Large Blue-Shift in the Optical Spectra of Fluorinated

Feb 16, 2008 - Manuel Piacenza,† Fabio Della Sala,*,† Gianluca M. Farinola,‡ Carmela Martinelli,‡ and. Giuseppe Gigli†. National Nanotechnol...
10 downloads 0 Views 361KB Size
Download PDF
2996

J. Phys. Chem. B 2008, 112, 2996-3004

Large Blue-Shift in the Optical Spectra of Fluorinated Polyphenylenevinylenes. A Combined Theoretical and Experimental Study Manuel Piacenza,† Fabio Della Sala,*,† Gianluca M. Farinola,‡ Carmela Martinelli,‡ and Giuseppe Gigli† National Nanotechnology Laboratory of CNR-INFM, Distretto Tecnologico ISUFI, UniVersita` del Salento, Via per Arnesano, I-73100 Lecce, Italy, and Dipartimento di Chimica, UniVersita` di Bari, Via Orabona 4, I-70126 Bari, Italy ReceiVed: October 10, 2007; In Final Form: December 6, 2007

Modifications of the optical properties of poly[2-methoxy-5-(2′-ethyl-hexyloxy)-1,4-phenylene vinylene] induced by fluorination of the vinylene units are investigated by means of time dependent density functional theory (TD-DFT) calculations and spectroscopic measurements in solution. The energy of the main absorption peak is blue-shifted by more than 0.8 eV in the fluorinated polymers. TD-DFT excitation energies for nonfluorinated and fluorinated oligomer structures of increasing number of monomers, employing fully relaxed geometries, are compared to the experimental absorption energies of the polymers. We found that the measured large blue-shift induced by the fluorination of the vinylene units is not caused by the electron-withdrawing effect of the fluorine substituents but it is related to a steric effect. The inter-monomer torsional angle of the fluorinated structures increases above 50°, while in the non-fluorinated systems it is below 20°. Further insight into the origin of the large blue-shift of the excitation energies is gained by a detailed analysis of the torsional potentials of non-fluorinated and fluorinated dihydroxystilbene. While for planar geometries the energy gap increases due to fluorination, it decreases for highly distorted geometries. In addition, we found that the torsional potential of dihydroxystilbene is rather flat, meaning that different isomers might, e.g., in the solid state, coexist.

1. Introduction A major drawback of many optoelectronic devices based on organic materials, e.g., polyphenylenevinylenes (PPVs),1,2 is the thermal and oxidative instability.3 Recent investigations have shown that fluorination generally enhances the stability of these compounds. (For a recent overview, see, e.g., ref 4 and references therein.) Most experimental5,6 and theoretical7-9 studies focus hereby on compounds, where some or all hydrogen atoms bound to the aromatic cores are substituted by fluorine atoms. A recent theoretical investigation10 of the fluorination effects on the optical and electronic properties of π-conjugated molecules, like oligoacenes, oligothiophenes, and oligophenylenevinylenes, has shown that this substitution generally stabilizes the frontier orbitals and depending on the individual shifts of each system can lead to either wider or narrower energy gaps and thus to blue- or red-shifts in the electronic spectra. All of these computational studies7,8,10 assumed planar geometries of the investigated structures, which is a reasonable approximation if no bulky substituents and/or groups that are carrying large partial charges are present directly at and in the vicinity of the vinylene bridges. In addition to the core substituted poly- and oligophenylenevinylenes, compounds that have been selectively fluorinated at the vinylene bridges have been proposed by Jin et al.,11 namely, the bare poly(p-phenylenedifluorovinylene) and the poly(2-dimethyloctylsilyl-p-phenylenedifluorovinylene). The * To whom correspondence should be addressed. E-mail: [email protected]. Phone: +39 0832 298202. Fax: +39 0832 298237. † CNR-INFM. ‡ Universita ` di Bari.

maxima of their main absorption bands are blue-shifted by 0.22 eV with respect to the non-fluorinated poly(2-dimethyloctylsilylp-phenylenevinylene). Frontier orbital energies and the resulting energetic gaps have been extracted from optical spectra and cyclo-voltametric measurements, but no theoretical description of the structural features or assignment of the observed bands has been made so far. In this study, we will investigate the optical properties of the recently synthesized12 poly(2-methoxy-5-(2′-ethylhexyloxy)-1,4phenylene-difluoro-vinylene) (MEH-PPVF) in comparison with poly(2-methoxy-5-(2′-ethylhexyloxy)-1,4-phenylenevinylene) (MEH-PPVH) (see Figure 1). We will report a spectroscopic characterization of the nonfluorinated MEH-PPVH and fluorinated MEH-PPVF polymer in solution at different concentrations and in different solvents, and a first-principles theoretical investigation for non-fluorinated and fluorinated oligomer model systems. From a computational point of view, we will investigate the fully optimized structures to achieve an accurate description of the influence of the individual substitution patterns on the molecular geometries and the thus induced impact on the excitation energies. Moreover, we will distinguish between this steric and a pure electronic effect induced by the fluorination of the vinylene groups. Most present day excited state calculations for small organic compounds are done using time dependent density functional theory (TD-DFT)13-15 employing gradient-corrected or hybrid density functional methods. It is well established that these approaches have severe shortcomings in the description of charge-transfer bands and Rydberg states,15-26 but they are considered as reliable tools for singlereference valence excitations. Besides semiempirical works,

10.1021/jp7098784 CCC: $40.75 © 2008 American Chemical Society Published on Web 02/16/2008

Optical Spectra of Fluorinated PPVs

J. Phys. Chem. B, Vol. 112, No. 10, 2008 2997

Figure 1. Chemical structures of the investigated systems.

