Cyclohexenylphenyldiazene: A simple surrogate of the azobenzene photochromic unit

Article abstract of DOI:10.1021/ja0668466

We have carried out an experimental and computational study on the ground- and excited-state photochemical and photophysical properties of (1-cyclohexenyl)phenyldiazene (CPD), a species formally derived from azobenzene in which one of the phenyl rings is

Full text of DOI:10.1021/ja0668466

Published on Web 02/23/2007
Cyclohexenylphenyldiazene: A Simple Surrogate of the
Azobenzene Photochromic Unit
Irene Conti,† Filippo Marchioni,† Alberto Credi,*,† Giorgio Orlandi,†
Goffredo Rosini,‡ and Marco Garavelli*,†
Contribution from the Department of Chemistry “G. Ciamician”, UniVersity of Bologna,
Via Selmi 2, 40126 Bologna, Italy, and Department of Organic Chemistry “A. Mangini”,
UniVersity of Bologna, Viale del Risorgimento 4, 40136 Bologna, Italy
Received September 22, 2006; E-mail: marco.garavelli@unibo.it; alberto.credi@unibo.it
Abstract: We have carried out an experimental and computational study on the ground- and excited-state
photochemical and photophysical properties of (1-cyclohexenyl)phenyldiazene (CPD), a species formally
derived from azobenzene in which one of the phenyl rings is replaced by a 1-cyclohexene substituent. The
results show that CPD does substantially behave like azobenzene, but with a higher (∼70%) ΦZfE (nπ*)
photoisomerization quantum yield, calling for CPD as an effective alternative of azobenzene itself with
new functionalization possibilities. By use of state-of-the-art ab initio CASPT2//CASSCF minimum energy
path computations, we have identified the most efficient decay and isomerization routes of the absorbing
singlet (ππ*), S1 (nπ*), T1, and S0 states of CPD. The resulting mechanistic scheme agrees with experimental
findings and provides a rationale of the observed photoisomerization quantum yields. Furthermore, this
study provides a deep insight on the photophysical and photochemical properties of compounds based on
the -NdN- double bond which supplies a general model for the photoreactivity of azobenzene-type
compounds in general. This is expected to be a useful guideline for the design of novel photoreactive azo
compounds.
Chart 1. Structural Formulas and E-Z Photochemical and
Thermal Isomerization Reactions of Azobenzene (AB) and
(1-Cyclohexenyl)phenyldiazine (CPD)
1. Introduction
The physicochemical and structural changes associated with
the E / Z photoisomerization of azobenzene derivatives1-3 have
been widely exploited to modify chemical systems by means
of light stimulation. For instance, such a reaction has been used
to modulate the structural and physicochemical properties of
polymers,4,5 liquid crystals,6 biomolecules,7 surfaces,8 and
dendritic9 and supramolecular10,11 systems, as well as to
construct multichromophoric compounds,10,12 molecular-level
memories,13 logic gates,14 devices, and machines.15,16
Because of the importance of azobenzene (AB, Chart 1) for
both basic science and applicative reasons, the mechanism of
† Department of Chemistry “G. Ciamician”, University of Bologna.
‡ Department of Organic Chemistry “A. Mangini”, University of Bologna.
(1) Rau, H. In Photochromism: Molecules and Systems; Du¨rr, H., Bouas-
Laurent, H., Eds.; Elsevier: Amsterdam, 2003; Chapter 4.
(2) Griffiths, J. Chem. Soc. ReV. 1972, 1, 481.
(3) Bortolus, P.; Monti, S. J. Phys. Chem. 1979, 83, 648 and references therein.
(4) (a) Sudesh Kumar, G.; Neckers, D. C. Chem. ReV. 1989, 89, 1915. (b)
Natansohn, A.; Rochon, P. Chem. ReV. 2002, 102, 4139.
(5) For a recent interesting example, see: Lu, Y.; Nakano, M.; Ikeda, T. Nature
2003, 425, 145.
(6) (a) Ikeda, T.; Kanazawa, A. In Molecular Switches; Feringa, B. L., Ed.;
Wiley-VCH: Weinheim, 2001; Chapter 12. (b) Boss, M. L.; Aramendia,
P. F. Photochem. Photobiol. Sci. 2002, 1, 507. (c) Langhoff, A.; Giessel-
mann, F. ChemPhysChem 2002, 3, 424.
(7) (a) Pieroni, O.; Fissi, A.; Angelini, N.; Lenci, F. Acc. Chem. Res. 2001,
34, 9. (b) Willner, I.; Willner, B. In Molecular Switches; Feringa, B. L.,
Ed.; Wiley-VCH: Weinheim, 2001; Chapter 6. (c) Ciardelli, F.; Pieroni,
O. In Molecular Switches; Feringa, B. L., Ed.; Wiley-VCH: Weinheim,
2001; Chapter 13.
(8) (a) Walter, D. G.; Campbell, D. J.; Mirkin, C. A. J. Phys. Chem. B 1999,
103, 402. (b) Janshoff, A.; Neitzert, M.; Oberdorfer, Y.; Fuchs, H. Angew.
Chem., Int. Ed. 2000, 39, 3212. (c) Sortino, S.; Petralia, S.; Conoci, S.; Di
Bella, S. J. Mater. Chem. 2004, 14, 811.
its E / Z photoisomerization reaction has been extensively
studied in the past.1 More recently, owing to the development
9
3198
J. AM. CHEM. SOC. 2007, 129, 3198-3210
10.1021/ja0668466 CCC: $37.00 © 2007 American Chemical Society

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Scheme 1. Synthetic Scheme for (1-Cyclohexenyl)phenyldiazine
(CPD) and Its Derivatives
arylazoalkenes with different ring sizes, too, which are prevented
in the parent compound azobenzene and could be exploited for
later applications.
Arylazoalkenes can be efficiently prepared from carbonyl
derivatives bearing a good leaving group X on the carbon atom
adjacent to the carbonyl function. The addition of arylhydrazine
gives the corresponding arylhydrazone, a labile intermediate,
that affords arylazoalkene by treatment with a mild base that
promotes a clean 1,4-elimination of HX.19 Typical reactions of
arylazoalkenes are: (i) the conjugate addition of nucleophiles19a,20
to the azo-ene moiety and (ii) the 4 + 2 cycloaddition reactions
with a variety of carbon-carbon dienophiles and heterodieno-
files, leading to the formation of six-membered heterocyclic
compounds.21
Here, we report the results of an investigation on the photo-
physical and photochemical properties of CPD, carried out by
electronic spectroscopic measurements and photochemical
experiments in solution and by computational methods. In
particular, fully unconstrained CASPT2//CASSCF/6-31G* mini-
mum energy path computations have been used to elucidate the
multistate dynamics and reaction coordinates, unveiling CPD’s
photochemical reactivity, as well as the energetics involved, and
have achieved a full rationale of its photochemical behavior.
For this purpose, the ground- and excited-state properties of
CPD in the singlet manifold as well as in the triplet T1 state,
with specific reference to its E / Z isomerization reactions,
are discussed in comparison to those of azobenzene.
of more powerful (time-resolved) spectroscopic and computa-
tional techniques, the photoisomerization of azobenzene has
been the subject of further detailed investigations with experi-
mental17 and theoretical17f,18 approaches.
While it is well-known1,2 that the photochemical behavior of
azobenzene derivatives is strongly influenced by the substituents
on the phenyl rings, not much information is available on the
photoreactivity of azobenzene-type compounds in which the
structure of the rings is changed. In particular, we are interested
to see if and how the spectroscopic and photochemical properties
change when the π-electron system adjacent to the -NdN-
double bond is reduced in size and the aromaticity has been
eventually removed in one of the two rings. For this reason,
we have synthesized and studied (1-cyclohexenyl)phenyldiazene
(CPD, Chart 1), a species formally derived from azobenzene in
which one of the phenyl rings is replaced by a cyclohex-1-ene
moiety. Besides representing an azobenzene-type compound
with reduced π-conjugation and aromaticity, its appeal resides
also in its new funtionalization potentialities as revealed by its
synthetic protocol (Scheme 1) that allows the preparation of
2. Experimental and Computational Details
(9) For representative examples, see: (a) Archut, A.; Vo¨gtle, F.; De Cola, L.;
Azzellini, G. C.; Balzani, V.; Ramanujam, P. S.; Berg, R. H. Chem. Eur.
