ARTICLE pubs.acs.org/JPCA
Paracyclophanes as Model Compounds for Strongly Interacting π-Systems, Part 3: Influence of the Substitution Pattern on Photoabsorption Properties Johannes Pfister,† Christof Schon,† Wolfgang Roth, Conrad Kaiser,‡ Christoph Lambert,‡ Katrin Gruss,§ Holger Braunschweig,§ Ingo Fischer,† Reinhold F. Fink,† and Bernd Engels*,† †
Institut f€ur Physikalische und Theoretische Chemie, Universit€at W€urzburg, Am Hubland, D-97074 W€urzburg, Germany Institut f€ur Organische Chemie, Universit€at W€urzburg, Am Hubland, D-97074 W€urzburg, Germany § Institut f€ur Anorganische Chemie, Universit€at W€urzburg, Am Hubland, D-97074 W€urzburg, Germany ‡
bS Supporting Information ABSTRACT: The structures and energetics of the ground and first excited states of [2.2]paracyclophane (PC) and its pseudopara- (p-DHPC) and pseudo-ortho-dihydroxy (o-DHPC) as well as monohydroxy derivates (MHPC) are investigated by quantum chemical calculations, X-ray crystallography, and resonance-enhanced multiphoton ionization spectroscopy (REMPI) in a free jet. We show that substitution of the aromatic hydrogens in PC causes significant changes of the structure and in particular its change between the ground and the excited state. The structural changes include a breathing mode as well as shift and rotation of the benzene moieties and are rationalized by the electronic structure changes upon excitation. Spin-component-scaled second-order MøllerPlesset perturbation method (SCS-MP2) reproduces the experimental X-ray structure correctly and performs significantly better than ordinary MP2 and the B3LYP methods. The parent propagation method, SCS-approximate coupled cluster second order (SCS-CC2), yields adiabatic excitation energies within 0.1 eV of the experimental values for PC and the investigated hydroxyl derivates as well as the related aromatic molecules benzene and phenol. It is shown that zero-point vibration energy corrections at the time dependent density functional (B3LYP) level are no more accurate enough for that level of theory and have to be substituted by SCS-CC2 values. While the structures of PC and o-DHPC are only slightly modified upon excitation, p-DHPC changes its structural parameters substantially. This is in line with [1 þ 1] REMPI-spectra of these substances, which are interpreted with the help of FranckCondon simulations.
1. INTRODUCTION π-Conjugated molecules are widely used as materials due to their applicability in organic solar cells,1 organic field effect transistors (OFET),2,3 organic light emitting diodes (OLED),4 and other (opto-)electronic devices. The material properties of these devices5,6 are owing to their capability of electronic switching by electric fields,7 electromechanical forces,8 or photoactive/ photochromic7,9 response. Advances in the design of these materials, known as molecular nanotechnology,10 depend on an improved control of the interplay between orientation of molecules relative to one another, the photophysics of the individual chromophores, as well as energy and charge transport in the bulk.11 Typically thin films12,13 of organic materials are used as device components. However, it is challenging to understand these amorphous materials in detail as the molecular subunits experience a range of environments and generally lack a periodic long-range order. Thus, small, well-defined model compounds are very well suited to understand fine details of the fundamental processes that give rise to the material properties. r 2011 American Chemical Society
[2.2]Paracyclophane (PC) and its derivatives are unique with respect to the strong “through space” coupling of the π-systems of their benzene moieties and are thus promising candidates as materials due to their specific photophysical properties1418 and their high electric conductivity.1921 PC is known since 194922 and has been widely modified.2329 These compounds can be regarded as models for closely packed π-systems bound together by organic bridges. The influence of the distance between the two benzene subunits on photoabsorption properties can be investigated by varying the bridge length and/or the substitution pattern at the π-systems. In this work we continue our research on ground state and excited state structures as well as the energetics of PC and its hydroxy derivatives monohydroxy-paracyclophane (MHPC), pseudo-ortho-dihydroxy[2.2]paracyclophane (o-DHPC), and Received: January 26, 2011 Revised: February 25, 2011 Published: March 29, 2011 3583