there are several theoretical investigations of polyphenylenevinylenes and related structures employing density functional theory.1,2,27,37 However, it has been shown that in some cases TD-DFT cannot correctly describe the chain length evolution due to the employed local or semilocal exchange-correlation (XC) kernel38,39 or due to the neglect of double excitations.32,40-42 Hybrid XC kernels with a different amount of Hartree-Fock exchange are often required to reach reliable prediction of the chain length dependence. Parac and Grimme report severe shortcomings of the TD-DFT approach for the description of the exited states of linear acenes.22 The authors state that “admixture of ‘exact’ HF exchange clearly improves the situation” and that “it can be expected that almost 50% exchange would be necessary to reach reliable predictions for the largest systems.” In a recent work, Tretiak et al.36 compared the applicability of several quantum chemical approaches for the calculation of excited states of phenylenevinylene oligomers including chain lengths of up to 20 monomer units. They also found that inclusion of a certain portion of “exact” exchange is strictly mandatory in the TD-DFT calculations to obtain reliable results. In the computational part of this work, we will investigate the structural and optical properties of MEH-PPVH and MEH-PPVF oligomers (dimer to pentamer) using the BHLYP43 XC functional. We found that the fluorinated oligomers are much more distorted and thus show up to 0.9 eV larger excitation energies than the non-fluorinated ones, in very good agreement with the measured absorption peak. The effects of the torsional potential upon the molecular and electronic structure, as well as upon excitation energies, are investigated in detail for non-fluorinated and fluorinated dihydroxystilbenes, that serve as model systems for the more complex molecules. The article is organized as follows: In section 2, we describe the experimental setup and the computational details. In section 3, we report the experimental and theoretical results. In section 4, the theoretical results are discussed and compared with the experimental data. 2. Methods 2.1. Computational Details. All calculations were performed using the TURBOMOLE 5.944 package. In particular, the modules DSCF,45 GRAD, and ESCF14 have been used. In the density functional theory (DFT) ground and excited state calculations, we employed the BH-LYP43 hybrid functional. The default m3 numerical quadrature grid was used in all DFT calculations. Basis sets of valence-triple-ζ quality with polarization functions TZVP46 taken from the TURBOMOLE basis set library47 have been employed.

Figure 2. First singlet TD-DFT excitation energies of planar oligophenylenevinylenes as a function of the number of monomer units (n), employing different density functionals. Experimental absorption energies are indicated as hollow circles.

2.2. Assessment of the Method. We checked the reliability of our computational approach by comparing the excitation energies of planar oligophenylenevinylenes obtained from TDDFT calculations employing density functionals with different amounts of Hartree-Fock exchange, namely, B-LYP48,49 (0% “exact” exchange), B3LYP49,50 (20% “exact” exchange), and BH-LYP43 (50% “exact” exchange) with experimental data. We optimized the structures of styrene, trans-stilbene (n ) 2), 1,4bis[(E)-styryl]benzene (DSB, n ) 3), 4,4′-distyrylstilbene (DSS, n ) 4), and 1,4-bis(4-styrylstyryl)benzene (BSSB, n ) 5) at the B3-LYP/TZVP level of theory and performed subsequent TD-DFT calculations with the three density functionals. The optimizations were restrained to planar geometries, to exploit the C2h symmetry of the oligomers. The results are shown in Figure 2. Experimental data is taken from refs 27, 31, 51, and 52. The employment of the BH-LYP functional leads to slightly overestimated excitation energies for the two smallest systems, but almost perfect agreement is achieved for DSB and DSS. The commonly applied B3LYP functional performs rather good for styrene and stilbene, but the excitation energy of DSB is already underestimated by almost 0.5 eV. This trend continues for DSS, where the calculated excitation energy is 0.6 eV too low. The same holds for the results presented in ref 9, where B3-LYP/SV(P) calculations severely underestimated the experimental excitation energies. Results obtained from B-LYP calculations are even worse, being 0.8 eV too low for DSS. These observations are in accordance with the conclusions drawn in ref 22, and the BH-LYP functional can be safely employed for calculating excitation energies of oligomers of the here investigated size. A similar investigation has been reported by Tretiak et al.36 They found good agreement of theoretical and experimental values, when the B3-LYP functional was employed. However, due to the large system sizes, their calculations were restricted to small (double-ζ) basis sets and planar geometries optimized using Hartree-Fock (HF) and also double-ζ basis sets. For pure planar hydrocarbons, these computational approaches might be appropriate, but as soon as more complex substitution patterns are present, like in the structures considered in this work, the

2998 J. Phys. Chem. B, Vol. 112, No. 10, 2008 application of correlated methods and larger basis set schemes might be desirable to obtain a more reliable description of the molecular systems. Concerning the molecular geometries, we have compared the B3-LYP/TZVP and BH-LYP/TZVP optimized structures of trans-2,2′-dihydroxystilbene and trans-R,β-difluoro-2,2′-dihydroxystilbene, which will also serve as model systems in the last section. The deviations of the molecular geometries obtained with the different functionals are very small. The vinylene single bond lengths do not differ by more than 0.004 Å, the vinylene double bond length by no more than 0.015 Å. The deviation of the torsional angle between the phenylene and vinylene units is less than 0.2° for the hydrogen substituted structure and less than 1.0° for the fluorine substituted structure. Moreover, we observed a slightly larger dependence of the geometrical parameters on the applied basis set. When, e.g., a double-ζ basis is used for the optimization, the differences of the bond distances increase up to 0.04 Å and the dihedral angle deviate by 4°. Regarding the small differences of the molecular geometries with respect to the applied hybrid DFT kernel, we will use for reasons of consistency BH-LYP/TZVP optimized structures exclusively, as according to the method comparison the latter approach is also superior for the description of the excitation energies of the here considered structures. 2.3. Experimental Setup. Absorption measurements were performed using an ultraviolet-visible (UV-vis) recording spectrophotometer SHIMADZU UV-2401PC and the photoluminescence measurements with a VARIAN Cary Eclipse fluorescence spectrophotometer. Experiments have been performed in CHCl3 solution for concentrations between 3.5 and 11.2 µg/L as well as in tetrahydrofurane (THF) and toluene at constant concentrations of 3.5 µg/L.

