Simulation of the Properties of a Photochromic Triad - American

Jun 22, 2010 - CNRS UMR 7086, Université Paris 7 - Paris Diderot, Bâtiment Lavoisier, 15 rue Jean Antoine de Baïf, 75205 Paris Cedex 13, France...
0 downloads 0 Views 3MB Size
pubs.acs.org/JPCL

Simulation of the Properties of a Photochromic Triad Denis Jacquemin,*,† Eric A. Perp ete,† Franc- ois Maurel,‡ and Aur elie Perrier*,‡ †

Unite de Chimie Physique Th eorique et Structurale (UCPTS), Facult es Universitaires Notre-Dame de la Paix, rue de Bruxelles, emes (ITODYS), 61, B-5000 Namur, Belgium, and ‡Laboratoire Interfaces, Traitements, Organisation et Dynamique des Syst CNRS UMR 7086, Universit e Paris 7 - Paris Diderot, B^ atiment Lavoisier, 15 rue Jean Antoine de Baïf, 75205 Paris Cedex 13, France

ABSTRACT To increase the contrast between the “on” and “off” states, photochromic units may be coupled together through delocalized bridges. However, such a procedure is practically limited by the impossibility to achieve the conversion of all photochromes. In this letter, we investigate the structures and properties of a dithienylethene trimer in order to apprehend its excited-state properties in the framework of a procedure combining the simulations of the electronic spectrum with time-dependent density functional theory and the analysis of the topology of the relevant molecular orbitals. Using a range-separated hybrid, this level of theory is able to perfectly reproduce the patterns of the absorption spectrum but only yields partially correct insights regarding the ring-closure of the three dithienylethene units. These theoretical simulations are a first step toward the development of more efficient molecular switches. SECTION Electron Transport, Optical and Electronic Devices, Hard Matter

M

olecular architectures encompassing two or more photochromic units have been the center of a growing number of investigations during the past years, as such structures pave the way toward efficient multiaddressable molecular switches. In the first category, one finds structures built with two photochromes belonging to different families,1-4 whereas systems composed of molecular switches of the same class constitute the second category.5-10 These latter entities often rely on dithienylethenes (DTs, also referred to as diarylethenes), that present both a colorless open form (O) and a colored closed form (C).11 In principle, the successive photocyclization of each DT could induce variations of the color (if the substituents of each photochromic unit are different),7 but in general, one mainly observes a doubling (or tripling) of the molar absorption coefficient for the band characteristic of the conjugated form.6,8-10 In other words, the coupled DT strategy is useful to significantly enhance the contrast between “on” and “off” states. However, in several cases, the anticipated full-photochromic activity cannot be reached experimentally, i.e., it is not possible to close all DT, so that only mixed open/closed compounds are present in solution (see ref 12 and references therein). The key factors governing the partial or complete activity of coupled photochromes remain quite unclear, as comparable connecting units lead to strongly dissimilar effects. Therefore, theoretical calculations able to deliver insights are especially welcome. However, an obvious pitfall of modeling approaches appears: the size of the compounds of practical interest definitively hampers the selection of the most accurate ab initio tools. In that framework, we have recently established that, for most DT dimers, the experimental behavior could be rationalized with the help of time-dependent density functional theory (TD-DFT).12-14 Indeed, a careful investigation of the topology

