Radiative depopulation of the excited intramolecular charge-transfer state of 9-(4-(N,N-dimethylamino)phenyl)phenanthrene

Article abstract of DOI:10.1021/ja953697a

Intramolecular photoinduced electron transfer in 9-(p-N,N-dimethylanilino)phenanthrene (9DPhen) has been studied in solution. The solvent dependence of the fluorescence spectra of 9DPhen indicates that the emission occurs from a highly polar excited state

Full text of DOI:10.1021/ja953697a

2892
J. Am. Chem. Soc. 1996, 118, 2892-2902
Radiative Depopulation of the Excited Intramolecular
Charge-Transfer State of
9-(4-(N,N-Dimethylamino)phenyl)phenanthrene
A. Onkelinx,† F. C. De Schryver,*,† L. Viaene,† M. Van der Auweraer,† K. Iwai,‡
M. Yamamoto,‡ M. Ichikawa,§ H. Masuhara,§ M. Maus,| and W. Rettig|
Contribution from the Department of Chemistry, Katholieke UniVersiteit LeuVen,
Celestijnenlaan 200F, B-3001 HeVerlee-LeuVen, Belgium, Department of Chemistry,
Faculty of Science, Nara Women’s UniVersity, Nara 630, Japan, Department of Applied Physics,
Faculty of Engineering, Osaka UniVersity, Suita, Osaka 565, Japan, and Institut fu¨r Physikalische
und Theoretische Chemie, Humboldt-UniVersita¨t, Bunsenstrasse 1, D-10117 Berlin, Germany
ReceiVed NoVember 2, 1995X
Abstract: Intramolecular photoinduced electron transfer in 9-(p-N,N-dimethylanilino)phenanthrene (9DPhen) has
been studied in solution. The solvent dependence of the fluorescence spectra of 9DPhen indicates that the emission
occurs from a highly polar excited state. The quantum yield of fluorescence (Φf) of 9DPhen is quite high and
increases with increasing solvent polarity. The radiative rate constant (kf), however, shows a maximum for solvents
of intermediate polarity, e.g., in butyl acetate a value of 2.3 × 108 s-1 is attained. These results are difficult to
explain within the “TICT” (twisted intramolecular charge transfer) model, which predicts a strongly forbidden
fluorescence caused by a minimum overlap of the orbitals involved in the transition. The above-mentioned trend as
a function of the solvent polarity is observed in particular donor-acceptor substituted arenes where the Lb state of
the corresponding arenes is lower in energy than the La state. The quantum chemical calculations actually could
1
1
explain this behavior on the basis of an ICT state which interacts with the lower lying La and Lb states of the
acceptor. The quantum mechanical mixing of states can occur by two pathways, namely orbital mixing and mixing
of configurations, and is modified by geometrical changes and by solvent polarity. The single exponential fluorescence
decay, obtained with time-correlated single-photon-timing, suggests emission from an excited charge-transfer state,
resulting from a solvent-induced rapid relaxation of the initial delocalized excited state of 9DPhen, obtained immediately
after picosecond pulsed excitation. Picosecond transient absorption spectra in acetonitrile show a rapid decay within
a few picoseconds from a less polar but delocalized excited state toward a more polar ICT state. Even the triplet
state of 9DPhen in isopentane at 77 K shows a significant polar character. As a reference compound,
9-phenylphenanthrene (9PhPhen) was also examined by means of stationary and time-resolved fluorescence
measurements as well as transient absorption experiments.
Introduction
donor and the acceptor π-electron systems into parallel planes
within a distance allowing through-space interactions and
Coulomb stabilization of the charge-transfer (CT) state.16-18 If
the driving force between the donor (D) and acceptor (A) is
appropriate, photoinduced electron transfer can occur even in
the stretched conformation.19-23
For molecules consisting of an electron donor and an electron
acceptor moiety coupled by a polymethylene chain,1-15 excita-
tion is often followed by a conformational folding bringing the
† Katholieke Universiteit Leuven.
‡ Nara Women’s University.
§ Osaka University.
(11) Hatano, Y.; Yamamoto, M.; Nishijama, Y. J. Phys. Chem. 1987,
82, 367.
| Humboldt-Universita¨t.
* To whom correspondence should be addressed.
X Abstract published in AdVance ACS Abstracts, March 1, 1996.
(1) Mataga, M.; Okada, T.; Masuhara, H.; Nakashima, N.; Sakata, Y.;
Misumi, S. J. Lumin. 1976, 12/13, 159.
(2) Okada,T.; Saito, T.; Mataga, N.; Sakata, Y.; Misumi, S. Bull. Chem.
Soc. Jpn. 1977, 50, 331.
(3) Migita, M.; Okada, T.; Mataga, N.; Sakata, Y.; Misumi, S.;
Nakashima, N.; Yoshihara, K. Bull. Chem. Soc. Jpn. 1981, 54, 3304.
(4) Okada, T.; Migata, M.; Mataga, N.; Sakata, Y.; Misumi, S. J. Am.
Chem. Soc. 1981, 103, 4715.
(12) Helsen, N.; Viaene, L.; Van der Auweraer, M.; De Schryver, F. C.
J. Phys. Chem. 1994, 98, 1532.
(13) Okada, T.; Fujita, T.; Kubota, M.; Masaki, S.; Mataga, N.; Hayashi,
T.; Ide, R.; Sakata, Y.; Misumi, S. Chem. Phys. Lett 1972, 14, 563.
(14) De Schryver, F. C.; Van der Auweraer, M.; Boens, N.; Declercq,
D.; Helsen, N.; DePaemelaere, S.; Van Haver, P.; Onkelinx, A.; Iwai; K.
Supramol. Chem. 1992, 313.
(15) De Schryver; F. C.; Declercq, D.; Depaemelaere, S.; Hermans, E.;
Onkelinx, A.; Verhoeven, J. W.; Gelan, J. J. Photochem. Photobiol. A:
Chem. 1994, 82, 171.
(5) Masaki, S.; Okada, T.; Mataga, N.; Sakata, Y.; Misumi, S. Bull. Chem.
Soc. Jpn. 1976, 49, 1277.
(6) Swinnen, A. M.; Van der Auweraer, M.; De Schryver, F. C.; Windels,
C.; Goedeweeck, R.; Vannerem, A.; Meeus, F. Chem. Phys. Lett. 1983,
95, 467.
(7) Swinnen, A. M.; Van der Auweraer, M.; De Schryver, F. C.; Chem.
Phys. Lett. 1984, 109, 574.
(8) Swinnen, A. M.; Van der Auweraer, M.; De Schryver, F. C. J.
Photochem. 1985, 28, 315.
(9) Swinnen, A. M.; Ruttens, F.; Van der Auweraer, M.; De Schryver,
F. C. Chem. Phys. Lett. 1985, 116, 217.
(10) Swinnen, A. M.; Van der Auweraer, M.; De Schryver, F. C.;
Nakatani, K.; Okada, T.; Mataga, N. J. Am. Chem. Soc. 1987, 109, 321.
(16) Hirayama, F. J. Chem. Phys. 1965, 42, 3163.
(17) Mimura, T.; Itoh, M. J. Am. Chem. Soc. 1976, 98, 1095.
(18) Davidson, R. S. In: AdVances in Physical Organic Chemistry; J.
Wiley: New York, 1983; Vol. 19, p 1.
(19) Wegewijs, B.; Hermant, R. M.; Verhoeven, J. W.; Kunst, A. G.
M.; Rettschnick, R. P. H. Chem. Phys. Lett. 1987, 140, 587.
(20) Van Stokkum, I. H. L.; Brouwer, A. M.; Van Ramesdonk, H. J.;
Scherer, T. Proc. K. Ned. Akad. Wet. 1993, 96, 43.
(21) Verhoeven, J. W. Pure Appl. Chem. 1990, 62, 1985.
(22) Hermant, R. M.; Wegewijs, B.; Verhoeven, J. W.; Kunst, A. G.
