Design and photochemical characterization of a biomimetic light-driven Z/E switcher

Article abstract of DOI:10.1021/ja038859e

Protonated Schiff bases (PSBs) of polyenals constitute a class of light-driven switchers selected by biological evolution that provide model compounds for the development of artificial light-driven molecular devices or motors. In the present paper, our pr

Full text of DOI:10.1021/ja038859e

Published on Web 07/08/2004
Design and Photochemical Characterization of a Biomimetic
Light-Driven Z/E Switcher
Diego Sampedro,† Annapaola Migani,† Alessandra Pepi,† Elena Busi,†
Riccardo Basosi,†,‡ Loredana Latterini,| Fausto Elisei,| Stefania Fusi,†
Fabio Ponticelli,† Vinicio Zanirato*,§ and Massimo Olivucci*,†,‡
Contribution from the Dipartimento di Chimica, UniVersita` di Siena, Via Aldo Moro 2,
I-53100 Siena, Italy, Centro di Studio dei Sistemi Complessi, Via Tommaso Pendola 37,
I-53100 Siena, Italy, Dipartimento di Scienze Farmaceutiche, UniVersita` di Ferrara,
Via Fossato di Mortara 17-19, I-44100 Ferrara, Italy, and Dipartimento di Chimica,
Laboratorio di Chimica Fisica, Via Elce di Sotto 8, I-06123 Perugia, Italy
Received October 3, 2003; E-mail: olivucci@unisi.it; znv@dns.unife.it
Abstract: Protonated Schiff bases (PSBs) of polyenals constitute a class of light-driven switchers selected
by biological evolution that provide model compounds for the development of artificial light-driven molecular
devices or motors. In the present paper, our primary target is to show, through combined computational
and experimental studies, that it is possible to approach the design of artificial PSBs suitable for such
applications. Below, we use the methods of computational photochemistry to design and characterize the
prototype biomimetic molecular switchers 4-cyclopenten-2′-enylidene-3,4-dihydro-2H-pyrrolinium and its
5,5′-dimethyl derivative both containing the penta-2,4-dieniminium chromophore. To find support for the
predicted behavior, we also report the photochemical reaction path of the synthetically accessible compound
4-benzylidene-3,4-dihydro-2H-pyrrolinium. We show that the preparation and photochemical characterization
of this compound (together with three different N-methyl derivatives) provide both support for the predicted
photoisomerization mechanism and information on its sensitivity to the molecular environment.
1. Introduction
of the above principle is the construction of single molecules
capable to convert light-energy in continuous unidirectional ro-
tary motion. Indeed, it has been shown that photochemical E/Z
switchers corresponding to chiral diarylidenes constitute examp-
les of light-driven molecular rotors.13-16 Here, the spatial asym-
metry (i.e., the chirality) of the molecular framework determines
the direction (either clockwise or counterclockwise) of the E
f Z and Z f E conformational changes leading, after sequential
absorption of two photons, to a complete (360°) rotation.
Chiral E/Z switchers are also known in photobiology. For
Molecular switchers based on photochemical E/Z isomeriza-
tion have been employed in different contexts to convert light-
energy into “mechanical” motion at the molecular level.1,2 In
basically all applications, the induced motion results in a
permanent or transient conformational change of a molecular
scaffold bounded to the switcher. Using this simple principle,
switchers based on the azobenzene chromophore have been used
to control properties such as ion complexation,3,4 electronic
properties5 and catalysis6 or to trigger folding/unfolding of
oligopeptide chains.7-12 A particularly sophisticated application
instance, the retinal chromophore of rhodopsin proteins,17-19
a
large class of trans-membrane photoreceptors, undergoes an
efficient unidirectional photoisomerization that, ultimately,
triggers a conformational change of the native protein scaffold.
† Dipartimento di Chimica, Universita` di Siena.
‡ Centro di Studio dei Sistemi Complessi.
§ Dipartimento di Scienze Farmaceutiche, Universita` di Ferrara.
| Dipartimento di Chimica, Laboratorio di Chimica Fisica.
(1) Sauvage, J.-P., Ed.; Molecular Machines and Motors; Springer-Verlag:
Berlin, London, 2001; Vol. 99.
(10) Rudolph-Bo¨hner, S.; Kru¨ger, M.; Oesterhelt, D.; Moroder, L.; Na¨gele, T.;
Wachtveitl, J. J. Photochem. Photobiol. A 1997, 105, 235-248.
(11) Renner, C.; Behrendt, R.; Spo¨rlein, S.; Wachtveitl, J.; Moroder, L.
Biopolymers 2000, 54, 489-500.
(2) Drexler, K. E. Nanosystems: Molecular Machinery, Manufacturing and
Computation; John Wiley & Sons: New York, 1992.
(3) Shinkai, S.; Kusano, Y.; Manabe, O.; Nakaji, T.; Nishida, Y.; Ogawa, T.
J. Am. Chem. Soc. 1980, 102, 5860-5865.
(12) Spo¨rlein, S.; Carstens, H.; Satzger, H.; Renner, C.; Behrendt, R.; Moroder,
L.; Tavan, P.; Zinth, W.; Wachtveitl, J. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 7998-8002.
(4) Shinkai, S.; Minami, T.; Kusano, Y.; Manabe, O. J. Am. Chem. Soc. 1983,
105, 1851-1856.
(13) Koumura, N.; Geertsema, E. M.; van Gelder, M. B.; Meetsma, A.; Feringa,
B. L. J. Am. Chem. Soc. 2002, 124, 5037-5051.
(5) Jousselme, B.; Blanchard, P.; Gallego-Planas, N.; Delaunay, J.; Allain, M.;
Richomme, P.; Levillain, E.; Roncali, J. J. Am. Chem. Soc. 2003, 125,
2888-2889.
(14) Koumura, N.; Zijistra, R. W. J.; van Delden, R. A.; Harada, N.; Feringa,
B. L. Nature 1999, 401, 152-155.
(15) Koumura, N.; Geertsema, E. M.; Meetsma, A.; Feringa, B. L. J. Am. Chem.
Soc. 2000, 122, 12 005-12 006.
(6) Cacciapaglia, R.; Stefano, S. D.; Mandolini, L. J. Am. Chem. Soc. 2003,
125, 2224-2227.
(16) Feringa, B. L. Acc. Chem. Res. 2001, 34, 504-516.
(17) Mathies, R. A.; Lugtenburg, J. In Handbook of Biological Physics; Stavenga,
D. G., de Grip, W. J., Pugh, E. N., Eds.; Elsevier Science: New York,
2000; Vol. 3, pp 56-90.
(7) Behrendt, R.; Renner, C.; Schenk, M.; Wang, F.; Wachtveitl, J.; Oesterhelt,
D.; Moroder, L. Angew. Chem., Int. Ed. Engl. 1999, 38, 2771-2774.
(8) Bredenbeck, J.; Helbing, J.; Sieg, A.; Schrader, T.; Zinth, W.; Renner, C.;
Behrendt, R.; Moroder, L.; Wachtveitl, J.; Hamm, P. Proc. Natl. Acad.
Sci. U.S.A. 2003.
(18) Kandori, H.; Shichida, Y.; Yoshizawa, T. Biochemistry (Moscow) 2001,
66, 1483-1498.
(9) Ulysse, L.; Cubillos, J.; Chmielewski, J. J. Am. Chem. Soc. 1995, 117,
8466-8467.
(19) Teller, D. C.; Okada, T.; Behnke, C. A.; Palczewski, K.; Stenkamp, R. E.
Biochemistry 2001, 40, 7761-7772.