Piacenza et al.

3. Results

Figure 3. Experimental absorption and photoluminescence spectra of polymers MEH-PPVH (upper panel) and MEH-PPVF (lower panel) at different concentrations in CHCl3.

3.1. Experimental Results. The absorption and photoluminescence spectra at different concentrations in CHCl3 for MEHPPVH and MEH-PPVF are reported in Figure 3. The main absorption peak of MEH-PPVF (3.44 eV) is blue-shifted by 0.89 eV with respect to the one of MEH-PPVH (2.55). The shape of the absorption spectra does not change for different concentrations, meaning that no aggregation effects are present in the polymer solution. A smaller blue-shift of 0.44 eV is observed in PL. This is due to the much larger Stokes shift of MEHPPVF (0.79 eV) with respect to MEH-PPVF, which is only 0.21 eV. The large Stokes shift of MEH-PPVF can be related to larger conformational changes in the excited state or exciton trapping.11 The PL spectrum of MEH-PPVH shows well resolved vibronic replicas (∆E ) 0.12 eV) that are not present in the MEH-PPVF case, probably indicating that the former is much more planar than the latter. In Figure 4, we report the same absorption and photoluminescence spectra recorded in different aprotic solvents at identical concentrations of 3.5 µg/L. All spectra are normalized and show no significant difference in their main absorption and emission bands with respect to the solvents, but for the PL vibrational resolved band of MEH-PPVH. 3.2. Theoretical Results. We modeled the MEH-PPVH and the MEH-PPVF as oligomers of increasing number of monomers n (PVH-n and PVF-n). Structures consisting of two to five monomer units have been considered in the calculations. In order to save computation time, the branched alkoxy substituents (R1 ) C2H5, R2 ) C4H9 in Figure 1) as present in the experimentally investigated polymers are truncated (R1 ) R2 ) CH3 in Figure 1) in the model system. This choice is justified as only isolated

molecules are considered in the theoretical study and no supramolecular orientation effects have been taken into account. In fact, test calculation for fully substituted dimers did not show any significant deviations in the molecular geometry from the truncated system. 3.2.1. Molecular Geometries. The molecular geometries of the PVH-n and PVF-n (n ) 2-5) oligomers have been optimized at the BH-LYP/TZVP level of theory. As representative examples, the structures of the non-fluorinated and fluorinated trimers PVH-3 and PVF-3 are shown in Figure 5. At this point, it should be mentioned that several local minima of the oligomers with respect to their tacticity can be found. In this study, isotactic and syndiotactic structures have been optimized and the energetically most favorable geometries have been used for the subsequent excited state calculations. We cannot exclude that also atactic isomers coexist, but nevertheless, we do not expect that the tacticity has a large influence on the excitation energies. In fact, for the fluorinated structures, several local minima for the here described oligomers have been found, most of them less than 0.5 kcal/mol above the most stable isomer. The maximum deviation for the TD-DFT excitation energies with respect to a close lying conformer of the trimer was 0.15 eV, which might serve as an estimate of error in the next section. Because in all here considered cases, the vinylene groups are selectively fluorinated, the most interesting geometric features for this study are the carbon-carbon single and double bond distances of the vinylene units and the dihedral angles between the planes of the six-membered rings and the vinylene bridge. Single bond distances of the hydrogen substituted oligomers

Optical Spectra of Fluorinated PPVs

J. Phys. Chem. B, Vol. 112, No. 10, 2008 2999

Figure 6. Distribution of dihedral angles for BH-LYP/TZVP optimized structures PVH-n and PVF-n. Figure 4. Experimental absorption and photoluminescence spectra of polymers MEH-PPVH (upper panel) and MEH-PPVF (lower panel) in different solvents (concentration 3.5 µg/L). All bands are normalized in this plot.

Figure 5. BH-LYP/TZVP optimized geometries of PVH-3 and PVF-3.

PVH-n are in a range between 1.456 and 1.461 Å, double bonds between 1.330 and 1.333 Å. No significant differences between dimers, trimers, tetramers, and pentamers with respect to the bond lengths have been found. The corresponding vinylene double bonds of the fluorinated structures are within 1.321-

1.324 Å, about 1 pm shorter, and the single bonds within 1.4611.466 Å, slightly longer. Thus, no major difference concerning these parameters is found between unsubstituted and fluorinated structures. The major change of the molecular geometries, induced by fluorination of the vinylene groups, concerns the dihedral angles between the planes of the aromatic rings and the bridging vinylene units. Torsional angles for fluorinated and nonfluorinated oligomers with increasing system size are shown in Figure 6. The torsional angle of PVH-2, PVH-3, PVH-4, and PVH-5 are found in a range between 0 and 22°. Thus, the structures are not completely planar but moderately twisted. The torsional angles increase for larger oligomers and are spread over a range of approximately 10° for the individual oligomers. This last point already bears the hint that the potential surface for the rotational motion of these structures is rather flat, which was also found in ref 1. It should be mentioned that, despite the torsion of the vinylene bridges, the aromatic rings adopt an almost perfect coplanar orientation in all structures (cf. Figure 5) and are not helically twisted. The dihedral angles of the corresponding fluorinated structures PVF-n are significantly larger. They are found in a range between 46 and 53°, and all structures are much stronger twisted, as can be seen for the optimized trimer in Figure 5. Neighboring six-membered rings are found in an approximate perpendicular orientation. The values of the torsional angle of the individual oligomers are more uniform than those for the hydrogen substituted species and do not differ by more than 5°. The large dihedral angle of the fluorinated compounds can be clearly attributed to a repulsive intramolecular interaction of the fluorine atoms with the neighboring alkoxy groups. Both substituents

3000 J. Phys. Chem. B, Vol. 112, No. 10, 2008

Piacenza et al.