J. 1998, 4, 699. (b) Junge, D. M.; McGrath, D. V. J. Am. Chem. Soc. 1999,
121, 4912. (c) Sebastian, R. M.; Blais, J. C.; Caminade, A. M.; Majoral,
J.-P. Chem. Eur. J. 2002, 8, 2172. (d) Uda, M.; Momotake, A.; Arai, T.
Photochem. Photobiol. Sci. 2003, 2, 845. (e) Nithyanandhan, J.; Jayaraman,
N.; Davis, R.; Das, S. Chem. Eur. J. 2004, 10, 689.
(10) (a) Balzani, V.; Scandola, F. Supramolecular Photochemistry; Horwood:
Chichester, 1991; Chapter 7. (b) Shinkai, S. In Molecular Switches; Feringa,
B. L., Ed.; Wiley-VCH: Weinheim, 2001; Chapter 9.
2.1. Experimental Section. Reagents and dry solvents were
purchased from Fluka, unless otherwise noted. E-Azobenzene was
purchased from Aldrich.
The absorption and luminescence spectra were recorded with a
Varian Cary 50 spectrophotometer and a LS 50 spectrofluorimeter,
respectively, on air-equilibrated acetonitrile or cyclohexane (Merck
Uvasol) solutions at room temperature, with concentrations ranging from
1 × 10-5 to 1 × 10-4 M, contained in 1-cm quartz cells. Experimental
errors: wavelengths, (1 nm; molar absorption coefficients, (5%.
Photochemical reactions were performed on the same solutions,
thoroughly stirred, by using the Xe lamp (150 W) of a Perkin-Elmer
650-10S spectrofluorimeter. The monochromator of the instrument
(bandpass ) 10 nm) was employed for the selection of the desired
irradiation wavelength. The number of incident photons, determined
by ferrioxalate actinometry22 in its micro version,23 was between 10-8
and 10-7 Nhν/min. The E f Z quantum yields for ππ* (λirr ) 325 nm
for CPD and 345 nm for AB; wavelengths of maximum conversion to
the Z isomer) and nπ* (λirr ) 430 nm) irradiation were determined
from the disappearance of the ππ* absorption band of the reactant at
low conversion percentages (<10%; if necessary, extrapolation to t )
(11) For early, representative examples, see: (a) Ueno, A.; Yoshimura, H.; Saka,
R.; Osa, T. J. Am. Chem. Soc. 1979, 101, 2779. (b) Shinkai, S.; Nakaji, T.;
Nishida, Y.; Ogawa, T.; Manabe, O. J. Am. Chem. Soc. 1980, 102, 5860.
(12) Representative examples: (a) Shinkai, S.; Manabe, O. Top. Curr. Chem.
1984, 121, 67. (b) Ritter, G.; Ha¨felinger, G.; Lu¨ddecke, E.; Rau, H. J. Am.
Chem. Soc. 1989, 111, 4627. (c) Tamaoki, N.; Koseki, K.; Yamaoka, T.
Angew. Chem., Int. Ed. Engl. 1990, 29, 105. (d) Vo¨gtle, F.; Mu¨ller, W.
M.; Mu¨ller, U.; Bauer, M.; Rissanen, K. Angew. Chem., Int. Ed. Engl. 1993,
32, 1295. (e) Ro¨ttger, O.; Rau, H. Mol. Cryst. Liq. Cryst. Sci. Technol.,
Sect. A 1994, 246, 143. (f) Bencini, A.; Bernardo, M. A.; Bianchi, A.;
Ciampolini, M.; Fusi, V.; Nardi, N.; Parola, A. J.; Pina, F.; Valtancoli, B.
J. Chem. Soc., Perkin Trans. 2 1998, 413. (g) Cisnetti, F.; Ballardini, R.;
Credi, A.; Gandolfi, M. T.; Masiero, S.; Negri, F.; Pieraccini, S.; Spada,
G. P. Chem. Eur. J. 2004, 10, 2011.
(13) Kawata, S.; Kawata, Y. Chem. ReV. 2000, 100, 1777.
(14) (a) Qu, D.-H.; Wang, Q.-C.; Tian, H. Angew. Chem., Int. Ed. 2005, 44,
5296. (b) Qu, D.-H.; Wang, Q.-C.; Ma, X.; Tian, H. Chem. Eur. J. 2005,
11, 5929.
(15) Balzani, V.; Credi, A.; Venturi, M. Molecular DeVices and Machines: A
Journey Into the Nano World; Wiley-VCH: Weinheim, 2003.
(16) For recent examples of optomechanical devices based on azobenzene, see
ref 14 and: (a) Hugel, T.; Holland, N. B.; Cattani, A.; Moroder, L.; Seitz,
M.; and Gaub, H. E. Science 2002, 296, 1103. (b) Ji, H.-F.; Feng, Y.; Xu,
X.; Purushotham, V.; Thundat, T.; Brown, G. M. Chem. Commun. 2004,
2532. (c) Wang, Q.-C.; Qu, D.-H.; Ren, J.; Chen, K.; Tian, H. Angew.
Chem., Int. Ed. 2004, 43, 2661. (d) Muraoka, T.; Kinbara, K.; Aida, T.
Nature 2006, 440, 512.
(17) (a) Rau, H.; Luddecke, E. J. Am. Chem. Soc. 1982, 104, 1620. (b) Fujino,
T.; Tahara, T. J. Phys. Chem. A 2000, 104, 4203. (c) Fujino, T.; Arzhantsev,
S. Y.; Tahara, T. J. Phys. Chem. A 2001, 105, 8123. (d) Fujino, T.;
Arzhantsev, S. Y.; Tahara, T. Bull. Chem. Soc. Jpn. 2002, 75, 1031. (e)
Hirose, Y.; Yui, H.; Sawada, T. J. Phys. Chem. A 2002, 106, 3067. (f)
Schultz, T.; Quenneville, J.; Levine, B.; Toniolo, A.; Martinez, T. J.;
Lochbrunner, S.; Schmitt, M.; Shaffer, J. P.; Zgierski, M. Z.; Stolow, A. J.
Am. Chem. Soc. 2003, 125, 8098. (g) Satzger, H.; Root, C.; Braun, M. J.
Phys. Chem. A 2004, 108, 6265. (h) Chang, C.-W.; Lu, Y.-C.; Wang, T.-
T.; Diau, E. W.-G. J. Am. Chem. Soc. 2004, 126, 10109. (i) Lu, Y.-C.;
Diau, E. W.-G.; Rau, H. J. Phys. Chem. A 2005, 109, 2090.
(18) (a) Cattaneo, P.; Persico, M. Phys. Chem. Chem. Phys. 1999, 1, 4739. (b)
Ishikawa, T.; Noro, T.; Shoda, T. J. Chem. Phys. 2001, 115, 7503. (c)
Gagliardi, L.; Orlandi, G.; Bernardi, F.; Cembran, A.; Garavelli, M. Theor.
Chem. Acc. 2004, 111, 363. (d) Diau, E. W.-G. J. Phys. Chem. A 2004,
108, 950. (e) Cembran, A.; Bernardi, F.; Garavelli, M.; Gagliardi, L.;
Orlandi, G. J. Am. Chem. Soc. 2004, 126, 3234. (f) Ciminelli, C.; Granucci,
G.; Persico, M. Chem. Eur. J. 2004, 10, 2327. (g) Tiago, M. L.; Ismail-
Beigi, S.; Louie, S. G. J. Chem. Phys. 2005, 122, 94311. (h) Ciminelli, C.;
Granucci, G.; Persico, M. J. Chem. Phys. 2005, 123, 174317.
(19) (a) Caglioti, L.; Rosini, G.; Rossi, F. J. Am. Chem. Soc. 1966, 88, 3865.
(b) Hassner, A.; Catsoulacos, P. J. Chem. Soc., Chem. Commun. 1967,
121. (c) Broda, S.; Simon, H. Chem. Ber. 1969, 102, 3647. (d) Buckingham,
J.; Guthrie, R. D. J. Chem. Soc. (C) 1968, 3079.
