The electronic transitions of dibenzothiophene: linear dichroism spectroscopy and quantum chemical calculations

The electronic transitions of dibenzothiophene: linear dichroism spectroscopy and quantum chemical calculations

Journal of Molecular Structure 661-662 (2003) 603–610 www.elsevier.com/locate/molstruc The electronic transitions of dibenzothiophene: linear dichroi...

199KB Sizes 16 Downloads 80 Views

Journal of Molecular Structure 661-662 (2003) 603–610 www.elsevier.com/locate/molstruc

The electronic transitions of dibenzothiophene: linear dichroism spectroscopy and quantum chemical calculations Jens Spanget-Larsen, Erik W. Thulstrup* Department of Life Sciences and Chemistry, Roskilde University, P.O. Box 260, DK-4000 Roskilde, Denmark Received 22 April 2003; accepted 23 June 2003 Dedicated to Professor Schrader for his outstanding contributions to spectroscopy over many years

Abstract The electronic states of dibenzothiophene are investigated by UV Linear Dichroism (LD) spectroscopy on samples partially aligned in uniaxially stretched polyethylene. The LD spectra reveal previously unobserved spectral features, leading to experimental identification and symmetry assignments of at least seven electronic singlet states in the region below 50,000 cm21. The observed transitions are well reproduced by Time-Dependent Density Functional Theory by using the B3LYP/6-31 þ G* functional, and by calculations in the semiempirical LCOAO method. The assignment of the observed absorption bands is further supported by comparison with previously published Magnetic Circular Dichroism spectra, and with the spectrum of the closely related compound dibenzofuran. q 2003 Elsevier B.V. All rights reserved. Keywords: Time-dependent density functional theory; Linear dichroism; Magnetic circular dichroism

1. Introduction Sulfur containing organic molecules have for years attracted considerable interest, for several reasons. Among these are both their use in a wide range of pharmaceutical drugs as well as their potential importance in molecular electronic devices, conducting polymers, etc. In the present investigation the UV spectrum of dibenzothiophene is studied. The optical (UV and IR) spectra of dibenzothiophene have been studied before on * Corresponding author. Tel.: þ45-46-742709; fax: þ 45-46743011. E-mail address: [email protected] (E.W. Thulstrup). 0022-2860/$ - see front matter q 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0022-2860(03)00509-X

several occasions, both experimentally and theoretically, and in several cases the molecule has served as a testing ground for quantum mechanical models, where questions like the possible importance of inclusion of d-orbitals on S in simple quantum mechanical calculations have been debated. More than 30 years ago, Bree and Zwarich [1] made a thorough study of the vibrations of dibenzothiophene, based on IR and Raman spectra of single crystals. They assigned the experimental spectra using normal coordinate calculations, performed on ‘large, high-speed computers’. Even 30 years later, these excellent spectra and the assignments were used for comparison with several theoretical studies of the structure and vibrations

604

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

of the molecule, see Lee [2]. Cai and Lim used luminescence studies to show the existence of an intermolecular triplet excimer of dibenzothiophene in fluid solutions [3], while Soscu´n et al. [4] determined the polarizability of dibenzothiophene, both theoretically and experimentally by refractometry. The UV spectrum of dibenzothiophene was reported in 1956 by Badger and Christie [5] and these results were used by Momicchioli and Rastelli [6] for their p-electron theoretical studies of electronic spectra of five-membered heterocycles, including dibenzothiophene, and later by Buemi, Zuccarello, and Romeo [7] in INDO and CINDO studies of excited state dipole moments for dibenzothiophene and other five-ring aromatics. The He(I) photoelectron spectra of a series of fluorene analogues, including dibenzo-thiophene, were published by Rusˇcˇic´ et al. in 1978 and interpreted in terms of simple p-electron theory [8,9]. Also the Magnetic Circular Dichroism (MCD) spectrum of dibenzothiophene was studied quite early; in 1981, Igarashi, Tajiri, and Hatano [10] measured MCD spectra of a series of conjugated O and S heterocycles, both experimentally and theoretically, using a PPP p-electron model. In connection with linear dichroism (LD) spectroscopic studies at room temperature of the alignments of a large number of different aromatic molecules, including dibenzothiophene, in stretched polyethylene film matrices, Thulstrup and Michl [11] and Radziszewski and Michl [12] determined those properties of the orientation distribution function that are available from absorption spectroscopy (see below). In the present communication we report complete UV LD spectra of dibenzothiophene down to 220 nm. The assignment of the observed electronic absorption bands is supported by the results of quantum chemical calculations. In particular, we shall use time-dependent density functional theory (TD-DFT) [13,14], which has been shown to lead to excellent predictions of electronic transitions for a variety of sulfur heterocycles [15, 16]. In addition, the investigation is supported by comparison with the previously published MCD spectrum [10], and by correlation with recent experimental and theoretical results for the closely related compound dibenzofuran [17].

