You are on page 1of 17

Chemical Physics 288 (2003) 309325 www.elsevier.

com/locate/chemphys

Solvent inuence on absorption and uorescence spectra of merocyanine dyes: a theoretical and experimental study
I. Baraldi, G. Brancolini 1, F. Momicchioli *, G. Ponterini, D. Vanossi
Dipartimento di Chimica, Universita  di Modena e Reggio Emilia, Via Campi 183, I-41100 Modena, Italy Received 8 November 2002

Abstract The solvatonCS INDO model, previously successfully used to describe the solvatochromic properties of merocyanines, has been extended to the study of the solvent inuence on the uorescence spectra (uorosolvatochromism) of these dyes. A ketocyanine (M1) and a stilbazolium betaine (M2) were chosen as representatives of positively and negatively solvatochromic behaviours, respectively. The gap of experimental knowledge concerning the emission properties of M2 was lled by a spectrouorometric analysis in a set of solvents covering a large range of the ET 30 scale. Solvato- and uorosolvatochromism were described by calculating the S0 eq: ! S1 FranckCondon and S1 eq: ! S0 Franck Condon transition energies as a function of a polarity factor related to the static dielectric constant of the solvent, and ranging from 0 to 1. The absorbing S0 eq: and emitting S1 eq: units (solute molecule + solvent cage) were approximated using the S0 and S1 geometries of the unsolvated molecule and the respective charge distributions tted to the current value of k e. The calculation results fully conrm that S0 and S1 states of merocyanines can be viewed as a mixture of a neutral and a zwitterionic structure whose composition is controlled by the solvent polarity. The plots of the calculated spectral data (absorption and emission maxima and corresponding Stokes shifts) vs k e are in fairly good agreement with those of N the experimental data over almost the entire range of the normalized ET values, thus showing that specic solvent interactions are at least partly simulated within the solvatonCS INDO scheme. The methodological prerequisites for a correct prediction of solvatochromic shifts are recalled with reference to previous conicting theoretical interpretations. 2003 Published by Elsevier Science B.V.

1. Introduction We have recently shown [1,2] that solvent eects on both ground-state properties and absorption spectra of classic donoracceptor dyes, such as
* Corresponding author. Tel.: +39-59-2055081; fax: +39-59373543. E-mail address: momicchioli.fabio@unimo.it (F. Momicchioli). 1 Present address: Dipartimento di Chimica G.Ciamician, Universit a di Bologna, Via F.Selmi 2, I-40126 Bologna, Italy.

merocyanines, can be fairly well accounted for within the CS INDO scheme [3]. Briey, the solutesolvent interactions were described by the simple solvaton model [4] and were incorporated in the CS INDO Hamiltonian according to previous basically equivalent all valence electron SCF approaches [511]. Our procedure, however, is characterized by a peculiar modelling of the solvaton set representing the polarized solvent surrounding the solute. We followed the basic widely accepted idea that the electronic structure of

0301-0104/03/$ - see front matter 2003 Published by Elsevier Science B.V. doi:10.1016/S0301-0104(03)00046-6

310

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

merocyanines can be described at the p level in terms of resonance between neutral and chargeseparated forms (Fig. 1) and that the solvatochromic behaviour can be traced back to the relative weights of the two structures in the ground state and their change upon vertical transition to the electronic excited state [12,13]. Such a scheme stresses the key role of the p-electron distribution and suggests that an eective solvent eld must be rst of all capable of correctly controlling the drift of p-electrons from the donor R2 N to the acceptor (CO) group. A very elementary way to simulate a polarized environment is to position one or more point charges at one or either end of the chromophore system, as some authors did by the middle of the 1990s [14,15]. In a more realistic way, following the solvaton model [4] we associated with each atom of the conjugated system, with p net charge Qp , a ctive particle with charge Qp interacting with all electron and core charges of the solute according to Borns law. The so-dened solvaton set reects the composition of the resonance hybrid (Fig. 1), depending on the nature of the donor and acceptor groups, and hence may eectively account for the p-electron redistribution induced by the solvent polarity. Using such a solvaton set within the CS INDO CI scheme [1,2], we were able to provide a satisfactory description of both the positive solvatochromism of two vinylogous streptomerocyanines (Fig. 1, n 2; 4) and the large negative solvatochromism of stilbazolium betaine [13] (Fig. 2, M2). Reasonable structural variations with solvent polarity, related to variations of the resonance hybrid composition, were also predicted. The same twofold problem had previously been addressed by Albert et al. [16] using the self-consistent reaction eld (SCRF) model within the INDO method, but no choice of the cavity-size parameter had yielded a reasonable prediction for the two opposite solvatochromic behaviours (for more details see [1,2]). To our knowledge, the two solvatochromic

Fig. 2. Investigated compounds. M1: 1,9-di-(N-phenyl-N-methyl)-4,6,dimethylene-nona-1,3,6,8-tetraen-5-one; M2: 40 -hydroxy-1-methylstilbazolium betaine.

Fig. 1. Neutral and charge-separated mesomeric structures of simple streptopolymethine merocyanines.

trends were qualitatively well reproduced only by Klamt [17] using the AM1/COSMO method. Other theoretical studies introducing the solvent dielectric eld through either a set of point charges [14,15] or the virtual charge model [10] dealt only with stilbazolium betaine which has attracted great attention in relation to its uncommonly large negative solvatochromism and the much-discussed solvatochromic reversal at low medium polarity [18]. Independently of the specic (continuum) solvent model, the majority of the cited theoretical studies [1,2,10,15,17] reproduced qualitatively the negative solvatochromism of M2. On the other hand, Morley [14] predicted the opposite trend combining AM1 structure optimization in the presence of the solvent and gas phase CNDOVS calculation of the transition energy for the solventdistorted molecular geometry. This result was interpreted by Morley as evidence that the zwitterionic (benzenoid) form, obtained in the polar medium, absorbs at the red of the neutral (quinonoid) form prevailing in non-polar medium. Such interpretation, however, is questionable since, no matter what geometry is used, the electronic structure yielded by an MO calculation corresponds to a mixture of VB structures. The zwitterionic form exists only in the presence of a polar medium and its electronic spectrum can be calculated only using the solvent-polarized MOs. As a matter of fact, all calculations complying with this condition [1,2,10,15,17] predicted solvato-