TABLE 1: BH-LYP/TZVP Absorption Energies, Oscillator Strengths (fL), and Main Single Particle Transitions (above 4%) for PVH-n and PVF-n Oligomers (n ) 2-5) E(ABS) structure

(eV)

(nm)

fL

PVH-2 PVH-3 PVH-4

3.89 3.29 3.07

PVH-5

2.90

PVF-2

4.46

278

PVF-3 PVF-4

4.11 3.94

302 315

1.024 1.624

PVF-5

3.84

323

2.238

transition

hydrogenated 318 0.831 HfL 377 1.646 HfL 403 2.423 HfL H-1fL+1 427 3.212 HfL H-1fL+1 fluorinated 0.522

HfL H-1fL HfL HfL H-1fL+1 HfL H-1fL+1

% 95.5 93.5 89.1 6.9 83.9 10.1 86.0 7.5 85.5 77.9 9.4 70.8 13.7

carry a negative partial charge, and the torsional motion increases the distance of both groups. 3.2.2. Optical Properties and Molecular Orbitals. A summary of the results from the TD-DFT calculations for the electric dipole allowed S1 transition of the considered oligomers can found in Table 1. The lowest excited states of the investigated structures is characterized by a π f π* transition from the highest occupied molecular orbital (HOMO) to the lowest unoccupied molecular orbital (LUMO). The excitation energies of the hydrogenated molecules decrease from 3.89 eV for the dimer (PVH-2) to 2.90 eV for the pentamer (PVH-5), while the oscillator strengths fL increase with the oligomer size. All corresponding states of the fluorinated structures are found at significantly higher energies, while the oscillator strengths are reduced by a factor of about 1.5. From the last two columns of Table 1, it becomes evident that the HOMO f LUMO π f π* transition is the dominant excitation. Contributions of single particle transitions below 4% are not reported. The corresponding eigenvalues of the frontier orbitals are plotted as a function of the oligomer sizes in Figure 7. HOMO and LUMO wavefunctions of PVH-3 and PVF-3 serve again as representative examples for the other oligomers and are depicted as an inset of Figure 7. Despite the torsional distortion, the orbitals of the fluorinated and non-fluorinated molecules are rather similar in shape. A small difference is found at the vinylene units, where small coefficients on the fluorine atoms of PVF-3, but no contribution of the corresponding hydrogen atoms of PVH-3 at the identical isosurface value can be seen. Regarding the eigenvalues, it is found that the HOMO energies of the fluorinated compounds are always below and the LUMO energies always above the corresponding orbital energies of the non-fluorinated structures. The LUMO energies of both classes of molecules decrease and the HOMO energies increase, but to a larger extent, with increasing system size. The latter effect is more strongly pronounced for the hydrogen substituted than for the fluorine substituted molecules. This can be traced back to the individual MO structures, because an antibonding contribution is located on the vinylene single bonds of the HOMO, whereas a bonding contribution is found at this position for the LUMO. Rotation around this bond, as apparently induced by fluorination, should consequently stabilize the HOMO and destabilize the LUMO. At a first glance, these results seem to contradict the previous theoretical10 and experimental11 findings, where a lowering of both HOMO and LUMO energies due to fluorination is

Figure 7. Eigenvalues of BH-LYP/TZVP frontier orbitals of nonfluorinated and fluorinated oligomers PVH-n and PVF-n (n ) 2-5). The corresponding wavefunctions of the frontier orbitals of PVH-3 and PVF-3 (0.04 au isosurfaces) are shown as an inset in the graph.

described. However, it must be pointed out that the study10 was employing planar and not fully relaxed structures and that torsional distortion should affect the frontier orbitals as described above. The structures investigated in ref 11 are carrying just one substituent on the phenylene core, which should reduce the impact of torsional distortion. Further insight into this phenomenon will be gained at the last part of the next section. 4. Discussion 4.1. Comparison between Theory and Experiment. In Figure 8, we plot the experimental and theoretical excitation energies against the inverse of the number of monomer units (1/n). The dotted lines are obtained from linear regression of the theoretical data and can be used as a first approximation to extrapolate the excitation energies for longer oligomers. As excitation energies achieve saturation for a very large number of monomers,9,36,53 more sophisticated extrapolation procedures9 must be applied to obtain more accurate results. Thus, our theoretical results for the largest systems, i.e., the pentamers, represent an upper bound for the deviation from the experimental values. Figure 8 shows that we obtained a satisfying agreement between the extrapolated theoretical results for the excitation energies and the experimental points, which correspond to system sizes of n ) 13 (PPVH) and n ) 94 for (PPVF).12 Concerning the blue-shift induced by fluorination, we obtained from the calculations shifts of 0.57, 0.82, 0.87, and 0.94 eV for the dimer, trimer, tetramer, and pentamer, respectively. The blue-shifts for the oligomers are in very good agreement with the experimental result of 0.89 eV. Thus, a short oligomer of three to five units is appropriate to describe the influence of fluorination upon the excitation energies.