2. Linear dichroism spectroscopy—experimental For an electric dipole transition, the molecular ^ l; transition moment is defined as M0f ¼ k0lMlf where k0landlf l indicate the wave functions for the ˆ is the initial and final states, respectively, and M electric dipole moment operator. The direction of the vector M0f (the absorption ‘antenna’) is characteristic for a given transition. If linearly polarized electromagnetic radiation with its electric vector along a direction 1 is used in the spectroscopic experiment, the probability of light absorption is proportional to ½M0f ·12 : The measurable properties of the average alignment of transition moments in a uniaxial sample may be conveniently expressed in terms of simple directional cosine square averages [18,19] Kf ¼ kcos2 ðU; M0f Þl

ð1Þ

where U is the unique sample axis (see below), and the brackets k l indicate an average of the cosine squares over all sample molecules, which in general will have different alignments. Similarly, the alignment of the molecular framework may be expressed in terms of K-values referring to the symmetry-determined molecular axes system: Ks ¼ kcos2 ðU; sÞl; s ¼ x; y; z: Due to the C2v symmetry of the dibenzothiophene molecule, all optically allowed electronic and vibrational transitions must have their transition moments along either x; y; or z: Note that for a description of experiments involving two or more photons (luminescence, Raman, etc.) additional orientational parameters are needed [18,19]. Although most spectroscopy is still performed on isotropic samples, aligned samples are common in nature, and especially biological samples offer many important examples. However aligned samples may also be produced in the laboratory. Among many alignment techniques [18,19], the use of stretched

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

polymers as anisotropic solvents for the molecules to be studied has turned out to be the most useful for a wide range of molecules. It is usually simple to align solute molecules of moderate dimensions in stretched polymer sheets; even near-spherical molecules can frequently be efficiently aligned [20,21]. Among the polymers, polyethylene, as in the present study, and poly(vinylalcohol) have been most often used. After stretching, these polymers produce uniaxial samples characterized by the stretching direction, the unique sample axis U; all directions perpendicular to U are thus equivalent, even in thin sheets. This assumption of uniaxiality has been verified by both X-ray diffraction studies on pure, stretched polyethylene sheets [22] and by optical spectroscopic studies of guest molecules in such sheets [23]. Dibenzothiophene was purchased from Aldrich (99 þ %). Linear dichroic UV absorption spectra of dibenzothiophene between 33,000 and 48,000 cm21 were recorded at room temperature on a Shimadzu MPS 2000 spectrophotometer with Glan polarizers inserted into both sample and reference beams. The partially aligned samples were prepared by dissolving the compound in a sheet of polyethylene, using chloroform as a solvent—chloroform swells the polyethylene, which allows the guest molecules to enter the polymer. Partially aligned samples of guest molecules in the sheet were obtained by stretching the sheet 400% [18,19] along laboratory axis U; where axes ðU; VÞ are in the plane of the sheet, W perpendicular to it. Aromatic molecules have been shown to be aligned with their ‘long axes’ predominantly in the stretching direction [11]. The probability for absorption by a molecule with its transition moment along the molecular ‘long axis’ y will thus be relatively large if the electric vector 1 of the radiation is directed along U and correspondingly low if it is directed perpendicular to U; e.g. along V; the other axis in the plane of the polymer film. If the transition moment for a third transition is along the shortest molecular axis, x; perpendicular to the molecular plane, the opposite will be observed: low absorbance when 1 is chosen along U and high absorbance when it is along V: The LD absorbance spectra obtained using linearly polarized light with the electric vector along U or V are called EU ðn~Þ and EV ðn~Þ; respectively. These