9
10.1021/ja038859e CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 9349-9359
9349

A R T I C L E S
Sampedro et al.
Scheme 1
Scheme 2
the path allows for redistribution of the photon energy. An
additional desirable property of photochemical switchers is the
stability of the isomers A and B with respect to thermal (i.e.,
ground state) Z/E isomerization. As shown in Scheme 1a (see
dashed energy profile), in an efficient switcher the barrier for
thermal Z/E isomerization must be high enough to restrain the
return of B to A.
In rhodopsin (Rh) itself (i.e., the visual receptor of superior
animals) the photoisomerization of the native 11-cis-retinal
chromophore (PSB11) to its all-trans form occurs with a 0.67
quantum yield.20 This value is larger than the 0.25 value
measured for the same chromophore in ethanol solution21,22 and
also larger than the 0.28 and 0.51 values measured for the
photoisomerization of trans- and cis-azobenzene, respectively.23
Due to its high photoisomerization efficiency, bacteriorhodopsin
(a member of the same photoreceptor family) and its mutants,
have been exploited to produce different light-driven molecular
devices (see ref 24 and references therein).
The ultrafast 200 fs reaction time scale27,28 and vibrational
coherence28 observed for the photoisomerization of PSB11 in
Rh, suggest that the reaction occurs via the mechanism shown
in Scheme 1a. In this model, the quantum yield of the reaction
and the nuclear velocity along the reaction coordinate must be
related by the Landau-Zener formula.29 The high quantum yield
is thus a consequence of the high rate of the 11-cis f all-trans
isomerization.17,28,30 This idea is supported by the fact that
13-demethyl-Rh (an isomer of Rh where the 13-methyl group
of PSB11 has been removed) and isorhodopsin (an isomer of
Rh containing a 9-cis retinal chromophore) isomerize more
slowly30,31 and have lower quantum yields32,33 than Rh.
PSB11 corresponds to the protonated Schiff base (PSB) of a
polyenal. The primary target of the present paper is to show,
through combined computational and experimental studies, that
certain biomimetic PSBs provide suitable “frameworks” for the
design of molecular switchers or motors. In the past, retinal
PSB models have been the subject of several computational
studies.34-44 The first example is provided by the PSB11 model
4-cis-γ-methylnona-2,4,6,8-tetra-enimminium (4-cis-γ-Me-C9H10-
NH2+). As reported in Scheme 2, the S1 isomerization path of
this molecule is sequentially dominated by two very different
It is apparent that the efficiency of the Z/E isomerization of
Rh must depend on the structure of the associated reaction path.
This idea is illustrated in Scheme 1 where a model reaction
path for an efficient (Scheme 1a) and a less efficient (Scheme
1b) switcher is reported. Accordingly, an efficient photoisomer-
ization would occur when the photoexcited reactant A* evolves
along a barrierless excited-state path, decays at a real surface
crossing (a conical intersection, CI)25,26 and finally relaxes to
the energy minimum corresponding to photoproduct B. Fur-
thermore, in an efficient switcher, the reaction coordinate
connecting A* to B should be as simple as possible i.e.,
dominated only by the reactiVe mode (e.g., the isomerization
mode) or by modes strongly coupled to it. As we will discuss
below, such a coordinate helps to limit the wastage of the photon
energy that has to be conVerted to reactive motion and not
redistributed among the remaining 3N - 7 internal degrees of
freedom of the system (here, N indicates the number of atoms
of A). In contrast, the reaction path of Scheme 1b belongs to
an inefficient switcher. In fact, the presence of excited state
and/or ground-state intermediates (I* and I respectively) along
(27) Schoenlein, R. W.; Peteanu, L. A.; Mathies, R. A.; Shank, C. V. Science
1991, 254, 412-415.
(28) Wang, Q.; Schoenlein, R. W.; Peteanu, L. A.; Mathies, R. A.; Shank, C.
V. Science 1994, 266, 422-424.
(29) Desouter-Lecomte, M.; Lorquet, J. C. J. Chem. Phys. 1979, 71, 3661-
3672.
(30) Wang, Q.; Kochendoerfer, G. G.; Schoenlein, R. W.; Verdegem, P. J. E.;
Lugtenburg, J.; Mathies, R. A.; Shank, C. V. J. Phys. Chem. 1996, 100,
17 388-17 394.
(31) Schoenlein, R. W.; Peteanu, L. A.; Wang, Q.; Mathies, R. A.; Shank, C.
V. J. Phys. Chem. 1993, 97, 12 087-12 092.
(32) DeLange, F.; Bovee-Geurts, P. H. M.; Van Oostrum, J.; Daniel Portier,
M.; Verdegem, P. J. E.; Lugtenburg, J.; de Grip, W. J. Biochem. 1999, 37,
1411-1420.
(33) Hurley, J.; Ebrey, T.; Honig, B.; Ottolenghi, M. Nature (London) 1977,
270, 540-542.
(34) Ben-Nun, M.; Molnar, F.; Schulten, K.; Martinez, T. J. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 1769-1773.
(35) Nonella, M. J. Phys. Chem. B 2000, 104, 11 379-11 388.
(36) Bonacic-Koutecky, V.; Kho¨ler, K.; Michl, J. Chem. Phys. Lett. 1984, 104,
440-443.
(20) Dartnall, H. Visio Res. 1967, 8, 339-358.
(21) Becker, R. S. Photochem. Photobiol. 1988, 48, 369-399.
(22) Becker, R. S.; Freedman, K. A. J. Am. Chem. Soc. 1985, 107, 1477-1485.
(23) Bortolus, P.; Monti, S. J. Phys. Chem. 1979, 83, 648-652.
(24) Birge, R. R.; Gillespie, N. B.; Izaguirre, E. W.; KUsnetzow, A.; Lawrence,
A. F.; Singh, D.; Song, W.; Schmidt, E.; Stuart, J. A.; Seetharaman, S.;
Wise, K. J. J. Phys. Chem. B 1999, 103, 10 746-10 766.
(37) Bonacic-Koutecky, V.; Kho¨ler, K.; Michl, J. Angew. Chem., Int. Ed. 1987,
26, 170-189.
(38) Tajkhorshid, E.; Suhai, S. J. Phys. Chem. B 1999, 103, 5581-5590.
(39) Birge, R. R.; Hubbard, L. M. J. Am. Chem. Soc. 1980, 102, 2195-2205.
(40) Weingart, O.; Buss, V. Phase Trans. 2001, 75, 19-29.
(41) Warshel, A. Nature 1976, 260, 769-683.
(25) Bernardi, F.; Olivucci, M.; Robb, M. A. Chem. Soc. ReV. 1996, 321-328.
(26) Michl, J.; Bonacic-Koutecky, V. Electronic Aspects of Organic Photo-
chemistry; Wiley: New York, 1990.
(42) Grossjean, M. F.; Tavan, P.; Schulten, K. J. Phys. Chem. 1990, 94, 8059-
8069.