Optical Spectra of Fluorinated PPVs

J. Phys. Chem. B, Vol. 112, No. 10, 2008 3001

Figure 8. Absorption energies of fluorinated and non-fluorinated oligomers plotted versus the inverse of the number of monomer units n. Experimental values are indicated as hollow symbols.

It is worth mentioning here that we are comparing the theoretical vertical excitation energies in a vacuum to the energy of the first absorption peak measured in solution. While Figure 4 shows that solvent polarization is negligible for these systems, temperature effects and vibronic couplings54-57 are not included in the TD-DFT calculations, thus preventing an exact comparison with experiments. However, according to the shape of the measured optical spectra (see Figure 3), a vibronic replica is not observed in the absorption spectra (probably hidden by conformationl disorder), so that vertical excitation energies are expected to be close (within 0.1-0.2 eV) to experimental absorption maxima. 4.2. Torsional versus Substituent Effects. As already mentioned above, the most relevant structural difference between the hydrogen and fluorine substituted oligomers has been found for torsional angles. It is well-known that an increase of the torsional angle of π-conjugated systems results in an increase of the excitation energies.58-61 The effect of fluorination on the inter-ring torsional angle of the systems and thus on their excitation energies could be characterized as a steric effect. However, fluorination does not only induce conformational changes but also causes an electron-withdrawing effect10 because of the higher electronegativity of fluorine compared to hydrogen. In order to separate the electron-withdrawing from the induced torsional effect, we repeated the excited state calculations, resubstituting the fluorine by hydrogen atoms but keeping the remaining atoms in fixed positions, as obtained from the optimization of the fluorinated structures. The resulting excitation energies and differences with respect to the corresponding fluorinated systems (∆E(ABS)) are shown in Table 2. All values for the hydrogenated systems are found between 0.07 and 0.11 eV above the values for the fluorinated ones. This means that fluorination itself induces a tiny red-shift, whereas the large blue-shift can be completely attributed to the change in the torsional angles as a consequence of increased steric repulsion. 4.3. Torsional Potential of Model Compounds. In order to gain further insight into the influence of fluorination on the molecular geometries and electronic properties, we computed the torsional potentials of trans-2,2′-dihydroxystilbene (DHS) and trans-R,β-difluoro-2,2′-dihydroxystilbene (DHSF) (see inset of Figure 9) from restrained geometry optimizations. Both

Figure 9. Torsional potential for disrotatorial rotation of both phenyl substituents of trans-2,2′-dihydroxystilbene (DHS, X ) H) and transR,β-difluoro-2,2′-dihydroxystilbene (DHSF, X ) F). The relative energies are obtained from restrained BH-LYP/TZVP geometry optimizations

TABLE 2: BH-LYP/TZVP Absorption Energies, Oscillator Strengths (fL), and Main Single Particle Transitions (above 4%) for Hydrogen Substituted Structures, Using the Optimized Geometries of Fluorinated Oligomers (PVF′-n, n ) 2-5)a structure

E(ABS) (eV)

∆E(ABS) (eV)

fL

transition

%

PVF′-2 PVF′-3 PVF′-4

4.53 4.21 4.05

+0.07 +0.10 +0.11

0.519 1.042 1.653

PVF′-5

3.95

+0.11

2.290

HfL HfL HfL H-1fL+1 HfL H-1fL+1

86.0 87.0 78.7 11.0 70.7 15.2

a

∆E(ABS) describes the differences with respect to PVF-n.

systems represent suitable model systems for the abovediscussed oligomers, as they feature all important chemical groups, that are expected to be relevant for the intramolecular interactions. The torsional potential of unsubstituted transstilbene, which is highly symmetric, is known well from the literature.62-64 We started from a fully planar configuration and rotated both phenyl substituents in steps of 15° in a disrotatorial manner (one group clockwise, the other anticlockwise). The two dihedral angles θ were kept fixed at these positions, whereas all other internal coordinates were allowed to relax. A conrotatorial movement (both substituents clockwise or anticlockwise) does not correspond to the fully optimized oligomers (PVH-n and PVF-n), because during this motion the phenyl rings would always remain in a coplanar orientation. Because of symmetry, which is at least C2 during rotation, a scan from 0 to 180° completely covers the torsional potential. As we are aiming for a merely qualitative picture, we did not optimize the transition states separately. The resulting relative energies with respect to the fully optimized structures (DHS, θ ) 22.0°; DHSF, θ ) 47.2°) are plotted in Figure 9. For both structures, two local minima are found on the torsional potential surface. For DHS, the first minimum is at 22° and the second minimum at 180° is 1 kcal/ mol less stable. Both are separated by a barrier of approximately 10 kcal/mol. The maximum at 0° torsion is only 0.11 kcal/mol

3002 J. Phys. Chem. B, Vol. 112, No. 10, 2008

Piacenza et al.

Figure 11. BH-LYP/TZVP frontier orbitals at 0° (a) and 90° (b) torsional angles for trans-2,2′-dihydroxystilbene (DHS, X ) H) and trans-R,β-difluoro-2,2′-dihydroxystilbene (DHSF, X ) F) (0.05 au isosurfaces). Figure 10. BH-LYP/TZVP frontier orbital energies (LUMO (a), HOMO (b), energy gap (c), and absorption energy (d)) in eV for disrotatorial rotation of both phenyl groups of trans-2,2′-dihydroxystilbene (DHS) and trans-R,β-difluoro-2,2′-dihydroxystilbene (DHSF).