M.; Rettscnick, R. P. H. Recl. TraV. Chim. Pays-Bas 1988, 107, 349.
(23) Wegewijs, B.; Verhoeven, J. W. J. R. Neth. Chem. Soc. 1995, 114,
1.
0002-7863/96/1518-2892$12.00/0 © 1996 American Chemical Society

Intramolecular Photoinduced Electron Transfer
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2893
In another type of molecules with the donor and the acceptor
groups directly connected by a single σ-bond, conformational
folding is excluded.24-29 It has been shown30-53 that primary
excited singlet states of p-(9-anthryl)-N,N-dimethylaniline
(ADMA) and related compounds undergo solvent-assisted
femto/picosecond relaxation to a polar fluorescent CT state. The
geometry and electronic structure of the emitting CT state,
however, are still a point of discussion.28,29,34,46 Okada and co-
workers30-32 first explained their results in terms of the TICT
(twisted intramolecular charge transfer) state model.34,54 Ac-
cording to this interpretation, the excited singlet state undergoes
an adiabatic intramolecular electron transfer. Two metastable
states are assumed to interconvert by a torsional motion, which
provides a possible reaction coordinate for the electron transfer.
More recent picosecond transient absorption studies however
show a difference between the electronic structure of the excited
state of ADMA or structurally similar molecules and that of
molecules with strongly orthogonal π systems of donor
and acceptor (e.g., 4-(9-anthryl)-N,N,2,6-tetramethylaniline)
(ATMA)).46,49,55 The transient absorption spectra of ADMA
in medium polar solvents cannot be represented by a linear
combination of the anthracene-like band in weakly polar, and
the anthracene anion radical band in strongly polar, solvents,
Figure 1. Structures of 9PhPhen and 9Dphen.
so doubts have been expressed as to the validity of the simple
two-state TICT model for ADMA and related compounds.
Thereby an aryl-amine torsion angle of less than 90° is
proposed for the excited states of these compounds.
A similar conclusion was also reached in recent quantum
chemical calculations56 where twist angle dependent values of
transition energies, oscillator strengths, and dipole moments
were calculated.
Similar to the behavior of ADMA and related compounds,
other large conjugated π-systems connected formally by a single
bond29,56-58 indicate a nonorthogonal geometry of the CT state
or a TICT state strongly vibronically coupled with a nearby
nonpolar state.59
A different model, not involving intramolecular twisting, has
recently been proposed to explain the dual fluorescence in
aminobenzonitriles, the compounds originally explained within
the TICT framework.34,54 In this model the dual emission is
interpreted as a solvent-induced pseudo-Jahn-Teller effect: dual
fluorescence and intramolecular charge transfer (ICT) are
observed when two excited levels S1 and S2 (CT) in N,N-
dimethyl-4-aminobenzonitrile (DMABN) have an energy gap
sufficiently small for vibronic coupling. It is argued that the
N-inversion of the amino group acts as a promoting mode which
decouples, even without rotational isomerization, the nitrogen
lone pair from the π-electrons of the phenyl ring.60,61
This interpretation was based on the observation of intramo-
lecular charge transfer in a series of 4-(dialkylamino)benzoni-
triles (methyl, ethyl, propyl, and decyl), even in apolar solvents.
The observation that the charge separation increased with
increasing length of the alkyl chain was difficult to explain
within the the former TICT model.
Recently, however, CASSCF (complete active space self-
consistent field) calculations62 seem not to support such an
alternative interpretation as the only contributive mechanism
to the CT state.
Also in another recent publication63 where solvent effects
have been analyzed using a microstructural solvation model
associated with the AM1 results of some parasubstituted N,N-
dimethylaniline derivatives, the same conclusion is obtained.
The dual fluorescence phenomena could be related to the
presence of a twisted internal charge-transfer state which is
stabilized even in a nonpolar solvent but cannot be related only
to the presence of a wagged internal charge-transfer state, even
in a strongly polar solvent.
(24) Rettig, W. Angew. Chem., Int. Ed. Engl. 1986, 25, 971.
(25) Hagopian, S.; Singer, L. A. J. Am. Chem. Soc. 1985, 107, 1874.
(26) Detzer, N.; Baumann, W.; Schwager; B.; Frohling, J.-C.; Brittinger,
C. Z. Naturforsch. 1987, A42, 395.
(27) Siemiarczuk, A.; Koput, J.; Pohorillo, A. Z. Naturforsch. 1982, A37,
598.
(28) Herbich, J.; Kapturkiewicz, A. Chem. Phys. 1991, 158, 143, and
references therein.
(29) Herbich, J.; Kapturkiewicz, A. Chem. Phys. 1993, 170, 221.
(30) Okada, T.; Fujita, T.; Kubota,M.; Masaki, S.; Mataga,N.; Ide, R.;
Sakata, Y.; Misumi, S. Chem. Phys. Lett. 1972, 14, 563.
(31) Okada, T.; Fujita, T; Mataga, N. Z. Phys. Chem. NF 1976, 101, 57.
(32) Masaki, S.; Okada, T.; Mataga, N.; Sakata, Y.; Misumi, S. Bull.
Chem. Soc. Jpn. 1976, 49, 1277.
(33) Chandross, E. A. In The Exciplex; Gordon, M.; Ware, W.R., Eds.;
Academic Press: New York, 1975; p 187.
(34) Grabowski, Z. R.; Rotkiewicz, K.; Siemiarczuk, A.; Cowley, D. J;
Baumann, W. NouV. J. Chim. 1979, 3, 443.
(35) Grabowski, Z. R.; Dobkowski, J. Pure Appl. Chem. ,1983, 55, 245.
(36) Grabowski, Z. R. Acta Phys. Pol. 1987, A71, 743.
(37) Grabowski, Z. R. In Supramolecular Photochemistry; Balzani, V.,
Ed.; Reidel: Dordrecht, The Netherlands, 1987; p 319.
(38) Siemiarczuk, A.; Grabowski, Z. R.; Kro´wczynski, A.; Asher, M.;
Ottolenghi, M. Chem. Phys. Lett. 1977, 51, 315.
(39) Grabowski, Z. R.; Rotkiewicz, K.; Siemiarczuk, A. J. Lumin. 1979,
18/19, 420.
(40) Siemiarczuk, A. Chem. Phys. Lett. 1984, 110, 437.
(41) Siemiarczuk, A.; Ware,W. R. J. Phys. Chem. 1987, 91, 3677.
(42) Baumann, W.; Petzke, F.; Loosen, K. D. Z. Naturforsch. 1979, A34,
1070.
(43) Baumann, W.; Bischof, H. J. Mol. Struct. 1982, 84, 181.
(44) Detzer, N.; Baumann, W.; Scwager, B.; Fro¨hling, J. C.; Brittinger,
C. Z. Naturforsch. 1987, A42, 395.
(45) Okada,T.; Kawai, M.; Ikemachi,T.; Mataga, N; Sakata,Y.; Misumi,
S.; Shionoya, S. J. Phys. Chem. 1984, 88, 1976.
(46) Okada,T.; Mataga, N.; Baumann, W.; Siemiarczuk, A. J. Phys.
Chem. 1987, 91, 4490.
(47) Baumann, W.; Schwager, B.; Detzer, N.; Okada, T.; Mataga, N. J.
Phys. Chem. 1988, 92, 3742.
(48) Mataga, N.; Yao, H.; Okada, T.; Rettig, W. J. Phys. Chem. 1989,
93, 3383.
(49) Mataga, N.; Nishikawa, S.; Asahi, T.; Okada,T. J. Phys. Chem. 1990,
94, 1443.
(50) Barbara, P. F.; Jarzeba, W. In AdVances in Photochemistry; Volman,
D. H., Hammond, G. S., Gollnick, C., Eds.; Wiley: New York, 1990; Vol.
15, p 1.
(51) Tseng, J. C. C.; Huang, S.; Singer, L. A. Chem. Phys. Lett. 1988,
153, 401.