9
9350 J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004

Biomimetic Light-Driven Z/E Switcher
A R T I C L E S
Scheme 3
To the best of our knowledge, attempts to synthesize
compounds containing the penta-2,4-dieniminium biomimetic
backbone have never been reported. As a consequence, their
synthesis represents an attractive target. Below, we apply the
methods of computational photochemistry48,49 to characterize
the prototype molecular switcher 4-cyclopenten-2′-enylidene-
3,4-dihydro-2H-pyrrolinium (1) and molecular rotor 5-methyl-
4-(5′-methylcyclopent-2′-enylidene)-3,4-dihydro-2H-pyrrolini-
um (5,5′-diMe-1) both containing the penta-2,4-dieniminium
unit. To find support for our predictions we also report the
computation of the photochemical reaction path of the syntheti-
cally accessible structure 4-benzylidene-3,4-dihydro-2H-pyrro-
lium (2H) where the terminal CdC double-bond of the penta-
2,4-dieniminium unit is replaced with a phenyl. We show that
the photochemical characterization of 2H (together with its
derivatives 4-Me-2H, 2, p-MeO-2, p-NO2-2) provide support
for the predicted photoisomerization mechanism and sensitivity
to the molecular environment.
molecular modes.45 The first describes the initial relaxation from
the Franck-Condon (FC) point to a planar intermediate FS and
is dominated by a double-bond-expansion/single-bond-contrac-
tion (i.e., stretching) mode. In contrast, the second mode de-
scribes the evolution of FS along the 4-cisfall-trans isomer-
ization mode ultimately leading to decay through a conical inter-
section (CI). Since the FC f FS relaxation is substantially un-
coupled to the isomerization, FS can be associated to I* of
Scheme 1a. Accordingly, the kinetic energy gained during the
initial relaxation may not be efficiently converted into Z f E
motion.
2. Experimental and Computational Methods
2.1 Computational Methods. The photoisomerization paths
of E-2H have been determined using fully unconstrained ab
initio quantum chemical computations in the framework of a
CASPT2//CASSCF strategy.33,34 This requires that the reaction
coordinate is computed at the complete active space self-
consistent field (CASSCF) level of theory and that the corre-
sponding energy profile is re-evaluated at the multiconfigura-
tional second-order Moller-Plesset perturbation theory level
(here we used the CASPT2 method implemented in MOLCAS-
5)50 to take into account the effect of electron dynamic
correlation. This approach allows for an accurate evaluation of
excited-state energy barriers.51 Thus, for 1, 5,5′-diMe-1 and
E-2H the photoisomerization coordinate is determined via
CASSCF minimum energy path (MEP) computations in mass-
weighted Cartesians, whereas an S1 single-bond twisting path
(see subsection 3.4.3) was built by linear interpolation of
CASSCF optimized geometries.
Isomerization path computations for the 2-cis-R-methyl-penta-
2,4-dieniminium (2-cis-R-Me-C5H6NH2+), a minimal model
of PSB11 with a penta-2,4-dieniminium (-CHdCH-CHd
CH-CHdNH(+)-) moiety, indicate that this system provides
a more suitable backbone.43,46 Indeed, such path shows no flat
S1 energy plateau (see framed region in Scheme 2) and the shape
of the potential energy surface corresponds to that given in
Scheme 3a. Due to the highly anharmonic shape of the S1 energy
For the CASPT2 calculations we used a zeroth-order three-
root (S0, S1, S2) state-average CASSCF wave function charac-
terized by a complete active space of 10-electrons in 10-orbitals
(10e/10o). For the CASPT2 computations of 1 and 5,5′-diMe-1
we used a zero-order two-root (S0, S1) state-average wave
function with a 6e/6o complete active space. The CASPT2
+
surface, the barrierless path of 2-cis-R-Me-C5H6NH2 (see
stream of arrows in Scheme 3a) is driven by highly coupled
stretching and isomerization modes. Thus, the reaction coordi-
nate becomes, to a certain extent, closer to the single-mode
coordinate of Scheme 1a, where one expects limited energy
redistribution and fast isomerization. This conclusion is sup-
ported by semiclassical dynamics computations that predict that
(43) Garavelli, M.; Celani, P.; Bernardi, F.; Robb, M. A.; Olivucci, M. J. Am.
Chem. Soc. 1997, 119, 6891-6901.
(44) Gonzales-Luque, R.; Garavelli, M.; Bernardi, F.; Merchan, M.; Robb, M.
A.; Olivucci, M. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9379-9384.
(45) Garavelli, M.; Vreven, T.; Celani, P.; Bernardi, F.; Robb, M. A.; Olivucci,
M. J. Am. Chem. Soc. 1998, 120, 1285-1288.
+
2-cis-R-Me-C5H6NH2 isomerizes in less than 100 fs.47
(46) Garavelli, M.; Bernardi, F.; Olivucci, M.; Vreven, T.; Klein, S.; Celani,
P.; Robb, M. A. Faraday Discuss. 1998, 110, 1-20.
(47) Vreven, T.; Bernardi, F.; Garavelli, M.; Olivucci, M.; Robb, M. A.; Schlegel,
H. B. J. Am. Chem. Soc. 1977, 119, 12 687-12 688.
(48) Olivucci, M.; Robb, M. A.; Bernardi, F. In Conformational Analysis of
Molecules in Excited States; Waluk, J., Ed.; Wiley-VCH: John Wiley and
Sons Inc.: New York, 2000; pp 297-366.
(49) Robb, M. A.; Garavelli, M.; Olivucci, M.; Bernardi, F. In ReViews in
Computational Chemistry; Lipkowitz, K. B., Boyd, D. B., Ed.; Wiley-
VCH: John Wiley and Sons Inc.: New York, Chichester, 2000, pp 87-
146.
(50) Andersson, K.; Barisz, M.; Bernhardsson, A.; Blomberg, M. R. A.; Cooper,
D. L.; Fleig, T.; Fu¨lscher, M. P.; de Graaf, C.; Hess, B. A.; Karlstro¨m, G.;
Lindh, R.; Malmqvist, P.-A° .; Neogra´dy, P.; Olsen, J.; Roos, B. O.;
Schmmelpfennid, B.; Schu¨ltz, M.; Sadlej, A. J.; Schu¨tz, M.; Seijo, L.;
Serrano-Andre´s, L.; Siegbahn, P. E. M.; Stårling, J.; Thorsteinsson, T.;
Veryazov, V.; Widmark, P.-O.; version 5.1 ed.: University of Lund,
Sweden, 2000.
(51) Page, C. S.; Olivucci, M. J. Comput. Chem. 2003, 24, 298-309.
9
J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004 9351

A R T I C L E S
Sampedro et al.
3. Results and Discussion
correction was applied only to CASSCF optimized geometries
(minima, transition states, and conical intersections). A detailed
comparison between the CASSCF and CASPT2 energy profiles,
absolute and relative energies for 1, 5,5′-diMe-1 and E-2H can
be found in the Supporting Information.
In subsection 3.1, we report the computed photochemical and
thermal reaction paths driving the isomerization of the E- and
Z- forms of 1. To investigate the possibility of prompting
unidirectional rotary motion, in subsection 3.2 we compare the
structure of the energy surface of 1 with that of its chiral
derivative 5,5′-diMe-1. In subsection 3.3, we demonstrate that
the photochemical reaction path of the synthetically accessible
chromophore 2H conforms to that seen for 1. Finally, in
subsection 3.4, we report the synthesis and the experimental
characterization of the derivatives 2, p-MeO-2, and p-NO2-2.
Unless otherwise stated, we shall only discuss CASPT2 energies.
Reaction paths are structural features of the potential energy
surface and do not provide direct information on the molecular
motion. However, since trajectory or quantum dynamics com-
putations on high-quality energy surfaces are presently impos-
sible for molecules of the size considered here, we use reaction
path analysis to discuss their possible dynamical behavior.
3.1 Prototype Switcher Based on a “Locked” Polyenal
Schiff Base (compound 1). In recent computational work,34,35,55
it has been shown that PSBs feature nearly competitive
photochemical Z/E isomerization paths corresponding to rotary
motion about the adjacent double bonds of the chromophore.