(20) Attanasi, O. A.; Filippone, P. Synlett 1997, 1128 and references cited therein.
(21) Bonini, B. F.; Maccagnani, G.; Mazzanti, G.; Rosini, G.; Foresti, E. J.
Chem. Soc., Perkin Trans. I 1981, 2322 and references therein.
(22) Hatchard, C. G.; Parker, C. A. Proc. R. Soc. London A 1956, 253, 518.
(23) Fischer, E. EPA Newsletter 1984, 20, 33.
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3199

A R T I C L E S
Conti et al.
0 was made). The yields were calculated by taking into account the
light absorbed by the sample at the irradiation wavelength. The
composition of the photostationary state in the E f Z photoisomer-
ization reactions was obtained by HPLC analysis; the absorption
spectrum of the photoproduct (Z isomer) was evaluated by subtraction
of the spectrum of the residual E isomer from the absorption spectrum
at the photostationary state. The Z f E photoisomerization experiments
were carried out with samples obtained by exhaustive irradiation of
solutions of E-CPD at 325 nm and containing at least 80% of the Z
isomer. The Z f E photoisomerization quantum yields were estimated3
from the composition of the photostationary state by using eq 1:
mizations and MEP computations at the CASSCF level26 (when
possible, single-state computations have been used; otherwise, when
root-flipping problems occurred, state averaged calculations have been
the choice), followed by single-point CASPT227 computations on the
optimized relevant structures to account for correlation energy.
For this purpose a state-averaged CASSCF procedure, with equal
weights between the involved states of the singlet manifold (the first
seven singlet states in E-CPD and the first 10 singlet states in Z-CPD),
has been used to generate molecular orbitals and the reference functions
for subsequent CASPT2 calculations. An “Imaginary Shift”28 value of
0.2 was employed for all state-averaged CASPT2 computations. This
option is used to eliminate intruder states problems. For the triplet state,
CASSCF single-state computations have been used in all cases, both
for the optimizations and CASPT2 calculations. The CASSCF state
interaction (CASSI)29 method has been used to calculate the transition
dipole moments. In the formula for the oscillator strength the CASSCF
transition moment and the energy difference obtained in the CASPT2
computation have been used. All computations have been performed
with the tools available in the Gaussian 9830 and the MOLCAS-531
quantum chemistry programs.
The active space choice is a crucial step in a CASSCF calculation.
Here the choice is straightforward, including all the orbitals and
electrons of the π system plus the two doubly occupied nitrogen lone
pairs, resulting in an active space of 12 orbitals and 14 electrons. A
6-31G* basis set of atomic orbitals was used. To improve vertical
excitation energies, computations in the FC region were also performed
by using the multistate-CASPT2 approach.32 Diffuse functions (using
a 6-31G+* basis set) or a generally contracted basis set of atomic
natural orbitals (ano-l with the contraction scheme 4s3p2d on C and
N, and 3s2p on H)31 were also used for this purpose.
ΦEfZ ꢀλZ [Z]PS
irr
)
(1)
ꢀEλ
irr
ΦZfE
[E]PS
At room temperature and in the time scale of the irradiation experiments,
the effect of the thermal Z f E isomerizationswhich is not taken into
account in eq 1scan be safely neglected. The values of ΦEfZ and the
molar absorption coefficients of the two isomers at the irradiation
wavelength, ꢀλE and ꢀλZ , were determined experimentally; d is the
optical path length of the cell (1 cm). In the present system, the
concentration of the E and Z species in solution can be calculated
from the absorbance values monitored at two different wavelengths,
λ1 and λ2:
irr
irr
Aλ1ꢀλZ2 - Aλ2ꢀλZ1
ꢀλE1ꢀZλ2 - ꢀλE2ꢀλZ1
[E] )
[Z] )
(2)
(3)
Aλ2ꢀλE1 - Aλ1ꢀλE2
ꢀλE1ꢀZλ2 - ꢀλE2ꢀλZ1
The search of conical intersections was performed by using
procedures developed in our laboratory or according to the prescriptions
of ref 33.
Four possible conformers exist for CPD. Two for the Z form: one
with the cyclohexene CdC double bond and the NdN double bond in
s-trans position (Z-MinS0), the other one with the two double bonds
in s-cis position (Z-MinS0′). Analogously, there are two E isomer
conformers: E-MinS0 and E-MinS0′. Through a preliminary CASPT2//
CASSCF calculation (see Table 5), we established that the most stable
conformer is the E-MinS0 one. All successive mechanism calculations
are relative to E-MinS0 for the E form isomerization and the
corresponding photoproduct Z-MinS0 for the Z form.
The concentrations of the two isomers at the photostationary state, [E]PS
and [Z]PS, were thus determined from eqs 2 and 3 for a solution at the
photostationary state. Experimental error on the E f Z quantum yield
values: (10% for ππ* irradiation, (20% for nπ* irradiation. For the
Z f E photoisomerization quantum yields the error was estimated to
be (30%.
Photochemical experiments with the use of triplet sensitizers were
carried out on solutions degassed by four freeze-pump-thaw cycles.
On the basis of the triplet energies obtained by calculations (see below),
biacetyl (ET ) 56.4 kcal mol-1 24 was employed as photosensitizer for
)
3. Results and Discussion
the T1 excited state of E-CPD. The concentration of E-CPD was 5.0 ×
3.1. Synthesis. (1-Cyclohexenyl)phenyldiazene (CPD). CPD
has been prepared according to a well-established procedure.
Phenylhydrazine (2.16 g, 20 mmol) was added at room
temperature to a stirred solution of 2-bromocyclohexanone (3.52
g, 20 mmol) in toluene (100 mL). After 15 min the orange
solution was washed first with 10% aqueous solution Na2CO3
and then with water, dried (Na2SO4), and concentrated under
reduced pressure. The residue was purified by flash chroma-
tography on a silica gel column eluting with toluene. A low-
melting red-orange solid was collected: mp 32-34 °C from
10-5 M, and that of the photosensitizer was 2 × 10-2 M.
The thermally activated Z f E isomerization was studied by
monitoring the absorption spectral changes in time for 5.0 × 10-5
M
solutions previously subjected to exhaustive irradiation at 325 nm and
containing at least 70% of the Z isomer. The temperature of the solution
was controlled in the range 20-50 °C by means of a thermostat, and
measured in the spectrophotometric cell by using a digital thermometer
with a probe. The concentration values of the E isomer at various times
were calculated by means of eq 2, and fitted according to a first-order
rate equation by means of the SPECFIT software.25 The Ea and A
activation parameters were calculated with the Arrhenius relation, k )
A exp(-Ea/RT). The values of ∆G*, ∆H*, and ∆S* were calculated
with the Eyring equation, k ) kBT/h[exp(-∆H*/RT)][exp(∆S*/R)], and
from ∆G* ) ∆H* - T∆S*. The error on the activation parameters
was estimated to be (5%.
(26) Roos, B.O. In AdVances in Chemistry and Physics: Ab Initio Methods in
Quantum Chemistry - II; Lawley, K. P., Ed.; John Wiley & Sons Ltd.:
Chichester, England, 1987; Vol. LXIX, p 399.
(27) Malmqvist, P.-Å.; Roos, B. O. Chem. Phys. Lett. 1989, 155, 189.
(28) Forsberg, N.; Malmqvist, P.-Å. Chem. Phys. Lett. 1997, 274, 196.
(29) Malmqvist, P.-Å.; Roos, B. O. Chem. Phys. Lett. 1989, 155, 189.
(30) Frisch, M. J.; et al. Gaussian 98, revision A.6; Gaussian, Inc.: Pittsburgh,
PA, 1998.
(31) Andersson, K.; et al. MOLCAS, Version 5.2; Lund University: Lund, 2001.
(32) Finley, J.; Malmqvist, P.-Å.; Ross, B. O.; Searrano-Andres, L. Chem. Phys.
Lett. 1998, 288, 299.
(33) (a) Ragazos, I. N.; Robb, M. A.; Bernardi, F.; Olivucci, M. Chem. Phys.