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

311

chromic shifts in qualitative agreement with experiment. 2 Qualitatively equivalent results were obtained by Benson and Murrell [19] within an SCF p-electron treatment where the eect of solvent was simulated by suitable choice of one-centre core and electron repulsion integrals for the nitrogen and oxygen atoms. From the above brief survey, our solvatonCS INDO scheme appears to have two main advantages: (i) both ground and excited-state properties are calculated for the solute molecule embedded in the solvent, (ii) the use of a solvaton set reecting the net p-electron charges enables the solvent interaction to be modelled in keeping with the VB description of the electronic structure of the solute. A disadvantage, shared by all continuum models, is that it formally leaves out specic solvent interactions. In principle, such eects can be accounted for only within semicontinuum type theories [20,21] or fully discrete type approaches as, for example, those based on statistical mechanics techniques [2224]. However, our very simple scheme, where the solvent interaction is introduced essentially through a variation of the diagonal elements of the Fock matrix corresponding to the AOs of the p system, may implicitly account for specic solvent interactions as rst argued by Benson and Murrell [19]. In the present work, the solvatonCS INDO scheme was subjected to further validation by studying the solvent eects on the uorescence spectra (uorosolvatochromism) 3 which have to date received relatively little attention from a theoretical point of view. As test compounds we chose two merocyanines, a ketocyanine dye (M1) and stilbazolium betaine (M2) (Fig. 2), and the solventinduced spectral shifts of both absorption and uorescence emission were investigated. Abundant experimental data concerning the absorption specA calculation procedure like the Morley one was applied in [16] where INDOSCRF optimized geometries were used in gas phase INDO/S type calculations of the spectra. This may explain the positive solvatochromism erroneously predicted for M2 in [16] when using physically reliable values of the cavity radius. 3 Hereafter, the solvent dependence of the position of the absorption and uorescence bands will be termed solvatochromism and uorosolvatochromism, respectively [25].
2

tra of these dyes in solvents of dierent polarities are available ([25], and references cited therein). On the other hand, to our knowledge solvent eects on the uorescence spectra have been reported only for M1 [26,27]. Thus, we rst of all carried out an experimental exploration of the absorption and emission solvatochromism of M2. In summary, the experimental data as a whole show that: (i) M1 exhibits strong positive solvatochromism in both absorption and emission [2527], (ii) M2, the absorption of which is characterized by one of the strongest negative solvatochromisms ever observed, features a markedly weaker negative uorosolvatochromism. The theoretical interpretation of the entire body of experimental evidence was undertaken by applying the solvatonCS INDO method within a usual scheme where the uorescence emission takes place from the equilibrium geometry of the lowest excited singlet state reached very quickly after vertical S0 ! S1 excitation of the equilibrium ground state. In principle, this should require geometry optimization of the solvated solute molecule in both the ground state S0 and the emitting S1 (pH p L; H HOMO, L LUMO) state. In practice, we simply used S0 and S1 geometries optimized in the gas-phase approximation and calculated the S0 eq: ! S1 FranckCondon and S1 eq: ! S0 Franck Condon transition energies as functions of the solvent polarity using solvaton sets reecting the p net charges of S0 eq: and S1 eq:, respectively. The calculation results will rst be thoroughly analysed by reference to the basic characteristics of the theoretical model and will then be subjected to a detailed comparison with experimental observations in solvents covering the whole scale of solvation power. It will be shown that both the solvatochromic and uorosolvatochromic behaviours of M1 and M2 are qualitatively well described within the solvatonCS INDO scheme.

2. Experimental investigation on dye M2 2.1. Materials, instrumentation and details of experiments M2 (4-[(1-methyl-4(1H)-pyridinylidene)ethylidene]-2,5-cyclohexadien-1-one) was purchased from

312

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

Aldrich and was used as received. All solvents (Merck and Lab-Scan) were of spectroscopic grade and were dehydrated with activated molecular sieves before use. The polar and hygroscopic ones were treated with solid KOH so as to dissolve the M2 betaine in the unprotonated form. For the same reason, measurements in water were carried out in 102 M NaOH. Absorption spectra were recorded on a Perkin Elmer k15 spectrophotometer, while a Spex-Jobin Yvon Fluoromax 2 spectrouorometer was employed for the uorescence measurements. All experiments were carried out at room temperature (1821 C). Maximum optical densities were, typically, between 0.1 and 1 (corresponding to sample concentrations from 3 106 to 3 105 mol dm3 ) for absorption measurements, and around 0.10.15 for uorescence measurements. In the latter, each sample was excited at, at least, two dierent wavelengths on the high-energy side of the absorption band. Emission and excitation spectra were corrected for the instrumental spectral response. Due to the very weak uorescence emission of M2, especially in low-polarity solvents, wide monochromator slits were used (68 nm spectral resolution) to improve the signalto-noise ratio. Fluorescence quantum yields UF were determined in methanol and water with respect to cresyl violet in methanol (UF 0:65 [28]) and eosin in methanol (UF 0:60 [29]) according to the usual expression: UF UF;r A=Ar n2 =n2 r ODr =OD, where r refers to the reference, A are the areas of the corrected emission bands, n are the solvent refractive indexes and the optical-density (OD) ratios at the excitation wavelengths were adjusted to unity. Inner lter eects were deemed negligible on both emission spectra and uorescence quantum yields because of the optical thinness of the samples employed and of the low to very low spectral overlap between absorption and emission. 2.2. Results and discussion The choice of test merocyanine dyes suitable to check thoroughly the capability of the solvaton CS INDO method of accounting for solvent eects on both absorption and uorescence spectra was

not a trivial aair. First, in compliance with the previous studies limited to eects on the absorption spectra [1,2], we needed two compounds characterized by opposite solvatochromisms. In [1,2] simple streptomerocyanines (Fig. 1) and stilbazolium betaine (M2 in Fig. 2) were taken as typical dyes with strong positive and negative solvatochromism, respectively. In both cases the intense colour band is due to the lowest p ! p (essentially pH ! p L ) transition and the observed solvent shifts are not aected by the presence of any forbidden np state at low energies. On the other hand, in order that the study may be extended to uorosolvatochromism the test compounds should be characterized by an eciently emitting lowest-lying pH p L state. Previous theoretical and photophysical studies [3032] have pointed out that in general streptomerocyanine derivatives have no such characteristics because of the presence of lowest-lying np states and the occurrence of fast transcis isomerization following direct S0 ! S1 pH p L excitation. Thus, in the present work we replaced the streptomerocyanines (Fig. 1) with the ketocyanine M1 (Fig. 2) which is known to be an ecient uorophore exhibiting large red shifts in both absorption and emission spectra upon increasing the solvent polarity [25]. In this case, the theoretical study (Section 3) was based on the abundant experimental data reported in the literature [2527]. The suitability of stilbazolium betaine M2 as test compound for negative solvatochromism in both absorption and uorescence required on the contrary further experimental investigation. As is well known, dye M2 exhibits a very large negative solvatochromism in absorption ($6500 cm1 on going from CHCl3 to H2 O solution) [25]. However, the absorption spectrum of M2 shows a change in shape on moving from low-polarity solvents, where a pronounced structure is observed, to highly polar and protic solvents, where the structure is blurred out. For merocyanines closely related to M2, the structured spectra observed in dierent solvents of relatively low polarity were attributed to the presence in solution of two variously identied species ([33], and references cited therein). Similar observations had led other authors to question the reliability of any