The existence of competing paths is clearly undesirable. For
this reason, 1 represents a good prototype for the design of
switchers based on the penta-2,4-dieniminium chromophore. In
fact:
(i) five member rings restrain isomerization about the initial
-CHdNH- and terminal -CHdCH- double bonds of the
structure;
(ii) five member rings are conformationally rigid so that the
possible existence of different conformers of the molecule is
avoided;
(iii) two -(CH2)2- saturated bridges create a scaffold for
functionalization of the switcher and for the introduction of
stereogenic centers (see subsection 3.2).
Notice that property (i) is also satisfied by the diarylene light-
driven molecular rotor13-16 mentioned above. However, these
molecules have “locks” based on cyclohexenylidene rather than
cyclopentenylidene rings and are therefore conformationally less
rigid. This flexibility is, in principle, a source of conformational
intermediates of the “I” type of Scheme 1b and are therefore
undesirable.
In all of the CASSCF and CASPT2 calculations we employed
the 6-31G* basis set except when otherwise stated. The S2 state
of the penta-2,4-dieniminium cation has been found to lie well
above (∼30 kcal mol-1) the lowest S1 state.43 For this reason,
the S2 state is expected not to contribute to the photocycle of 1
and 5,5′-diMe-1 and therefore is not considered. For comparison
with the experimental absorption spectrum the vertical excitation
energies for compound E-2H were computed at the CASPT2
level for the three lowest singlet excited states by averaging
four roots (ground state and three excited states). To assess the
effect of basis set expansion, the CASPT2 calculations at FC
were repeated using the 6-31G** basis set.
The molecular dipole moments and charge distribution
(Mulliken charges) along the backbone of E-2H are determined
at the CASSCF level of theory (see Supporting Information for
details). The structure of all conical intersections have been
optimized applying the methodology included in GAUSSIAN
98.52 The initial relaxation direction (IRD) method53,54 was used
to locate the steepest-descent direction to be follow in the MEP
calculation when starting at FC or CI points.
2.2 Synthesis and Photochemical Studies. The E- and
Z-isomers of 2, p-MeO-2, p-NO2-2 were easily obtained starting
from the corresponding neutral imines. Because the E-sterei-
somers are largely predominant in the synthesized neutral forms
2neut, p-MeO-2neut, p-NO2-2neut, the corresponding Z-
stereoisomers were obtained through irradiation of the E form.
The two isomers could be separated by chromatography. While
reaction of the E- and Z-stereisomers of 2neut, p-MeO-2neut,
and p-NO2-2neut with trifluoroacetic acid yielded the corre-
sponding PSBs, the reaction with methyl trifluoromethane-
sulfonate led quantitatively to the formation of the E- and
Z-forms of 2, p-MeO-2, and p-NO2-2. These methylated
derivatives were found to be more stable and tractable than the
protonated forms. In particular, we observed that the protonated
Z-stereisomers (2H, p-MeO-2H, and p-NO2-2H) tended to
transform thermally to the corresponding E-stereisomers in the
presence of impurities. The higher stability of N-methyl
compounds is qualitatively rationalized by the additional
stabilization of the positive charge at N and by the decreased
mobility of the alkyl group with respect to the proton. Full
synthetic procedures and characterization of new compounds
are reported in the Supporting Information.
While the five-membered ring “locks” of structure 1 are
responsible for the favorable properties i-iii, they also introduce
an angular (Bayer) strain that modifies the values of the N1-
C5-C4 and C1′-C2′-C3′ angles of the penta-2,4-dieniminium
moiety. Similarly, the inductive effects due to alkyl substitution
at the N1, C4, C1′, and C3′ must change the S1 and S0 positive
charge distribution along the chromophore framework. In
principle, these effects may change the favorable properties of
the penta-2,4-dieniminium cation seen in Schemes 2 and 3a.
Thus, to assess whether these properties are still satisfied by 1,
we computed the photoisomerization paths for both the E-1 f
Z-1 and Z-1 f E-1 processes.
(52) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian,
Inc.: Pittsburgh, PA, 1998.
Inspection of Figure 1 indicates that the computed E f Z
and Z f E paths conform to the model energy surface of
Scheme 3a and, thus, display an energy inflection point in
(53) Celani, O.; Robb, M. A.; Garavelli, M.; Bernardi, F.; Olivucci, M. Chem.
Phys. Lett. 1995, 243, 1-8.
(54) Garavelli, M.; Celani, P.; Fato, M.; Bearpark, M. J.; Smith, B. R.; Olivucci,
M.; Robb, M. A. J. Phys. Chem. A 1997, 101, 2023-2032.
(55) De Vico, L.; Page, C. S.; Garavelli, M.; Bernardi, F.; Basosi, R.; Olivucci,
M. J. Am. Chem. Soc. 2002, 124, 4124-4134.
9
9352 J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004

Biomimetic Light-Driven Z/E Switcher
A R T I C L E S
(S0) tip of the conical intersection splits the ground-state energy
surface in two different valleys.
3.2 A Chiral Derivative (compound 5,5′-diMe-1). The
comparison of the symmetric energy surface of Scheme 3a with
the asymmetric energy surface of Scheme 3b provides a pictorial
demonstration that the breaking of the torsional symmetry of
the S1 potential energy surface of 1 must prompt E f Z (for
the E-1 isomer) or Z f E (for the Z-1 isomer) unidirectional
rotary motion. In these conditions the initial relaxation involves:
(i) immediate coupling of the stretching and torsional
coordinate (as the gradient at FC will now have a small
component along the torsional deformation).
(ii) higher probability of clockwise (or counterclockwise)
rather than counterclockwise (or clockwise) motion since the
local force field will now be asymmetric with respect to the
two directions of torsional deformation.
In principle, such breaking can be induced at FC by
strategically placed substituents changing 1 to a chiral derivative,
such as 5,5′-diMe-1.
In Figure 2 we report an analysis of the energy relationship
between FC (i.e., the Z-1, 5,5′-diMe-E-1 and 5,5′-diMe-Z-1
equilibrium structures), the S1 transition structure (i.e., TSEX
Z
and TSEXE) and the conical intersection structures connected
to FC through clockwise and counterclockwise deformation (CI
and CI′ or CIP and CIM). As expected the FC, TSEXZ structures
of Z-1 (see Figure 2a) occurs at 0° torsion and, consistently
with Scheme 3a, the two ca. 90° twisted conical intersections,
CI and CI′, are mirror images (i.e., enantiomers). In contrast,
the FC and TSEXZ points of structure 5,5′-diMe-Z-1 (see Figure
2b) are displaced from planarity by -2° and -1°, respectively.
Consistently with Scheme 3b, the two conical intersections, CIP
and CIM, are not enantiomers but diastereoisomers (also notice
the difference in the absolute value of their torsional angles and
energies). Similarly, the FC and TSEXE points of structure 5,5′-
diMe-E-1 are displaced from planarity (the torsional angles are
174° and -171°, respectively).
Since TSEXZ defines the lowest energy point along the ridge
separating the clockwise and counterclockwise slopes of the S1
energy surface (see the “ridge” in Scheme 3b), then a molecule
relaxing from FC will be accelerated toward the steepest slope
directing the Z/E isomerization motion in a specific direction.
For this reason the R enantiomer of 5,5′-diMe-Z-1 (that of Figure
2b) is expected to direct the motion toward CIP (as pointed out
by the arrows). This idea is rigorously demonstrated by
computing the initial parts of the excited-state relaxation paths
of the E and Z forms of 1 and 5,5′-diMe-Z-1 that are, in fact,
pointing toward CIP. The corresponding data are given in the
Supporting Information.