Lett. 1992, 197, 217. (b) Bearpark, M. J.; Robb, M. A.; Schlegel, H. B.
Chem. Phys. Lett. 1994, 223, 269.
2.2. Computational Section. The PES for the E / Z thermal- and
photoisomerization have been studied using fully unconstrained opti-
(24) Montalti, M., Credi, A., Prodi, L., Gandolfi, M. T., Eds. Handbook of
Photochemistry, 3rd ed.; CRC Press: Boca Raton, FL, 2006.
(25) Binstead, R. A. SPECFIT; Spectrum Software Associates: Chapel Hill,
NC, 1996.
9
3200 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Table 1. Absorption Data (Acetonitrile Solution, Room
Temperature)
S
2 r S0
(ππ*)
S
1 r S0 (n
π
*)
λmax
(nm)
ꢀ
λmax
(nm)
ꢀ
1
1
compound
isomer
(M-1 cm-
)
(M-1 cm-
)
CPD
E
Z
E
Z
304
274
316
279
22000
5100
21000
4300
430
429
445
428
350
1200
550
AB
1200
methanol;19c bp 110-120 °C (0.05 mmHg). Spectroscopic and
analytical data agree with those previously reported.34
3.2. Absorption Spectra: The Singlet Manifold.
Experiments. The spectroscopic data for CPD and AB,
obtained at room temperature in acetonitrile solution, are
gathered in Table 1. The absorption spectrum of the E isomer
of CPD is shown in Figure 1 together with the spectrum of
E-AB. The absorption spectrum of E-CPD is qualitatively
similar to that of E-AB, and exhibits two bands in the 250-
520 nm region. In analogy with azobenzene, the low-energy
band of E-CPD is attributed to S1 r S0 (nπ*) transitions, and
the higher-energy band is assigned to S2r S0 (ππ*) transitions.
Both absorption bands, however, are noticeably shifted to higher
energy on going from E-AB to E-CPD (∼800 and ∼1200 cm-1
for the nπ* and ππ* bands, respectively). This observation is
not unexpected, considering the reduced size of the π-electron
system in CPD with respect to AB. The molar absorption
coefficient of the S2r S0 band is nearly the same as that of
azobenzene, while in the case of the S0r S1 band the molar
absorption coefficient of E-CPD is somewhat smaller than that
of E-azobenzene.
Figure 1. Absorption spectra in acetonitrile solution at room temperature
of the E isomers of CPD (full line) and AB (dashed line).
Figure 2. Absorption spectra in acetonitrile solution at room temperature
of the Z isomers of CPD (full line) and AB (dashed line).
by 6.6 and 2.8 kcal/mol respectively for E-CPD and Z-CPD
(as compared with the data in Table 1). Computed oscillator
strengths (∼10-5 for E-CPD and ∼10-2 for Z-CPD) are
consistent with a forbidden nπ* excitation in E-CPD (ꢀ ) 350
M-1 cm-1), but a more intense (i.e., slightly allowed) transition
in Z-CPD (ꢀ ) 1200 M-1 cm-1). While these data are acceptable
and in fairly good agreement with the experiments, more serious
problems arise in reproducing excitation energies for the bright
S2r S0 (ππ*) transitions (i.e., the higher-energy band of the
absorption spectra). Our computations overestimate these ener-
gies by 19.2 kcal/molsassigning this state as S3 (S5 at the
CASSCF level)sand 16.3 kcal/molsassigning this state as S5
(S9 at the CASSCF level)sfor E-CPD and Z-CPD, respectively,
while the computed oscillator strengths (0.3 for E-CPD and 0.1
for Z-CPD) are roughly in line with the measured molar
absorption coefficients of a spectroscopically allowed transition
(22000 M-1 cm-1 for E-CPD and 5100 M-1 cm-1 for Z-CPD).
These errors can be only slightly reduced by adding diffuse
functions to the basis set (6-31G+*), while nπ* transitions
remain substantially unchanged (see Supporting Information).
Interestingly, if we correct these energies to match experimental
values, we can see that the bright ππ* state become S2 in both
cases, with nearby (i.e., nearly degenerate) πbenzπ* states
involving benzenic excitations, as already known for AB. We
confidently think this should be the correct energy order in the
singlet manifold.
The absorption spectrum of the Z isomer of CPD is almost
identical to that of Z-AB (Table 1 and Figure 2) and exhibits a
two-band pattern with well separated nπ* and ππ* absorption
features. The only noticeable difference between the two spectra
is a modest (∼700 cm-1) shift of the S2r S0 band of Z-CPD to
higher energy, compared to the same band in Z-AB, and a
slightly larger intensity. The shift can again be interpreted in
terms of a larger extension of the π-electron system in
azobenzene.
The spectral features of both isomers of CPD and AB do not
change appreciably on going from acetonitrile to a less polar
solvent like cyclohexane. E- and Z-CPD do not exhibit
fluorescence either in fluid solution at room temperature or in
a frozen glass at 77 K, in line with the behavior of azoben-
zene.1,2,35
Calculations. CASPT2//CASSCF/6-31G* vertical transitions
and oscillator strengths (f) have been computed in vacuo for
the first low-energy states of the singlet manifold (the first 7
singlet states in E-CPD and the first 10 singlet states in Z-CPD),
as reported in Tables 2 (E-MinS0) and 3 (Z-MinS0). It turns
out that the computed S1r S0 (nπ*) transitions slightly
underestimate the experimentally detected excitation energies
in solution (i.e., the low-energy band of the absorption spectra)
(34) Caglioti, L.; Grasselli, P.; Rosini, G. Tetrahedron Lett. 1965, 4545.
(35) A very weak fluorescence band (Φfl ≈ 10-6), displaying an ultrafast
(subpicosecond) decay, has been recently observed for AB at room
temperature in DMSO (see: Satzger, H.; Sporlein, S.; Root, C.; Wachtveitl,
J.; Zinth, W.; Gilch, P. Chem. Phys. Lett. 2003, 372, 216) and hexane
solution (see: Fujino, T.; Arzhantsev, S. Y.; Tahara, T. J. Phys. Chem. A
2001, 105, 8123). Such an emission band, if present for CPD, has an
intensity below the detection limit of our instrumentation (Φfl < 10-5).
While the computed energies for the bright ππ* excitation
are far from being satisfactory, we will see below that these
errors do not affect at all our Discussion and Conclusions on
the reactivity and photochemical/photophysical properties of
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3201

A R T I C L E S
Conti et al.
Table 2. CASPT2 and CASSCF Relative (∆E) Energies and Sn r S0 Oscillator Strengths (f) for All the Relevant Points Discussed in the
Text along the Photochemical Paths of the E Isomera
CASPT2
CASSCF
CASPT2
CASSCF
∆
E
∆
E
∆
E
∆E
1
1
structure
state
ω
(kcal mol-1) [nm]
(kcal mol-1) [nm]
f
structure
state
ω
(kcal mol-
)
(kcal mol-
)
E-MinS0
S0
S1
S2
S3
S4
S5
S6
.64
.63
.63
.62
.63
.63
.62
0.0
0.0
E-MinS1
S0
S1
S2
S0
S1
S2
S0
S1
S0
S1
S0
S1
S2
S1
S2
S3
S2
S3
S0
S1
S0
S1
S0
S1
.64
.63
.63
.64
.63
.63
.63
.64
.63
.63
.63
.63
.63
.63
.63
.63
.63
.63
.63
.64
.63
.64
.63
.63
18.4
46.4
95.5
28.5
49.2
81.0
45.7
53.5
31.6
36.1
39.3
50.5
60.8
68.0
91.9
98.6
89.5
105.4
63.9
76.5
58.3
64.6
66.8
67.3
17.4
58.7
113.3
31.9
63.0
93.6
55.5
61.1
52.4
58.1
47.3
67.4
78.6
88.8
98.6
109.6
98.6
113.5
75.6
76.7
68.8
73.7
78.7
76.5
59.9 [477]
99.0 [289]
112.9 [253]
115.0 [249]
119.1 [240]
126.8 [225]
74.6 [383]
108.1 [264]
150.0 [190]
133.7 [214]
133.3 [214]
150.4 [190]
∼0
0.004
0.325
0.005
0.008
0.0185
E-TSS1
b
b
E-CItors1
E-CItors2
MIN-TS2
E-TSS2
E-MinS2
b
E-CId_inv
b
E-CIinv1
E-CIinv2
a Singlet states (Sn) are ordered according to CASPT2 results. Energy values are state averaged through the first 7 singlet states. The energy for the ground
state minimum is chosen as the zero. Basis set: 6-31G*. Active space: 14 electrons in 12 MOs. The weight (ω) of the zeroth order CASSCF wave function
has also been reported. b The nature of singlet sates is already exchanged: S0 is an nπ* state, and S1 has the ground-state configuration.