dx.doi.org/10.1021/jp200823q | J. Phys. Chem. A 2011, 115, 3583–3591
The Journal of Physical Chemistry A pseudo-para-dihydroxy[2.2]paracyclophane (p-DHPC). Preceeding work30,31 addressed already the comparison between PC, MHPC, and o-DHPC. In the present work we investigate the influence of symmetric dihydroxy substitution in the example p-DHPC and show that this compound behaves rather differently than the other paracyclophanes. Our theoretical results are compared for the case of p-DHPC and o-DHPC with resonance-enhanced multiphoton-ionization (REMPI) in a free jet. Comparable REMPI investigations of PC32 and MHPC30 have been published elsewhere. A quantitative description of the interaction of π-systems is already a challenge for standard quantum chemical methods, even if only ground states are considered.33 We employ the spincomponent-scaled second-order MøllerPlesset perturbation method (SCS-MP2)34 for the ground states and the related spin-component-scaled variant of the approximate coupled cluster second order method (SCS-CC2) for the excited states. Both approaches have been successfully applied to π-conjugated systems.3538 The results provide insight into geometric as well as electronic structures of the four model compounds and their changes upon excitation. The paper is structured as follows: in the Methods section the synthesis and crystal structure determination of p-DHPC and o-DHPC are described as well as computational details and the setup of the REMPI experiment. In the Results section experimental and calculated structures, excitation energies, and REMPI spectra are compared with each other. Additionally, two-dimensional potential energy surfaces for the two lowest vibrational modes corresponding to a twist and a shift of the benzene units are presented. The influence of o-DHPC impurities in p-DHPC is addressed by NMR and REMPI mixture experiments. The Discussion and Conclusion section explains the structure and structural changes between ground and excited states based on the frontier molecular orbitals as well as electrostatic potentials. The article closes with a summary.
2. METHODS 2.1. Synthesis and Crystal Structure Determination. The synthesis of p-DHPC via iron-catalyzed dibromination of PC followed by dilithiation with butyllithium and oxidative quenching with nitrophenol has been carried out as described in the literature.39 For purification of p-DHPC flash chromatography was employed. To remove possible remaining traces of o-DHPC the substances were recrystallized from ethanol. NMR experiments (see Supporting Information) showed that after purification the contamination by o-DHPC is less than 0.3%. Racemic o-DHPC (97% purity) was acquired from ABCR GmbH. Single crystals suitable for X-ray analysis were obtained by recrystallization from methanol. For both compounds the crystal structure data were collected at a Bruker D8 diffractometer with an Apex CCD area detector and graphite monochromated MoKR radiation. The structure was solved using direct methods, refined with the Shelx software package40 and expanded using Fourier techniques. All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were assigned idealized positions and were included in structure factors calculations. Crystal data for p-DHPC 3 (EtOH)2: C20H28O4, Mr = 332.42, colorless block, 0.37 0.36 0.35 mm3, monoclinic space group P21/c, a = 8.2875(11) Å, b = 13.1716(18) Å, c = 8.3272(11) Å, β = 102.735(2)°, V = 886.6(2) Å3, Z = 2, Fcalcd = 1.245 g 3 cm3, μ = 0.085 mm1, F(000) = 360, T = 168(2) K, R1 = 0.0457,
ARTICLE