3.3 Photoisomerization Path of a Synthetically Accessible
Derivative (structure 2H). Compounds 1 and 5,5′-diMe-1 are
presently unknown. Thus, while their synthesis represents an
attractive research target, the properties predicted above cannot
be experimentally assessed. For this reason, we have looked
for a analogue that could be synthesized. The synthesis of 2H
has been reported (see below). This compound features the same
five-membered heterocycle seen in 1. However, the cyclopen-
tenylidene ring is replaced with a benzylidene unit. While 2H
features an augmented conformational freedom due to the
presence of the unlocked single bond holding the phenyl ring,
it represents an analogue of the achiral switcher 1. We will now
Figure 1. Energy profiles along the four MEPs describing the relaxation
from the FC points of the E-1 and Z-1 stereoisomers and their (common)
CI point. Full and light arrows indicate the E f Z and Z f E photoreaction,
respectively. The structures (parameters in Å and degrees) document the
molecular structure change along the FC(E-1) f CI coordinate. The framed
S1 energy profile is located on an energy surface of the type seen in Scheme
3a. Accordingly, TSEXE and TSEXZ correspond to the transition states
separating the clockwise and counterclockwise relaxation. TSGSI and TSGSII
are transition states driving ground state Z/E isomerization. The CASSCF
S1 energies of all points along the S1 and S0 MEPs have been scaled69 to
match the S1 CASPT2//CASSCF energies computed for E-1, Z-1 and CI.
correspondence of a change in direction of the reaction
coordinate (first dominated by stretching and then by twisting
deformations). The fact that the FC regions of E-1 and Z-1 have
the shape seen in Scheme 3a, is also demonstrated by the
existence of two planar S1 transition states (TSEXE and TSEXZ).
Notice that the existence of these transition states implies that
the energy surface is symmetric with respect to an out-of-plane
deformation and therefore S1 clockwise and counterclockwise
twisting motions have the same probability to occur.
For both the E f Z and Z f E paths S1 f S0 decay occurs
in the region of a conical intersection (CI) displaying a ca. 90°
central bond. The data of Figure 1 also indicate that, similarly
to S1 relaxation, S0 relaxation does not involve formation of
any intermediate and the system relaxes directly to the energy
minimum corresponding to the photoproduct well. Since Z-1 is
the photoproduct of E-1 and this is the photoproduct of Z-1,
the two paths in Figure 1 describe a complete photocycle leading
to a return to the original material after absorption of two
photons of similar wavelengths. Finally, notice that two different
ca. 90° twisted transition states (TSGSI and TSGSII) both
controlling E-1 f Z-1 and Z-1 f E-1 thermal isomerization
have been located ca. 40 kcal mol-1 above the S0 equilibrium
structures. Such barriers guarantee that the thermal isomerization
will not compete with the photochemical process in these
species. As previously reported55 the general presence of two
different transition states driving the same thermal Z/E isomer-
ization of a PSBs is a consequence of the fact that the lower
9
J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004 9353

A R T I C L E S
Sampedro et al.
Figure 2. Relationship between relative energy and C1′-C4 torsion for the S1 stationary points and conical intersections of: (a) Z-1, (b) 5,5′-diMe-Z-1, and
(c) 5,5′-diMe-E-1. In 1, the planar FC and TSEXZ are located at the center of the plot. Clockwise and counterclockwise relaxation lead to two enantiomeric
conical intersections, CI and CI′. In 5,5′-diMe-1, FC for the Z- and E-stereisomers, TSEXZ and TSEXE are displaced from planarity. Clockwise and
counterclockwise relaxation are not equivalent and lead preferentially to one of the two diasteromeric conical intersections CIP or CIM.
show that, indeed, the electronic and molecular structure of this
species presents several of the features seen for 1.
stretching (the central double bond expands from 1.36 Å to 1.39
Å) coupled with a limited 4° torsional deformation about the
central double bond and a 3° deformation about the adjacent
unlocked single bond.
As displayed in Figure 3, E-2H is planar. In contrast Z-2H
features a 6° twist about the central double bond and a 45° twist
about the adjacent unlocked single bond. Presumably, the phenyl
ring is twisted out-of-plane by the steric interaction with the
heterocycle hydrogen. The corresponding destabilization also
accounts for the higher stability (-1.2 kcal mol-1) of E-2H with
respect to Z-2H.
The TSGSI connecting the two forms features a ca. 80° twisted
central double bond, and is located 43.9 kcal mol-1 above E-2H.
Thus, as for compound 1, such a high barrier indicates that the
E- and Z-forms will not thermally interconvert at room
temperature.
The excited-state surfaces of 2H and 1 present remarkable
differences. In fact, 2H has two low-lying ππ* excited states
separated by less than 2 kcal mol-1. At FC, the CASSCF level
of theory yields a spectroscopic state corresponding to S2,
whereas the lower S1 state corresponds to a dark state located
2.4 kcal mol-1 lower in energy. When dynamic correlation is
included, the spectroscopic state is pushed down by 5.3 kcal
mol-1 (see Figure 3), while the dark state is pushed up by only
0.3 kcal mol-1 Thus, at the CASPT2 level, the energy order of
the spectroscopic and dark state is inverted. Notice that due to
this inversion the S2/S1 crossing found at the CASSCF level
does no longer occur. Thus, at the CASPT2 level the spectro-
scopic state is the lowest excited state along the entire reaction
coordinate.
The analysis of the computed reaction coordinate reveals
some other differences between 2H and 1. In particular, the
S1 path features a minimum MINEX and a transition state
TSMINEXfCI (located 1.6 kcal mol-1 higher in energy) connect-
ing MINEX to a conical intersection CIS1/S0. Comparison of
structures E-2H and MINEX reveals that, similar to compound
1, the initial relaxation of the molecule is dominated by a
The evolution of MINEX is dominated by twisting about both
the central double bond and the adjacent unlocked single bond.
Such deformation causes the hydrogen atom in ortho-position
to approach one of the hydrogen atoms placed at C4 of the
pyrroline ring. The distance between these two atoms changes
from 2.22 Å at MINEX to 2.02 Å in TSMINEXfCI. This steric
interaction is partially avoided by the twisting of the phenyl
ring (3° in MINEX, 13° in TSMINEXfCI) that slowly returns to
planarity once surpassed the transition state. The lowest energy
point of the S1 energy surface corresponds to the conical
intersection structure CIS1/S0 featuring a ca. 90° twisted central
double bond.41 Despite the small computed barrier, TSMINEXfCI
may allow partial or complete energy redistribution at MINEX
.
In practical terms the ∼10 kcal mol-1 photon energy separating
FC and MINEX may be, at least partially, redistributed. Thus,
the photoisomerization path of E-2H may be consider closer to
that given in Scheme 1b due to the presence of a transient I*
corresponding to MINEX. In contrast, we have not been able to
locate any intermediate I along the S0 relaxation path.