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

313

analysis of the solvatochromism of these dyes [34]. In order to check the importance of any extra contribution to the absorption of M2 in solvents of low polarity, we carried out a spectrouorometric analysis in a 1:1 dichloromethanetoluene mixture. The emission and excitation spectra showed only a very modest dependence on, respectively, the excitation and monitoring wavelength. Representative spectra are shown in Fig. 3. A close correspondence was observed between the absorption and excitation spectra, apart from a weak solvent Raman band (around 560 nm in Fig. 3) ca. 3000 cm1 from the monitoring wavenumber, and some blurring of the vibronic structure possibly due to poor instrumental spectral resolution (8 nm). We conclude that the absorption and emission spectra of M2 in low-polarity solvents are due to essentially one and the same species. Moreover, the rather small Stokes shift and the mirror image relationship between the absorption and emission peaks suggest that uorescence emission takes place from the local minimum of the initially excited 1 pH p L state. This is in keeping with the recently reported femtosecond dynamics of dye M2 [35]. According to [35], direct S0 ! S1 excita-

tion of unprotonated stilbazolium betaine M2 does not result in transcis isomerization but it is followed by fast thermally activated relaxation (1.1 ps) to a non-emitting conformational intermediate. Thus, although emission is predicted to occur from the local minimum of the S1 pH p L state, the uorescence quantum yield should be very small. As a matter of fact, we found the uorescence quantum yield of M2 to be only 2:0 103 in methanol, 1:5 103 in water and even lower values were measured in solvents of low polarity. From the foregoing considerations, it ensues that M2, even though weakly uorescent, may be used to investigate solvatochromism in both absorption and emission. For this purpose, the absorption and uorescence emission maxima of M2 were measured in nine solvents with ET 30 solvent polarity parameter [25] ranging from 39.1 (chloroform) to 63.1 (water), and the results are reported in Table 1. Both absorption and emission spectra exhibit a negative solvatochromism, and shift quite regularly to the blue with increasing the value of ET 30 which represents the overall solvation power of the solvents. On the other hand, the static dielectric constant e, reecting only

Fig. 3. Absorption (full line), uorescence excitation (dashed line, kem 660 nm) and emission (dotted line, kexc 580 nm) spectra of M2 in a 1:1 dichloromethanetoluene mixture.

314

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

Table 1 Absorption and emission maxima of M2 in solvents of dierent characteristics Solvent Chloroform Dichloromethane Acetone DMFe DMSOf Acetonitrile 2-Propanol Methanol Water Dmmax cm1
a b

ET 30a (kcal/mol) 39.1 40.7 42.2 43.2 45.1 45.6 50.7 55.4 63.1

N ET

ec 4.8 8.9 20.7 37.0 46.7 37.5 19.9 32.7 78.4

Absorptiond , mmax cm1 16,310 16,470 17,040 17,150 17,390 17,590 18,450 20,700 22,570 )6260

Fluorescenced , mmax cm1 15,850 15,820 16,130 16,140 16,160 16,260 16,700 17,120 17,190 )1340

0.259 0.309 0.355 0.386 0.444 0.460 0.546 0.762 1.000

Solvent polarity parameter of Dimroth and Reichardt measured at 25 C and 1 bar [25]. N Normalized ET value. c Dielectric constant at room temperature (20 or 25 C) [13]. d Typical uncertainties: 30 cm1 on absorption, 60 cm1 on emission. e Dimethylformamide. f Dimethyl sulfoxide.

non-specic solvent eects, fails to account for the quite dierent band shifts induced by aprotic and protic solvents of comparable polarity. The maximum solvatochromic shift found for the absorption, Dmabs max mchloroform mwater 6260 cm1 , is in agreement with data of other authors ([25], and references cited therein). Much smaller hypsochromic shifts were observed for 1 emission, Dmflu max 1340 cm . We note that a qualitatively similar, yet less pronounced, solvent dependence of the absorption and uorescence spectra has recently been reported for another negatively solvatochromic merocyanine (MC 540) [36].

3. Theoretical investigation 3.1. Method and calculation details Incorporation of the solvaton model into the CS INDO method has already been described in detail [1,2]. Here, we only summarize the main points concerning the calculation of the solvatochromic shifts. Briey, the Fock matrix general element is expressed as
s 0 Flm Flm dlm k e N X B0

QB0 cAB0

l 2 A;

0 is the CS INDO matrix where the rst term Flm element in the absence of solvent [3], while the second term accounts for the electronsolvaton interaction eects. In particular, k e is a solvent polarity factor related to the static dielectric constant e. In keeping with Costanciel and Tapias virtual charge model the polarity factor p [9], where p is set equal to e 1= e, we assumed k e to range from 0 (when e is 1) to 1 (for e ! 1). QB0 is the charge of the solvaton associated with atom B and cAB0 is taken equal to the electron repulsion integral cAB . As pointed out in Section 1 (see also [1,2]), solvatons were only associated with the atoms contributing to the p system (C, N, O) and their charges were taken equal to the negative of the respective net p-electron charges. The determination of the solvent eects on the position of the rst absorption bands required the HartreeFock equations incorporating solute solvaton interactions to be resolved for the equilibrium ground-state geometries and the corresponding vertical S0 ! S1 pH p L transitions to be calculated at dierent values of k e. As suggested by the experimental observations (Section 2.2) uorescence emission of both M1 and M2 dyes takes place from the excited S1 pH p L state after geometrical relaxation of the solute and reorganization of the solvent molecules in the solvation shell. According to this scheme, the uorescence

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

315

transitions were calculated at dierent k e values using geometries and solvaton charges optimized for the excited state. For the sake of simplicity, in the present work geometrical distortions induced by the solvaton eld on the solute molecule were disregarded. In other words, we simply used geometries optimized in the absence of solvent k e 0. This simplication appears to be acceptable, since solvatochromic shifts are mainly determined by polarization and electrostatic eects that are not very sensitive to small geometry modications. However, the solvaton charges corresponding to the equilibrium geometries of the ground and the excited states were adjusted by an iterative procedure, i.e., by performing a sequence of SCF (for S0 ) and SCF CI (for S1 ) calculations until convergence of the QB0 values. To summarize, we carried out the following set of calculations: (i) The S0 -state equilibrium geometries of both dyes were calculated by minimizing the total CS INDO SCF energy of the isolated molecule, as a function of the internal coordinates of the chromophoric group. With these geometries, SCF calculations were carried out, for dierent k e values, using solvaton charges corresponding to the S0 net p-electron charges. The MOs produced at each k e value by the iteratively adjusted solvaton set were then used for the CI calculation of the absorption spectrum, thus determining the absorption spectral shifts. (ii) An analogous procedure was carried out to calculate uorosolvatochromic shifts. First, for each molecule, the equilibrium geometry of the S1 pH p L state of the isolated molecule was calculated by minimizing the excited-state energy ES1 ES0 DES0 S1 derived from the CI calculation as a function of the internal coordinates of the chromophoric group. The so obtained geometries were used to generate, at dierent k e values, optimized solvaton sets corresponding to the excited-state net p-electron charges and the MOs to be used in the CI calculation of the emission spectra and solvent-induced uorescence shifts. Since the excited state responsible for both the absorption colour band and the uorescence emission, S1 pH p L , is fully represented by the