wR2 = 0.1248, 1758 independent reflections [2θ e 52.16°], and 112 parameters. Crystal data for o-DHPC 3 0.5MeOH: C16.5H18O2.5, Mr = 256.31, colorless block, 0.43 0.26 0.185 mm3, monoclinic space group C2/c, a = 24.937(5)Å, b = 7.8461(14)Å, c = 13.183(2)Å, β = 95.655(3)°, V = 2566.8(8)Å3, Z = 8, Fcalcd = 1.326 g 3 cm3, μ = 0.088 mm1, F(000) = 1096, T = 172(2) K, R1 = 0.0644, wR2 = 0.1961, 3210 independent reflections [2θ e 56.68°], and 182 parameters. Crystallographic data have been deposited with the Cambridge Crystallographic Data Center as supplementary publication no. CCDC-809116 (p-DHPC) and CCDC-809117 (o-DHPC). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc. cam.ac.uk/data_request/cif. 2.2. Computational Details. The structures of all molecules were optimized in the ground and first singlet excited states. A variety of density functional and wave function based ab initio methods have been employed. The best results are provided by the SCS-MP234 method for the ground states and with the related SCS-CC235,4143 method for the excited states. In this work these approaches are used for the interpretation of the experimental results, while the other methods are only mentioned for comparison. All calculations were performed with the TURBOMOLE program suite Version 5.10.44 The TZV45 basis sets of Sch€afer et al. were augmented with the (2df,p) polarization functions for the MP2 and CC2 calculations and the (2df,2pd) set of polarization functions for density functional calculations.46 All MP2 and CC2 calculations employed the resolution of the identity (RI) approximation47,48 using the auxiliary basis sets of Weigend et al.46 The zero-point vibrational energies of the ground and excited states were calculated at the (SCS)-CC2 level of theory using a SV(P) basis set.49 The vibrational frequencies were scaled by a factor of 0.95, which is commonly applied on this level of theory to account for anharmonicity of the potentials and errors of the method.5052 Electrostatic potentials have been generated by the RICC2 module of the TURBOMOLE program suite. Simulation of the REMPI spectra was done with the parallel mode approximation53 as described by Malagoli et al.54 Neglect of the Duschinsky effect makes it possible to factorize the FranckCondon factors and to evaluate the intensity of a multidimensional vibrational transition very efficiently as a simple product of one-dimensional FranckCondon integrals (FCI).55 The square of the FCI is referred to as the FranckCondon Factor (FCF). As we consider only transitions from the vibrational ground state, the contribution of a single vibrational mode to the band is given by a standard Poisson distribution.54 A convolution of these distributions provides the total simulated spectrum. The specification of the irreducible representations of the D2 and D2h symmetric species is ambiguous as the three C2 axes in these point groups are equivalent such that the main axis is not defined. Following the idea of the proposal of Mulliken56 and the recommendation of the IUPAC convention,57 we define the C2 axis through the center of the (CH2)2 bridge as z, the axis through the two benzene rings as y, and the remaining axis as x. 2.3. Setup of the REMPI Experiment. The experimental setup has been described previously. For detailed information we refer to our previous publication.30,31 In brief, the experiments were carried out in a differentially pumped jet apparatus equipped with a time-of-flight mass spectrometer. p-DHPC, synthesized as described above, and racemic o-DHPC, 3584
dx.doi.org/10.1021/jp200823q |J. Phys. Chem. A 2011, 115, 3583–3591
The Journal of Physical Chemistry A
Figure 1. Experimental REMPI spectra of o-DHPC (a) and p-DHPC (b) as well as the calculated spectrum of p-DHPC (c). The intensity is given in arbitrary units for the experimental spectra and based on FranckCondon factors for the simulated spectrum of p-DHPC. Dagger symbols denote peaks from a fragmentation of the water cluster.
commercially obtained from ABCR, were spread out on glass wool and placed in the sample compartment of a modified solenoid valve. The source was heated to approximately 165 °C (150 °C) to evaporate the p-DHPC (o-DHPC), which was seeded in 1.1 bar of Ar. The molecular beam was formed and cooled by expanding the mixture of cyclophane and carrier gas into the vacuum through a 0.5 mm diameter nozzle. For REMPI, a nanosecond laser system consisting of a 10 Hz Nd:YAG laser pumping a frequency-doubled grazing incidence dye laser was used. The UV energy of the laser was varied between 0.5 and 5 mJ/pulse. Wavenumbers were calibrated with an optogalvanic lamp, filled with neon.
ARTICLE
Figure 2. Experimental REMPI spectra of (a) pure p-DHPC (intensitiy multiplied by a factor of 25 for a better comparability) and (b) a 6:1 mixture between p-DHPC and o-DHPC. Small amounts of o-DHPC dominate the p-DHPC spectrum. The daggers indicate signals that are due to water clusters.