r 2010 American Chemical Society

of the molecular orbitals implied in the most relevant UV electronic transitions of the mixed open/closed structures often allows to gain insights regarding the ring-closure of the second photochrome. Of course, such a first-order procedure is unable to simulate the excited-state potential energy surface, nor the conical intersections that are important for DT.15 Therefore, this static method, the only tractable in practice, can only be viewed as a first screening scheme and it may fail to give the correct prediction.14 In 2002, Irie's group presented the first successful synthesis of quasi-symmetric trimers and tetramers, in which the DT units are coupled through ethynyl bridges.6 For the trimer, represented in Scheme 1, continuous irradiation with a 320 nm light of the structure presenting only open DT (referred to as OOO in the following) led to the formation of three additional products, which were separated by high-performance liquid chromatography (HPLC) and identified by NMR and UV-vis spectroscopies.6 These three compounds were, on the one hand, the two possible singly closed trimers, namely, OCO and COO, and, on the other hand, a small fraction of the doubly closed COC derivative. On the contrary, no structures presenting two successive closed DT (CCO and CCC) could be detected within the experimental setup. 6 In the present letter, we investigate the structure and properties of these photochromic triads with TD-DFT. The relative energies, carbon-carbon bond lengths and absorption wavelengths computed with DFT and TD-DFT are Received Date: May 21, 2010 Accepted Date: June 17, 2010 Published on Web Date: June 22, 2010

2104

DOI: 10.1021/jz1006753 |J. Phys. Chem. Lett. 2010, 1, 2104–2108

pubs.acs.org/JPCL

Scheme 1. Representation of the Investigated Trimer (CCC Form)

Table 1. Relative Gibbs Free Energies (G in kcal 3 mol-1), Distances Separating the Reactive Carbon Atoms (di in Å) and Relevant Vertical Transition Wavelengths (λ in nm and the Corresponding Oscillator Strengths, f )a theory

Figure 1. CAM-B3LYP simulated spectra for the six species. A broadening Gaussian with a full width at half-maximum (fwhm) of 0.35 eV was used.

experiment

form

G

d1

d2

d3

λ

f

λ

ε

CCC

38.7

1.53

1.53

1.53

777 402

1.42 0.61

CCO

25.3

1.53

1.53

3.80

725 396 347

1.18 0.63 0.71

COC

25.9

1.53

3.82

1.53

585 367 363 337

0.96 0.51 0.47 0.75

584

2.9

OCO

12.8

3.80

1.53

3.77

652 386 345

0.86 0.55 1.23

635

2.4

COO

12.9

1.53

3.82

3.78

580 365 337 330

0.53 0.53 1.37 0.33

584

1.4

OOO

0.0

3.79

3.80

3.77

342 333

2.08 0.35

325

4.6

chromes composing the trimer. On the contrary, this distance is more sensitive for the open switches. Indeed, it is systematically larger for the central DT than for the peripheral DT. In addition, the CC length increases, although by a small amount, when the vicinal DTs are closed: it is maximal for COC. For comparison, in the corresponding monomer, the spacing between the two reactive centers is slightly smaller: 3.77 Å. A 3.82 Å separation also significantly exceeds the one found in typical coupled-DT architectures presenting a complete photochromic activity (3.5-3.6 Å),14 and we can state that this parameter is probably more significant than the total energies for explaining the partial photochromism of the triad, although the differences between the six systems remain limited. Figure 1 provides a comparison between the computed absorption spectra for the six possible states of the trimer. With CAM-B3LYP, the agreement between theory and experiment is highly satisfying, as can be deduced from Table 1. In the Supporting Information (SI), results obtained with other functionals (PBE, PBE0, B3LYP) are also presented. Indeed, for the four CAM-B3LYP λmax reported in ref 6, we obtain a mean absolute deviation of 10 nm (0.07 eV), with accurate evolutions of the oscillator strengths. For instance, the experimental fact that COO and COC absorb at the same wavelength but the latter presents a twice as large intensity is perfectly reproduced by TD-DFT (Figure 1). The same holds for the twin bathochromic and hyperchromic shift noticed when comparing the terminal (COO) and central (OCO) singly closed triad. One can therefore be confident that the evolution given by theory for the two unseen species (CCO and CCC) are correct, at least qualitatively. It turns out that a third electro-cyclization of the doubly closed COC system would induce both a large bathochromic shift and a 50% increase of the molar absorption coefficient. Such structure could therefore be attractive for practical applications. For all forms, the first absorption band collated in Table 1 mainly corresponds to a highest occupied molecular orbital (HOMO) to lowest unoccupied molecular orbital

a The two former values were obtained at the PCM-PBE0/6-311G(d,p) level, whereas the latter was computed with the PCM-TD-CAM-B3LYP/6311þG(2d,p) approach. Form indicates the closed or open nature of the successive DT in the trimer. Experimental wavelength (nm) and molar absorption coefficient (104 M-1 cm-1) are from ref 6.