(52) Tseng, J. C. C.; Singer, L. A. J. Phys. Chem. 1989, 93, 7092.
(53) Rettig, W; Baumann, W. In: Photochemistry and Photophysics;
Rabek, J. F., Ed.; CRC Press: Boca Raton, FL, 1992; Vol. 6, p 79.
(54) Rettig, W. Electron transfer I. In Topics in Current Chemistry, Vol.
169; Mattay, J., Ed.; Spinger Verlag: Berlin, 1994; p 253.
(55) Okada, T. Proc.sIndian Acad. Sci., Chem. Sci. 1992, 104, 173.
(56) Klock, A. M.; Rettig, W. Polish. J. Chem. 1993, 67, 1375.
(57) Van Damme, M.; Hofkens, J.; De Schryver, F. C.; Ryan, T. G.;
Rettig, W.; Klock, A. M. Tetrahedron 1989, 45, 4693.
(58) Herbich, J.; Waluk, J. Chem. Phys 1994, 188, 247.
(59) Dobkowski, J.; Grabowski, Z.; Paeplow, B.; Rettig, W.; Koch, K.;
Mu¨llen, K.; Lapouyade, R. New. J. Chem. 1994, 18, 525.
(60) Schuddeboom, W.; Jonker, S. A.; Warman, J. M.; Leinhos, U.;
Ku¨hnle, W.; Zachariasse, K. J. Phys. Chem. 1992, 96, 10809.
(61) Zachariasse, K.; Von Der Haar; T.; Hebecker, A.; Leinhos, U.;
Ku¨hnle, W. Pure Appl. Chem. 1993, 65, 1745.
(62) Serrano-AndrC¸ s, L.; Merchan, M.; Roos, B. O.; Lindh, R. J. Am.
Chem. Soc. 1995, 117, 3189.
(63) Gorse, A. D.; Pesquer, M. J. Phys. Chem. 1995, 99, 4039.

2894 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Onkelinx et al.
Absorption spectra were recorded on a Perkin Elmer Lambda 6 UV/
vis spectrophotometer. Corrected fluorescence spectra were obtained
with a SPEX Fluorolog 212. The fluorescence quantum yields were
determined using quinine sulfate in 0.1 N H2SO4(Φf ) 0.54) as a
reference.68
The present contribution concerns the photoinduced electron
transfer in phenanthryl-aniline systems. These compounds
have some important advantages compared to other aryl systems,
e.g. a low tendency to form intermolecular aggregates, the
absence of excimer formation,64 and a high quantum yield of
fluorescence.
In this paper we focus on the solvent dependence of the
spectral position of the CT fluorescence maxima (ν˜max), the
quantum yield (Φf), the excited state depopulation kinetics ( τf )
and resulting radiative rate constant ( kf ).
In order to get more information about the nature of the
fluorescent excited states, single photon timing experiments are
combined with picosecond transient absorption experiments in
a polar and an apolar solvent. To obtain more information about
the nature of the triplet state and, more importantly, about the
different triplet states in 9DPhen and 9PhPhen, nanosecond
transient absorption measurements are executed in combination
with phosphorescence experiments in a rigid glass.
Because of the size of the 9DPhen compound, it was
practically impossible to perform CASSCF calculations in
combination with multiconfigurational second-order perturbation
theory (CASPT2) to calculate transition energies (absorption
and emission) and structural and electronical properties of the
ground and excited states and to compare these findings with
available experimental information.
Stationary and time-resolved fluorescence measurements were
performed on samples degassed by several freeze-pump-thaw cycles.
The picosecond laser photolysis system69-70 as well as the nano-
second laser photolysis system71 has been described in detail elsewhere.
Time-resolved fluorescence decays were obtained by the single-
photon-timing (SPT) technique, and the fluorescence decays were
analyzed by the global analysis as described, using the reference
convolution method.72-75 For the SPT experiments an excitation
wavelength of 295 nm (fwmh ) 40 ps) was used and decays were
measured over the whole emission range (isooctane, 370-470 nm;
dibutyl ether, 380-460 nm; diisopropyl ether, 380-460 nm; diethyl
ether, 380-460 nm; butyl acetate, 380-460 nm; ethyl acetate, 390-
540 nm; tetrahydrofuran, 390-500 nm; ethanol, 440-480 nm; aceto-
nitrile, 380-530 nm). All fluorescence decay curves were observed
at the magic angle (55°), contained 104 peak counts, and were collected
in 511 channels of the multichannel analyzer. Time increments of 60,
62, and 81 ps/channel were used for 9DPhen and 327 ps/channel for
9PhPhen, respectively.
For 9DPhen in acetonitrile the previous experiments were completed
with measurements using a time increment of 5 ps as well as a scatter
solution (LUDOX) instead of a reference.
POPOP [p-bis[2-(5-phenyloxazoyl)]benzene in methylcyclohexane],
PPO (2,5-diphenyloxazole) in methylcyclohexane], and MSQ (meth-
oxysulfopropylquinolinium in water), having a decay time (τ) of
respectively 1.1, 1.4, and 27 ns at room temperature, were used as
reference compounds. The reference samples were degassed by several
freeze-pump-thaw cycles before use.
Instead, the semiempirical CNDO/S method of Del Bene and
Jaffe modified to cope with larger molecules and to calculate
excited-state dipole moments was used with two different
parametrizations65 to calculate transition energies, oscillator
strengths, and dipole moments.
The molecular conformation of 9DPhen has been investigated by
X-ray diffraction, and the dihedral angle between the two aromatic ring
systems in crystalline 9DPhen was found to be 64.2°.76
Experimental Section
For the quantum chemical calculations, configuration interaction with
50 singly excited configurations was included. The latter were
generated from the usual canonical molecular orbitals which are
virtually localized on the submoieties for 90° twist but can extend over
the whole π-system for other twist angles (see below Figure 12). The
input geometries have been optimized with the AM1 Hamiltonian within
the AMPAC program.77 For comparison, idealized input geometries
were also used (regular fully planar hexagons for the phenyl and
phenanthrene groups with 1.4 Å R(CC), 1.08 Å R(CH), and R(CC) )
1.50 Å between these two groups and R(CN) ) 1.37 Å and R(N-CH3)
) 1.46 Å for the dimethylamino group). The solvation energies of
the excited states were calculated from their dipole moments µ
Synthesis. 9-(4-(N,N-Dimethylamino)phenyl)phenanthrene was syn-
thesized by the Grignard coupling reaction66 of (4-(N,N-dimethylamino)-
phenyl)magnesium bromide with 9-bromophenanthrene using dichloro-
[1,3-bis(diphenylphosphino)propane]nickel(II) catalyst in dry THF. The
crude product was purified by column chromatography on silica gel
with benzene eluent and recrystallized from hexane: mp 155.5-156.5
°C. EIMS (20 eV): m/e (relative intensity) 297 (M+, 100) and 282
(4). 1H NMR (CDCl3): δ 3.03 (6H, s), 6.71-8.84 (13H, m). 9PhPhen
was similarly synthesized from 9-phenanthrylmagnesium bromide and
phenyl bromide.
The crude product was purified by column chromatochraphy on silica
gel with benzene eluent and recrystallized three times from methanol
to yield a white powder: mp 103.5-104.5 °C [lit.67 104-104.5 °C].
All solutions used for stationary and time-resolved fluorescence
measurements were further purified on HPLC with 10:90 V% ethyl
acetate, with hexane as the eluent.
Solvents. The solvents used (isooctane, Merck; cis-decaline,
Aldrich; dibutyl ether, Merck; diethyl ether, Lab scan; butyl acetate,
Janssen Chimica; ethyl acetate, Merck; tetrahydrofuran, Rathburn;
butanol, Merck; acetonitrile, Merck) were of spectroscopic grade and
were used as received. The solvent diisopropyl ether (Riedel De Hae¨n)
was checked for fluorescence before use.