To assess if structure 2H could be used as the basic
chromophore for the development of a single-molecule rotor
analogue to 5,5′-diMe-1, we have investigated the R enantiomer
of the chiral structure 4-Me-E-2H where a methyl group has
been inserted at C4. Although the computation of the full
photoisomerization path of 4-Me-E-2H goes beyond the scope
of the present work an analysis of its optimized S0 structure
reveals a firm deviation from planarity. In fact, the dihedral
angle between the planes defined by the two cycles has a value
of 4.4°, whereas the phenyl plane is twisted by 17.4° with
respect to the central double bond. This structural deformation
9
9354 J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004

Biomimetic Light-Driven Z/E Switcher
A R T I C L E S
Figure 3. Energy profiles along the MEP describing the excited-state relaxation (FC(E-2H) f MINEX f TSMINEXfCI f CIS1/S0) of E-2H. Full and open
diamonds indicate the CASSCF S1 and S0 energies, respectively; full and open squares indicate the CASPT2 S1 and S0 energies; open triangles indicate the
CASSCF S2 energies. The structures (geometrical parameters in Å and degrees) document the progression along this path and the equilibrium structures of
the E-2H and Z-2H stereoisomers.
is consistent with a helical structure of the molecule that suggests
a S1 potential energy surface of the type seen in Scheme 3b.
principle, they could result from an aldol-like condensation
between 1-pyrroline and appropriate aldehydes. This last ap-
proach seemed to us the preferable route since it leads to the
target compounds in a straightforward way, although, due to
its tendency to trimerize, the material referred to as 1-pyrroline62
presents special difficulties in the isolation.62-65 (In no case a
suitable method for isolation of 1-pyrroline in pure form was
developed.)
3.4 Syntheses of the Neutral Imines 2neut, p-MeO-2neut,
p-NO2-2neut. The search for synthetically available molecules
containing the penta-2,4-dieniminium unit, led us to select the
3-alkylidene-1-pyrrolines, for which a number of synthetic
methods56-61 have been reported to date. In particular, in the
first phase of the project, we turned our attention to simple para-
substituted-3-benzylidene-1-pyrrolines whose structures incor-
porate the crucial part of the wanted π-conjugated system and
allow for a change in the nature of the para-substituent on the
phenyl ring that could be used to adjust the electron density
within the π-system. (As we will discuss below, this simple
modification has also been exploited to “tune” the photochemical
behavior of these molecules.)
In 1982, it was developed a more convenient route to
1-pyrroline trimer by silver (I) catalyzed oxidation of pyrrolidine
with peroxodisulfate and, for the first time, it could be unam-
biguously assigned the structure of 1,6,11-triazatetracyclo-
[10.3.0.0.2,607,11]pentadecane66 (see Scheme 4). Interestingly, the
same authors found that this material reacted with benzaldehyde
to furnish 3-benzylidene-1-pyrroline 2neut in acceptable yield.67
Most of the synthetic approaches to these targets feature a
cationic cyclization58-61 as the key step even if, at least in
(62) Fuhlhage, D. W.; Vander Werf, C. A. J. Am. Chem. Soc. 1958, 80, 6249-
6254.
(63) Cragg, J. E.; Herbert, R. B.; Jackson, F. B.; Moody, C. J.; Nicolson, I. T.
J. Chem. Soc, Perkin Trans. 1 1982, 2477-2485.
(56) Meyers, A. I.; Ritter, J. J. J. Org. Chem. 1958, 23, 1918-1922.
(57) Sugasawa, S.; Ushioda, S. Tetrahedron 1959, 5, 48-52.
(64) Jakoby, W. B.; Fredericks, J. J. Biol. Chem. 1959, 234, 2145-2150.
(65) Richards, J. C.; Spenser, I. D. Tetrahedron 1983, 39, 3549-3568.
(66) Ogawa, K.; Nomura, Y.; Takeuchi, Y.; Tomoda, S. J. Chem. Soc., Perkin
Trans. 1 1982, 3031-3035.
(58) Gawley, R. E.; Termine, E. J. J. C. S. Chem. Commun. 1981, 568-569.
(59) Sakane, S.; Matsumura, Y.; Yamamura, Y.; Ishida, Y.; Maruoka, K.;
Yamamoto, H. J. Am. Chem. Soc. 1983, 105, 672-674.
(60) Gawley, R. E.; Chemburkar, S. Tetrahedron Lett. 1986, 27, 2071-2074.
(61) Gawley, R. E.; Chemburkar, S. Heterocycles 1989, 29, 1283-1292.
(67) Nomura, Y.; Bando, T.; Takeuchi, Y.; Tomoda, S. Bull. Chem. Soc. Jpn.
1983, 56, 3199-3200.
9
J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004 9355

A R T I C L E S
Sampedro et al.
Scheme 4
Table 1. UV/Vis Absorptions (λMax) of Compounds 2, p-MeO-2,
and p-NO2-2
structure
λmax, nm
ꢀ, M-1cm-1
λ
max, nm
ꢀ, M-1cm-1
E-2H
326
30600
225
10300
Z-2Ha
E-2
Z-2
p-MeO-E-2
p-MeO-Z-2
p-NO2-E-2b
p-NO2-Z-2b
327
328
364
365
322
310
17400
11800
6600
6600
13000
19000
254
248
265
265
223
224
26000
21000
10100
13600
8500
18000
a Unstable. b Introduction of EW-NO2 in compound 2 results in a
relatively small difference in the UV/Vis absorption spectra for the Z- and
E-isomers. In fact p-NO2-E-2 absorbs at λmax)322 nm (ꢀ)1300) and p-NO2-
Z-2 has a higher-intensity absorption at 310 nm (ꢀ)1900).
the experimental absorption has a maximum at 326 nm, with
an approximate oscillator strength of 0.7. Although S2 was found
to be nearly degenerate to S1, it has f ) 0.029. The S0 f S2
transition is thus expected not to contribute to the qualitative
aspect of the spectrum since the corresponding weak band will
be hidden by the close and intense band associated with the
first transition. For the S0 f S3 transition we compute a λmax
value of 253 nm with an oscillator strength of 0.16. This values
compare reasonably well with the observed λmax ) 225 nm and
f ) 0.1.
The computed spectra are found to be red-shifted with respect
to the spectral bands observed in acetonitrile. The deviation
between the theoretical results and the experimental data -0.34
eV for the S0 f S1 transition and -0.61 eV for the S0 f S3
transition) is larger than the expected CASPT2 error (typically
0.2 eV) on excitation energies. Furthermore, we find that the
vertical excitation energy values obtained using the 6-31+G*
basis set are very close to those computed with the 6-31G* basis
set (see Table SI2 in the Supporting Information) supporting a
very limited basis set effect. We conclude that the observed
deviation is most probably due to the effect of the (solvated)
negative counterion present in solution on the different electronic
structure (charge distribution) of the ground and excited states
of 2H. In fact, as previously reported,43 the S1 state of the penta-
2,4-dieniminium chromophore has a dominant charge-transfer
(i.e., ionic) character while the S2 state has a diradical (i.e.,
covalent) character. Due to the more extended π-system (i.e., a
phenyl group is replacing a double bond) the S1, S2, and S3
states of 2H all feature a charge-transfer character that can be
quantified in terms of the total charge of the benzylidene and
heterocyclic moieties (see the Supporting Information for the
details). Accordingly, the main resonance formulas describing
these states are as follows:
They showed how the predominant trimer has to be in
equilibrium with 3,4-dihydro-2-H-pyrrole, that is in turn able
to react with benzaldehyde in an aldol-like condensation. The
main advantage of this reaction is the possibility of carrying
out the synthesis of the target molecules without the use of any
basic or acid catalysts, a need due to the extreme chemical
instability of 1-pyrroline.62
The application of this protocol led us to obtain the unknown
derivatives p-MeO-2neut and p-NO2-2neut simply by combin-
ing the trimer with p-methoxy- or p-nitro benzaldehyde,
respectively. However, in the latter case, adopting usual reaction
conditions, only traces of the desired p-nitro benzylidene
derivative are formed; instead, a solid and unexpectedly polar
compound (TLC) is dominant in the reaction mixture. Due to
the lack of olefin proton signal on the 1H NMR spectrum this
was tentatively assigned to the intermediate aldolic reaction
product. Consequently, we decided to force the removal of water
by heating a methanolic solution of this intermediate in the
presence of AcOH/AcONa buffer, furnishing the desired
compound in more than 40% yield.