Table 3. CASPT2 and CASSCF Relative (∆E) Energies and Sn r S0 Oscillator Strengths (f) for All the Relevant Points Discussed in the
Text along the Photochemical Paths of the Z Isomera
CASPT2
CASSCF
CASPT2
CASSCF
∆
E
∆
E
∆
E
∆E
1
1
structure
state
ω
(kcal mol-1) [nm]
(kcal mol-1) [nm]
f
structure
state
ω
(kcal mol-
)
(kcal mol-
)
b
b
Z-MinS0
S0
S1
S2
S3
S4
S5
S6
S7
S8
S9
.64
.63
.62
.63
.62
.60
.62
.62
.62
.62
0.0
0.0
Z-CItors1
S0
S1
S0
S1
S1
S2
S2
S3
S4
S0
S1
S0
S1
S0
S1
S0
S1
.63
.64
.63
.64
.63
.63
.63
.63
.62
.64
.63
.63
.63
.63
.63
.64
.63
30.2
37.8
29.2
33.9
59.3
63.3
77.3
87.8
100.1
59.0
67.4
56.6
57.9
60.2
64.0
2.5
36.4
40.7
34.2
38.7
79.7
77.5
89.8
99.8
117.8
62.7
71.1
62.2
67.6
67.6
70.1
2.7
63.8 [448]
104.5 [274]
105.9 [270]
117.0 [244]
120.6 [237]
128.4 [223]
133.9 [214]
139.2 [205]
141.6 [202]
77.5 [369]
124.0 [231]
112.7 [254]
139.0 [206]
178.0 [161]
152.8 [187]
162.2 [176]
161.1 [177]
166.7 [172]
0.008
0.005
0.0004
0.003
0.108
0.016
0.002
0.007
0.001
Z-CItors2
Z-CIS2/S1
INTs2
Z-CIinv1
b
Z-CIinv2
Z-CId_inv
INTs1
58.7
72.6
a Singlet states (Sn) are ordered according to CASPT2 results. Energy values are state averaged through the first 10 singlet states. The energy for the
ground-state minimum is chosen as the zero. Basis set: 6-31G*. Active space: 14 electrons in 12 MOs. The weight (ω) of the zeroth order CASSCF wave
function has also been reported. b The nature of singlet sates is already exchanged: S0 is an nπ* state, and S1 has the ground-state configuration.
CPD. Anyway, in the attempt to improve these energies and
correct such a surprising mismatch (which should not generally
occur at such a correlated high level of theory), multistate-
CASPT2 (MS-CASPT2) computations have also been per-
formed in conjunction with a larger basis set (ano-l). Notably,
the results lead to significantly improved excitation energies
(97.9 kcal/mol and 112.9 kcal/mol for the bright ππ* excitation
of the E and Z isomers, respectively) in much better agreement
with the experimental values (94.0 and 104.3 kcal/mol, see also
Supporting Information for details). Anyway, this computational
level has not been further used due to its computational cost.
3.3. E / Z Photochemical and Thermal Isomerization.
Direct Irradiation Experiments. The quantum yields for
E f Z and Z f E photoisomerization at selected irradiation
wavelengths in acetonitrile solution are reported in Table 4. As
a control experiment the photoisomerization of azobenzene was
also studied under our experimental conditions. The quantum
yield values determined for AB are in agreement with those
found in earlier accurate studies.3
The absorption changes observed upon irradiation of E-CPD
in its S2 r S0 (ππ*) absorption band at 325 nm are shown in
Figure 3. Such changes are fully consistent with the E f Z
9
3202 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Table 4. Data for the E f Z and Z f E Photoisomerization Processes at Room Temperature and the Z f Z Isomerization Reaction
(acetonitrile Solution)
E−Z photoisomerization
ΦE
ΦE
fE
Z
f
E thermal isomerization
f
Z
a
b
c
b
1
1
1
1
1
compd
ππ
*
n
π
*
ππ
0.3c
0.35d
*
n
π
*
Ea (kcal mol-
)
A (s-
)
∆
H* (kcal mol-
)
∆
S* (cal K-1 mol-
)
∆
G*298 (kcal mol-
)
CPD
AB
0.17
0.14
0.30
0.31
0.7c
22.5
8.2 × 1011
21.9
-6.1
-8.6f
22.3
0.55d
23.9e
2.43 × 1011 c
23.2f
25.8f
b
a λirr ) 325 nm for CPD and 345 nm for AB. λirr ) 430 nm. c Values calculated by eq 1. d From ref 36. e From ref 39a. f Calculated from the data
reported in ref 39a: ∆H* ) Ea - RT; ∆S* ) [(ln A) - 1.00 - ln(kBT/h)]; ∆G*298 ) ∆H* - 298∆S*.
isomerization of the -NdN- unit. Neat isosbestic point are
observed at 261 and 365 nm, indicating the occurrence of a
clean transformation. A photostationary state that contains
∼85% of the Z-CPD isomer is reached within 30 min of
irradiation in our conditions. Irradiation in the S1r S0 (nπ*)
absorption band at 430 nm causes the same spectral changes
observed for 325-nm excitation; the photostationary state,
however, is much less rich in the Z form (∼10%), mainly
because of the unfavorable ratio of the molar absorption
coefficients of the E and Z isomers (see Experimental Section).
Similarly to azobenzene, either nπ* or ππ* excitation of the
Z isomer of CPD leads to its conversion back to the E isomer,
as confirmed by absorption spectral changes opposite to those
observed on irradiation of the E form. In all cases the
photoisomerization quantum yield values are remarkably higher
for nπ* irradiation compared to those obtained on ππ*
irradiation (Table 4). This is again in line with the behavior of
azobenzene (see Table 2).1-3,36 Although CPD does substantially
behave as AB, it is noteworthy that the former gives a much
higher (nπ*) ΦZfE quantum yield.
us for the S1 (nπ*) state of AB,18e clearly account for the
existence in both isomers of efficient S1 f S0 radiationless
deactivations, which are triggered by easily accessible low-
energy S1/S0 real crossing points (i.e., conical intersections) with
a fully twisted (∼90°) central -N-N- bond (E-CItors2 and
Z-CItors1), prompting the isomerization of the system upon decay
to the ground state. Interestingly, as already pointed out in AB,18e
a barrierless path exists for the Z isomer, which directly connects
the FC region on S1 to a twisted low-energy S1/S0 conical
intersection (CI) point, while an extended energy plateau (with
a very shallow planar minimum E-MinS1 and a tiny energy
barrier at ∼50° of CNNC torsion, E-TSS1) exists along the S1
path of the E isomer.
Interpretation of Photoisomerization Quantum Yields.
These results clearly indicate that for both the E and Z isomers
the S1 state internal conversion involves mainly the torsion route,
in agreement with the experiments and photoisomerization
quantum yields measurements (Table 4). To determine quan-
titatively the E / Z isomerization quantum yields, a more
complete characterization of the S1 potential energy surface
(PES) around these CIs and molecular dynamics simulations
would be required. Nevertheless, at a qualitative level, the
barrierless versus the barrier-controlled twisting path accounts
for the observed higher yield (∼70%) of the Z f E, with respect
to the E f Z (30%), photoisomerization process. Moreover,
the lowest-energy S1/S0 real crossing points found from both
the E and Z form (which are basically the same structure), are
placed slightly on the E side (i.e., the -N-N- twisting angle
is around 93°), which accounts for a ground-state relaxation
process leading preferentially to the E isomer formation. Note
also that the sum of the ΦZfE (nπ*) and ΦEfZ (nπ*) quantum
yields equals one, which suggests a common deactivation point
for both the E and Z forms as computations reveal.