above the fully optimized structure, and the potential between -30 and 30° is extremely flat. The potential energy plot for the fluorinated structure (DHSF) looks rather different. The two minima are almost isoenergetic and shifted to 45 and 135°. Moreover, the barrier separating the minima is reduced to approximately 3 kcal/mol and thus interconversion of the two rotamers should be facilitated compared to DHS. Planar conformations are clearly disfavored because maxima of 7.6 and 13.4 kcal/mol are found at 0 and 180°, respectively. The differences in the potential curves show that fluorination of the vinylene groups leads to major changes in the molecular structure. The hydrogen substituted species are approximately planar, whereas the fluorinated systems are strongly tilted with 45° torsion between the aromatic rings and the vinylene linkers and neighboring phenylenes adopt a perpendicular orientation. In order to investigate how far these geometric differences have an effect on the electronic structure and thus the optical properties, it is necessary to have a closer look at the changes of the frontier orbital energies during rotation. In Figure 10, the eigenvalues of the LUMO (panel a) and HOMO (panel b), their difference, i.e., the energy gap (panel c), and the absorption energies (panel d) of DHS and DHSF are plotted as a function of the torsional angle. Corresponding isosurface plots at 0 and 90° torsion are shown in Figure 11. From the orbital energy plots, we can make several observations:

(i) The eigenvalues of both frontier orbitals are for a given torsional angle always lower for the fluorinated than for the non-fluorinated structure. This is in good agreement with previous findings10 from DFT calculations that describe a systematic stabilization of the frontier orbitals of oligoacenes, oligothiophenes, and oligophenylenevinylenes (all treated as planar systems) due to fluorination. However, if we focus on the orbital energies of the equilibrium structures, the situation changes and an explanation for the seemingly contradicting trend of the PVH and PVF eigenvalues (as shown in Figure 7 and discussed there below) can be found. The HOMO energy of DHS at 0° is above the HOMO energy of DHSF. Because the fluorination induced torsion further decreases EHOMO, the difference is even enlarged for the relaxed DHSF (≈45° torsion). At 0° torsion, the ELUMO value of DHSF is below the eigenvalue of DHS, but because the LUMO energy increases for larger torsional angles, a larger eigenvalue for DHSF than for DHS at their corresponding equilibrium geometries can be found. (ii) The HOMO energies of both structures are lowered during rotation, until a minimum is reached at 90°, whereas the corresponding LUMO energies are destabilized and a maximum is found at 90° torsion. This finding can be easily explained from the orbitals’ character, which are very similar in shape to the corresponding orbitals of the oligomer systems described above. In the HOMO of the planar conformations (upper part of Figure 11), there is a binding π contribution located on the vinylene double bond, whereas the interaction of this orbital fragment with the orbital fragments on the six-membered rings is of antibonding character. The two resulting nodal planes are located on the carbon-carbon single bonds, centering the two dihedral angles that are changed during rotation. Thus, rotating the lateral rings reduces the antibonding interaction until finally at 90° (lower part of Figure 11) zero overlap and a minimum

Optical Spectra of Fluorinated PPVs of the eigenvalues is reached. For the LUMOs, the situation is opposite. Here, a nodal plane is found in the center of the vinylene bridge and binding π fragments are located on the vinylene-ring bonds. Rotation weakens this bonding interaction and destabilizes the orbitals. (iii) The decrease of HOMO energies and the increase of LUMO energies, when the phenyl substituents are rotated, adds up to an increase of the HOMO-LUMO gap size from ≈6 eV at 0° to ≈8 eV at 90° (see Figure 11c). The lowest TD-DFT excitation energy (see Figure 11d) follows the HOMO-LUMO energy-gap variation with the torsional angle, and thus, the increasing gap size explains the blue-shift of structures featuring large torsional distortion. (iv) While the HOMO energies of the fluorinated and nonfluorinated systems change rather uniformly during rotation (the distance between the two curves being almost constant), the LUMO of DHS becomes more destabilized than the LUMO of its fluorinated counterpart DHSF. This effect has an impact on the energetic gap. At torsional angles smaller than 45°, the gap of DHS is found at lower values than the gap of DHSF, and at angles larger than 45°, it is opposite. Noteworthy, this does not change back between 90 and 180°. In the first order approximation, this finding means that, at small torsion angles, fluorination yields a blue-shift while, at larger torsion angles, a red-shift. This finding explains the results reported in Table 2.

J. Phys. Chem. B, Vol. 112, No. 10, 2008 3003 angles, is about 1 order of magnitude larger than the pure substituent effect. Thus, fluorination of PPV turned out to be a successful structural modification for achieving blue emission from PPV polymers. The here reported investigation demonstrates that this effect can be completely attributed to major changes in the conjugated backbone molecular geometry induced by the presence of the fluorine atoms. This effect can only be reproduced in theoretical calculations, if fully relaxed geometries and no planar model systems are employed. The experimentally made observations could be explained in detail by the DFT calculations. An interesting challenge for a further investigation of the fluorination effect is the synthesis of short oligomers and their photophysical characterization including absorption and photoluminescence behavior as well as the direct comparison with calculations of excitation energies in the ground and excited state geometry. Acknowledgment. We thank R. Ahlrichs for providing us with the TURBOMOLE program package and G. Aloisio for his support. Calculations have been carried out on a HP XC6000 cluster of the SPACI Consortium (Lecce). This work is partially funded by the MIUR FIRB 2003 ‘SYNERGY’ grant. References and Notes