singly excited 1 UpH p L conguration, we limited ourselves to standard S-CI calculations. The MO active spaces included all p and p molecular orbitals of the chromophoric groups and, for M1, of the benzene rings. In view of the model character of this study, the solvaton charges suitable for the equilibrium excited state were simply derived from the net p-electron charges of the 1 UpH p L conguration. The geometry optimizations were carried out introducing in our CS INDO CI programs an optimization code where the gradient is calculated numerically within a FletcherPowell-type optimization algorithm [37]. The way the gradient is computed makes the program easily adaptable for calculations of excited-state geometries. The parameters peculiar to the CS INDO method [3] were chosen as follows: (i) the screening constants, klm , where the indexes refer to two hybrid atomic orbitals of the CS INDO basis set, were given the values: krr 1, kpp 0:50, krp 0:65, knp 0:60, knr 0:72, knn 0:68; (ii)  the atomic pair parameters, aAB (a.u.) and R0 AB (A), entering the calculation of the core repulsion energy, were assigned the values aCC 1:35, 0 aCN 1:40, aCO 1:20, R0 CC 1:7, RCO 2:10. The remaining pair parameters were given the usual values, aAB 1:50 and R0 AB 2:00. Finally, two-electron repulsion integrals were calculated according to OhnoKlopman [38]. 3.2. Results and discussion 3.2.1. Ground- and excited-state equilibrium geometries After exhaustive test calculations on ground and excited state geometries of prototypic systems, not reported here for conciseness [39], the CS INDO based approach described in the Section 3.1 was applied to M1 and M2 in both S0 and S1 states. The optimized bond lengths are reported in Table 2. The structural dierences between the two dyes in the ground state and the geometry changes induced by the S0 ! S1 excitation can be illustrated by analysing the extent of bond length alternation (BLA) in the conjugated bridge connecting the donor (amino) and the acceptor (carbonyl) groups, which reects the relative

316

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

Table 2 Bond lengths and bond length alternation (BLA) parameters for the ground state S0 and the rst pp excited state S1 of dyes M1 and M2 Bonds M1 S0 a b c d e f g h i l BLA 1.374 1.398 1.446 1.398 1.458 1.245 S1 1.368 1.411 1.430 1.415 1.450 1.253 M2 S0 1.365 1.396 1.447 1.419 1.436 1.418 1.455 1.401 1.450 1.250 0.038b 0.017c S1 1.362 1.402 1.444 1.427 1.421 1.430 1.447 1.406 1.447 1.254 0.024b )0.008c

0.055a

0.027a

For labelling of the bonds, see Fig. 2. BLA M1 c e b d=2. b BLA M2 c e g i b d f h=4. c BLA M2 e d f =2.
a

weights of the neutral polyenic and zwitterionic structures in the VB resonance hybrid (Fig. 1). All such characteristics are well expressed by the value of the so called BLA parameter (last row of Table 2), dened as the dierence between the average length of the single bonds and that of the double bonds [16]. The single and double bonds were identied by reference to the neutral VB structures (see Fig. 2), so high positive (negative) value of the BLA parameter indicates predominance of the neutral (zwitterionic) mesomeric structure. A vanishing BLA parameter, on the other hand, indicates a cyanine-like structure characterized by similar weights of the two forms. On this basis, the calculation results of Table 2 indicate that: (i) in the ground state, the chargeseparated form is more important for M2 than for M1 BLA M1; S0 > BLA M2; S0 , (ii) in both merocyanines bond-length alternation decreases upon electronic excitation, i.e., the contribution of

the neutral VB structure becomes lower on moving from S0 to S1 BLA M1; S0 > BLA M1; S1 ; BLA M2; S0 > BLA M2; S1 . The dierences between the two dyes become more evident if the BLA parameters of M2 are calculated in terms of the bonds (d, e, f) of the central polymethinic fragment. As a matter of fact, in this case BLA is already rather small in the ground state (0.017) and changes sign in S1 , thus reecting an opposite bond alternation with respect to S0 . The goodness of our theoretical predictions should now be veried by comparing the calculated ground-state geometries of Table 2 with the experimentally determined structures as well as the results of previous theoretical studies. However, as far as we know, structure determinations and calculations have been reported only for dye M2. From the crystal structure determination of M2 trihydrate [40] the molecule was found to be almost planar and to have a zwitterionic (benzenoid) structure with an essentially double CC central bond (1.346 ) connected to the six-membered rings by bonds A , and a CO bond (1.304 A ) rather of $1.44 A ). As is longer than a double C@O bond (1.215 A evident, such structure is at variance with the CS INDO optimized geometry of the isolated molecule (Table 2) which corresponds to a mixture of the two VB forms, with a slight prevalence of the neutral (quinonoid) one (Fig. 4). This is not surprising since in the crystal the zwitterionic form is imposed by the formation of hydrogen bonds between the carbonyl end of the chromophore and the crystal water. Thus, the crystal structure is assimilable to that of M2 in aqueous solution. The CS INDO optimized geometry of M2 is in qualitative agreement with previous theoretical studies [10,14,16,19] in that all calculations predict the quinonoid form to be (more or less) prevailing in the isolated molecule. However, with reference to the central polymethine fragment (e, d, f CC bonds), our geometry exhibits the smallest BLA

Fig. 4. Neutral (quinonoid) and zwitterionic (benzenoid) forms of dye M2.

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

317

value (present work: 0.017; [19]: 0.030; [16]: 0.038; [14]: 0.058; [10]: 0.077). In other words, our description reects a resonance hybrid with similar weights of the two limiting structures, while the other theoretical predictions favour more decidedly the quinonoid structure. We do not deny that our calculation may have underestimated the contribution of the quinonoid form, but we can say that this result is consistent with the observed solvatochromic behaviour of M2, i.e., an initial small red shift followed by a large blue shift of the rst absorption band on going from low polar to highly polar solvents [1,2,13,18]. 3.2.2. Absorption and uorescence solvatochromism Using the S0 and S1 optimized geometries, the vertical S0 eq: ! S1 and S1 eq: ! S0 transition energies of M1 and M2 were calculated in a medium of changing polarity (k e 00.9) by the CS INDO CI procedure described in Section 3.1. The results DE are collected in Tables 36, together with the oscillator strengths of the transitions (f), the electric dipole moments (l), and the net charges on the nitrogen and oxygen atoms in the S0 QN;O and S1 Q N;O states. First of all, let us point out some essential characteristics of the theoretical description. Tables 3 and 4 show that on going from k e 0 (gas phase) to k e 0:9 (highly polar medium), both the absorption and uorescence maxima of M1 are predicted to undergo marked bathochromic shifts (DmS0 ! S1 3550 cm1 and DmS1 ! S0 5000 cm1 ) 4, while the oscillator strengths are found to be both quite high (>2) and not very sensitive to solvent polarity. The bathochromic shifts are related to the increase of the solute dipole moment on the S0 ! S1 transition, resulting in a solvent-induced stabilization of the excited state relative to the ground state, which increases with increasing solvent polarity. The polarization eects induced by the solvaton eld on the electronic structure of the solute result in a quite regular growing of both lS0 and lS1 on going from k e 0 to 0.9. Such growing is little higher when using the equilibrium geometry of the excited state (Table 4), but Dl lS1 lS0 is predicted to
4

Dm mnon-polar solvent mpolar solvent.