Methods. The absorbance of the sample solutions used for stationary
and time-resolved fluorescence measurements equals 0.1 at the excita-
tion wavelength for the fluorescence experiments (λexcitation ) 295 nm)
and 1.5 for the nanosecond transient absorption measurements (λexcitation
) 320 nm).
(69) Ichikawa, M.; Fukumura, H.; Masuhara, H. J. Phys. Chem. 1994,
98, 12211. Ichikawa, M.; Fukumura, H.; Masuhara, H.; Koide, E.;
Hyakutake, H. Chem. Phys. Lett. 1995, 232, 346.
(70) Fukazawa, N.; Fukumura, H.; Masuhara, H.; Prochorow, J. Chem.
Phys. Lett. 1994, 220, 461.
(71) Viaene, L.; Van Mingroot, H.; Van haver, P.; Van der Auweraer,
M.; De Schryver, F. C.; Itaya, A.; Masuhara, H. J. Photochem. Photobiol.
A: Chem. 1992, 66, 1.
(72) Khalil, M. M. H.; Boens, N.; Van der Auweraer, M.; Ameloot, M.;
Andriessen, R.; Hofkens, J.; De Schryver, F. C. J. Phys. Chem. 1991, 95,
9375.
(73) O’Connor, D. V.; Phillips, D. Time-Correlated Single Photon
Counting; Academic Press: London, 1984.
(74) Boens, N.; Janssens, L. D.; De Schryver, F. C. Biophys. Chem. 1989,
33, 77.
(75) Boens, N. Luminiscence Techniques in Chemical and Biochemical
Analysis; De Keukeleire, W. R. G., Korkidis, K., Eds.; Marcel Dekker:
New York, 1991; pp 21-45.
(76) Van Meervelt, L Acta Crystallogr. 1993, C49, 593.
(77) AMPAC 5.0, Semichem, 7128 Summit, Shawnee, KS 66216, 1994.
(78) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486.
(79) Jaffe´, H.; Orchin, M. Theory and applications of ultraViolet
spectroscopy; John Wiley and Sons: New York, 1966; p 81.
(80) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486.
(81) Lippert, E. Z. Naturforsch. 1955, 10A, 541.
(64) Chandross, E. A.; Langworth, J. W.; Visco, R. E. J. Am. Chem.
Soc. 1965, 87, 3259.
(65) CNDO/S: Del Bene, J.; Jaffe´, H. H. J. Chem. Phys. 1968, 48, 1807;
1968, 49, 1221; 1969, 50, 1126. CNDUV: QCPE program no. 333 with
the original paramaters was used.
(66) Kumada, M. Pure Appl. Chem. 1980, 52, 669 and references therein.
(67) Koussini, R.; Lapouyade, R.; de Violet, P. F. J. Am. Chem. Soc.
1978, 100, 6679.
(68) Morris, J. V.; Mahaney, M. A.; Huber, J. R. J. Phys. Chem. 1976,
80, 969.
(82) Mataga, N.; Kaifu, Y.; Koizumi, M. Bull. Chem. Soc. Jpn. 1955,
28, 690.
(83) McRae, E. G. J. Phys. Chem. 1957, 61, 562.

Intramolecular Photoinduced Electron Transfer
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2895
Figure 3. Fluorescence emission spectra of 9DPhen as a function of
the solvent polarity at room temperature. (excitation wavelength is 295
nm): (s) isooctane, (- - -) dibutyl ether, (- - -) diethyl ether, (- -)
tetrahydrofuran, and (- - -) acetonitrile.
Figure 2. Absorption spectra of 9DPhen in isooctane (s) and
acetonitrile (- - -).
(calculated in the gas phase) and the Onsager cavity radius taken as
0.56 nm for 9DPhen using the following equations:78
Esolv(Sn) ) Egas(S0) + ∆Egas(S0fSn) + ∆Esolv(Sn)
(1)
(2)
µ2(Sn) ꢀr - 1
1
∆Esolv(Sn) ) -
a3
4πꢀ0
2ꢀr + 1
where ꢀr is the static dielectric constant of the solvent. These
calculations were executed on a HP 715 workstation.
Results and Discussion
Absorption and Emission Experiments as a Function of
the Solvent Polarity. The absorption spectra of 9DPhen in
isooctane and acetonitrile, presented in Figure 2, consist of two
absorption bands. The long-wavelength absorption band at 312
nm exhibits a slight bathochromic shift of 700 cm-1 upon
increasing the solvent polarity from isooctane to acetonitrile.
The 1Lb transition of phenanthrene (1Lb: 348 nm) is not observed
because of its symmetry-forbidden nature and the more intense
absorption band around 312 nm. This observation suggests that
there is already an interaction between the dimethylanilino and
the phenanthryl components in the ground state. The molar
extinction coefficient of 9DPhen in isooctane has a value of
15 200 L mol-1 cm-1 in the maximum of the absorption band
(312 nm) and 11 500 L mol-1 cm-1 in the maximum of the
absorption band in acetonitrile, respectively (319 nm).
9DPhen shows a fluorescence emission, which is highly
solvatochromic at room temperature as can be seen in Figure
3. The emission spectra lose structure, broaden, and shift to
the red with increasing solvent polarity, indicating a substantial
dipole moment of the excited state. The emission maxima of
9DPhen obtained in solvents of different polarity are reported
in Table 1 together with the photophysical constants and the
derived kf and reduced kf values according to eq 3. The
Figure 4. Absorption and emission spectra of 9PhPhen in isooctane
at room temperature.
Table 1. Solvent Effects on the Spectral Position of the CT
Fluorescence Maxima, Quantum Yields for Fluorescence, and Decay
Parameters of 9DPhen and 9PhPhen in Solution
ν˜max
( kf /n3ν˜3) ×
solvent
(cm-1
)
Φf τ (ns) kf /107 s-1 10-6 s-1 cm3
9DPhen
isooctane
dibutyl ether
25 939 0.36
25 316 0.44
diisopropyl ether 24 631 0.49
9.2
3.9
19.4
22.2
22.7
22.7
18.1
20.7
12
0.8
4.4
5.5
6.2
6.5
5.6
6.0
5.1
2.42
2.21
2.16
2.73
3.6
diethyl ether
butyl acetate
ethyl acetate
24 570 0.49
23 474 0.62
23 255 0.65
tetrahydrofuran 23 148 0.68
2.7
7
acetonitrile
21 321 0.85
9PhPhen
isooctane
acetonitrile
25 641a 0.23 41.1
25 641a 0.39 38.4
0.56
1.02
0.12
0.25
a 0-0 transition.
64π4
3h
390 nm) is also comparable with the emission of 9-ethylphenan-
threne79 and shows virtually no solvent dependence (Table 1).
For 9DPhen, the dipole moment of the highly polar state can
be determined by the fluorescence solvatochromic shift
method80-86 (Figure 5), given that the excited state lives
sufficiently long with respect to the orientational relaxation time
of the solvent, smaller than 0.1 ns in every case, and assuming
a point dipole situated in the center of the spherical cavity.
Neglecting the dipole moment of the ground state and the
n3ν˜3max|M|
(3)
2
kf )
wavelength shift between isooctane and acetonitrile corresponds
to an energy difference of 4600 cm-1 and indicates a highly
polar emitting state. The absorption and emission spectra of
9DPhen were compared to those of 9PhPhen.
The positions of the absorption maxima of 9PhPhen in
isooctane (λmax: 334, 342, and 351 nm) resemble quite well
those of 9-ethylphenanthrene,79 and the extinction coefficient
of 9PhPhen in isooctane (351 nm) has a value of 340 L mol-1
cm-1. The emission spectrum of 9PhPhen (λmax: 355, 373, and
(84) Bakshiew, N. G. Opt. Spektrosk. 1961, 10, 717.
(85) Bilot, L.; Kawski, A. Z. Naturforsch. 1962, 17A, 621.
(86) Liptay, W. Z. Naturforsch. 1965, 20A, 1441.

2896 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Onkelinx et al.