listed and compared to experimental values6 in Table 1. The energetical penalty related to the ring-closures is in the line of previously calculated values12 and remains nearly independent of the position of the closed photochrome and practically proportional to the number of closed DTs. Indeed, the CCC relative Gibbs free energy is þ38.7 kcal 3 mol-1, whereas one would have predicted 38.6 kcal.mol-1 (12.9  2 þ 12.8) on the basis of the singly closed systems. Therefore, as for dimers,14 the energy required for successive electrocyclizations is not a discriminating factor allowing one to explain the partial photochromism of the triad. This can also be concluded by comparing the G values of CCO and COC that are similar, though only the latter is observed experimentally. For closed DT, the distance between the two reactive carbon atoms, typical of a single carbon-carbon bond (1.53 Å), is independent of the open/closed nature of the other photo-

r 2010 American Chemical Society

2105

DOI: 10.1021/jz1006753 |J. Phys. Chem. Lett. 2010, 1, 2104–2108

pubs.acs.org/JPCL

Figure 2. Molecular orbitals (HOMO left and LUMO right). A contour threshold of 0.035 au has been used. See also the SI.

(LUMO) transition (Figure 2). Once at least one DT is closed, the HOMO and LUMO present the awaited topologies:12 the former (latter) is mainly located on the double (single) bonds of the closed unit. Both orbitals are also delocalized on the ethynyl bridge, but present trifling contributions on the open photochrome(s). This explains the bathochromic shift between COO and OCO, the former (latter) having frontier orbitals delocalized on one (two) triple bond(s). In the same way, COC essentially corresponds to two noninteracting DTs, which is consistent with both the intensity increase and the absence of bathochromic shift with respect to COO. On the contrary, in CCC, the frontier orbitals imply all three photochromes, therefore reducing the gap and increasing the λmax. Let us now turn toward the inspection of the frontier molecular orbitals (see SI for complete orbital plots of all six forms). Our goal is to determine whether the transitions listed in Table 1 induce electronic promotion to virtual orbitals presenting a photochromic topology,12,14 that is, orbitals with a significant density on (at least one of) the two reactive carbon atoms and a bonding character16 for the to-be-formed carbon-

r 2010 American Chemical Society

carbon bond. For the fully open derivative, the two largest components in the strong 342 nm absorption correspond to HOMO f LUMO and HOMO-1 f LUMOþ1 transitions, whereas several smaller (but non-negligible) contributions include the LUMOþ2, LUMOþ3, and LUMOþ4 virtuals. The LUMO (LUMOþ1) presents a bonding character for all three (the two external) DTs; the three next unoccupied orbitals, well localized on one of the three DTs, clearly display the photochromic shape. Consequently, the experimental finding that irradiation at 320 nm induces the ring-closure of central or peripheral DTs can be straightforwardly explained. For OCO, the two virtual orbitals possessing a topology allowing one to efficiently close the two external DTs are the LUMOþ2 and LUMOþ3 (see SI). Unfortunately, these orbitals do not contribute to the 386 and 345 nm peaks listed in Table 1, nor to any of the weaker absorptions computed >320 nm: all exclusively involve the LUMO, the LUMOþ1, and the LUMOþ4. Both the LUMO and the LUMOþ4 present an antibonding character for the external DT, whereas one of the two open DTs has a bonding character in the LUMOþ1, but with a trifling orbital component for the to-be-formed CC

2106

DOI: 10.1021/jz1006753 |J. Phys. Chem. Lett. 2010, 1, 2104–2108

pubs.acs.org/JPCL

Figure 3. LUMOþ1, þ2, þ3 and þ4 of COO. The LUMO is sketched in Figure 2.