be almost the same in absorption (Table 3) and emission (Table 4) at all values of k e. However, it is worth noting that the permanent dipole moment of M1 is directed along the twofold axis containing the C@O bond (see Fig. 2), while the S0 S1 transition is polarized perpendicularly to the twofold axis 1 B, i.e., is associated to charge transfer along the longitudinal molecular axis. Thus, the dipolemoment change occurring on the S0 S1 transition is not as eective a parameter of the solvatochromic eect as in the case of the parent streptopentamethinemerocyanine [1,2] where both permanent and transition dipole moments are aligned with the chromophoric chain. As a matter of fact, the S0 and S1 dipole moments of the ketocyanine M1 (and their increase with increasing k e) turn out to be decidedly smaller than those of its pentamethinemerocyanine moiety, while the contrary occurs for oscillator strength and solvent shift of the S0 ! S1 transition (see [2]). For a more precise comparison, we carried out CS INDO SCI test calculations for pentamethinemerocyanine (Fig. 1; n 2, R Me) and its N-phenyl derivative in the absence of solvent. The calculation results (Table 7) indicate that the phenyl substitution produces a small red shift and intensity increase of the transition, while lS0 and lS1 remain almost unchanged and substantially higher than those of the ketocyanine (Table 3, rst row). The absorption data in dichloromethane, also reported in Table 7, corroborate the theoretical prediction even if transition energies and spectral shift are a little overestimated because of the adopted simplications (essentially, singly excited CI and coplanarity of the phenyl ring). A more correct correlation between the solvatochromic behaviour of ketocyanine M1 and that of its parent chromophore emerges by changing from the dipole moments (i.e., global electrical properties) to the atomic net charges reecting the local charge distributions. As a matter of fact, Table 3 shows that the net charges on the nitrogen and oxygen atoms in the ground state, as well as their changes due to an increase of k e and/or S0 ! S1 excitation, closely resemble those previously found for streptopentamethinemerocyanine [2]. In particular, in S0 an increase of k e results in a drift of electrons from the two nitrogen atoms (donors) to the

318

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

Table 3 Calculated S0 ! S1 absorption properties of M1 as functions of the solvent polarity factor k e: vertical transition energy DE, os cillator strength (f), ground and excited-state dipole moments lS0 , lS1 and net charges on nitrogen QN ; Q N and oxygen QO ; QO atoms k e 0.0 0.2 0.4 0.6 0.8 0.9 DE (eV) 3.258 3.163 3.060 2.952 2.855 2.818 f 2.378 2.367 2.353 2.321 2.263 2.223 lS0 D 5.59 6.11 6.64 7.10 7.55 7.79 lS1 D 8.03 8.65 9.27 9.87 10.47 10.78 QN )0.13 )0.10 )0.06 )0.01 +0.07 +0.12 QO )0.72 )0.81 )0.91 )1.01 )1.09 )1.13 Q N )0.03 0.00 +0.04 +0.09 +0.16 +0.21 Q O )0.77 )0.86 )0.94 )1.03 )1.10 )1.14

Table 4 Calculated S1 ! S0 emission properties of M1 as functions of the solvent polarity factor k e (see caption of Table 3 for the meaning of symbols) k e 0.0 0.2 0.4 0.6 0.8 0.9 DE (eV) 3.118 2.968 2.823 2.686 2.558 2.498 f 2.411 2.384 2.362 2.345 2.331 2.326 lS0 (D) 5.67 6.44 7.17 7.86 8.53 8.86 lS1 (D) 8.10 8.99 9.81 10.58 11.30 11.64 QN )0.12 )0.08 )0.04 0.00 +0.05 +0.07 QO )0.73 )0.83 )0.91 )0.98 )1.05 )1.08 Q N )0.03 +0.01 +0.05 +0.10 +0.14 +0.16 Q O )0.78 )0.87 )0.95 )1.01 )1.07 )1.10

Table 5 Calculated S0 ! S1 absorption properties of M2 as functions of the solvent polarity factor k e (see caption of Table 3 for the meaning of symbols) k e 0.0 0.2 0.4 0.6 0.8 0.9 DE (eV) 2.782 2.766 2.790 2.944 3.184 3.291 f 2.075 1.904 1.702 1.425 1.234 1.163 lS0 (D) 17.92 22.09 26.88 33.80 39.91 43.12 lS1 (D) 18.36 21.30 24.73 30.58 37.54 41.81 QN )0.04 +0.02 +0.09 +0.21 +0.35 +0.45 QO )0.73 )0.84 )0.94 )1.04 )1.13 )1.16 Q N +0.01 +0.05 +0.11 +0.22 +0.37 +0.47 Q O )0.71 )0.81 )0.91 )1.02 )1.11 )1.15

Table 6 Calculated S1 ! S0 emission properties of M2 as functions of the solvent polarity factor k e (see caption of Table 3 for the meaning of symbols) k e 0.0 0.2 0.4 0.6 0.8 0.9 DE (eV) 2.732 2.725 2.724 2.732 2.750 2.763 f 2.227 2.126 2.007 1.884 1.757 1.699 lS0 (D) 19.37 21.67 24.49 27.56 30.91 32.53 lS1 (D) 17.56 19.51 21.79 24.27 26.98 28.33 QN )0.02 +0.02 +0.08 +0.15 +0.23 +0.28 QO )0.74 )0.83 )0.92 )1.02 )1.11 )1.15 Q N +0.01 +0.05 +0.10 +0.16 +0.23 +0.28 Q O )0.70 )0.79 )0.89 )1.00 )1.09 )1.13

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

319

Table 7 Calculated and experimental properties of the S0 ! S1 transition of the streptopentamethinemerocyanine and its N-phenyl derivative Molecule Calculateda DE (eV) Me2 NCH@CH2 CHO PhNMeCH@CH2 CHOc
a b

lS0 (D)

lS1 (D)

Experimentalb DE (eV)

103 emax

4.087 3.797

1.211 1.360

8.64 8.49

13.43 13.17

3.430 3.337

51 65

At k e 0. In dichloromethane, [41]. c The phenyl ring was assumed to be coplanar with the chomophore chain and to have the benzene structure.

oxygen atom (acceptor), traceable to an increased weight of the two equivalent zwitterionic forms

the absorption maximum of M2 with increasing k e, is accompanied by a substantial decrease of

giving rise to a nonamethinecyanine-like electronic structure relative to the neutral form (M1 in Fig. 1). Moreover, the comparison between Q N, Q and Q , Q provides an estimate of the real N O O charge transfer occurring during the S0 ! S1 transition at each k e. Table 4 shows that, inverting the direction of the charge transfer, similar considerations apply to the transition from the relaxed excited state to the FranckCondon ground state, S1 eq: ! S0 . In conclusion, the ketocyanine M1 provides an example of a system where the solvatochromic eects may be described within the solvaton model better than by using SCRF models adopting the Onsager dipolar approximation [16]. Merocyanine M2 is predicted to have a very dierent solvatochromic behaviour both in absorption and in emission (Tables 5 and 6). In going from k e 0 to 0.9, the absorption maximum is found to undergo a small bathochromic shift (until k e 0:2) followed by a large hypsochromic shift leading to a global strongly negative solvatochromism Dm 4100 cm1 (Table 5). This behaviour follows that found in our previous theoretical study [1,2], even if the transition energies and the solvatochromic shift turn out to be somewhat reduced because of some dierences in the calculation procedure concerning the parametrization (kpp 0:5 instead of 0.55) and the use of xed geometries optimized at k e 0. The blue-shift of

the oscillator strength attributable to the increasing importance of the zwitterionic form (Fig. 4). In this case, where permanent and transition dipole moments are both parallel to the longitudinal molecular axis, the S0 and S1 dipole moments should be fully representative of the solvatochromic behaviour. Briey, Table 5 shows that lS0 is quite high at k e 0, due to a substantial contribution of the zwitterionic VB structure, and grows rapidly with increasing k e owing to growing solvent-induced polarization. At k e 0, lS1 is slightly higher than lS0 but, from k e 0:2 on, the FranckCondon S1 state is 23 D less dipolar than S0 . Thus, both the small red shift between k e 0 and 0.2, and the successive strong blue shift can be rationalized in terms of net stabilization of S1 or S0 induced simply by dipolar solutesolvent interactions, as expected for a polar solute in a polar solvent [20,25]. Similar considerations can be made by analysing the net atomic charges. Table 5 shows, in particular, that in both S0 and S1 an increase in solvent polarity gives rise to an increase in electron population on the oxygen atom, while the vertical S0 ! S1 transition results in a reduction of such electron-charge accumulation. The latter phenomenon, which is the opposite of that found with M1 (see Table 3), suggests that S0 ! S1 excitation produces an enrichment of the resonance hybrid in the quinonoid form. The same set of calculations performed for