Table 1. Relevant Structure Parameters of the Ground (gs) and First Excited States (es) of PC, o-DHPC, and p-DHPCa molecule parameter experiment B3LYP MP2/CC2 SCS-MP2/CC2 PC gs/es
159/---
161/158
159/157
159/159
278/--310/---
283/253 316/278
274/253 304/278
276/257 307/283
R(C8C15)
310/---
316/278
304/278
307/283
θtwist
9/---
0/10
14/10
13/7
θshift
0/---
0/0
0/0
0/0
157/ ---
160/162
156/156
159/157
R(C3C14)
277/---
282/279
273/253
274/258
R(C4C13)
302/---
307/306
293/272
296/280
R(C8C15) θtwist
308/--9/---
320/299 12/17
309/281 16/13
311/286 16/13
θshift
6/---
1/2
1/0
0/0
159/---
160/158
159/157
159/157
R(C3C14)
276/---
281/265
271/252
274/256
R(C4C13)
305/---
313/293
298/276
302/281
R(C8C15)
305/---
312/298
297/280
301/286
θtwist
0/---
0/0
0/15
0/12
θshift
13/---
12/4
21/4
20/2
o-DHPC R(C1C2) gs/es
3. RESULTS 3.1. Experimental Results. In Figure 1 the REMPI spectra of p-DHPC and o-DHPC are shown. For comparison a simulated spectrum of p-DHPC based on the parallel mode approximation is also shown. Surprisingly, the experimental spectra of o-DHPC and p-DHPC look very much alike. The REMPI and spectral hole burning (SHB) spectra of o-DHPC were already discussed in our earlier publication.31 Thus, only the main results are briefly summarized here: The intense band at 31483 cm1 (3.90 eV) was assigned to the S1rS0 origin of o-DHPC. The signals marked with a dagger in Figure 1 are due to o-DHPC 3 H2O which is formed under the evaporation conditions as a water molecule can bind to o-DHPC via two hydrogen bonds resulting in a particularly large binding energy. As visible, all major peaks of the o-DHPC spectrum are also visible in the p-DHPC spectrum. In addition the experimental spectrum of p-DHPC shows a large number of closely spaced peaks with comparably small intensities. It seems that the p-DHPC spectrum actually shows signals that are due to contaminations of o-DHPC. To test the hypothesis that only the closely spaced peaks in the REMPI spectra of p-DHPC are actually due to the substance while the major peaks result from o-DHPC impurities, we recorded a REMPI spectrum of a 6:1 mixture of p-DHPC and o-DHPC; see Figure 2. As visible, the signal intensity of the major peaks increases by a factor of
R(C1C2) R(C3C14) R(C4C13)
p-DHPC R(C1C2) gs/es
a
Distances (R) are given in pm and angles (θ, definition see text) in degrees.
25 upon adding 16% of o-DHPC. This confirms that the carrier of the intense bands is indeed o-DHPC. The peaks that can be assigned to p-DHPC are lost in the background noise of the mixture experiment. Note that small impurities of the para isomer do not affect the spectroscopy of the ortho isomer. The spectral features of p-DHPC in the REMPI spectrum are obviously extremely weak which can be readily explained by the significant relaxation of that compound upon excitation (see below). The first band of p-DHPC that appears reproducibly above the noise level was found at 31272 cm1 (3.88 eV). This band is absent in the o-DHPC spectrum. However, the calculated strong geometry change of p-DHPC upon excitation (see below and Table 1) and the slowly rising intensity indicate that the 3585
dx.doi.org/10.1021/jp200823q |J. Phys. Chem. A 2011, 115, 3583–3591
The Journal of Physical Chemistry A
ARTICLE
Figure 4. Molecular structure and atom labels of the EZ rotamer of p-DHPC.
Figure 3. PC, MHPC (E rotamer), o-DHPC (EZ rotamer), and p-DHPC (EE rotamer) shown in different perspectives in their groundand excited-state structures.