bond. Therefore, the simple orbital picture figures out that no (or only an extremely inefficient) electro-cyclization of the peripheral DT can be attained once the central structure is closed. For COO, the 365 nm peak incorporates the LUMO and LUMOþ4, which are both unhelpful (see Figures 2 and 3): they present negligible contributions on one of the reactive carbon atoms, and more importantly have an antibonding character on the central DT. The three major orbital contributions in the 337 and 330 nm transitions are (in decreasing order of importance) the HOMO-4 f LUMO, HOMO-1 f LUMOþ2, and HOMO-1 f LUMOþ1. While the LUMOþ2 is also antibonding, the shape of the LUMOþ1 should allow the electrocyclization of the inner photochrome. In practice, such cyclization does not take place, as a rapid electronic transfer from the excited open DT to the adjacent close DT takes place.6 Such dynamic effect can hardly be predicted by a simple examination of the molecular orbitals, a clear limitation of our approach. Experimentally, the COO f COC process is inefficient: after 11 min of irradiation with a 320 nm input light, the relative ratio of the COC form is as small as 3%, whereas 54% of OCO is detected.6 This suggests either a completely different pathway (which we cannot foresee) or a similar but much less efficient mechanism. In the simple orbital picture, one should promote the electron to the LUMOþ3 in order to close the external open DT of COO. This can be achieved by absorption at 309 and 302 nm, both peaks presenting weakly allowed probabilities (f of 0.06 and 0.03, respectively). While these transitions appear at larger energies than the input beam, they remain within 20 nm of the experimental threshold, and such error is certainly not impossible with TD-DFT. An intensity borrowing phenomena with absorptions at smaller transition energies, or the existence of a less-efficient two-photon phenomena could also possibly explain this ring-closure. Finally, for COC, transitions toward the LUMO and LUMOþ1 should not induce electro-cyclization of the central unit, as these virtual orbitals possess almost no density on the two reactive carbon atoms. The same conclusion holds for an electronic promotion to the LUMOþ3 and LUMOþ4 that present a significant density but exhibit an antibonding

r 2010 American Chemical Society

character. In fact, only the LUMOþ2 seems reactive among the five first virtual orbitals. As both the 367 and 363 nm absorptions only involve the LUMO and LUMOþ1, and as there is no LUMOþ2 contribution in the 337 nm peak [nor in the weaker absorption bands at 331 nm ( f = 0.22), 325 nm ( f = 0.10), and 323 nm ( f = 0.12)], the orbital picture correctly foresees that the last ring-closure is inhibited for COC. Using TD-DFT, we have simulated the properties of the photochromic triad synthesized in 2002 by Kaieda et al.6 and constituted of three DTs coupled through triple bonds. The theoretical simulations show that the energy associated with the successive ring-closure is almost additive, whereas the distance between the two reactive carbon atoms in the open photochromes tends to slightly increase when the nearby DT(s) is (are) closed. This is probably detrimental for the photochromic activity. Using a range-separated hybrid (CAMB3LYP), we accurately reproduce the positions and intensities of the visible bands of the electronic absorption spectra. On the contrary, the simple frontier orbital picture that was efficient for photochromic dimers only provides partially correct insights in the present case: it restores most of the experimental findings [(i) OOO to OCO and (ii) OOO to COO transformations as well as the impossibility to close (iii) the terminal DTs in OCO and (iv) the central DT in COC], but provides a clear “false positive” prediction for the ring-closure of the central DTof the COO structure. This failure is due to the use of a static approach (the only affordable scheme for such large molecules) that cannot account for the electronic transfer from one DT to another,6 nor locate conical intersections.15 This contribution therefore illustrates, on the one hand, the efficiency of TD-DFT in reproducing electronic absorption spectra of very large organic photochromic entities, and, on the other hand, the limitations of the orbital picture for “crystal balling” all photochromic paths of complex architectures.

