Normal modes that are used as basis for the structure fit in the electronically excited state. 

Context in source publication

Context 1

... can be assigned ͑ 6 b and 1 ͒ as well as combination bands of the modes with intermolecular vibrations. Therefore we suggest that this spectrum arises from excitation of a combination band, possibly 6 b + ␴ . The second trace depicts the emission spectrum, obtained by excitation of 0 , 0 + 744 cm −1 . This spectrum consists of rather few bands, thus making the assignment of the excitation band very uncertain. The first trace of Fig. 5 shows the fluorescence emission spectrum, obtained via excitation of the vibrationless origin 0,0. The assignments given in Figs. 5 and 6 are based on the ab initio calculations described in Sec. III A. The calculated frequencies for the ground-state vibrations that were used for the assignment are summarized in Table II. The strongest band in the emission spectrum after excitation at 0 , 0 + 165 cm −1 ͓ trace ͑ d ͒ of Fig. 5 ͔ is found at 120 cm −1 and can be assigned on the basis of anharmonic MP2/cc-pVDZ calculations to the intermolecular ␤ 1 vibration. The corresponding computed frequency value from the MP2 calculation is quite close ͑ 131 cm −1 ͒ ͑ see Table II ͒ . The ␤ 1 mode can be described as an in-plane wag mode of the water moiety with respect to the 7AI moiety. A progression of this mode can be seen up to the third overtone as well as combination bands with the vibrations ␴ , 6 b , and 1. The assignment of the 165 cm −1 band in the absorption spectrum to ␤ 1 in the S 1 state therefore appears reasonable. Excitation at 185 cm −1 ͓ trace ͑ g ͒ of Fig. 5 ͔ results in a long progression up to the third overtone of a band at 157 cm −1 and combination bands with the modes ␤ , ␳ , 6 b , and 1. The MP2 calculations allow a straightforward assignment of the S 1 vibration at 185 cm −1 to mode ␴ . Upon excitation of 0 , 0 + 199 cm −1 the emission spectrum shows a very strong feature at 182 cm −1 and combinations with this band ͓ trace ͑ a ͒ of Fig. 6 ͔ . Comparison with the results of MP2 calculations show that this band can be assigned to the intermolecular vibration ␳ 2 . This mode can be described as a wag mode of the out-of-plane hydrogen. Also in this case, the propensity rule allows for an unequivocal assignment of the transition at 199 cm −1 to the intermolecular mode ␳ 2 in the S 1 state. The strongest band after excitation at 0 , 0 + 795 cm −1 ͓ trace ͑ d ͒ of Fig. 6 ͔ is lo- cated at 764 cm −1 and can be assigned to the ring breathing mode 1. The corresponding value obtained via the ab initio calculation is 735 cm −1 . Furthermore, the overtone of mode 1 as well as many combination bands can be assigned. These findings result in the assignment of the band 0 , 0 + 795 cm −1 to mode 1. The intensities of the emission bands after excitation of the described S 1 -state vibrations are summarized in Table V. They show clearly how the strongest transition ͑ marked by an asterisk ͒ appears upon excitation of the corresponding vibration in the excited state, illustrating the propensity rule. The change of a molecular geometry upon electronic excitation can be determined from the intensities of absorption or emission bands using the FC principle. The fit procedure 5 has been explained in detail in a previous publication. In a first step the geometry and Hessian matrix of both the ground and excited states are calculated at the TDDFT level ͑ B3LYP/cc-pVTZ ͒ . Using the recursion formula given by 3,4 35 Doktorov et al. the Franck-Condon factors of the observed and assigned transitions are calculated and a simulated intensity distribution is obtained. In subsequent steps, the S 1 -state geometry is displaced along selected normal coordinates and the resulting intensity pattern is calculated. The displacements are iterated until the observed intensity pattern matches the simulated one. If experimental data for rotational constants in both states are available ͑ possibly for several isotopomers ͒ their changes upon electronic excitation can be used as an additional part of the overall FC fit. The fit has been performed using the program FCFII , which has been 5 developed in our group and described previously. The use of a selected subensemble of normal modes as basis for the displacements is forced by the difficulty to assign a sufficient number of vibrations in the electronically excited state. This selection has to be performed carefully, in order to avoid artificial displacement effects by consideration of too similar modes. Emission spectra have been obtained from excitations of 0,0, ␤ 1 , ␴ , ␳ 2 , and 1. These modes were taken as displacement vectors for the fit. Additionally, the ring modes 6 b and 18 b were included in the fit. The mode 6 b shows up in all emission spectra and the 18 b vibration shows up very promi- nent in the emission spectrum of the 0,0 transition. Thus, the six motions which form the basis for the displacements upon electronic excitation are ␤ 1 , ␴ , ␳ 2 , 6 b , 1, and 18 b . All of them are intermolecular or in-plane modes and are depicted in Fig. 8. First of all, a simulation of the intensities of the emission bands was performed, using the geometries and the Hessian matrices from the TDDFT B3LYP/cc-pVTZ calculations. We decided to take the results from the ͑ TD ͒ DFT calculations because the CASSCF wave function cannot properly de- scribe the intermolecular bonds due to the lack of dynamic electron correlation. DFT calculations give more reliable structures because a correlation functional is contained in the B3LYP functional. Hence CASSCF underestimates the bond strength between the chromophore and the water molecule resulting in too long bond lengths for the hydrogen bonds. Additionally the rotational constants for the ground state obtained via CASSCF are not particularly good, see especially the rotational constant B , in Table I. From this it follows that the changes in the rotational constants are described poorly on the CASSCF level, cf. Table VII. The simulations from the ͑ TD ͒ DFT calculations are shown in the traces which follow the respective experimental emission spectrum in Figs. 5 and 6. Although the overall performance of these simulations seems not to be too bad, there are severe deviations between the experimental and simulated intensities. In the emission spectrum via the excitation of 0,0 the intensity of mode 1 is underestimated, while the intensities of the overtones of ␴ are strongly overestimated. The largest deviations are observed for the spectrum obtained after the excitation of mode 1. In the simulation the intensity of the combination bands 6 a + ␴ and 1 2 1 + ␴ are strongly overestimated. Both bands are more intensive in the simulation than the pumped band 1. The experimental trace, however, shows a completely different situation, with weak 6 a + ␴ and 1 2 1 + ␴ combination bands and 1 as the strongest transition. After the simulations with unchanged geometries, we performed FC fits of the intensities of the vibrations, overtones, and combination bands in the spectra of Figs. 5 and 6 by displacing the S 1 -state geometry along the six normal modes described above. The results are shown in the traces, which follow the respective FC simulations in Figs. 5 and 6. Close inspection of all emission spectra shows that the intensity pattern is well reproduced upon displacement of the S 1 geometry. As a representative example let us compare the experimental spectrum, the simulation, and the fit of the emission spectrum upon ␴ excitation ͓ Fig. 5, traces ͑ g ͒ – ͑ i ͔͒ . The intensities of the first and the second overtones of mode ␴ are too weak in the simulation, what is corrected in the fit. The fundamental of this mode is overestimated in the simu- lation and is corrected by the fit as well. Also the quality of the simulation of the ␴ progression is quite bad. The experimental spectrum shows a maximum intensity at the first overtone, while the fundamental of the ␴ mode is calculated to be the strongest band of this progression in the simulation. An obvious improvement concerns the intensity of mode 6 a , c.f. trace ͑ g ͒ in Fig. 5, for example. In the simulation this band is much too strong, compared with the experimental spectrum. The coupling of the ring breathing mode 1 to mode ␴ and its overtones is described wrong. The simulation shows a maximum intensity of the fundamental of mode ␴ whereas the experiment exhibits the maximum at the first overtone. All these intensities are very well reproduced in the FC fit of the ␴ emission spectrum, cf. Fig. 5, trace ͑ i ͒ . Table VII shows the results for the S 1 displacement obtained from the Franck-Condon fit described above. The first column gives the results for the geometry changes from the 7AI monomer obtained from the Franck-Condon fit as dis- 10 cussed in an earlier paper. The second and third columns present the results for the geometry changes upon electronic excitation as obtained from the CASSCF and ͑ TD ͒ DFT/ B3LYP calculations. The fourth column shows the results for the geometry changes from the Franck-Condon fit to 105 fluorescence emission bands and to the changes of the rotational constants of one isotopomer of 7AI-water cluster. The fifth column gives the experimentally determined rotational constant changes. The geometry changes of 7AI water upon electronic excitation are displayed in Fig. 9. The main changes of the 7AI-water geometry can be described as a distinct shortening of both the O ̄ H and the H ̄ N hydrogen bonds ͑ −11.5 and −5.4 pm, respectively ͒ . This reasoning follows directly from the experimental finding that most of the Franck-Condon active vibrations are intermolecular ones, both in absorption as well as in emission. While for the monomer the pyridine ring expands and the pyrrole ring distorts unsymmetrically upon electronic excitation, the situation is different for the cluster. Here, both moieties are distorted unsymmetrically, cf. Table VII. This deformation of the pyridine moiety in the cluster can be described as a shortening ...