605

spectra have been baseline-corrected for scattering and absorption due to the sheet itself, by subtracting, for each polarization direction, the absorbance spectrum of an empty, stretched polyethylene sheet. If EU ðn~Þ and EV ðn~Þ are different, the sample is said to exhibit LD (‘it has an LD’). It may be added that LD spectra may be obtained in the IR region in a similar fashion, but in most cases much thicker polyethylene sheets must be used because of the lower absorbance in this region [18,19]. The absorption experiments performed on a uniaxial sample thus produce two, baseline corrected, linearly independent spectra: X EU ðn~Þ ¼ kcos2 ðU; M0f ÞlAf ðn~Þ ð2Þ f

EV ðn~Þ ¼

X

kcos2 ðV; M0f ÞlAf ðn~Þ

ð3Þ

f

A third spectrum, with light polarization along W; is difficult to record on the present sample, but for a uniaxial sample it would be equal to EV ðn~Þ; anyway (provided the number of molecules in the light path were the same). The sums run over f ; which describes transitions contributing to the spectrum; Af ðn~Þ is defined as three times the contribution from transition f to the absorbance that would be measured on an unaligned sample of the same molecules. The two spectra, EU ðn~Þ and EV ðn~Þ; are called LD spectra and the LDðn~Þ is frequently defined as the difference between EU ðn~Þ and EV ðn~Þ: 2.1. Calculations The molecular equilibrium geometries and electronic structures of dibenzothiophene and dibenzofuran were computed by using B3LYP density functional theory (DFT) [13,14] and the 6-31 þ G* basis set [13,14]. Electronic transitions to excited singlet states were computed with TD-DFT, [13,14, 15], by using the B3LYP/631 þ G* functional. The results are given in Tables 1 and 2. These calculations were performed with the GAUSSIAN 98 software package [24]. In addition, electronic transitions and MCD B-terms [19] for dibenzothiophene were computed with the semiempirical ‘Linear Combination of Orthogonalized Atomic Orbitals’ (LCOAO) method [25].

606

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

Table 1 Observed and calculated electronic transitions for dibenzothiophene: n~: wavenumber in 103 cm21, 1 : molar absorbance in M21 cm21, f : oscillator strength, pol: polarization direction TD-B3LYP/6-31 þ G*

Observed a

b

a

n~

log 1

pol

Term

A B

30.6 34.9

3.54 4.15

z y

C D E

38.0 39 42.0

4.00 4.75

y z y

F G

43.1 47

4.73 4.23

z y

2 1 1 2 3 3 1 2 4 4 2 3 3 5 5 4 4

a b

1

A1 B2 1 A2 1 B2 1 A1 1 B2 1 B1 1 B1 1 A1 1 B2 1 A2 1 B1 1 A2 1 B2 1 A1 1 A2 1 B1 1

n~

f

Leading configurations

32.9 36.6 39.1 39.7 40.0 41.6 42.9 44.0 44.1 45.4 46.0 46.2 47.5 47.6 48.8 48.8 49.5

0.029 0.13 0 0.003 0.010 0.550 0.001 0.009 0.165 0.154 0 0.0001 0 0.058 0.029 0 0.014

87% 5b1 ! 6b1, 9% 3a2 ! 4a2 67% 3a2 ! 6b1, 20% 5b1 ! 4a2 92% 5b1 ! 19b2, 4% 5b1 ! 20b2 61% 3a2 ! 7b1, 20% 2a2 ! 6b1, 15% 5b1 ! 5a2 76% 5b1 ! 7b1, 15% 3a2 ! 4a2 52% 5b1 ! 4a2, 15% 3a2 ! 6b1, 14% 2a2 ! 6b1 97% 3a2 ! 19b2 94% 5b1 ! 23a1 61% 3a2 ! 4a2, 12% 5b1 ! 7b1, 6% 4b1 ! 6b1 52% 2a2 ! 6b1, 22% 3a2 ! 7b1, 6% 5b1 ! 5a2 98% 3a2 ! 23a1 94% 5b1 ! 24a1 90% 5b1 ! 20b2 80% 5b1 ! 5a2, 7% 4b1 ! 4a2 48% 3a2 ! 5a2, 44% 4b1 ! 6b1 98% 3a2 ! 24a1 66% 5b1 ! 25a1, 16% 3a2 ! 20a2, 11% 2a2 ! 19b2

Stretched polyethylene, this work. Heptane solution [28].