METHOD We use a similar procedure as in recent works devoted to photochromic dimers.12-14 Therefore, the computational

2107

DOI: 10.1021/jz1006753 |J. Phys. Chem. Lett. 2010, 1, 2104–2108

pubs.acs.org/JPCL

steps are only summarized here; the interested reader can find additional information and bibliographic references in recent TD-DFT investigations.17,18 We used the Gaussian09 program throughout19 and selected default algorithms and parameters except when noted. In a first step we optimized, at the PBE0/6-311G(d,p) level,20 the ground-state geometry of all derivatives until the residual mean square force was smaller than 1  10-5 a.u. These optimizations were performed without imposing any symmetry constraints. In a second stage, the vibrational spectrum of each species was determined analytically at the same level, and it was checked that all structures correspond to true minima of the potential energy surface. In a third step, the first fifteen low-lying excited-states weren determined within the vertical TD-DFT approximation using the CAM-B3LYP/6-311þG(2d,p) level of approximation,21 which delivers accurate transition energies in DT.13,14,18 Test calculations were performed with other functionals (see SI). The bulk solvent (n-hexane, see ref 6) effects were included at all stages, by means of the polarizable continuum model (PCM).22

(4)

(5)

(6)

(7)

(8)

(9)

(10)

SUPPORTING INFORMATION AVAILABLE Comparisons of the TD-DFT results obtained with different functionals, complete frontier orbital plots, and Cartesian coordinates for all investigated systems. This material is available free of charge via the Internet at http://pubs.acs.org/.

(11) (12)

AUTHOR INFORMATION Corresponding Author:

(13)

*To whom correspondence should be addressed. E-mail: denis. [email protected] (D.J.); [email protected] (A.P.).

(14)

ACKNOWLEDGMENT D.J. and E.A.P. thank the Belgian National

(15)

Fund for Scientific Research for their research associate and senior research associate positions, respectively. This research used resources of the Interuniversity Scientific Computing Facility located at the University of Namur, Belgium, which is supported by the F.R. S.-FNRS under Convention No. 2.4617.07. The collaboration between the Belgian and French group is supported by the WallonieBruxelles International, the Fonds de la Recherche Scientfique, the Ministere Franc-ais des Affaires  etrang eres et europ eennes, and the Ministere de l'Enseignement sup erieur et de la Recherche in the framework of Hubert Curien Partnership.

(16)

(17)

(18)

REFERENCES (1)

(2)

(3)

Mrozek, T.; G€ orner, H.; Daub, J. Towards Multifold Cycloswitching of Biphotochromes: Investigation on a Bond-Fused Dihydroazulene/Vinylheptafulvene and Dithienylethene/ Dihydrothienobenzothiophene. Chem. Commun. 1999, 16, 1487–1488. Favaro, G.; Levi, D.; Ortica, F.; Samat, A.; Guglielmetti, R.; Mazzucato, U. Photokinetic Behaviour of Bi-photochromic Supramolecular Systems Part 3. Compounds with Chromene and Spirooxazine Units Linked Through Ethane, Ester and Acetylene Bridges. J. Photochem. Photobiol. A: Chem. 2002, 149, 91–100. Frigoli, M.; Mehl, G. H. Multiple Addressing in a Hybrid Biphotochromic System. Angew. Chem., Int. Ed. Engl. 2005, 44, 5048–5052.

r 2010 American Chemical Society

(19)

(20)

(21)

(22)