2µe2
hcν˜CmTax ) hcν˜max(0) -
3f ′′(ꢀr,n)
(6)
CT
4πꢀ0hca0
n2 - 1
ꢀr - 1
2ꢀr + 1
2
(
)
(
)
2(2n + 1)
f ′′(ꢀr,n) )
-
(7)
(n2 - 1)
ꢀr - 1
1 -
1 -
(
)
3(2n2 + 1)
3(2ꢀr + 1)
Taking into account this polarizability there is still a deviation
from linearity (Figure 5). The nonlinear behavior, even with
the correction for the polarizability, could be explained by an
equilibrium between two excited states or by a continuous
change in the excited-state wave function as a function of solvent
(vide infra).
Figure 5. Lippert-Mataga plot: fluorescence emissiom maximum as
a function of the solvent polarity parameter f - 0.5f ′ (without solute
polarizability) and f - 0.5f ′′ (including solute polarizability): (9) f -
0.5f ′ and (b) f - 0.5f ′′.
While the change of slopes clearly indicates some relaxation
process depending on solvent polarity, the absolute values of
the excited-state dipole moments should be viewed with caution.
One should take into account that the solvatochromic method
is rather crude in predicting absolute values of µCT as there is
a certain arbitrariness in the choice of the Onsager radius value.
For 9DPhen, the determined fluorescence quantum yields
strongly depend upon the solvent polarity as can be seen in
Table 1. The higher the solvent polarity, the larger the quantum
yield of fluorescence.
In contrast with the solvent dependence of 9DPhen, the
fluorescence quantum yields of 9PhPhen are quite solvent
independent. The quantum yield of 9PhPhen did amount to
0.23 in isooctane and in dibutyl ether and to a value of 0.26 in
diethyl ether and tetrahydrofuran. Only in acetonitrile is a small
increase of the quantum yield (0.39) observed (Table 1).
In a rigid matrix (isopentane, 77 K), the structure of the
emission band of 9DPhen was very similar in shape with the
emission spectra of 9DPhen in isooctane at room temperature.
Also in a rigid matrix of a more polar solvent ethanol, the
structure of the emission band of 9DPhen was very similar in
shape to the emission spectra of 9DPhen in isooctane at room
temperature.
polarizability (R) of the solute in the states involved in the
transition, one can derive the following equations:87,88
2µe2
1
2
hcν˜CmTax ) hcν˜max(0) -
f - f ′
(4)
(5)
CT
(
)
3
a0 4πꢀ0hc
1
2
1 n2 - 1
ꢀr - 1
2ꢀr + 1
f - f ′ )
-
(
)
(
)
(
2
)
2
2n + 1
ν˜max and νmax(0) correspond, respectively, to the wavenumber
of the emission maximum in a solvent with dielectric constant
ꢀ and refractive index n and the wavenumber of the emission
maximum in the gas phase. ꢀ0, h, c, and a0 correspond to the
permitivity of vacuum (8.85 × 10-12 J-1 C2 m-1), the Planck
constant (6.62 × 10-34 J), and the velocity of light in a vacuum
(2.99 × 108 m/s). For 9DPhen we estimate an Onsager radius
a0 of the solvent cavity of 5.7 × 10-10 m by analogy with the
Onsager radius calculated for the compound 1-(N,N-diethyl-
anilino)pyrene.52 µe (Cm-1) is the permanent dipole of the
excited state.
In this rigid matrix system also a stuctureless phosphorescence
was observed which differed distinctly in shape and position
from the phosphorescence of 9-ethylphenanthrene and 9-phen-
ylphenanthrene as can be seen in Figure 6.
Because of this noticeable difference in phosphorescence also
nanosecond transient absorption measurements were performed
for 9DPhen and 9PhPhen in diethyl ether. As can be seen in
Figure 7, the difference in the transient absorption spectra of
the two compounds spectra is quite obvious.
As can be seen in Figure 5, the experimental results do not
follow the linear relationship predicted by eq 4. The change in
slope is observed between the solvents dibutyl ether and diethyl
ether. Assuming that µg equals zero, a value of 24 500 cm-1 is
2
3
obtained for the ratio of -2µe /a0 4hcπꢀ0 (µe ) 21.2 D) in the
region of highly polar solvents. In the region of less polar
solvents, a linear relationship with a slope of 4800 cm-1 can
be drawn from which a dipole moment of 9.3 D can be derived.
The deviation from the linear equation could be caused by
the large contribution of the induced dipole moment in the
excited state involved in the transition. In the former equation,
the polarizability of the solute was neglected but the influence
of the induced dipole moment can be taken into account by
postulating the polarizability Rg ≡ Re ≡ F3/3 for respectively
the ground and the excited state.89-90
The obtained solvent polarity parameter f - 0.5f ′′ (eq 7) leads
to a new plot of ν˜max as a function of the solvent polarity (Figure
5).
When the transient absorption spectra of 9DPhen were
measured in solvents of different polarity a clear shift of the
absorption bands was observed as can be seen in Table 2. In
very polar solvents (acetonitrile), however, the shape of the
transient absorption band is changed compared to that of the
medium polar solvents, and at long delay times (50 µs), an
absorption at 436 nm remains, the contribution of which
increases with increasing laser energy. This suggests that the
remaining absorption could be caused by photoionization.
Single Photon Timing Experiments. Using single curve
analysis, the fluorescence decay of 9DPhen in isooctane, dibutyl
ether, diisopropyl ether, diethyl ether, butyl acetate, ethyl acetate,
tetrahydrofuran, ethanol, and acetonitrile could be analyzed as
a single exponential decay and the fluorescence lifetime does
not depend upon the analysis wavelength. Also by global
analysis, the fluorescence decays obtained at different emission
wavelengths could be analyzed as a single exponential decay
(87) Liptay, W. In Excited States; Lim, E. C., Ed.; Acadamic Press: New
York, 1974; p 129.
(88) Baumann, W.; Nagy, Z.; Maiti, A. K.; Reis, H.; Rodriges, S. V.;
Detzer, N. Yamada Conference XXIX on Dynamics and Mechanisms of
Photoinduced Electron Transfer and related Phenomena, Osaka, Japan;
Mataga, N., Okada N., Masuhara, H., Eds.; Elsevier: Amsterdam, 1992; p
211.
(89) van der Zwan, G.; Hynes, J. T. J. Phys. Chem. 1985, 89, 4181.
(90) Suppan, P. Chem. Phys. Lett. 1983, 94, 272.
2
(Zø < 3).

Intramolecular Photoinduced Electron Transfer
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2897
Figure 6. Emission spectra of 9DPhen (- - -), 9-ethylphenanthrene
(s), and 9PhPhen (- - -) in a rigid matrix: isopentane, 77 K.
Figure 8. (a) Picosecond transient absorption spectra of 9DPhen in
isooctane. (b) Transient absorption spectrum of 9PhPhen in isooctane.
(I) A quantum chemical mixing (configuration interaction)
leading to mixed wave functions for a single conformation. This
mixing will be strongly angle dependent. For a 90° twist, the
ICT state converges to the classical TICT state.34,54
(II) A broadened potential minimum due to a rotational
distribution of conformers in the emitting state which changes
with solvent polarity. This distribution will change with solvent
polarity if the excited-state potential is solvent polarity depend-
ent.
Figure 7. Subnanosecond transient absorption spectra of 9DPhen
(- - -) and 9PhPhen in diethyl ether (delay time and pulse width of 100
ns).
Table 2. Spectral Shift of the Transient Absorption Bands of
9DPhen in Solvents of Different Polarity (laser energy: 27 mJ)
(III) Solvent dependency of the relative population of the
minumum. In the equilibrium model, the relative population
of the minimum can be solvent dependent if these possess
different excited state dipole moments.