Table 2 Observed and calculated electronic transitions for dibenzofuran: n~ : wavenumber in 103 cm21, 1 : molar absorbance in M21 cm21, f : oscillator strength, pol: polarization direction TD-B3LYP/6-31 þ G*

Observed a

A B C D E

a b

b

a

n~

log 1

pol

Term

33.0 35.5 40.0 43.6 45.4

3.69 4.24 4.33 – 4.60

z y y z y

2 1 2 3 3 1 1 2 4 4 2 5

Stretched polyethylene [17]. Light petroleum solution [29].

1

A1 B2 1 B2 1 A1 1 B2 1 A2 1 B1 1 B1 1 B2 1 A1 1 A2 1 A1 1

n~

f

Leading configurations

35.8 36.9 42.2 43.7 44.8 46.1 47.1 47.4 47.5 47.7 48.3 49.4

0.029 0.299 0.032 0.078 0.287 0 0.009 0.006 0.443 0.026 0 0.074

80% 4b1 ! 5b1, 16% 3a2 ! 4a2 76% 3a2 ! 5b1, 20% 4b1 ! 4a2 50% 2a2 ! 5b1, 39% 3a2 ! 6b1 57% 3a2 ! 4a2, 22% 4b1 ! 6b1, 8% 4b1 ! 5b1 63% 4b1 ! 4a2, 13% 2a2 ! 5b1, 7% 3a2 ! 6b1 98% 3a2 ! 21a1 82% 4b1 ! 21a1, 15% 3a2 ! 6b2 84% 3a2 ! 16b2, 14% 4b1 ! 21a1 37% 3a2 ! 6b1, 21% 2a2 ! 5b1, 16% 4b1 ! 4a2 49% 4b1 ! 6b1, 27% 3b1 ! 5b1, 9% 3a2 ! 4a2 94% 4b1 ! 16b2 51% 3a2 ! 5a2, 22% 3b1 ! 5b1, 9% 4b1 ! 6b1

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

3. Results and discussion The UV absorption spectra, EU ðn~Þ and EV ðn~Þ; recorded with linearly polarized light on the uniaxially aligned sample as described above, are shown in Fig. 1. If we introduce the orientation factors K defined in Eq. (1) into Eqs. (2) and (3), we obtain EU ðn~Þ ¼

X

Kf Af ðn~Þ

ð4Þ

f

1X EV ðn~Þ ¼ ð1 2 Kf ÞAf ðn~Þ 2 f

ð5Þ

Eq. (5) follows from the equivalence of axes V and W; EV ðn~Þ P ¼ EW ðn~Þ; and the fact that EU ðn~Þ þ EV ðn~Þ þ EW ðn~Þ ¼ Af ðn~Þ: The two spectra are now conveniently expressed in a form that separates the effects of molecular alignment from the spectral properties [11,12]. Note that in an isotropic sample all Kf are equal to 1/3 and EU ðn~Þ ¼ EV ðn~Þ in accordance with the definition of Af ðn~Þ: It is important to stress that Kf contains the exact orientational information about transition f that is

Fig. 1. UV linear dichroic absorbance curves for dibenzothiophene partially aligned in stretched polyethylene at room temperature. The curves indicate EU ðn~Þ (full line) and EV ðn~Þ (dashed line) recorded with the electric vector of the linearly polarized radiation parallel and perpendicular, respectively, to the polymer stretching direction U:

607

available from an absorption experiment, and that no assumption, except about uniaxiality, has been made about the molecular orientation distribution in the sample. A Kf value may be determined for each recognizable spectral feature (peak or shoulder) by the so-called TEM method [18,19]. In this, series of linear combinations of EU ðn~Þ and EV ðn~Þ are formed, for example [26] ri ðn~Þ ¼ ð1 2 Ki ÞEU ðn~Þ 2 2Ki EV ðn~Þ