In all cases, the preparations of 2neut, p-MeO-2neut, and
p-NO2-2neut yielded a major E-stereisomer and only minor
quantities of the Z-stereisomer. The geometry of their exocyclic
double bonds was easily inferred both on the base of charac-
teristic53 coupling between allylic methylene protons and the
exocyclic olefin protons and, on NOE experiments. It is
noteworthy that the E isomers partially convert to Z isomers
by the action of UV light and that by chromatography we could
separate the diastereisomers eventually transformed to the
targeted E- and Z-isomers of imines 2, p-MeO-2, and p-NO2-2
(see Supporting Information).
3.4.1 Absorption Spectra. The synthesis of 2H allows the
validation of our computational strategy (see subsection 2.1)
by comparing computed (gas-phase) and measured (solution)
spectroscopic quantities such as the absorption maximum (λmax).
As reported in Table 1 we observed two main bands (see the
spectrum of 2H in Figure SI4 of the Supporting Information)
with λmax at 225 nm and at 326 nm, respectively.
The long wavelength band in the recorded spectrum can be
related to the computed S0 f S1vertical transition. This transition
has a λmax ) 358 nm with an oscillator strength of 0.82, whereas
9
9356 J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004

Biomimetic Light-Driven Z/E Switcher
A R T I C L E S
Table 2. PSS Composition for Compounds 2, p-MeO-2, and
p-NO2-2
In solution, the counterion will be statistically closer to the
locus of the positive charge of the S0 structure (i.e., the
-CHdNH- moiety). Upon S0 f S1 excitation the positive
charge is translocated in the region of the phenyl ring and
therefore away from the counterion. It is therefore clear that, in
the presence of the counterion the S0-S1 energy gap will
increase due to preferential stabilization of S0 relative to S1
yielding a blue-shifted λmax. The same reasoning leads to the
conclusion that also the S0-S3 energy gap will be increased by
the presence of the counterion.
ratioa at PSS
compound
λ (nm)
E
Z
2
313
313
313
300
310
320
330
360
223
1
1
1
1
1
1
1
1
1
0.3
0.1
1
0.1
0.7
1
1
1.5
0
p-MeO-2
p-NO2-2
p-NO2-2
p-NO2-2
p-NO2-2
p-NO2-2
p-NO2-2
p-NO2-2
The spectra of 2, p-MeO-2 and p-NO2-2 have the same two-
band structure characterizing the spectrum of the parent
compound 2H. The comparison between the spectroscopic data
for 2H and 2 shows virtually no differences regarding the first
photoactive band, which occurs at an almost identical wave-
length in the two compounds (λmax ) 326 nm in 2H compared
λmax ) 327 nm in 2). These results suggest that the excited-
state behavior of 2 and 2H will be substantially equal.
Accordingly, below we focus exclusively on the more experi-
mentally tractable methylated compound 2 and its derivatives.
1
a Ratios determined by H NMR peak integral.
investigate the substituent and wavelength dependence of the
PPS. The effect of the substituent was investigated using a fixed
irradiation wavelength (λmax g 313 nm) for 2, p-MeO-2 and
p-NO2-2. Although compound 2 showed a 3:1 ratio with E-2
as the main product, the ratio for p-MeO-2 was raised to 10:1
and for p-NO2-2 was 1:1. This effect cannot be explained only
on the basis of different values of the λmax as in 2 and p-MeO-2
these values are substantially identical for the E and Z forms
(see Table 1). Nevertheless, with respect to the unsubstituted
compound, the ER substituent of p-MeO-2 induces a shift of
the composition toward the E-form. This implies that the
quantum yields for the E f Z and Z f E processes depend on
the substituent and that the Z f E process is enhanced by the
-OCH3 group. Such complex effects cannot be explained on
the basis of qualitative considerations and require further
computational/mechanistic studies to be correctly interpreted.
According to Table 1 compoundp-NO2-2 features slightly
different long-wavelength λmax for the E- and Z-forms. Thus in
this compound one would predict a sensitivity of the PPS
composition to the irradiation wavelength. In fact, our experi-
mental data (see Table 2) show that, increasing the wavelength
from 300 to 360 nm, the PPS E/Z ratio inverts from 1:0.1 to
1:1.5. (i.e., from 91% p-NO2-E-2 and 9% p-NO2-Z-2 to 40%
and 60%, respectively).
The data in Table 1 shows that, for both the E- and
Z-stereisomers, the λmax values of the first and second band are
sensitive to the electron-withdrawing (EW) or electron-releasing
(ER) character of the substituent in para-position. In compound
p-MeO-2 that carries an ER group the two bands are red-shifted
relative to 2, while in compound p-NO2-2 that instead carries
the EW group they are blue-shifted. These spectral changes are,
again, readily explained on the basis of the electronic structure
of the S0, S1, and S3 states. Accordingly, the ER groups (e.g.,
-OCH3) that stabilize the phenyl ring positive charge will
stabilize the S1 and S3 state relative to the S0 state leading to a
red-shifted λmax of the first and second band. In contrast, the
EW groups (e.g., -NO2) that destabilize the positive charge
on the phenyl ring will yield blue-shifted λmax. The comparison
of the spectra for p-MeO-2 and p-NO2-2 with that of 2 is thus
fully consistent with the computed S3, S1, and S0 electronic
structures.
3.4.2 Photoisomerization and Composition of the Photo-
stationary States. According to the reaction path of Figure 3,
upon irradiation E-2 is expected to produce Z-2. Similarly Z-2
is expected to produce E-2. Because these species have
comparable λmax and ꢀ values, one expects generation of a
photostationary state (PPS) whose composition (i.e., ratio
between the E- and Z-forms) is independent from the starting
stereisomer and controlled by the differences in Z-2 and E-2
extinction coefficients and by the differences in quantum yields
of the Z f E and E f Z photoreactions. Indeed, Pyrex-filtered
irradiation of both Z-2 and E-2 yielded after 1-2 h fully
equivalent mixtures of both stereisomers in a 3:1 ratio (deter-
As mentioned in Section 3.4.1, our compounds present a
second band at ca. 225 nm assigned to the S0 f S3 transition.
We selectively irradiated this band in p-NO2-E-2. After several
hours of irradiation at 223 nm no traces of the Z-isomer could
be found. However, irradiation of the Z-isomer in the same band
yielded the more stable E-isomer with complete conversion.
Since no side-reaction disrupting the photocycle could be
detected, this result suggests that photoisomerization from S3
follows a different (but well defined) reaction pathway involving
three excited-state energy surfaces. The computational investiga-
tion of such complex pathway goes beyond the scope of the
present work.
1
mined via H NMR), always with the E form as the main
Notice that the degree of photodegradation of 2 (a property
of great importance for the applications of molecular switchers)
was assessed irradiating the photostationary mixture for 24 h
product. This fact corroborates that Z-2 is the photoproduct of
E-2 and, in turn, this is the photoproduct of Z-2. Thus,
consistently with the mechanism shown in Figure 1 for the
simpler chromophore 1, irradiation of 2 must lead to a
photocycle.
1
in the same reaction conditions. We found that the H NMR
spectrum remained unaltered after that time, showing that the
photochemical decomposition of these molecules is minimal.