2108

Sevez, G.; Gan, J.; Delbaere, S.; Vermeersch, G.; Sanguinet, L.; Levillain, E.; Pozzo, J. L. Photochromic Performance of a Dithienylethene-Indolinooxazolidine Hybrid. Photochem. Photobiol. Sci. 2010, 9, 131–135. Matsuda, K.; Irie, M. Photoswitching of Intramolecular Magnetic Interaction Using a Diarylethene Dimer. J. Am. Chem. Soc. 2001, 123, 9896–9897. Kaieda, T.; Kobatake, S.; Miyasaka, H.; Murakami, M.; Iwai, N.; Nagata, Y.; Itaya, A.; Irie, M. Efficient Photocyclization of Dithienylethene Dimer, Trimer, and Tetramer: Quantum Yield and Reaction Dynamics. J. Am. Chem. Soc. 2002, 124, 2015–2024. Higashiguchi, K.; Matsuda, K.; Tanifuji, N.; Irie, M. Full-Color Photochromism of a Fused Dithienylethene Trimer. J. Am. Chem. Soc. 2005, 127, 8922–8923. Areephong, J.; Browne, W. R.; Feringa, B. L. Three-State Photochromic Switching in a Silyl Bridged Diarylethene Dimer. Org. Biomol. Chem. 2007, 5, 1170–1174. Kim, H. E.; Jang, J. H.; Choi, H.; Lee, T.; Ko, J.; Yoon, M.; Kim, H. J. Photoregulated Fluorescence Switching in Axially Coordinated Tin(IV) Porphyrinic Dithienylethene. Inorg. Chem. 2008, 47, 2411–2415. Roberts, M. N.; Carling, C. J.; Nagle, J. K.; Branda, N. R.; Wolf, M. O. Successful Bifunctional Photoswitching and Electronic Communication of Two Platinum(II) Acetylide Bridged Dithienylethenes. J. Am. Chem. Soc. 2009, 131, 16644–16645. Irie, M. Diarylethenes for Memories and Switches. Chem. Rev. 2000, 100, 1685–1716. Jacquemin, D.; Perp ete, E. A.; Maurel, F.; Perrier, A. Ab Initio Investigation of the Electronic Properties of Coupled Dithienylethenes. J. Phys. Chem. Lett. 2010, 1, 434–438. Jacquemin, D.; Michaux, C.; Perp ete, E. A.; Maurel, F.; Perrier, A. Photochromic Molecular Wires: Insights from Theory. Chem. Phys. Lett. 2010, 488, 193–197. Jacquemin, D.; Perp ete, E. A.; Maurel, F.; Perrier, A. Doubly Closing or Not? Theoretical Analysis for Coupled Photochomes. J. Phys. Chem. C 2010, 114, 9489–9497. Asano, Y.; Murakami, A.; Kobayashi, T.; Goldberg, A.; Guillaumont, D.; Yabushita, S.; Irie, M.; Nakamura, S. Theoretical Study on the Photochromic Cycloreversion Reactions of Sithienylethenes: On the Role of the Conical Intersections. J. Am. Chem. Soc. 2004, 126, 12112–12120. When the electron density on one of the two reactive carbon atoms was trifling, we used a smaller contour threshold to check the bonding/antibonding character. Jacquemin, D.; Perp ete, E. A.; Ciofini, I.; Adamo, C. Accurate Simulation of Optical Properties in Dyes. Acc. Chem. Res. 2009, 42, 326–334. Jacquemin, D.; Wathelet, V.; Perp ete, E. A.; Adamo, C. Extensive TD-DFT Benchmark: Singlet-Excited States of Organic Molecules. J. Chem. Theory Comput. 2009, 5, 2420–2435. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; et al. Gaussian 09, revision A.02; Gaussian Inc.: Wallingford, CT, 2009. Adamo, C.; Barone, V. Toward Reliable Density Functional Methods Without Adjustable Parameters: The PBE0Model. J. Chem. Phys. 1999, 110, 6158–6170. Yanai, T.; Tew, D. P.; Handy, N. C. A New Hybrid ExchangeCorrelation Functional Using the Coulomb-Attenuating Method (CAM-B3LYP). Chem. Phys. Lett. 2004, 393, 51–56. Tomasi, J.; Mennucci, B.; Cammi, R. Quantum Mechanical Continuum Solvation Models. Chem. Rev. 2005, 105, 2999–3094.

DOI: 10.1021/jz1006753 |J. Phys. Chem. Lett. 2010, 1, 2104–2108