ð6Þ

with Ki ranging fromP0 to 1. It follows from Eqs. (4) and (5) that ri ðn~Þ ¼ f Af ðn~ÞðKf 2 Ki Þ: Hence, spectral features due to transitions with orientation factor equal to Ki will disappear in the combination ri ðn~Þ and the K-values can be determined by visual inspection. An example of application of the TEM method is shown in Fig. 2. As mentioned above, for molecules of C2v symmetry only three different, mutually perpendicular transition moment directions, corresponding to the symmetry axes of the molecules, are expected. Assuming a planar structure for the molecule, the present spectral region will be dominated by p – p* transitions which have always transition moments in the molecular plane; thus most observed intensity will correspond to y-or z-polarization.

Fig. 2. Reduced absorbance curves rK ðn~Þ ¼ ð1 2 KÞEU ðn~Þ 2 2KEV ðn~Þ with K ranging from 0 to 1 in steps of 0.1 (see text).

608

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

We have for these (as for any set of three orthogonal axes): Kx þ Ky þ Kz ¼ 1: We obtain the following values for ðKy ; Kz Þ : (0.59,0.28), in excellent agreement with the results (0.59, 0.29) earlier obtained by UV spectroscopy [11], but demonstrating a somewhat better alignment than that (0.49,0.32) obtained in the thick polyethylene sheet which had to be used in the earlier IR spectroscopic studies [12]. A recent IR reinvestigation of dibenzothiophene in a thick polyethylene sample yielded ðKy ; Kz Þ ¼ ð0:48; 0:32Þ [27]. In spectral regions where only contributions to EU ðn~Þ and EV ðn~Þ from two different transition moment directions are present, for example corresponding to y and z; it is possible to separate these contributions into two parts, Ay ðn~Þ Pand Az ðn~Þ; which represent the contributions to Af ðn~Þ from all transitions with transition moments along y and z; respectively. From Eqs. (4) –(6) we have: Ay ðn~Þ ¼ ðKy 2 Kz Þ21 rz ðn~Þ

ð7Þ

Az ðn~Þ ¼ ðKz 2 Ky Þ21 ry ðn~Þ

ð8Þ

The two spectra, Ay ðn~Þ and Az ðn~Þ representing the sums of y-and z-polarized contributions, are shown in Fig. 3, together with the corresponding results of the theoretical calculations. Table 1 gives a summary of the experimental and theoretical results. Based on a comparison of experimental and theoretical results for UV transition energies, intensities, and transition moments, we obtain the following assignments: The first, medium intense transition A at 30,600 cm21 is z-polarized and is followed by a more intense, y-polarized transition B at 34,900 cm21. They are easily assigned to the 2 1A1 and 1 1B2 states predicted at 32,900 and 36,600 cm21 (Table 1). In the region of band A; weaker, y-polarized absorption is observed from around 31,000 cm21. This absorption can probably be assigned to vibrational components of A gaining intensity by vibronic coupling with transition B: In the related case of dibenzofuran, the 2 1 A1 and 1 1B2 states are much closer in energy (Table 2) and the strong vibronic coupling between them has been a matter of some debate; see the discussion in Ref. [17] and literature cited therein. The application of Eqs. (7) and (8) reveals a more complicated picture of the following part of

Fig. 3. Partial absorbance curves Ay ðn~Þ (full line) and Az ðn~Þ (dashed line) corresponding to Eqs. (7) and (8) with ðKy ; Kz Þ ¼ ð0:59; 0:28Þ: The curves indicate absorbance polarized along the molecular symmetry axes y and z:

the spectrum. The medium-intense y-polarized band C with peaks at 38,000 and 39,200 cm21 overlap weaker, z-polarized absorption D starting around 39,000 cm21. We assign C and D to the 2 1B2 and 3 1 A1 states computed at 39,700 and 40,000 cm21. The predicted transitions to these states are relatively weak, but the shapes of the bands C and D indicate that part of the observed intensity may be borrowed from the stronger bands E and F by vibronic coupling. E and F have maxima at 42,000 and 43,100 cm21 and are the most intense transitions in the spectrum. E is y-polarized and must be assigned to the optically intense state 3 1B2, calculated at 41600 cm21. The zpolarized transition F can be assigned to 4 1A1 predicted at 44,100 cm21. Finally, near 47,000 cm21, we observe the last transition in the investigated spectral region, the y-polarized G; probably corresponding to the 4 1B2 state calculated at 45,400 cm21. It may be added that several x-polarized (B1) transitions are predicted (generally of s– p* type), but all have low intensities. Additional experimental information is provided by the MCD spectrum of dibenzothiophene published