3.4.3 Photophysical and Photochemical Properties of
p-NO2-2. To elucidate the S1 deactivation mechanism of the
compounds under investigation and to collect quantitative
information about the quantum efficiency of the photoreaction,
we carried out a photophysical study of p-NO2-2 including flash-
The basic photocycle proposed for our class of switchers
implies that, at a given wavelength, the differences in extinction
coefficient (roughly, the differences in absorption maxima) of
the Z and E forms can be exploited to control the PPS
composition. Thus, to support the computed mechanism, we
9
J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004 9357

A R T I C L E S
Sampedro et al.
Figure 4. S1 reaction paths controlling the evolution and decay of 2H. The MINEX f TSMINEXfCI f CIS1/S0 path is documented in Figure 3. The MINEX
f MINTWIST f CITWIST is associated with rotation about the single-bond connecting the phenyl moiety to the central double bond. The bold numbers along
the energy profile document the relative change in energy. The numbers in italic document the change in molecular dipole moment. The structures (parameters
in Å and degrees) document the progression along the single bond twisting path.
photolysis experiments with ns-resolution. In this section, we
report the preliminary results of such investigation. First of all,
no effective radiative decay channel seems to be operative in
p-NO2-E-2 and p-NO2-Z-2 in acetonitrile solutions as demon-
strated by the absence of fluorescence emission in the UV-vis
region. Taking into account the instrumental sensitivity, an upper
limit of 10-4 was estimated for the fluorescence quantum yields
of the two isomers. Upon laser excitation at 355 nm of p-NO2-
E-2 in acetonitrile solutions at room temperature, no transients
were detected in the investigated UV-vis region (300-750 nm)
by ns-laser flash photolysis experiments. On the basis of this
experiment the presence of either an excited-state or a ground-
state intermediate with lifetimes above 20 ns can be excluded.
The E f Z photoisomerization process was spectrophotometri-
cally followed in acetonitrile solutions by irradiation of the
p-NO2-E-2 at 320 and 360 nm. Small E f Z photoisomerization
quantum yields (ca. 6 × 10-3) were measured in these
experimental conditions indicating that the majority of the lowest
excited state (S1) population does not decay through the Z/E
reactive channel.
energy surface of the parent system 2H. As reported in Figure
4 the calculations revealed that a flat conformational path exists
along the unlocked single bond torsion leading to a new excited-
state intermediate (MINTWIST) featuring a ca. 90° twisted phenyl
ring and lying 1.2 kcal mol-1 below MINEX. Most importantly
a new conical intersection (CITWIST) structure connected to
MINTWIST was located ca. 4 kcal mol-1 higher in energy than
MINEX. Inspection of the CITWIST structure (close to that
documented for polyenes68) reveals that despite the highly
distorted hetherocycle the double bond is still substantially
planar while the single bond features a large twisting. As a result
decay at CITWIST may only lead to production of a “degenerate”
photoproduct, in the sense that twisting about the single bond
will only regenerate the starting material.
In the gas phase, the MINEX f MINTWIST f CITWIST
channel is predicted to be less efficient than the MINEX f TSEX
f CIS1/S0 channel due to a ca. 2 kcal mol-1 larger energy barrier.
Accordingly the discovery of such an alternative decay route
is, again, not consistent with a low E f Z quantum yield.
However, the computations also reveal that the change in dipole
moment occurring along MINEX f MINTWIST should favor
population of this route when the molecule is embedded in a
polar environment. Indeed, as reported in Figure 4, twisting
about the single bond yields a 4 D increase of the dipole moment
while twisting about the double bond does not lead to a
substantial dipole moment change. The low quantum yield
measured in acetonitrile is thus consistent with a >2 kcal mol-1
The facts that p-NO2-E-2 does not show any fluorescence
and no long-lived excited- or ground-state transients are detected
are consistent with the photoisomerization path presented in
Figure 3 (the only source of fluorescence may be MINEX
.
However, even a simple Arrhenius picture of its decay via
CIS1/S0 would yield an excited-state lifetime of ca. 2 ps that is
well below our 20 ns resolution). In contrast, the observed very
limited p-NO2-E-2 f p-NO2-Z-2 quantum yield is at variance
with the proposed gas-phase mechanistic picture. Indeed, decay
at a 90° twisted conical intersection located at the very bottom
of the S1 energy surface should lead to an efficient photoi-
somerization with a quantum yield >0.5. For this reason, we
shall conclude that, in the chosen experimental conditions,
p-NO2-E-2 must decay via an unreactiVe channel. The nature
of this channel was investigated via further mapping of the S1
(68) Celani, O.; Garavelli, M.; Ottani, S.; Bernardi, F.; Robb, M. A.; Olivucci,
M. J. Am. Chem. Soc. 1995, 117, 11 584-11 585.
(69) The energies along the S1 FC(E-1) f CI and FC(Z-1) f CI MEPs are
two-root (S0, S1) state average CASSCF energies relative to the S0 energy
of FC(Z-1). The S0 energies along the CI f Z-1 and CI f E-1 MEPs are
single-root (S0) CASSCF energies relative to the S0 energy of FC (Z-1)
S0. These relative S0 and S1 energies have been scaled to match the CASPT2
S0 f S1 vertical energy separation at FC(Z-1). The energies of S0 and S1
transition structures are instead true CASPT2 energies (see Table SI2 in
Supporting Information).
9
9358 J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004

Biomimetic Light-Driven Z/E Switcher
A R T I C L E S
stabilization of the single bond twisting route and therefore with
a strong decrease of the E f Z isomerization quantum
efficiency. While further computational and, in particular,
experimental investigations are required to confirm the existence
of the competitive decay pathway illustrated above, it is likely
that the synthesis of “single bond locked” derivatives such as 3
should yield compounds displaying highly efficient Z f E
photoisomerization. Such synthesis is currently pursued in our
laboratory.
ab initio CASSCF and CASPT2. The results provide the basis
for the development of a class of biomimetic switchers related
to the retinal protonated Schiff base chromophore of rhodopsin
proteins. In particular, the structure of the photoisomerization
paths of the prototype compounds 1 and 5,5′-diMe-1 satisfy
the criteria required for an efficient molecular switcher and
motor, respectively. In turn, the successful synthesis and
photochemical characterization of three different analogues of
1 (i.e., 2, p-MeO-2, p-NO2-2) clearly call for an increased
research effort based on tightly coupled computer modeling and
lab synthesis work.
Acknowledgment. Funds have been provided by the Uni-
versita` di Siena (Progetto di Ateneo 02/04) and HFSP (RG 0229/
2000-M). D.S. thanks the Spanish MECD for his fellowship.
We thank CINECA for granted calculation time.
Supporting Information Available: (i) Computational details.
(ii) Comparison between E f Z and Z f E photoisomerization
energy profiles for compounds 1 and 5,5′-diMe-1. (iii) Com-
parison between CASSCF and CASPT2 energy profiles for 2H.
(iv) Cartesian coordinates for compounds 1, 5,5′-diMe-1, and
2H. (v) Charge distribution and dipole moments for different
structures of compound 2H. (vi) Full synthetic procedures and
photochemical characterization data for compounds 2, p-MeO-2,
and p-NO)-2. This material is available free of charge via the
Internet at http://pubs.acs.org.
4. Conclusions
Despite the increasing interest in the development of novel
and more effective light-driven molecular devices, ab initio
quantum chemical methods have not been routinely employed
for the design of these functional materials. Among other
reasons, the specialized methods required for the correct
investigation of excited states, where the photon energy is
initially stored, have precluded such a desirable development.
Above, we have presented the result of a computational exercise
employing state-of-the-art quantum chemical methods such as
JA038859E
9
J. AM. CHEM. SOC. VOL. 126, NO. 30, 2004 9359