The photoreactivity of both isomers of CPD and AB do not
change appreciably on going from acetonitrile to a less polar
solvent such as cyclohexane.
Computational Analysis of the Photochemistry upon nπ*
Excitation. The results of CASPT2//CASSCF/6-31G* minimum
energy path (MEP) mapping from the Franck-Condon (FC)
region (i.e., the region initially populated by vertical excitation)
of the S1 (nπ*) state of the E and Z isomers of CPD are
summarized in Figures 4 and 5 (blue arrows) respectively, while
Tables 2 and 3 collect the relative energies of all the relevant
points discussed (for absolute energy values see Tables S3 and
S4 in the Supporting Information). The reaction paths, which
appear to be analogous to the ones previously documented by
Computational Analysis of the Photochemistry upon ππ*
Excitation. The results of CASPT2//CASSCF/6-31G* MEP
mapping from the FC point of the first bright ππ* singlet state
of the E and Z isomers of CPD are summarized in Figures 4
and 5 (red arrows) respectively, while Tables 2 and 3 collect
the relative energies of all the relevant points discussed (for
absolute energy values see Tables S3 and S4 in the Supporting
Information). As mentioned above, this state is overestimated
in energy for both isomers. Anyway, as already seen in AB,
both isomers display an unbound state (characterized by a double
excitation), which is higher in energy in the FC point but
suddenly crosses the ππ* state along the MEP, getting lower
in energy. Therefore, an ultrafast depopulation/deactivation of
the bright state is expected in favor of this double-excitation
Figure 3. Absorption changes upon 325-nm irradiation of a 7.5 × 10-5
M
acetonitrile solution of E-CPD at room temperature. Irradiation times are
0, 1, 2, 4, 6, 8, 10, 13, 16, 20, 25, 30, and 35 min. (Inset) Detail of the
changes in the S1 r S0 (nπ*) absorption band.
(36) Siampiringue, N.; Guyot, G.; Monti, S.; Bortolus, P. J. Photochem. 1987,
37, 185.
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3203

A R T I C L E S
Conti et al.
Figure 4. E-CPD computed reaction paths upon nπ* and ππ* excitations. Blue arrows show the nπ* state (solid diamonds) MEP from the Franck-Condon
region to the low-energy S1/S0 crossing point (-NdN- torsions are mainly involved); red arrows show the MEP from the bright ππ* state (solid asterisks).
The higher double-excitation state (solid squares) suddenly crosses the ππ* state and drives the relaxation of the excited molecule. The dotted red points
zone represents the extended S2/S1 crossing seam. The red dotted line is the bright ππ* state scaled to match the experimental absorption value. All the
energy profiles have been scaled to match CASPT2 values (Table 2).
(nπ*)2 state, which thereafter drives the relaxation of the excited
molecules.37
a significant S2/S1 energy gap exists for E at this point,
suggesting a different (and hopefully more efficient) radia-
tionless decay route to S1. Remarkably, an extended S2/S1
crossing region exists which is almost degenerate and parallel
to the computed MEP on the S2 state (see Figure 4 and Table
S3 in the Supporting Information for the energies of the
computed S2/S1 crossing points, from 1-CIS2/S1 to 8-CIS2/S1).
We have proved that this crossing seam can be easily accessed
by moving orthogonally to the S2 MEP along symmetric
CNN bending modes (CNN bendings must get smaller to reach
this crossing region; see the geometrical parameters reported
in Figure 4) and that this motion is substantially barrier-
less. Furthermore, the extended and continuous S2/S1 crossing
region, which runs parallel and isoenergetic to almost the whole
S2 MEP, grants a high probability for the excited molecules to
enter this space by normal thermal vibrations (see Scheme 2).
Hence, decay to the S1(nπ*) state at -N-N- twisting angles
that are far from being fully twisted, must be very likely and
efficient.
Interestingly, initial relaxation from the ππ* FC point of the
Z isomer does suddenly involve the -N-N- bond torsion
leading (through a barrierless and steep path) to a partially
twisted (65°) S2(nπ*)2/S1(nπ*) surface crossing (Z-CIS2/S1),
which easily accounts for a rapid S2 f S1 nonradiative decay.
In contrast, -N-N- bond torsion is not initially involved in
the ππ* relaxation of the E isomer, which is characterized by
a more general skeletal rearrangement leading to an increase
in N-N bond length and a decrease in C-N bonds. Only
after this skeletal rearrangement has occurred, torsion around
the N-N bond starts, still spanning a flat and extended
energy region before reaching the fully twisted (96°) minimum
(MIN-TS2) on the S2 (nπ*)2 state (see Figure 4 and Table S3
in the Supporting Information for the energies of the optimized
points along the S2 MEP, from 1-HMS2 to 4-HMS2). Notably,
(37) This mechanism should remain substantially unchanged even if we place
the bright state at the experimentally correct energy level (see red dotted
lines in Figures 4 and 5 for experimentally scaled energies); also in this
case the double-excitation state, which gets lower in energy, is easily
populated directly by crossing the ππ* state or by mediation of an excited
state of benzenic character (πbenzπ*), which is almost degenerate to the
spectroscopic state. In conclusion, whatever be the specific case considered,
excited-state relaxation from the bright ππ* state should end up in its rapid
deactivation to an unbound double-excited (nπ*)2 state.
Exploration and Analysis of the S1/S0 Crossing Region.
Once the S1(nπ*) state has been populated, this will drive
deactivation to the ground state according to the accessibility
of the S1/S0 crossing regions which trigger S1(nπ*) f S0
9
3204 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Figure 5. Z-CPD reaction paths upon nπ* and ππ* excitation. Blue arrows show the nπ* state (solid diamonds) MEP from the Franck-Condon region to
the low-energy S1/S0 crossing point (-NdN- torsions are mainly involved); red arrows show the MEP from the bright ππ* state (solid asterisks). The
higher double excitation state (solid squares) suddenly crosses the ππ* state and drives the relaxation of the excited molecule. The red dotted line is the
bright ππ* state scaled to match the experimental absorption value. All energy profiles have been scaled to match CASPT2 values (Table 3).
Scheme 2. Cartoonlike Drawing Showing the Extended S2/S1 Crossing Seam Running Almost Parallel to the Computed S2 MEPa
a The crossing seam can be accessed by vibrations (e.g., symmetric CNN bending modes) orthogonal to the MEP.
radiationless decay. For this reason, to access the feasibility of
multiple deactivation channels, an extended investigation and
analysis of the S1/S0 crossing seam has been performed, from
low-energy to higher-energy conical intersection points. In
particular, these former points may become accessible in a
condition of excess vibrational energy, as it is when the molecule
is decaying from the higher-energy ππ* state.
substantially identical in the two isomers, and spans very
different structures, going from the lowest-energy geometries
with a fully twisted -N-N- bond (see E-CItors1, E-CItors2
,
Z-CItors1, and Z-CItors2), to higher-energy inVerted (i.e., one of
the CNN bending angles is linear, while the other is much
smaller, thus calling for an inVersion mechanism in the
photoisomerization, see E-CIinv1, E-CIinv2, Z-CIinv1, Z-CIinv2
)
A complex S1/S0 crossing subspace exists, with energies that
all lie below the extended S2/S1 crossing region from which
decay to S1 occurs. The topology of this crossing space is
or symmetric (i.e., identical CNN bendings, see E-CId_inv
Z-CId_inv) structures. More precisely, the S1/S0 crossing seam,
which is schematically depicted in Scheme 3, has a symmetric
,
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3205

A R T I C L E S
Conti et al.
Scheme 3. Cartoonlike Drawing Representing the Topology of the
S1/S0 Crossing Region
involve -N-N- torsions at all, and under these conditions no
excess vibrational energy is expected to populate this mode in
the short lifetime of S2.