Taking into account broad angular distributions, the effective
kf can be written more accurately as
solvent
λ(max1) (nm)
λ(max2) (nm)
Imax2/Imax1
isooctane
diethyl ether
ethyl acetate
THF
530
542
555
562
564
448
446
446
441
440
1.30
1.32
1.41
1.13
1.04
butanol
k ) k(θ) ηθ dθ
(8)
∫
f
θ
In acetonitrile some additional experiments were performed
at the blue edge of the emission spectrum, down to 380 nm. In
these experiments, a time increment of 5.3 ps was used and a
scatter solution as a reference. For these conditions also a single
exponential decay was obtained which could be analyzed
together globally with the earlier experiments.
Combining the results of stationary and time-resolved measure-
ments, the rate constant of fluorescence kf and kf /n3ν˜3 can
be obtained using eq 9 because emission in 9DPhen is supposed
to occur from a charge-transfer state formed more quickly than
the time resolution of the instrumental equipment.
In Table 1, the decay times of 9DPhen and 9PhPhen are
reported as a function of the solvent polarity.
Φf
kf )
(9)
In analogy with the spectral similarity between 9PhPhen and
9-ethylphenanthrene, the decay time did not vary that much upon
increasing the solvent polarity; in isooctane, it amounted to 57.6
and 41.1 ns for 9-ethylphenanthrene and 9PhPhen, respectively.
In acetonitrile the decay times amounted to 55.2 and 38.4 ns,
respectively. Judging from these experiments the first excited
states of 9-ethylphenanthrene and 9PhPhen seem quite similar
(Table 1).
τ
The high quantum yields of fluorescence and the relatively
short decay times correlate with the highly allowed nature of
the emissive CT state59 and contrast with the rather forbidden
emission which is associated with a TICT state.34,54
Picosecond Transient Absorption Measurements. The
results from the picosecond transient absorption measurements
of 9DPhen and 9PhPhen in the solvent isooctane are represented
in Figure 8. The transient absorption spectrum of 9PhPhen is
similar to that of 9-ethylphenanthrene,91 and the shift of the
maximum is about 10 nm. This suggests that the substitution
effect of the phenyl group in the phenanthrene electronic
The single exponential decay of 9DPhen in all solvents can
be explained by the presence of a single excited state, the nature
of which is changing as a function of the solvent polarity or by
a fast equilibration between two excited-state minima with
different properties. Three origins for the change with solvent
polarity could be suggested:
(91) Tamai, N.; Masuhara, H.; Mataga, N. J. Phys. Chem. 1983, 87, 4461.

2898 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Onkelinx et al.
Figure 10. Normalized transient absorption spectra of 9DPhen in
acetonitrile at 3 ps after excitation and at 100 ps after excitation and a
difference spectrum (thick line) obtained by subtracting the spectrum
100 ps after excitation from the spectrum 3 ps after excitation.
Figure 9. Picosecond transient absorption spectra of 9DPhen in
acetonitrile.
structure is not large. The decay time of this species is too
long to be determined accurately within the used experimental
setup by the change in intensity of the absorption band.
On the other hand, 9DPhen in isooctane shows an absorption
band at 520 nm, assigned to the S1-Sn absorption of a
delocalized excited state.
The bathochromic shift of 30 nm between the S1-Sn
absorption of 9-ethylphenanthrene (maximum of absorption at
490 nm in THF)91 and the S1-Sn absorption of 9DPhen in
isooctane again is an indication for a substantial coupling
between the phenanthrene and the dimethylanilino unit in the
excited state, even in nonpolar solvents such as isooctane.
The decay time of 9DPhen deduced from these measurements
(10 ns) is in good agreement with the the value obtained in the
single photon timing experiments (9.1 ns).
In the polar solvent acetonitrile, the charge-transfer absorption
band of 9DPhen is situated at 460 nm, which could be a sum
of the cation of the DMA group and the anion of the
phenanthrene group.92,93 This suggests that a polar electronic
structure is formed. The value of the decay time (5 ns) is in
good agreement with the value obtained from the analysis of
the fluorescence decay (7 ns), confirming that there was no
photoionization during these experiments.
If the transient absorption spectra of 9DPhen in acetonitrile
are investigated in more detail, a fast component of a few
picoseconds is distinguished, which on the basis of the time
resolution in SPT can be estimated to be less than 15 ps. Also
if the normalized difference spectra of 9DPhen in acetonitrile
are obtained after 3 and 100 ps delay times, this difference
spectrum is quite similar to the spectrum of 9DPhen in isooctane.
Thus, in acetonitrile, the spectrum even at an early stage (first
few picoseconds) contains the contribution of the less polar
electronic structure as observed in isooctane and then changes
to the more polar species.
Quantum Chemical Calculations on 9DPhen. Figure 11
shows the ground-state twist potential of 9DPhen (regarding
the dihedral angle of the X-ray structure C14-13-15-16). In
agreement with the X-ray structure where a value of 64.2° was
measured, the most probable twist angle æ is 62°, with a small
barrier to rotation for æ ) 90° (0.3 kcal/mol or 0.013 eV) and
Figure 11. S0 gas phase twist potential of 9DPhen using the AM1
method and full geometrical optimization at each twist angle. The
population distribution according to Boltzmann is also given for two
different temperatures.
a large one for planarity (20 kcal/mol or 0.87 eV). The
accessible angular range is thus 30-150° at room temperature,
with a sizable population at 90° (see Figure 11). For angles
deviating from 90°, the molecule relieves additional sterical
strain mainly by the following two geometrical deformations:
(1) bending of the whole molecule with respect to the linking
bond and, (2) loss of planarity of the phenanthrene moiety to a
boatlike structure. These effects are already clearly visible at
the minimum-energy geometry (æ ) 62°) and get stronger for
more planar geometries. They have an evident influence on
the excited-state properties and lead to a larger coupling between
the relevant excited states, which is discussed below (see Table
3 with a comparison of optimized and idealized geometries).
The challenging question for the quantum chemical calcula-
tions is to give some information on the nature of the lowest
excited singlet state which is observed experimentally connected
with an allowed and solvent polarity dependent emissive
transition. Figure 12 depicts the relevant molecular orbitals
(MO’s) of 9DPhen involved in the low-energy transitions at
90° and at the equilibrium twist angle between the electron donor
unit dimethylanilino (D) and the acceptor unit phenanthryl (A).
It can be seen that for 90° the orbitals are localized on the
subchromophores, and consequently, the transitions on one
subunit can be assigned to local transitions of phenanthrene (A)
and dimethylaniline (D). The local singlet transition of lowest
(92) Land, E. J.; Porter, G. Trans. Faraday Soc. 1963, 59, 2027.
(93) Shida, T.; Iwata, S. J. Am. Chem. Soc. 1973, 95, 3473.

Intramolecular Photoinduced Electron Transfer
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2899
Figure 12. Molecular orbitals (MO’s) of 9DPhen at perpendicular and
equilibrium geometries.
Figure 13. State energies S1-S4 in the gas phase (a) and the
corresponding oscillator strengths (b) calculated with CNDO-S param-
etrization. The energies calculated using the CNDUV parametrization
are similar with a maximum deviation of about 0.1 eV. The broken
lines indicate the angular range that is not populated in S0. The crosses
indicate the oscillator strengths for S4 calculated with idealized
geometries.
1
energy is the Lb(A) consisting of the one-electron transitions
1 f -2 and 3 f -1.The second local singlet transition is the
1La(A) connected with 1 f -1 and 3 f -2 configurations.
Transitions between the subunits are connected with a full
intramolecular charge transfer (1ICT (DA)), and the one of the
lowest energy consists of the 2 f -1 or HOMO(D) f LUMO-
(A) configuration. For deviations from 90°, the transitions are
linked with the same configurations but the orbitals become
delocalized over the subunits, especially the occupied MO’s 1,2
and 3 (Figure 12). As a result all of the transitions get
delocalized character and mix in some contributions of the
configuration 2 f -1. Nevertheless, they can be regarded as
keeping their substantial (spectroscopic) properties over the
whole angular range and their directions and they can be
the ground state, with the exception of S2 (1La(A) at 90°), which
is stabilized for planar conformations around 30°. There is a
common tendency for shifting the energetic minima to values
smaller than 60°.