320

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

the vertical S1 eq: ! S0 transition, i.e., using geometry and solvaton charges of the relaxed excited state, led to similar trends of the dipole moments and net charges (Table 6). Two points, however, have to be noted: (i) lS1 lS0 is negative over the whole range of k e, (ii) the increase of lS1 and lS0 with increasing k e is less marked than that found for the vertical S0 eq: ! S1 transition (Table 5). Such trend, which is contrary to that exhibited by M1 (Tables 3 and 4) (where lS1 > lS0 ), is related to the fact that the solvaton eld acting on the equilibrium excited state is weak compared with that acting on the equilibrium ground state. As a consequence, the S1 eq: ! S0 transition energy turns out to be little aected by solvent polarity, with a global negative solvatochromism of just 250 cm1 ()0.031 eV) (Table 6). The above discussion has shown how the adopted theoretical scheme is reected in the calculation results. Now, these results have to be compared with the observed solvato- and uorosolvatochromic behaviours of M1 and M2. In order to make such comparison as concise and comprehensive as possible, we have plotted in a same diagram the calculated and experimental band frequencies (in cm1 ) using, respectively, the N k e and ET [25] scales, both ranging from 0.0 to 1.0. This way, we obtained four diagrams corresponding to the absorption and uorescence properties of M1 (Figs. 5 and 6) and M2 (Figs. 7 and 8). Let us comment rst on Figs. 5 and 6. As is evident, the calculated mmax values are overestimated with respect to the experimental values. However, the deviation is nearly the same for absorption (Fig. 5) and emission (Fig. 6), and keeps almost constant over the entire range of solN vent polarity ($0.5 eV at k e, ET 0:1 and $0.45 N eV at k e, ET 0:9). Thus, apart from the systematic blue-shift of the calculated S0 eq: ! S1 and S1 eq: ! S0 transitions (essentially due to the adopted CI truncation), our simple solvatonCS INDO scheme describes fairly well the absorption and uorescence properties of ketocyanine M1 as well as their dependence on the solvent characteristics. As a matter of fact, the calculated Dmmax values mk 0 mk 0:9 for absorption 3550 cm1 and emission 5000 cm1 are in good qualitative agreement with the experimental

Fig. 5. Wavenumber of the calculated and experimental absorption maximum of M1 vs, respectively, the k e and norN malized ET values of solvent polarity. The experimental mmax values in eight solvents (toluene, THF, acetone, DMF, 2-proN panol, ethanol, methanol and water) with ET ranging from 0.090 to 1.000, are reported from [26,27].

Fig. 6. Wavenumber of the calculated and experimental emission maximum of M1 vs, respectively, the k e and normalized N ET values of solvent polarity. The experimental mmax values are reported from [26,27] (see caption of Fig. 5 for the solvent characteristics).

N N 0:1 mET 1:0 ([26]: 4010, ones mET 1 3980 cm ; [27]: 3600, 3880 cm1 ). Interestingly, the agreement becomes even better if the comparison is made over the same interval (0.10.9) of solvent-polarity parameter. In this case, the solvatochromic shifts, derived by simple linear interpolations, were found to be: Dmabs =cm1 3170 calcd:, 3260 [26], 2800 [27], Dmfluo 4350

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

321

Fig. 7. Wavenumber of the calculated and experimental abN sorption maximum of M2 vs, respectively, the k e and ET values of solvent polarity. The experimental data in 25 solvents N were taken from [18]. The mmax covering the entire range of ET values of Table 1 are also reported.

are aligned almost parallely to the theoretical plots (Figs. 5 and 6). The same thing emerges from the fluo plots of the Stokes shifts (mabs max mmax Dmabs fluo) vs the normalized solvent-polarity parameters (Fig. 9). In spite of the large uctuations of the experimental data, it is evident that calculated and observed Stokes shifts of M1 exhibit similar plots N until ET 0:7 . Beyond this value, the theoretical plot goes on increasing, while the experimental ones rst bend and then undergo a dramatic drop down N in water (ET 1:0). We can conclude that: (i) due to the water peculiarity, any discussion considering only the overall spectral shifts between water N N (ET 1:0) and toluene (ET 0:1) may lead to N wrong interpretations, (ii) except for ET P 0:7, in M1 solvato- and uorosolvatochromic shifts seem to be to a great extent determined by non-specifc solutesolvent interactions. Now, let us consider the stilbazolium betaine M2 (Figs. 7 and 8). In low polarity solvents, the absorption and emission maxima of M2 are redshifted with respect to those of M1 of more or less 0.5 eV. This is correctly predicted by calculations (e.g., at k e 0 the calculated red shifts are 0.48 and 0.38 eV, respectively; see Tables 3 and 5 and Tables 4 and 6) even if the individual transition energies are rather overestimated ($0.7 eV at k e, N ET 0:3) due to the adopted CI limitations. Fig. 7 shows that both the initial small red shift and the

Fig. 8. Wavenumber of the calculated and experimental emission maximum of M2 vs, respectively, the k e and normalized N ET values of solvent polarity. The experimental mmax values are those of Table 1.

calcd:, 3810 [26], 3710 [27]. Such estimates N indicate that excepting solvents with very high ET , where specic solutesolvent interactions become very important, in M1 the uorescence is more sensitive than the absorption to solvent polarity. This behaviour, originating in the marked orientational stabilization of the emitting state S1 [20] is evidenced by the fact that, leaving out solvents N with ET > 0:7, the remaining experimental points

Fig. 9. Calculated and experimental Stokes shift of M1 vs, reN values of solvent pospectively, the k e and normalized ET larity. Experimental values were derived from [26,27] (see caption of Fig. 5 for the solvent characteristics).