Another interesting point concerns the difference of photo-
isomerization quantum yields observed in the ππ* and nπ*
states (i.e., the Kasha rule is violated), with the former invariably
larger than the latter, as in AB. Since we have just shown that
all the photoexcited molecules are ultimately collected in the
S1 nπ* state (which then drives the deactivation onto the ground
state), such an observation can appear somewhat surprising. This
suggests that we are dealing with a dynamical problem, as
recently shown by Persico and co-workers for AB.18f Therefore,
to correctly account for this phenomenon, molecular dynamical
simulations would be required which is beyond the scope of
this work. Nevertheless, a qualitative (but reasonable) explana-
tion may be drawn if we focus into the complex S1/S0 crossing
region described above. As we have seen, S2 f S1 decay occurs
at energies that are invariably higher than the S1/S0 crossing
region, with this subspace extending widely below the S2/S1
intersection space. It is likely that, following S2 f S1 decay,
molecules on the S1 state may intercept the S1/S0 crossing region
well above the lowest-energy -N-N- fully twisted CIs: for
instance, at crossing points with CNNC angles which are only
partially twisted, i.e., on the reactant side (as shown in Scheme
4). Actually, we may predict that such a situation is quite
common and favorable, since this region extends for a wide
portion below the S2 state. This should trigger the population
of ground-state relaxation channels leading preferentially to
reactant back formation, which accounts for the smaller quantum
yield observed from the ππ* state.
E / Z Photoisomerization by Triplet Sensitization. In
order to explore the reactivity of the T1 (nπ*) triplet excited
state, the E / Z photoreaction of CPD has been studied by
using a triplet photosensitizer. On the basis of the triplet energies
obtained by calculations (see Table 5), we have selected biacetyl
(ET ) 56.4 kcal mol-1) as a photosensitizer for the T1 excited
states of E-CPD. The complete quenching of the biacetyl
phosphorescence in the condition employed for the experiments
(carefully deaerated acetonitrile solutions; for more details see
the Experimental Section) is an indication that quantitative
energy transfer to the relevant triplet excited state of CPD takes
place. Irradiation has been carried out at 420 nm, where a
substantial fraction of light is absorbed by the photosenzitizer;
however, direct nπ* irradiation of CPD cannot be avoided. For
this reason, the photoreaction has also been studied on the same
solutions equilibrated to air, under otherwise identical experi-
mental conditions. Since the concentration of oxygen present
in the air-equilibrated solvent is largely sufficient to afford
complete quenching of the photosensitizer triplet excited states,
thereby switching off the photosensitized route, these experi-
ments have been taken as a control for the direct E-Z
photoisomerization of CPD.
shape with a ridge of symmetric (i.e., identical CNN bendings)
conical intersections separating a two-fold channel (one col-
lecting conical intersections with a CNN bending angle that is
larger than the other, such as Z,E-CItors1 and Z,E-CIinv1, and
the other collecting the corresponding pseudo-enantiomeric
structures, i.e., with CNN bending angles of inverted order
such as Z,E-CItors2 and Z,E-CIinv2).38 Remarkably, this
crossing subspace is wholly connected, i.e., one can freely move
from one conical intersection to the other without never exiting
the intersection space (see Supporting Information for details).
In general, identical bending angles (i.e., symmetric CIs) lead
to the highest energies (i.e., the ridge), while a decrease in
energy is observed when the symmetry is broken (i.e., different
bending angles) and the CNNC angle gets more and more
twisted.
In conclusion, the lowest-energy points in the S1/S0 crossing
seam are the asymmetric CIs described in the previous sec-
tion for nπ* photochemistry, which possess a fully twisted
-N-N- bond. These points are the only ones accessible for
cold systems, i.e., in conditions of a small excess of vibrational
energy, while very different S1/S0 CIs can be potentially entered
on decaying from higher-energy regions, such as from the ππ*
state.
Interpretation of Photoisomerization Quantum Yields. At
a qualitative level, the barrierless and steep S2 relaxation path
found for the Z isomer (which is dominated by the -N-N-
torsional motion), versus the comparatively flat twisting path
found for the E isomer, accounts for the observed higher yield
(30%) of the Z f E, with respect to the E f Z (17%),
photoisomerization process. In fact, while along the first path
decay to S1 can occur around a CNNC twisting of 65° (i.e., the
S2/S1 CI, see Figure 5) with an impulsive motion about this
torsional mode (therefore providing the driving force triggering
Z f E photoisomerization), S2 f S1 decay for the E isomer
can occur at much less twisted structures (see discussion above).
Moreover, initial relaxation from the ππ* FC region does not
Upon irradiation of degassed solutions containing E-CPD and
the photosensitizer, very little spectral changes are observed,
whereas irradiation of the same solutions in the presence of
oxygen leads to spectral changes consistent with the E f Z
isomerization, as expected for nπ* irradiation of E-CPD (see
Supporting Information). Taken together, these observations
indicate that photosensitization of the first triplet excited state
of E-CPD does not lead to efficient E f Z isomerization; rather,
(38) It is worth noting that the presence of different rings, namely, phenyl (φ)
or cyclohexene (c), does not seem to affect very much the CI structures,
as CIs with bigger CNN angle/smaller NNCf angle (e.g., Z,E-CItors1 and
Z,E-CIinv1) and smcaller CNN angle/bigger NNCf angle (e.g. Z,E-CItors2
and Z,E-CIinv2) exist (asc well as with identical CNN and NNCf angles).
c
9
3206 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Scheme 4. Predicted Photoisomerization Quantum Yields for Z-CPD (a) and E-CPD (b). The Rationale Is Based on Where the
Radiationless Decay Occurs along the Photochemical Reactions Channels, i.e., the Geometry of the Crossing Points (with CNNC Angles
Partially or Fully Twisted) Triggering Internal Conversiona
a Bigger arrows show the most favorite paths. The red color identifies ππ* decay paths, while the blue color identifies nπ* decay paths.
population of T1 causes the Z f E isomerization very efficiently,
thereby preventing accumulation of the Z isomer which is indeed
formed with a substantial quantum yield (Table 4) upon direct
irradiation of CPD. As it will be discussed below, this
observation is consistent with the computational results.
Thermally Activated Z f E Isomerization. When a solution
of E-CPD subjected to exhaustive irradiation at 325 nm and
containing >80% of the Z form is left in the dark, absorption
spectral changes are observed which indicate the occurrence of
a back Z f E isomerization reaction. This behavior is similar
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3207

A R T I C L E S
Conti et al.
Table 5. CASPT2 and CASSCF Single-State Relative (∆E)
Energies for All the Relevant Points Discussed in the Text for the
PES of S0 and T1 in CPDa
CASPT2
E (kcal mol-
CASSCF
1
1
structure
state
ω
∆
)
∆
E (kcal mol-
)
E-MinS0
S0
T1
T2
S0
S0
T1
S0
S0
T1
S0
T1
S0
T1
S0
T1
.64
.63
.61
.64
.64
.63
.64
.63
.63
.63
.63
.63
.63
.63
.63
0.0
48.4
65.0
7.0
0.0
62.9
71.2
3.0
E-MinS0′
Z-MinS0
17.8
60.5
17.4
33.9
30.8
36.5
32.6
34.8
33.5
29.8
31.3
17.0
77.5
16.5
40.8
39.7
44.0
41.2
43.1
43.5
35.6
40.0
Z-MinS0′
MinT1
TSS0
CR1
CR2
Figure 6. Absorption changes monitored at 304 nm as a function of time
for an acetonitrile solution containing a mixture of E- and Z-CPD left in
the dark at 20.0 °C. The starting (t ) 0) solution was obtained by exhaustive
irradiation of a 5.0 × 10-5 M solution of E-CPD at 325 nm. The full line
represents the best fit to a first-order kinetic scheme. (Inset) Arrhenius plot
shows the temperature dependence of the Z f E isomerization rate constant.
a The energy for the ground state of the E form is chosen as the zero.
Basis set 6-31G*. Active space: 14 electrons in 12 MOs. The weight (ω)
of the zeroth order CASSCF wave function has also been reported.
to that of azobenzene1,2,39 and is consistent with the fact that
the E isomer is the thermodynamically stable form of CPD.