5. Conclusions In this article, experimental absorption spectra of fluorinated polyphenylenevinylenes and theoretical calculations for fluorinated oligophenylenevinylenes in comparison with their nonfluorinated counterparts have been presented. The employment of the BH-LYP hybrid functional has been justified by comparing three different density functionals, including different amounts of “exact” exchange with experimental results. Both experimental absorption spectra and TD-DFT calculations show that fluorination of the vinylene bridges leads to a large blue-shift (about 0.9 eV) of the S1 excitation energy and a loss of intensity. This blue-shift is present in both absorption and emission spectroscopy measurements in solution of different concentrations and solvents. TD-DFT calculations of oligomer model systems allow an estimation of the optical properties of the polymer. According to the DFT calculations, the blue-shift can be clearly attributed to torsional distortion, which is induced by steric repulsion between the fluorine atoms and their neighboring alkoxy substituents. The electron-withdrawing effect induced by the fluorine atoms is very small and negative (about -0.1 eV); i.e., it reduces the excitation energies. The torsional potential of the model trans-stilbene systems revealed that the two local minima of the hydrogenated systems are found at 0° and at 180° torsion. Both are separated by a rather high barrier of ≈10 kcal/mol. The local minima of the fluorinated structures are shifted to 45 and 135°, and the intermediate barrier is lowered to ≈3 kcal/mol. Such a small energy barrier can be easily overcome due to environmental effects. Thus, we expected that in the solid state different polymorphs of MEH-PPVF can be obtained by the variation of the individual growth conditions,59 yielding quite different absorption and photoluminescence spectra. Investigation of the HOMO-LUMO energy gaps has shown that the pure substituent effect of the fluorination leads at small torsional angles to a blue-shift and at larger torsional angles to a red-shift. The blue-shift, induced by the change in the dihedral

(1) Hrenar, T.; Mitric´, R.; Meic´, Z.; Meier, H.; Stalmach, U. J. Mol. Struct. 2003, 661-662, 33-40. (2) Han, Y.-K.; Lee, S. U. J. Chem. Phys. 2004, 121, 609-611. (3) Scurlock, R. D.; Wang, B.; Ogilby, P. R.; Sheats, J. R.; Clough, R. L. J. Am. Chem. Soc. 1995, 117, 10194. (4) Babudri, F.; Farinola, G. M.; Naso, F.; Ragni, R. Chem. Commun. 2007, 10, 1003-1022. (5) Sakamoto, Y.; Komatso, S.; Suzuki, T. J. Am. Chem. Soc. 2001, 123, 4643-4644. (6) Sakamoto, Y.; Komatso, S.; Suzuki, T. Synth. Met. 2003, 133134, 361-363. (7) Strehmel, B.; Sarker, A. M.; Malpert, J. H.; Strehmel, V.; Seifert, H.; Neckers, D. C. J. Am. Chem. Soc. 1999, 121, 1226-1236. (8) Chen, H.-Y.; Chao, I. Chem. Phys. Lett. 2005, 401, 539-545. (9) Gierschner, J.; Cornil, J.; Egelhaaf, H.-J. AdV. Mater. 2007, 19, 173-191. (10) Medina, B. M.; Beljonne, D.; Egelhaaf, H.-J.; Gierschner, J. J. Chem. Phys. 2007, 126, 111101. (11) Jin, Y.; Kim, J.; Lee, S.; Kim, J. Y.; Park, S. H.; Lee, K.; Suh, H. Macromolecules 2004, 37, 6711-6715. (12) Babudri, F.; Cardone, A.; Farinola, G. M.; Martinelli, C.; Naso, F. Unpublished results. (13) Casida, M. E. Recent AdVances in Density Functional Methods; World Scientific: Singapore, 1995; Vol. 1. (14) Bauernschmitt, R.; Ahlrichs, R. Chem. Phys. Lett. 1996, 256, 454. (15) Marques, M. A. L., Ullrich, C. A., Noguera, F., Rubio, A., Burke, K., Gross, E. K. U., Ed. Time-Dependent Density Functional Theory; Springer: Berlin, Heidelberg, 2006. (16) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys. 1998, 108 (11), 4439-4449. (17) Tozer, D. J.; Handy, N. C. J. Chem. Phys. 1998, 109 (23), 1018010189. (18) Cai, Z.-L.; Sendt, K.; Reimers, J. R. J. Chem. Phys. 2002, 117 (12), 5543-5549. (19) Deleuze, M. S. J. Chem. Phys. 2002, 116 (16), 7012-7026. (20) Dreuw, A.; Fleming, G. R.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2003, 5, 3247-3256. (21) Dreuw, A.; Fleming, G. R.; Head-Gordon, M. J. Phys. Chem. B 2003, 107, 6500-6503. (22) Grimme, S.; Parac, M. ChemPhysChem 2003, 3, 292. (23) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. J. Chem. Phys. 2003, 119 (6), 2943-2946. (24) Dreuw, A.; Head-Gordon, M. J. Am. Chem. Soc. 2004, 126, 40074016. (25) Gritsenko, O.; Baerends, E. J. J. Chem. Phys. 2004, 655 (2), 655. (26) Deleuze, M. S. J. Phys. Chem. A 2004, 108, 9244-9259. (27) Cornil, J.; Beljonne, D.; Shuai, Z.; Hagler, T. W.; Campbell, I.; Bradley, D. D. C.; Bre´das, J. L.; Spangler, C. W.; Mu¨llen, K. Chem. Phys. Lett. 1995, 247, 425-432.