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

by Igarishi, Tajiri, and Hatano [10]. However, these authors arrive at a different assignment of the lowenergy transitions. They assign band A to two overlapping electronic transitions, 2 1A1 and 1 1B2, and band B to 2 1B2. No direct experimental evidence for the contribution of two overlapping transitions to band A was found, the assignment was based on correlation with transitions in dibenzofuran and by the results of PPP p-electron calculations [10]. As discussed above, we prefer the assignment of the three bands A; B; and C to the three electronic states 2 1 A1, 1 1B2, and 2 1B2. This assignment is consistent with the MCD data: Negative, positive, and positive MCD B-terms are observed for A; B; and C [10], in agreement with the results of our LCOAO [25] calculation that predicts B-terms equal to 2 2.8, þ 3.9, and þ 1.0 (in units of 1023 be Debye2/cm21) for the 2 1A1, 1 1B2 and 2 1B2 transitions, respectively (calculated by LCOAO at 32,700, 36,000, and 39,600 cm21). Towards higher wavenumbers, the MCD spectrum becomes more complex with strongly overlapping contributions [10]. The absorption band D is not

609

clearly resolved in the MCD spectrum. We believe that the maximum in the MCD curve close 41,000 cm21, where the negative MCD comes close to zero, may be due to transition D (negative MCD corresponds to positive B-term, and vice versa [19]). This implies that the MCD associated with D is positive (corresponding to a negative B-term), but overlapped by negative MCD contributions from the neighboring bands C and E: We conclude this discussion by a brief comparison with transitions for dibenzofuran [17] (Table 2). The transitions predicted by TD-B3LYP/6-31 þ G* for the two compounds are shown in Fig. 4. When passing from dibenzofuran to dibenzothiophene, the predicted transitions are generally red-shifted by 2,000 – 4,000 cm21, except for 1 1B2 that is hardly shifted at all. As a result, the two lowest states 2 1A1 and 1 1 B2, which are near-degenerate in dibenzofuran, are split by about 4,000 cm21 in dibenzothiophene. The predicted shifts are consistent with those observed for bands A and B; which lends further support to our assignment of the two transitions (these shifts, however, are poorly reproduced by a number of

Fig. 4. Graphical representation of electronic transitions in dibenzofuran and dibenzothiophene predicted with TD-B3LYP/6-31 þ G*.

610

J. Spanget-Larsen, E.W. Thulstrup / Journal of Molecular Structure 661-662 (2003) 603–610

semiempirical procedures [6,7,10]). For the third optically allowed state, 2 1B2, a marked decrease in intensity is predicted when moving from dibenzofuran to dibenzothiophene. The y-polarized band C can probably be assigned to this state, but comparison of the observed spectra in the region of band C seems to be complicated by the influence of strong vibronic coupling, particularly in dibenzothiophene, as mentioned above. The shape of band C is thus quite different in the two spectra (actually, band C of dibenzofuran [17] has a profile similar to that of band B of dibenzothiophene). A somewhat similar situation applies to the state 3 1A1, probably corresponding to the z-polarized band D: In both spectra, characterization of this band is complicated by overlap with the much stronger, y-polarized band E: The latter is the strongest band in both spectra and can be assigned to the 3 1B2 state. A red-shift of 3,200 cm21 and a significant increase in intensity is predicted for this state when passing from dibenzofuran to dibenzothiophene, in excellent agreement with the observed shifts of band E: The last two transitions observed in the present investigation of dibenzothiophene are F and G: The corresponding transitions in dibenzofuran are predicted to occur at higher wavenumbers and they were not clearly observed within the spectral region covered in the previous investigation of this compound [17].