origin of the S1rS0 transition in p-DHPC transition lies most likely at lower excitation energies. This value of 31272 cm1 thus constitutes an upper bound for the adiabatic transition of p-DHPC, which lies in any case at significantly lower excitation energies than that of o-DHPC. Above about 31700 cm1 p-DHPC gives rise to closely spaced signals in the REMPI spectrum that form almost a continuous background such that no individual bands can be assigned. This background is an order of magnitude smaller in the spectrum of o-DHPC. A possible explanation for the appearance of o-DHPC in the p-DHPC spectrum could be a para-to-ortho isomerization under the evaporation conditions in the REMPI experiment. We therefore sublimated and recondensated p-DHPC and recorded a NMR spectrum of this sample. The spectrum recorded after this procedure was identical to the original one, which rules out that a significant amount of p-DHPC isomerizes to the ortho compound. The purity of p-DHPC was determined by NMR experiments (see Supporting Information) showing that the contamination of o-DHPC in the measured samples of p-DHPC must be certainly less than 0.3%. In conclusion, experiments are insufficient to determine the properties of the first excited state of p-DHPC, since the p-DHPC spectrum is dominated by traces of o-DHPC. Thus, a theoretical approach is necessary for an understanding of the electronic excitation. 3.2. Computational Results. 3.2.1. Ground-State Properties. The ground-state structures are shown in the lower part of Figure 3, while the most relevant structural parameters are collected in Table 1. Figure 4 gives the numbering of the atoms represented by p-DHPC. Although the geometry of the carbon framework is well-defined in o-DHPC and p-DHPC, three rotational isomers are possible, differing in the orientation of the hydroxyl group relative to the (CH2)2 bridge: They can be termed ZZ, EZ, and EE isomers where E (Z) designates a hydrogen atom pointing away from (toward) the bridge. For p-DHPC the EE isomer was calculated to be 3 and 6 kJ/mol more stable than the EZ and ZZ isomers, respectively. Thus, the EE isomer represents the ground-state minimum structure and will be the isomer we discuss throughout the rest of this article. Since its energy is substantially lower than that of the EZ and ZZ isomers it will most likely be dominant at the temperatures of the
free jet expansion experiment. For o-DHPC on the other hand EZ is the most stable conformation. At the SCS-MP2 level it is 0.4 and 4.8 kJ/mol lower in energy than the EE and ZZ isomers, respectively. This difference of 0.4 kJ/mol between the EZ and EE isomers is too small to exclude a contribution of the EEisomer to the REMPI spectra.31 To describe the ππ interaction in the paracyclophanes the distances R(C3C14), R(C4C13), and R(C8C15) are going to be considered. Calculated results agree very well with the X-ray structures.34,5860 In all paracyclophanes the two carbon atoms standing via-a-vis are located only 2.83.1 Å apart from each other, which is below the van der Waals distance of π-systems (about 3.5 Å). PC is distorted from the eclipsed structure with D2h symmetry by a counter-rotation of the benzene rings58,61 resulting in a D2 symmetry. The EZ and EE isomers of o-DHPC have a similar structure as PC. In addition to the twisted distortion, the hydroxyl-functions in o-DHPC attract each other resulting in a small tilt between the phenol rings. The EZ and EE isomers belong to the C1 and C2 point groups, respectively. The structure of p-DHPC differs from PC and o-DHPC. Its rings are not counter-rotated but shifted, and therefore p-DHPC has Ci symmetry. Such structures have been observed previously only for different pseudo-para-[2.2]paracyclophanes2,62 but not for their constitutional isomers pseudo-gem-, pseudo-ortho-, or pseudo-meta-[2.2]paracyclophanes. The latter molecules all show the D2 skeleton of PC with twisted (CH2)2 bridges.6365 To quantify the structural difference between the compounds, we calculated two-dimensional potential energy surfaces as a function of the two internal coordinates θtwist and θshift that correspond to the twist and shift motion of the benzene rings. As in our preceding work30,31 these motions are represented by appropriate linear combinations of dihedral angles as 1 θtwist ¼ ½θðC1 C2 C3 C4 Þ þ θðC1 C2 C3 C8 Þ 8 þ θðC2 C1 C14 C13 Þ þ θðC2 C1 C14 C15 Þ þ θðC10 C9 C6 C5 Þ þ θðC10 C9 C6 C7 Þ þ θðC9 C10 C11 C12 Þ þ θðC9 C10 C11 C16 Þ ð1Þ and 1 θshift ¼ ½θðC1 C2 C3 C4 Þ þ θðC1 C2 C3 C8 Þ 8 þ θðC2 C1 C14 C13 Þ þ θðC2 C1 C14 C15 Þ θðC10 C9 C6 C5 Þ θðC10 C9 C6 C7 Þ θðC9 C10 C11 C12 Þ θðC9 C10 C11 C16 Þ ð2Þ 3586
dx.doi.org/10.1021/jp200823q |J. Phys. Chem. A 2011, 115, 3583–3591
The Journal of Physical Chemistry A
ARTICLE
Figure 5. Ground-state potential energy surfaces for the angles θtwist and θshift for the PC (top-left), MHPC (bottom-left), p-DHPC (top-right), and oDHPC (bottom-right) molecules.