The oscillator strengths (Figure 13b) indicate forbidden
character for the 1Lb states at all twist angles and for the charge-
1
transfer state ICT(DA) at the perpendicular geometry. For
deviations from 90°, the 1ICT(DA) state becomes allowed
through a better overlap of the MO’s involved. Similar to
perpendicular geometry’s, below 60°, the above-mentioned
geometric deformations also disturb the overlap of the MO’s
concerned and therefore lead to a decrease of the oscillator
strength. In contrast to the optimized geometry, the oscillator
strength for the idealized geometry (with subunit structures
without deformations) increases up to an angle of 20°. This
indicates the necessity of a good orbital overlap for the MO’s
involved in ICT to obtain high oscillator strengths of the S0 f
S4 transition. Moreover geometric deformations also have a
considerable influence on the mixing of the states through
configuration interaction (CI). For example, the contribution
of the HOMO(D) f LUMO(A) configuration to the ICT
transition S4 changes from about 53% for the idealized to about
27% for the optimized geometry at an equilibrium twist angle
accompanied by an increase of this contribution to S1 (1Lb(A))
from 19% to 23% (see Table 3).
1
1
correlated with the “pure” La, Lb, etc., states of D and A at
90°.
We want to express this by calling these states 1La-type, 1Lb-
type, etc., for angles deviating from 90°. Due to the orbital
and CI mixing, S1-S3 increase their dipole moments from 90°
to 62°, and that of ICT (S4) decreases, so that the dipole
moments of the four lowest singlet states become rather similar
at 62° (see below Figure 14a). The transition moments of the
1Lb states (S1 and S3) remain quite low, and the strongest angular
changes for f are found for S4 (ICT) (Figure 13b).
Figure 13a summarizes the energetic angular dependence of
the lowest singlet excited states in the gas phase as derived from
adding the excited-state energy difference, calculated within
CNDO/S, to the ground-state energy and using the appropriate
optimized geometry for every twist angle. Most of the excited
states are energetically flat in the region 30°-150°, similar to

2900 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Onkelinx et al.
Table 3. Contribution of the HOMO(D) f LUMO(A)
Configuration (%) to the First Four Excited Singlet States for the
Perpendicular and Equilibrium Geometrya
æ ) 90°
æ ) 62°
CNDUV CNDUV CNDO-S CNDUV CNDUV CNDO-S
state nature
ideal
opt
opt
ideal
opt
opt
S1 1Lb(A)
S2 1La(A)
S3 1Lb(D)
0
0
0.28
0
0.01
0.4
0.001
0.02
0.4
19.3
2.1
0.07
52.7
23
2.4
0.03
26.8
11.1
2.6
0.01
37.9
S4 1ICT(DA) 88.4
85.8
84.1
a Idealized and optimized geometries as well as two different
parametrizations are compared.
A complete CI expansion of the wave functions for the lowest
four states is given in the following equations.
CI expansion for S1 to S4 (contributions g 0.1) (from CNDO-
S/opt. calculation):
90°
S1 (1Lb(A)) ) -0.003ø(2-1) - 0.69ø(1-2) -
0.58ø(3-1) + 0.31ø(1-1) - 0.19ø(3-2) (10)
S2 (1La(A)) ) +0.013ø(2-1) - 0.84ø(1-1) +
0.37ø(3-2) - 0.30ø(1-2) - 0.20ø(3-1) (11)
S3
(1Lb(D)) ) +0.064ø(2-1) + 0.87ø(2-3)
(12)
S4 (ICT) ) +0.92ø(2-1) + 0.23ø(2-7) +
0.22ø(2-2) (13)
62°
Figure 14. Dipole moments of the various excited states as a function
of twist angle æ (a) and the derived solvation corrected excited state
energies in n-hexane (b) and acetonitrile (c).
S1 (1Lb(A)-type) ) +0.33ø(2-1) + 0.64ø(1-2) +
0.47ø(3-1) - 0.34ø(2-2) + 0.27ø(1-1) (14)
1
which constitutes a large part of the ICT (DA) state S4 with
S2 (1La(A)-type) ) +0.16ø(2-1) - 0.87ø(1-1) +
0.29ø(1-2) + 0.21ø(3-2) + 0.16ø(2-2) (15)
1
1
the La(A)-type (S2) or the Lb(D)-type (S3) states than with
the 1Lb(A)-type (S1) and furthermore with the 1Bb(A)-type (S6)
states, the transition moments of which are polarized in the short
axis of the phenanthrene unit and thus are polarized ap-
proximately in the same direction (RLb/ICT ) 30°) as the ICT
transition represented by the HOMO(D) f LUMO(A) config-
uration. On the other hand the transition moments of the
S3 (1Lb(D)-type) ) -0.01ø(2-1) + 0.66ø(1-3) +
0.41ø(2-3) - 0.37ø(3-3) + 0.14ø(1-1) (16)
S4 (ICT) ) -0.62ø(2-1) - 0.25ø(3-2) -
0.24ø(3-4) + 0.24ø(1-5) + 0.23ø(1-2) +
0.23ø(1-6) + 0.21ø(2-5) + 0.21ø(3-5) -
0.16ø(2-6) + 0.16ø(3-1) - 0.10ø(2-2) -
0.10ø(2-4) (17)
1
1
1La(A), Ba(A), and Lb(D) states are close to perpendicular to
that of 1ICT(DA) (RLa/ICT ) 60°). Hence, although there is no
symmetry in the molecule, in the case of the gas phase, there is
1
1
enhancement of CI mixing between the ICT(DA) and Lb(A)
states because of their roughly parallel transition moments.
To get an idea how solvation affects the various excited states,
dipole moments were calculated and the excited-state energies
derived for different solvents by adding the solvation energy
(eq 2) to the gas phase energies. The results, presented in Figure
14a, show that the excited-state dipole moments are medium
sized (<10 D) and tend to minimize for the perpendicular
In these equations, the ICT configuration (ø(2-1)) has been
included in all cases as the first member in the expansion. Note
that it does not contribute to S1-S3 at 90° but constitutes nearly
purely the S4 state whereas it contributes to most states to a
nonnegligible extent at 62°. A very strong mixing between a
large number of configurations is obvious at 62°.
A simplified discussion is possible by considering the percent
contribution for the various states for ø(2-1), the HOMO(D)
f LUMO (A) configuration, which is contained in Table 3.
The comparison of the contribution of the HOMO(D) f
LUMO(A) configuration to the various excited states (Table 3)
reveals that, at 90°, S4 is represented by this configuration
alonespure ICT corresponding to the classical TICT state,34,54
whereas for the equilibrium angle of 62°, it is distributed over
several states which can mix with the 1ICT(DA) state S4.
Moreover from Table 3 it follows that at 62° there is less
significant mixing of the HOMO(D) f LUMO(A) configuration
1
conformations, with the exception of the S4 state (pure ICT-
(DA) at 90°), which exhibits the opposite behavior. For angles
æ deviating strongly from 90°, this state is of low dipole moment
due to a loss of HOMO(D) f LUMO(A) contribution (Table
3) and due to the enhanced delocalization of the MO’s involved
(Figure 12). The large dipole moments for the more twisted
conformations lead to a strong energetic lowering for the ICT
state (Figure 14 b,c). As a consequence, in highly polar
solvents, a nearly pure ICT state with perpendicular conforma-
tion becomes the lowest excited state, whereas in weakly and
medium polar solvents, the energetic closeness should result in

Intramolecular Photoinduced Electron Transfer
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2901
an enhanced mixing of the 1ICT(DA) with the 1Lb(A) and
furthermore with the La(A) state.
evidence that the excited state emitting in high-polarity solvents
has a pronounced charge-transfer character.
1
The nonlinear dependence of the maximum of the emission
as a function of the solvent polarity and the solvent dependences
of kf and kf/ν˜3n3, are an indication that emission in all solvents
occurs from a fluorescent CT state with a solvent-dependent
electronic structure. In polar media the spectrum differs
markedly between the frozen and liquid solvents, which can be
an indication of a change in the conformational distribution after
excitation or can be explained by solvent relaxation alone.