Acknowledgements The authors are grateful to D. Liang, who helped measure the initial UV spectra of dibenzothiophene, and to Eva Karlsen for performing an IR LD investigation of the compound.

References [1] [2] [3] [4]

A. Bree, R. Zwarich, Spectrochim. Acta 27A (1971) 599. S.Y. Lee, J. Phys. Chem. A105 (2001) 8093. J. Cai, E.C. Lim, J. Chem. Phys. 97 (1992) 3892. H. Socsu´n, Y. Alvarado, J. Herna´ndez, P. Herna´ndez, R. Atencio, A. Hincliffe, J. Phys. Org. Chem. 14 (2001) 709. [5] G.M. Badger, B.J. Christie, J. Chem. Soc. (1956) 3438. [6] F. Momicchioli, A. Rastelli, J. Chem. Soc. B (1970) 1353. [7] G. Buemi, F. Zuccarello, G. Romeo, J. Mol. Struct. 94 (1983) 115.

[8] B. Rusˇcˇic´, B. Kovac´, L. Klasinc, H. Gu¨sten, Z. Naturforsch 33a (1978) 1006. [9] R. Gleiter, J. Spanget-Larsen, Top. Curr. Chem. 86 (1979) 139. [10] N. Igarashi, A. Tajiri, M. Hatano, Bull. Chem. Soc. Jpn 54 (1981) 1511. [11] E.W. Thulstrup, J. Michl, J. Am. Chem. Soc. 104 (1982) 5594. [12] J.G. Radziszewski, J. Michl, J. Am. Chem. Soc. 108 (1986) 3289. [13] F. Jensen, Introduction to Computational Chemistry, Wiley, Chichester, 1999. [14] W. Koch, M.C. Holthausen, A Chemist’s Guide to Density Functional Theory, Wiley-VCH, Weinheim, Germany, 2000. [15] J. Fabian, Theor. Chem. Acc. 106 (2001) 199. [16] J. Spanget-Larsen, Spectrochim. Acta A58 (2002) 2245. [17] J. Spanget-Larsen, D. Liang, E. Chen, E.W. Thulstrup, Asian Chem. Lett. 4 (2000) 121. [18] E.W. Thulstrup, J. Michl, Elementary Polarization Spectroscopy, VCH, New York, 1989. [19] J. Michl, E.W. Thulstrup, Spectroscopy with Polarized Light. Sample Alignment by Photoselection, in Liquid Crystal, Polymers, and Membranes; (paperback edition), VCH, New York, 1995. [20] J. Spanget-Larsen, R. Gleiter, R. Haider, Helv. Chim. Acta 66 (1983) 1441. [21] J. Spanget-Larsen, J. Uschmann, R. Gleiter, J. Phys. Chem. 94 (1990) 2334. [22] E.W. Thulstrup, J. Michl, J. Phys. Chem. 74 (1970) 3868. [23] F.R. Steenstrup, K. Christensen, C. Svane, E.W. Thulstrup, J. Mol. Struct. 408 (1997) 139. [24] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery, Jr, R.E. Stratmann, J.C. Burant, S. Dapprich, J.M. Millam, A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G.A. Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, P. Salvador, J.J. Dannenberg, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Ortiz, A.G. Baboul, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, J.L. Andres, C. Gonzalez, M. HeadGordon, E.S. Replogle, J.A. Pople, GAUSSIAN 98, Revision A.11, Gaussian, Inc., Pittsburgh PA, 2001. [25] J. Spanget-Larsen, Theor. Chem. Acc. 98 (1997) 137. [26] F. Madsen, I. Terpager, K. Olskær, J. Spanget-Larsen, Chem. Phys. 165 (1992) 351. [27] E. Karlsen, J. Spanget-Larsen, E.W. Thulstrup, unpublished results. [28] H.-H. Perkampus, I. Sandemann, C.J. Timmons (Hrsg.), DMS-UV-Atlas, Chemie, Butterworth, London, Weinheim, 1966–1971. Spektrum H15/1. [29] H.-H. Perkampus, I. Sandemann, C.J. Timmons (Hrsg.), DMSUV-Atlas, Verlag Chemie/Butterworth, London, Weinheim, 1966–1971. Spektrum H14/1.