322

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

successive strong blue shift of the absorption band N of M2 observed on increasing the ET value are qualitatively well reproduced by the calculations. Quantitatively speaking, the large negative solvaabs tochromism of M2 (mabs chloroform mwater 6260 1 1 cm (Table 1), 6490 cm [18,25]) is somewhat underestimated by the solvatonCS INDO calcuabs 1 lations (mabs k e0:2 mk e0:9 4230 cm , see Table 5 and Fig. 7). According to the previous discussion on the values and solvent-induced changes of dipole moments, in this case lS1 < lS0 uorosolvatochromism should be expected to be weaker than solvatochromism, contrary to what occurs in the case of M1 lS1 > lS0 . This is fully conrmed by our uorescence study on M2 in various solfluo 1 vents showing that mfluo chloroform mwater 1340 cm (Table 1 and Fig. 8), i.e., about a fth of the blueshift observed in absorption. Calculations underestimate uorosolvatochromic eects to the same extent as the solvatochromic ones, so mfluo k e0:2 1 mfluo reduces to $ 300 cm (Table 6, Fig. 8). k e0:9 In other words, the calculation results agree with experiment as regards the direction of the spectral shifts of absorption and uorescence maxima, but in both cases the eects are predicted to be weaker than the experimental ones. This allows for a much better agreement between calculated and observed Stokes shifts (Dmabs fluo), as clearly shown by Fig. 10. As is evident, the theoretical and experi-

mental plots of Dmabs fluo are in very good agreement over the experimentally investigated N values, and indicate that the Stokes range of ET shift of M2 increases rapidly on increasing the solvent polarity. Interestingly, contrary to what is found for M1 (Fig. 9), in this case the experimental plot displays no inversion 5 and remains parallel to N the theoretical one until ET 1:0. Figs. 9 and 10 lend themselves to the following concluding remarks. Although dyes M1 and M2 exhibit opposite solvatochromic and uorosolvatochromic behaviours, in both systems the Stokes shift increases considerably (with the only exception of M1 in water) with increasing the solvation N power of the solvent (as expressed by the ET value). This occurs for opposite reasons in the two dyes: in M1 the Stokes shift raises because the bathochromic shift of the uorescence is greater than that of the absorption; on the contrary, in M2 the same phenomenon is due to the hypsochromic shift of the absorption being greater than that of the uorescence. Both phenomena are well described within the solvatonCS INDO scheme and can be rationalized in terms of the dierent freemolecule electronic structures (i.e., the dierent relative weights of the neutral and zwitterionic VB structures) and of their evolution with increasing the solvent polarity. Moreover, it should be noted (see also Figs. 58) that the solvatonCS INDO approach, formally including only electrostatic solutesolvent interactions, actually accounts fairly well for the eects of media characterized by N quite high ET values where specic (H-bond) interactions are expected to be as important as dipolar interactions. The reason for this somewhat surprising result lies in the way the solutesolvaton interactions were included in the molecular HamFig. 10 shows that the solvation power of hydrogen-bond donor solvents towards M2 (in both the ground and the excited N states) is well described by the ET 30 ET parameter. On the contrary (Fig. 9), the solvation power of such solvents (primarily water) towards M1 is poorly described within the N ET 30 ET ) solvent scale. This is likely due to the fact that the probe used to develop this scale (pyridinium N-phenolate betaine) [25] is chemically more similar to M2 than to M1. A better general description could be achieved with proper multiparameter approaches [25,36], but such correlation analysis does not fall within the purposes of this work.
5

Fig. 10. Calculated and experimental Stokes shift of M2 vs, N values of solvent porespectively, the k e and normalized ET larity. The experimental values derive from the absorption and emission data of Table 1.

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

323

iltonian. As reminded in Sections 1 and 3.1, our procedure acts prevalently on the p-electron system by correction terms linearly related to the solvation parameter k e (Eq. (1)). Within this s scheme, the modications of the Fpp elements at high k e values may well account for the eects of H-bond interactions occurring in the molecular plane (e.g., those involving the sp2 -like oxygen lone pairs). This interpretation is plausible since it has been shown [19] that the eects of protic solvents on merocyanine spectra may be described within p-electron theory using heteroatom parameters corresponding to the molecule protonated on the oxygen. A complete validation of the calculated ground and excited-state dipole moments was not possible for lack of experimental data. However, using the solvatochromic comparison method, Banerjee et al. [27] evaluated the ratio lS1 =lS0 to be equal to 1.4 for dye M1. Table 5 shows that for M1 the theoretical prediction is in very good agreement with experiment (at k e 0, lS1 =lS0 1:44).

4. Summary and conclusions The main purpose of the present paper was to appraise the capability of the solvatonCS INDO model to arrange both solvatochromism and uorosolvatochromism of merocyanine dyes. In order to carry out an exhaustive test, we searched for two sample merocyanines characterized by opposite solvent eects. As a prototypic dye exhibiting large positive solvatochromism on both absorption and emission, we chose the ketocyanine M1 which has been the subject of extensive experimental work. The stilbazolium betaine M2 was chosen for the large negative solvatochromism of its absorption band, which has been widely studied both experimentally and theoretically. Since, to our knowledge, the emission properties of M2 had never been reported before, we carried out an experimental investigation on the matter. We found that M2 is rather weakly uorescent and exhibits negative uorosolvatochromism, even if the emission is much less sensitive to a change of solvent polarity than the absorption. Moreover, the analysis of the uorescence emission and excitation

spectra led us to conclude that, as it happens in the case of the strongly uorescent M1, the emission of M2 originates from the local minimum of the solvated S1 pH p L state. According to this body of experimental evidence, the theoretical descriptions of solvatochromism and uorosolvatochromism were undertaken within the same conventional scheme, i.e., calculating the shifts induced by a solvent-polarity change on the vertical S0 eq: ! S1 FranckCondon and S1 eq: ! S0 FranckCondon transitions. Such a scheme was directly applied within the solvatonCS INDO method. The equilibrium absorbing and emitting units (molecule + solvaton pattern) and the corresponding S0 ! S1 and S1 ! S0 vertical transitions were studied for M1 and M2 as a function of a polarity factor formally related to the dielectric p p constant, k e e 1= e. For the sake of simplicity, the molecular geometries in the equilibrium S0 and S1 states were approximated by those optimized in the absence of solvent, and the respective solvaton patterns, setup using the subset of the net p-electron charges, were adjusted iteratively starting from those corresponding to the unsolvated molecules. The solvent-dependent composition of the resonance hybrid between the covalent and the zwitterionic VB structures was taken as the key to interpret the calculated ground and excited-state properties of the two merocyanines at the various k e values. The analysis of the gas-phase optimized geometries in terms of BLA parameter showed that in the ground state M1 has a markedly covalent character while in M2 the covalent character is nearly balanced by the zwitterionic one. On moving from S0 to S1 , the contribution of the charge-separated structure increases in both dyes, even if the phenomenon is less marked in M2 than in M1. Introduction of a polar solvent stabilizes the zwitterionic forms but, due to the dierence in the starting electronic structures, it results in opposite solvatochromic behaviours in both absorption and emission. The use of solvaton sets reecting the net p-electron charges proved to be most eective to rule the evolution of the resonance hybrid composition of S0 and S1 as the solvent polarity increases. The description of the solutesolvent interaction in terms of local charge distributions

