 , the fact that the transition from D 0 17b 1 ) S 1 17b 2 is favored could be interpreted as a  replanarisation of the molecular geometry in going from S 1 to D 0 along that mode. Also during ionization via the vibrationless level of the rst excited state, the transition D 0 17b 1 ) S0 0 is clearly favored over D 0 0 0 ) S 1 0 0 (see MATI via 0 0 Fig. 2). A major role of the 17b mode during ionization is aggravated by the fact the 17b also appears as combination bands (17b 1 9a 1 in each of the recorded MATI spectra, 17b 1 6b 1 in the MATI via 17b 2 ). The strong activity of the mode 16b 1 in the MATI via 17b 2 and the general, unusual appearance of multiple other out-of-plane modes throughout the MATI spectra give rise to the assumption that the geometry in the S 1 is a product of a distortion along several modes. The CC2 computed geometry optimization seem to support the statements derived from the experimental data: the geometry shown in Fig. 5 is obviously not just the product of a distortion along a single (out-of-plane) normal mode ( Table 2).
The quantum chemical calculations predict a S 1 (p* ) p) transition into the rst excited state for 1,3,2-DCFB. However, due to s*-orbitals localized on the Halogen-Carbon bond, it is possible for modes of appropriate symmetry to induce coupling to (ps*)-states. In accordance with the assumption of a vibronic coupling effect we prevalently observed a substantial decrease in frequency comparing the S 0 state and the S 1 state of the neutral for formal forbidden modes with strong inuence of halogen atoms on their displacement pattern. For the modes 15, 6b, 3 (b 2 symmetry) and 16a (a 2 symmetry) we observed such a decrease.
The decrease in frequency for the modes 15, 6b, 3 is explainable in accordance with the Herzberg-Teller (HT) effect. However, the frequency lowering of the symmetry allowed mode 17b 2 could not be described in terms of the HT effect in a sufficient way. It seems that the distortion along 17b blurs the symmetry differences between the (ps*)-and (pp*)-states which would be an indication for a pseudo Jahn-Teller effect.

1,3,2-DFCB
In 1,3,2-DFCB the vibration 15 appears in the REMPI spectrum as a progression-forming and largely frequency lowered mode. Remarkably, the progression reveals a negative anharmonicity. The occurrence of the 15 as a b 2 symmetric mode is explained in terms of the Herzberg-Teller effect. Besides the mode 15, also the modes 6a and 9a exhibit a largely lowered frequency in the rst excited state S 1 . Both modes are likely to contribute in pseudo Jahn-Teller distortion in going from the electronic ground to excited state or from excited state to ionic ground state, respectively. This is seen from the different MATI spectra as well as the different calculated structures given in Tables 3-8. The MATI spectrum via the electronic origin exhibits a harmonic, three-membered progression of mode 9a (9a 1 , 9a 2 , 9a 3 ), whereby the band 9a 1 is almost as intense as the 0-0 transition. Such a shi of the Franck-Condon maximum is a clear indication for a distortion along the eigenvector of this mode during ionization.
A similar situation is found in the MATI spectrum via 9a 1 . Here, in contradiction to the Dv ¼ 0 propensity rule, the transition into the S 1 6a 1 state is the most favorable while the vertical transition into the 9a 1 is scarcely observable. Nevertheless the mode 9a is strongly present in overtones (9a 1 6a 1 , 9a 1 1 1 , 9a 2 6a 1 ). The geometry found by the TDDFT calculations ( Fig. 6a; Tables 6 and 7) support the experimental ndings in predicting a C 2vsymmetric structure distorted along the 6a. The CC2 calculation suggests a 17b-like distorted structure ( Fig. 6b; Table 8).
The CC2 calculation suggests a 17b-like distorted structure (Fig. 6b). Considering the vanishing low activity of 17b in the MATI spectra, the suggestion seems incorrect. Nevertheless the substantial decrease in frequency for 6a 1 and 9a 1 comparing the D 0 state and the S 1 state is accurately reproduced by all quantum chemical methods employed. Comparing experimental and calculated results for 1,3,2-DFCB, the unusual, striking deviation (up to 49%) for a 1 -symmetrical S 1 -frequencies was particularly noticeable. A poor reproduction of the molecular equilibrium structure in the S 1 (wrong minimum on the PES) could be one possible explanation.
A contrary indication is the fact that we nd very similar frequencies for widely differing structures predicted by TDDFT (planar structure) and CC2 method (out-of-plane-structure). The planar structure resembles the experimental observations well, but since all frequency analyses were performed in harmonic approximation, a strong anharmonicity for this particular vibrations could be another explanation for this ndings.

Conclusion
Comparing the results from two compounds presented in this paper with results obtained from different isomeric trichlorobenzenes, 1 the increasing number of uorine atoms lead to a progressive decrease in some of the observed frequencies in going from S 0 to S 1 and from D 0 to S 1 , respectively. This is especially true for modes with a strong uorine participation in their vibrational pattern like 17b (cf. Table 1). We interpreted this as a strong indication for an out-of-plane distortion during excitation or a replanarisation during ionization along the eigenvector of those modes. This phenomenon could be attributed to an above, bounding S(ps*) state which is stabilized by an increasing number of uorine atoms. 12 It can be concluded that the uorine atoms contribute a signicant share in form of s* ) p character to the transition.  (BP86/TZVPP) and 4.44 eV (CC2/cc-pVTZ). The CC2 value reproduces the experimental value of 4.52 eV with a deviation of 0.08 eV best.
The experimental value for the ionization energy (IE) of 9.32 eV is underestimated by both DFT methods with 9.00 eV (B3LYP/TZVPP) and 8.97 eV (BP86/TZVPP). (The coupled cluster method CC2 is not suited for the ionic species!). In this case the frequency analysis performed by the CC2 method showed to be most appropriate to reproduce the experimental frequencies.
Furthermore the S 1 ) S 0 electronic excitation energies (EE) and D 0 ) S 0 adiabatic ionization energies (IE) could be determined very exactly: EE (