According to the mixing of the wave functions on the one
hand and the conformational distribution on the other hand
which both contribute to the change of the electronic structure
with increasing solvent polarity, kf is first strongly increasing,
due to the narrowing of the energy difference of ICT and S1 or
S2, and then decreasing slightly again in highly polar solvents,
explainable by a larger contribution of more twisted conforma-
tions. For 9PhPhen, with no low-lying ICT state, kf is not
polarity sensitive.
The highly allowed CT emission properties, even in aceto-
nitrile, however, lead to the questioning of a classical TICT
model assuming a relaxed state of perpendicular geometry. On
the other hand, a simple geometry relaxation toward planarity
in polar solvents cannot totally explain the high dipole moments
and the observed variation of kf.
A more realistic model would involve rather broad angular
distributions with significant population of nonperpendicular
conformations possessing increased transition moments even if
the energetic minimum would be located at 90°.
The calculation indicates strong angular changes of f mainly
in the neighborhood of 90°, i.e. for the more twisted conforma-
tions.
At low twist angles there is no solvent-induced state mixing
possible.
The single exponential decay obtained by SPT experiments
and the rapid decay of the delocalized less polar state absorption
in acetonitrile suggests a solvent-assisted relaxation on a time
scale of a few picoseconds to a single polar fluorescent charge-
transfer state or to an ensemble of excited-state species with a
different angular distribution on a time scale that is faster than
the resolution of the experimental setup of SPT.
The relative mixing strength with these two LE states is
expected to be strongly polarity dependent. Such a behavior is
indicated by the curved nature of the solvatochromic plot (Figure
5) which yields a calculated dipole moments of 18 D in medium
and highly polar and 9 D in weakly polar solvents, respectively.
The latter value is in good agreement with the calculated dipole
moment of about 8 D of the lowest excited singlet state at
1
equilibrium geometry which is of Lb(A) nature. But the fact
that the oscillator strength of this state is strongly underestimated
indicates that the allowed ICT character is underestimated by
the calculation. It has to be kept in mind that the quantum
chemical calculations yielding the f values, are done for the
gas phase, not for the solvated phase which are the conditions
of the experiments. To treat the problem properly, a program
which includes the solvation energy in the Hamiltonian should
be used. Two sources of mixing which possibly are considered
insufficiently by the gas phase calculations may serve to describe
the experimental observations: (a) solvent-dependent config-
uration interaction (CI) between configurations with LE and ICT
character and (b) mixing of donor and acceptor MO’s yielding
delocalized orbitals.
As can be seen from Figure 12 the second occupied MO
HOMO(D) (MO 2 at 90°) strongly interacts with the neighboring
MO’s because they are very close in energy. The MO mixing
1
of the HOMO(D) to MO’s 1 and 3 involved in the Lb(A) as
well as the 1La(A) transition (Figure 12) gets enhanced for more
planar geometries diluting the localized character of S1 and
introducing ICT without necessitating CI.
In an attempt to evaluate this orbital coupling, we adjusted
the parametrizations such that the HOMO(D) became energeti-
cally situated above the HOMO(A) at 90° (compare Figure 12
where HOMO(D) is situated energetically below HOMO(A)).
The resulting mixed MO’s at 62° are then qualitatively different
from those in Figure 12, namely orbital 3 loses its donor
contribution whereas MO’s 1 and 2 both contain a large weight
of HOMO(D). Consequently, the 1 f -1 transition, which
contributes to the 1La(A) state, gets a much stronger ICT
character leading to a 1La(A)/ICT state that is energetically close
to S1 (1Lb(A)) and connected with an increased dipole moment
and oscillator strength. It cannot be excluded that this change
of orbital coupling, which we simulated here by a change of
parameters, may be induced also by differently polar solvents.
Already medium-polar solvents may then be expected to lead
to a level crossing of S1 (1Lb(A)) and S2 (1La(A)/ICT).
From the quantum chemical calculations and the comparison
with the experiments, the following conclusions can be drawn:
In the ground state, there is a significant population of
perpendicular twist angles and conformations below 20° are
unavailable. According to the gas phase calculations, the state
which is strongly angle and polarity dependent is the state which
is correlated to S4, the 1ICT(DA) state for twisted conformations.
Its involvement is confirmed experimentally by the solvato-
chromic experiments.
The principle of MO mixing can also explain the 1Lb character
of S1 in 9PhPhen. Here, the HOMO of the phenyl unit is
situated considerably below the two highest occupied MO’s of
phenanthrene. Thus the MO interaction is only very weak and
the lowest transition is of rather pure 1Lb(A) nature at all twist
angles. The increased influence of MO mixing in 9DPhen gets
obvious by the comparison of the calculated oscillator strengths
1
The first excited singlet state S1 is calculated as a Lb state
of phenanthrene but gains ICT character by two mixing
phenomena, i.e. orbital mixing and configuration interaction.
The mixed 1La(A)/ICT character of the calculated S2 state
strongly depends on the parametrization, and thus, polar solvents
may lead to a solvent-induced level crossing of S1 and S2. This
mixing model can also explain the curved solvatochromic plot
and the variation of kf with solvent polarity. In weakly polar
solvents 1ICT(DA) mixes into 1Lb(A) and/or 1La(A), leading to
the enhanced kf compared to that of phenanthrene 1Lb emission.
In polar solvents, CI mixing becomes stronger because of a
sizably reduced energy gap between the first two locally excited
1
for the Lb states. The relative weights for the gas phase are
f9DPhen:f9PhPhen:fPhen ) 6.6:3.3:1.
Conclusions
For 9DPhen, the spectral shift of the second absorption band
as a function of the solvent polarity indicates the interaction
between the dimethylanilino and the phenanthrene unit in the
ground state. This could correspond to the mixing (through
1
1
orbital overlap and CI) of the La and Lb states with the CT
state related to the formation of delocalized orbitals.
1
(LE) singlet states and ICT(DA), and furthermore, a solvent-
induced level reversal of S1 and S2 may take place.
The bathochromic shift of the emission spectra and the
resulting dipole moment value in polar solvents are clear
(94) Onkelinx, A.; De Schryver, F. C.; Iwai, K. To be published.

2902 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Onkelinx et al.
In the related molecule 9PhPhen, kf is not solvent dependent
and S1 retains its forbidden 1Lb(A) character as there is no low-
lying ICT state.
If, based on the quantum mechanical calculations, a popula-
tion distribution of different torsional angles is assumed which
is changed with the polarity of the solvent, the observed emission
can be understood as being due to more than one species but in
fast equilibrium and kf is a weighted average of the values of kf
for the different species. In this case it is more accurate to speak
of an average kf value kf ) Φf/ τ .
Acknowledgment. A.O. thanks the I.W.T. for financial
support. M.V.D.A. is an “Onderzoeksleider” of the Belgian
“Fonds voor Kollektief Fundamenteel Onderzoek”. The con-
tinuing support of the Belgian “Fonds voor Kollektief Funda-
menteel Onderzoek” and the DWTC through IUAP-II-16 and
UIAP-II-040 is gratefully acknowledged. We thank Prof. Th.
Bally (Fribourg) for the program MOPLOT.
The angular relaxation in the excited state most probably
occurs toward planarity in low-polarity surroundings because,
in this case, 9DPhen possesses no low-lying ICT similar to that
of 9PhPhen. However, in medium polar solvents, the ICT of
9DPhen becomes low lying, leading to an enforced coupling
between the LE states with the ICT(DA). As a consequence,
potentials with polarity-dependent minima may arise.
In highly polar solvents relaxation is, as suggested by
preliminary experiments on sterically hindered derivatives,
restricted to the neighborhood of 90° twist,94 which indeed
indicates a strong reduction of kf consistent with the calculated
low oscillator strengths for the perpendicular conformations.
JA953697A