324

I. Baraldi et al. / Chemical Physics 288 (2003) 309325

makes the solvaton model free from symmetry imposed restraints, this being at variance with all solvaton models (e.g., SCRF) where the solute is treated as a point dipole. Thus, the fact that M1 has moderate S0 and S1 dipole moments perpendicular to the conjugated chain (the longitudinal components being equal to zero by symmetry) while M2 has large dipole moments aligned with the long molecular axis leads to quite dierent reaction elds (and solutesolvent interaction energies) within the dipolar approximation in spite of the basic similarities of the two systems. On the contrary, the solvaton model provides fully consistent descriptions for the M1 and M2 solvatochromisms. A thorough comparison with the experimental data was made possible by the construction of diagrams where calculated and observed frequencies of the absorption and emission maxima were N plotted against the k e and ET solvent polarity parameters, both ranging from 0 to 1. Apart from a systematic overestimation (0.50.7 eV) of the transition energies the solvatonCS INDO calculations provided good descriptions of the opposite solvatochromic behaviours of M1 and M2 in both absorption and emission. Quantitatively speaking, the solvent shifts of the absorption and uorescence spectra were correlated better for the positively solvatochromic ketocyanine than for the negatively solvatochromic stilbazolium betaine, where the solvatochromic ranges (mnon-polar solvent mhighly polar solvent) of absorption and emission were both somewhat underestimated. However, in very good agreement with experiment the Stokes shifts (Dmabs fluo) of both M1 and M2 were predicted to substantially increase (especially in the case of M2) on increasing the solvent polarity. We notice that, except for the peculiar behaviour of M1 in water, the solvatonCS INDO model is capable of accounting for the solvation eects on merocyanine spectra over almost the N entire range of the normalized ET values. This happens since the polarization of the p-electron system induced by specic (H-bond) solvent interactions taking place in the molecular plane is implicitly accounted for within our formulation of the solvaton model. To sum up, the solvatonCS INDO method was applied to the study of solvent eects on both

absorption and uorescence spectra of positively and negatively solvatochromic merocyanines. Such a severe test has been passed rather well, at least as far as the main qualitative aspects are concerned (but also the sizes of the solvent shifts were in general well reproduced). An elementary condition underlying these results is that the electronic transitions of the solute were generated in the presence of the polarized solvent. Quite surprisingly, the controversy that arose some years ago in the interpretation of the solvatochromism of M2 (see Section 1) came from theoretical treatments devoid of this prerequisite.

Acknowledgements This research was jointly supported by the MURST (Roma) and the University of Modena and Reggio Emilia within the Programmi di Ricerca di Interesse Nazionale. We are indebted to Dr. J.P. Flament for providing us with his geometry optimization program and to Prof. G. Berthier for valuable discussions.

References
[1] I. Baraldi, F. Momicchioli, G. Ponterini, D. Vanossi, Chem. Phys. 238 (1998) 353. [2] I. Baraldi, F. Momicchioli, G. Ponterini, D. Vanossi, Adv. Quantum Chem. 36 (1999) 121. [3] F. Momicchioli, I. Baraldi, M.C. Bruni, Chem. Phys. 82 (1983) 229. [4] G. Klopman, Chem. Phys. Lett. 1 (1967) 200. [5] H.A. Germer, Theor. Chim. Acta 34 (1974) 145. [6] H.A. Germer, Theor. Chim. Acta 35 (1974) 273. [7] S. Miertus, O. Kysel, Chem. Phys. 21 (1977) 27. [8] S. Miertus, O. Kysel, Chem. Phys. Lett. 65 (1979) 395. [9] R. Costanciel, O. Tapia, Theor. Chim. Acta 48 (1978) 383. [10] A. Botrel, A. Le Beuze, P. Jacques, H. Straub, J. Chem. Soc. Faraday Trans. II 80 (1984) 1235. [11] A. Botrel, B. Aboad, F. Corre, F. Tonnard, Chem. Phys. 194 (1995) 101. [12] J. Griths, Colour and Constitution of Organic Molecules, Academic Press, London, 1976. [13] C. Reichardt, Solvent Eects in Organic Chemistry, Verlag Chemie, Weinheim, 1979. [14] J.O. Morley, J. Mol. Struct. (Theochem) 304 (1994) 191. [15] L. da Silva, C. Machado, M.C. Rezende, J. Chem. Soc. Perkin Trans. II (1995) 483.

I. Baraldi et al. / Chemical Physics 288 (2003) 309325 [16] I.D.A. Albert, T.J. Marks, M.A. Ratner, J. Phys. Chem. 100 (1996) 9714. [17] A. Klamt, J. Phys. Chem. 100 (1996) 3349. [18] P. Jacques, J. Phys. Chem. 90 (1986) 5535. [19] H.G. Benson, J.N. Murrell, J. Chem. Soc. Faraday Trans. II 68 (1972) 137. [20] L. Salem, Electrons in Chemical Reactions: First Principles, Wiley, New York, 1982 (Chapter 8). [21] J. Tomasi, M. Persico, Chem. Rev. 94 (1994) 3027. [22] J.T. Blair, K. Krogh-Jespersen, R.M. Levy, J. Am. Chem. Soc. 111 (1989) 6948. [23] K. Coutinho, S. Canuto, M.C. Zerner, J. Chem. Phys. 112 (2000) 9874. [24] K.J. de Almeida, K. Coutinho, W.B. de Almeida, W.R. Rocha, S. Canuto, Phys. Chem. Chem. Phys. 3 (2001) 1583. [25] C. Reichardt, Chem. Rev. 94 (1994) 2319. [26] M.A. Kessler, O.S. Wolfbeis, Spectrochim. Acta 47A (1991) 187. [27] D. Banerjee, A. Kumar Laha, S. Bagchi, J. Photochem. Photobiol. A: Chem. 85 (1995) 153. [28] S.J. Isak, E.M. Eyring, J. Phys. Chem. 96 (1992) 1738. [29] G.R. Fleming, A.W.E. Knight, J.M. Morris, R.J.S. Morrison, G.W. Robinson, J. Am. Chem. Soc. 99 (1977) 4306.

325

[30] I. Baraldi, S. Ghelli, Z.A. Krasnaya, F. Momicchioli, A.S. Tatikolov, D. Vanossi, G. Ponterini, J. Photochem. Photobiol. A: Chem. 105 (1997) 297. [31] P. Milli e, F. Momicchioli, D. Vanossi, J. Phys. Chem. B 104 (2000) 9621. [32] I. Baraldi, F. Momicchioli, G. Ponterini, A.S. Tatikolov, D. Vanossi, Phys. Chem. Chem. Phys. 5 (2003). [33] J.O. Morley, R.M. Morley, A.L. Fitton, J. Am. Chem. Soc. 120 (1998) 11479. [34] J. Catal an, E. Mena, W. Meutermans, J. Elguero, J. Phys. Chem. 96 (1992) 3615. [35] C. Burda, M.H. Abdel-Kader, S. Link, M.A. El-Sayed, J. Am. Chem. Soc. 122 (2000) 6720.  underl  ikurov [36] B. C kov a, L. S a, Chem. Phys. 263 (2001) 415. [37] R. Fletcher, M.J. Powell, Comput. J. 6 (1963) 163. [38] K. Ohno, Theor. Chim. Acta 2 (1964) 219; G. Klopman, J. Am. Chem. Soc. 86 (1964) 4550. [39] G. Brancolini, Doctorate thesis, University of Modena and Reggio Emilia, 2002. [40] D.J.A. De Ridder, D. Heijdenrijk, H. Schenk, R.A. Dommisse, G.L. Lemi ere, J.A. Lepoivre, F.A. Alderweireldt, Acta Crystallogr. C 46 (1990) 2197. [41] S.S. Malhotra, M.C. Whiting, J. Chem. Soc. (1960) 3812.

You might also like