The reaction can be followed spectrophotometrically by fol-
lowing the recovery time of the S2 r S0 band of the E isomer
(λmax ) 304 nm). Figure 6 shows the absorption changes at
304 nm as a function of time for an acetonitrile solution of the
Z and E forms (total CPD concentration, 5.0 × 10-5 M) at 20.0
°C; the inset of Figure 6 shows the temperature dependence of
the rate constant (Arrhenius plot). The activation parameters
obtained from the Arrhenius and Eyring equations (see Experi-
mental Section) are gathered in Table 4. The recovery of the E
isomer of CPD in the dark is somewhat faster compared to that
of AB. It is worthwhile to note that the activation energy for
the Z f E process decreases slightly on going from AB to CPD.
The difference in the values of the preexponential factor and
∆S* also suggest a different entropic contribution to the Z f E
isomerization in the two cases.
Computational Analysis of the Isomerization in the
Ground and Triplet T1 State: S0-T1 Intersystem Crossing.
According to CASPT2 calculations, the S0 potential energy
curve along the CNNC coordinate shows a barrier of 36.5 kcal/
mol (TSS0) while the T1 potential energy curve presents a
minimum (MinT1) of 30.8 kcal/mol in the same twisted region
(CNNC twisting of 104°); i.e. it is below the S0 state. A
qualitative picture of the S0 and T1 potential energy curves is
shown in Figure 7 (Table 5 collects the energies of all the
relevant points discussed). The two curves cross each other at
about 75°(CR1) and 115° (CR2) of CNNC twisting. The energy
of the former crossing, which is closer to the Z form, is the
(39) (a) Halpern, J.; Brady, G. W.; Winkler, C. A. Can. J. Res., Sect. B 1950,
28, 140. (b) Brown, E. V.; Granneman, G. R. J. Am. Chem. Soc. 1975, 97,
621.
Figure 7. S0 and T1 potential energy profiles along the -NdN- isomerization path. Black arrows show the Z f E thermal isomerization by the torsion
mechanism via S0-T1-S0 non-adiabatic path involving the triplet state (solid triangles). Energy profiles have been scaled to match CASPT2 values (see
Table 5).
9
3208 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007

CPD: Surrogate of the Azobenzene Photochromic Unit
A R T I C L E S
Scheme 5. Cartoonlike Drawing Summarizing E-CPD Photochemical Reaction Channels Triggered by nπ* (Blue Arrows) and ππ* (Red
Arrows) Excitationsa
a The semi-transparent surface represents the S1/S0 crossing region: it spans very different structures along the -NdN- torsional and the asymmetric
bending coordinates. Empty arrows show the less probable paths.
higher and is 34.8 kcal/mol above the E isomer, while the latter
is 29.8 kcal/mol above the E isomer. Thus, CPD can easily cross
from S0 to T1, or vice versa, in the twisted region, provided a
significant S0-T1 spin-orbit coupling exists.
forbidden processes (indeed, this would suggest a much smaller
pre-exponential factor than the recorded one of about 1011 s-1),
we have already shown in AB18e that the computed S0-T1 spin-
orbit coupling values lead to an S0 f T1 intersystem crossing
rate that is of the same size of the pre-exponential factors
reported for the thermal Z f E isomerization of AB and CPD
(see Table 4). Therefore, the present calculation indicates that
thermal isomerization occurs by the non-adiabatic torsion
mechanism and predicts kinetic parameters in agreement with
the experimental results.
Interpretation of Photoisomerization Quantum Yields in
the Triplet State. According to the experimental results, the E
f Z photosensitized isomerization is inefficient (or does not
occur at all), whereas the Z f E photosensitized process appears
to be very efficient. These observations are accounted for by
the fact that the lower-energy S0/T1 crossing (where the T1 f
S0 intersystem crossing should occur) is on the E side (CNNC
torsion of ∼104°) and is almost degenerate with the T1 minimum
(see Figure 7). Therefore, T1 decay should always lead to the E
isomer on the ground state, in agreement with the observations.
These findings support the idea that the decay of T1 can occur
more efficiently via the torsion mechanism with a rate depending
on the strength of the relevant spin-orbit couplings. In our
previous work on AB18e (which substantially presents an
identical topology for the S0 and T1 states), we have shown
that these couplings are large enough to ensure that the rate of
the T1 f S0 intersystem crossing process is of the order of 1011
s-1. This rate indicates that AB can decay from T1 to the ground
state in a picosecond time scale and that the torsion mechanism
provides indeed a very efficient mechanism for the T1 decay
and isomerization. Because of the analogy of the two systems,
we suggest a similar situation for CPD with a very short T1
lifetime that should lead to a difficult spectroscopic detection
of this state.
These findings support also the idea that the thermal
isomerization of CPD can proceed by the torsion mechanism
via the S0- T1- S0 non-adiabatic path involving the triplet state.
Remarkably, also the calculated Z f E activation energy along
this triplet-mediated path is consistent with the measured kinetic
parameter. In fact, our calculated barrier (from the Z isomer to
the higher crossing point) is 17 kcal/mol, which is in fairly good
agreement with the activation enthalpy of 21.9 kcal/mol
measured for the thermal isomerization. Although it may appear
quite surprising that thermal isomerization does involve spin-
4. Conclusions
The main purpose of this work has been to analyze the
isomerization routes of CPD in the ground and excited (nπ*
and ππ*) singlet states and in the triplet state by means of
computational and experimental tools. By using high-level
quantum chemical methods, we have searched for the photo-
chemical reaction channels leading from the E and Z excited
9
J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007 3209

A R T I C L E S
Conti et al.
Scheme 6. Cartoonlike Drawing Summarizing Z-CPD Photochemical Reaction Channels Triggered by nπ* (Blue Arrows) and ππ* (Red
Arrows) Excitationsa
a The semi-transparent surface represents the S1/S0 crossing region: it spans very different structures along the -NdN- torsional and the asymmetric
bending coordinates.
isomers to the funnels determining the decay of the system,
eventually leading to the final photoproducts on the ground state.
We have also matched experiments with computational predic-
tions, finding a reasonable agreement between calculations and
observations and drawing a fully comprehensive reaction scheme
for the photochemistry of CPD, which accounts for the observed
photophysical/photochemical properties. A cartoonlike drawing
summarizing these results is reported in Scheme 5 (E isomer)
and Scheme 6 (Z isomer).
To clarify further the interplay (and decay) of the photo-
chemically relevant states involved in the reactive processes,
we have mapped the S3/S2, S2/S1, and S1/S0 CI hyperlines
connecting the different branches of the photochemical paths.
The results show the existence of complex and extended
intersections spaces whose accessibility and structural properties
allow an understanding of the observed photochemical outcome.
In conclusion we have shown that, despite the difference in
molecular structure, the photophysical and photochemical
properties of CPD in the singlet and triplet states are substan-
tially the same as those in azobenzene. Nevertheless, CPD is a
better Z f E photoisomerizing system, leading to a higher
(∼70%) ΦZfE (nπ*) quantum yield. Therefore, CPD appears
as a simpler but more efficient substitute of azobenzene, still
retaining all its main reactivity features but with new promising
functionalization opportunities. Furthermore, due to the photo-
chemical/photophysical similarity of the two systems, CPD can
be seen as a reliable model for the reactivity of azobenzene
itself. In particular, because of the reduced size of its π-system
compared to that of AB, CPD has been fully studied with a
complete (i.e., unreduced) active space allowing highly accurate
CASPT2//CASSCF mappings of the potential energy surfaces
of the photochemically relevant states, and reliable computa-
tional descriptions at a level of accuracy that is currently much
more expensive for AB. The computed reaction channels provide
a detailed understanding of the photochemical reactivity of CPD,
which is in agreement with the experiments and can be regarded
as a general model for the photoreactivity of azobenzene-type
compounds in general.
A similar study for AB is under way and will be presented
elsewhere.
Acknowledgment. Support from the University of Bologna
(funds for selected research topics), from MURST PRIN 2005
(Project: “Trasferimenti di energia e di carica a livello mole-
colare”) and FIRB (RBAU01L2HT) are gratefully acknowl-
edged.
Supporting Information Available: Cartesian coordinates of
all the optimized structures discussed in the text, additional
tables and figures and complete refs 30 and 31. This material
is available free of charge via the Internet at http://pubs.acs.org.
JA0668466
9
3210 J. AM. CHEM. SOC. VOL. 129, NO. 11, 2007