3004 J. Phys. Chem. B, Vol. 112, No. 10, 2008 (28) Cornil, J.; Beljonne, D.; Heller, C. M.; Campbell, I. H.; Laurich, B. K.; Smith, D. L.; Bradley, D. D. C.; Mu¨llen, K.; Bre´das, J. L. Chem. Phys. Lett. 1997, 278, 139-145. (29) van der Horst, J.-W.; Bobbert, P. A.; Michels, M. A. J.; Brocks, G.; Kelly, P. J. Phys. ReV. Lett. 1999, 83 (21), 4413. (30) Albota, M.; Beljonne, D.; Bre´das, J.-L.; Ehrlich, J. E.; Fu, J.-Y.; Heikal, A. A.; Hess, S. E.; Kogej T.; Levin, M. D.; Marder, S. R.; McCordMaughon, D.; Perry, J. W.; Ro¨ckel, H.; Rumi, M.; Subramaniam, G.; Webb, W. W.; Wu, X.-L.; Xu, C. Science 1998, 281, 1653-1656. (31) Renak, M. L.; Bartholomew, G. P.; Wang, S.; Ricatto, P. J.; Lachicotte, R. J.; Bazan, G. C. J. Am. Chem. Soc. 1999, 121, 7787-7799. (32) Kwasniewski, S. P.; Deleuze, M. S.; Franc¸ ois, J. P. Int. J. Quantum Chem. 2000, 80, 672-680. (33) Grozema, F. C.; Telesca, R.; Snijders, J. G.; Siebbeles, L. D. A. J. Chem. Phys. 2003, 118, 9441-9446. (34) Tretiak, S.; Saxena, A.; Martin, R. L.; Bishop, A. R. Phys. ReV. Lett. 2002, 89 (9), 097402. (35) Masunov, A.; Tretiak, S. J. Phys. Chem. B 2004, 108, 899-907. (36) Tretiak, S.; Igumeshchev, K.; Chernyak, V. Phys. ReV. B 2005, 71, 033201. (37) Ng, M.-F.; Sun, S. L.; Zhang, R. Q. J. Appl. Phys. 2005, 97, 103513. (38) Weimer, M.; Heiringer, W.; Della Sala, F.; Go¨rling, A. Chem. Phys. 2005, 309. (39) Fabiano, E.; Della Sala, F.; Cingolani, R.; Weimer, M.; Go¨rling, A. J. Phys. Chem. A 2005, 109, 3078. (40) Grimme, S.; Waletzke, M. J. Chem. Phys. 1999, 111, 5645. (41) Cave, R. J.; Zhang, F.; Maitra, N. T.; Burke, K. Chem. Phys. Lett. 2004, 389, 39-42. (42) Della Sala, F.; Vitale, V.; Mazzeo, M.; Cingolani, R.; Gigli, G.; Barbarella, G.; Favaretto, L. J. Non-Cryst. Solids 2006, 352, 2461. (43) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (44) Ahlrichs, R.; Ba¨r, M.; Baron, H.-P.; Bauern-schmitt, R.; Bo¨cker, S.; Ehrig, M.; Eichkorn, K.; Elliott, S.; Furche, F.; Haase, F.; Ha¨ser, M.; Horn, H.; Huber, C.; Huniar, U.; Kattannek, M.; Ko¨lmel, C.; Kollwitz, M.; May, K.; Ochsenfeld, C.; O ¨ hm, H.; Scha¨fer, A.; Schneider, U.; Treutler, O.; von Arnim, M.; Weigend, F.; Weis, P.; Weiss, H. TURBOMOLE, version 5.9; Universita¨t Karlsruhe: Karlsruhe, Germany, 2007. See also http:// www.turbomole.com.

Piacenza et al. (45) Treutler O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346. (46) Scha¨fer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829. (47) The here applied basis sets are available from the TURBOMOLE homepage via the FTP Server Button (in the subdirectories basen, jbasen, and cbasen). See http://www.turbomole.com. (48) Becke, A. D. Phys. ReV. A 1988, 38, 3098. (49) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37:785. (50) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (51) Walsh, A. D. Proc. R. Soc. London, Ser. A 1947, 191 (1024), 3238. (52) Champagne, B. B.; Pfanstiel, J. F.; Plusquellic, D. F.; Pratt, D. W.; van Herpen, W. M.; Meerts, W. L. J. Phys. Chem. 1990, 94, 6-8. (53) Peeters, E.; Marcos Ramos, A.; Meskers, S. C. J.; Janssen, R. A. J. J. Chem. Phys. 2000, 112 (21), 9445-9454. (54) Kwasniewski, S. P.; Franc¸ ois, J. P.; Deleuze M. S. Int. J. Quantum Chem. 2001, 85, 557-568. (55) Kwasniewski, S. P.; Franc¸ ois, J. P.; Deleuze, M. S. J. Phys. Chem. A 2003, 107, 5168-5180. (56) Dierksen, M.; Grimme, S. J. Phys. Chem. A 2004, 108, 1022510237. (57) Santoro, F.; Lami, A.; Improta, R.; Barone, V. J. Chem. Phys. 2007, 120, 3544. (58) Della Sala, F. H.; Heinze, H.; Go¨rling, A. Chem. Phys. Lett. 2001, 339, 343. (59) Viola, I.; Della Sala, F.; Piacenza, M.; Favaretto, L.; Gazzano, M.; Anni, M.; Barbarella, G.; Cingolani, R.; Gigli, G. AdV. Mater. 2007, 19, 1597. (60) Piacenza, M.; Della Sala, F.; Fabiano, E.; Maiolo, T.; Gigli, G. J. Comput. Chem., 2008, 29, 451-457. (61) Kauffmann H.-F.; Lukes, V.; Aquino, A. J. A.; Lischka, H. J. Phys. Chem. B 2007, 111, 7954-7962. (62) Kwasniewski, S. P.; Franc¸ ois, J. P.; Deleuze, M. S. J. Phys. Chem. A 2003, 107, 5168-5180. (63) Kwasniewski, S. P.; Claes, L.; Franc¸ ois, J. P.; Deleuze, M. S. J. Chem. Phys. 2003, 118 (17), 7823-7836. (64) Catala´n, J. Chem. Phys. Lett. 2006, 421, 134-137.