where θ(CaCbCcCd) represents the dihedral angle between the respective atoms;66 see Figure 4 for the numbering of atoms. This definition ensures that the twisting and shifting motions are performed in a symmetric fashion. Figure 5 shows the potential energy surfaces for all compounds obtained by optimizing the structure for fixed θtwist and θshift at the SCSMP2/TZV(2df,p) level of theory. The global minima can also be taken from Table 1. The distortion from the eclipsed structure (D2h, θtwist = 0° and θshift = 0°) to the corresponding minimum results in an energy gain of 1.2, 4.6, 5.3, and 7.2 kJ/mol for PC, MHPC, o-DHPC, and p-DHPC, respectively. These numbers and Figure 5 demonstrate that PC itself is very flexible with respect to the twist and shift motions. This has already been pointed out in previous investigations.33,61,67,68 It should be noted, however, that substitution leads to significantly more rigid derivatives. Figure 6a shows the ground-state electrostatic potential in the plane between the benzene subunits of PC in its eclipsed D2h structure. X denotes the positions of the hydroxyl groups in pDHPC. Positive charge is accumulated around the (CH2)2 bridges and negative charge is located between the nonbridge aryl C atoms and its bonded H atoms. 3.2.2. Excited-State Properties. In the first singlet excited states of all compounds the benzene subunits move toward each other. According to our SCS-CC2/TZV(2df,p) calculation the interring distances decrease by 1624 pm.31,33 In PC and o-DHPC beside the interring distance only θtwist decrease from 13 to 7° and 16 to 13°, respectively. Nevertheless, both molecules retain their ground state symmetry. In contrast to this, the equilibrium structure of p-DHPC changes dramatically upon excitation. On the one hand the shift is essentially removed (θshift = 20° f θshift = 2°), on the
other hand the excited state structure exhibits a twist distortion (θtwist = 12°, see also Table 1) that does not exist in the electronic ground state. The electrostatic potentials of the ground and excited states of PC in its eclipsed D2h structure are shown in parts a and b of Figure 6. As discussed below, for the substituted compounds the structural changes after excitation can be most easily discussed with the electrostatic potential difference shown in figure 6c. It illustrates that the excitation goes along with an increase in negative charge in the bonds between the nonbridge aryl C atoms and their corresponding H atoms, while negative charge decreases in the symmetry center of this D2h structure. In Table 2 the S1rS0 experimental and calculated excitation energies of the investigated paracyclophanes as well as benzene and phenol are collected. All values include zero-point vibrational energies (ZPE) as evaluated at the corresponding level with a SV(P) basis set. The SCS-CC2 method provides adiabatic excitation energies of 3.71 eV for EE-p-DHPC and 3.87 eV (31204 cm1) for the most stable rotamer of o-DHPC, while the calculated excitation energy of the less stable EE (ZZ) isomer is 53 cm1 larger (119 cm1 smaller) than this. ZPE corrections for the SCS-CC2 method are shown in Table 3: The excitation energies of all compounds are predicted within better than 0.05 eV, with the exception of p-DHPC, where the S1rS0 origin cannot be determined from the experimental data. An analysis of the contributions to the adiabatic excitation energies shows that although p-DHPC has the larger vertical excitation energy, its adiabatic excitation energy is the lowest due to large relaxation energy λes in the excited state (0.45 eV, whereas λes for the other compounds ranges between 0.27 and 0.35 eV). 3587
dx.doi.org/10.1021/jp200823q |J. Phys. Chem. A 2011, 115, 3583–3591
The Journal of Physical Chemistry A
ARTICLE
Table 2. Adiabatic Excitation Energies Including ZPE (All Values in eV) method
benzene
phenol
PC
o-DHPC
p-DHPC
TD-B3-LYP
5.70
4.61
3.89
3.94
3.73
CC2 SCS-CC2
4.95 4.79
4.67 4.56
3.71 3.78
3.80 3.87
3.67 3.71
experiment
4.7273
4.5174
3.81a
3.9031
