Solvent-polarity tuning excited-state charge coupled proton-transfer reaction in p-N,N-ditolylaminosalicylaldehydes

Article abstract of DOI:10.1021/jp048415k

Detailed insight into the excitation behavior for charge versus proton transfer in p-N,N-ditolylaminosalicylaldehyde (Ia) has been gained via luminescence spectroscopy and femtosecond dynamics. In cyclohexane, following an ultrafast rate (～2.0 × 10¹²

Full text of DOI:10.1021/jp048415k

J. Phys. Chem. A 2004, 108, 6487-6498
6487
Solvent-Polarity Tuning Excited-State Charge Coupled Proton-Transfer Reaction in
p-N,N-Ditolylaminosalicylaldehydes
Pi-Tai Chou,* Wei-Shan Yu, Yi-Ming Cheng, Shih-Chieh Pu, Yueh-Chi Yu, Yu-Chung Lin,
Chien-Huang Huang, and Chao-Tsen Chen*
Department of Chemistry, National Taiwan UniVersity, Taipei, Taiwan, ROC
ReceiVed: April 10, 2004; In Final Form: May 31, 2004
Detailed insight into the excitation behavior for charge versus proton transfer in p-N,N-ditolylaminosalicyl-
aldehyde (Ia) has been gained via luminescence spectroscopy and femtosecond dynamics. In cyclohexane,
following an ultrafast rate (∼2.0 × 1012 s-1) of excited-state intramolecular proton transfer (ESIPT), fast
equilibrium takes place between normal (N*) and tautomer excited states (T*), resulting in dual fluorescence
maximized at 450 and 540 nm, respectively, with a common population decay rate of 360 ps-1. The normal
emission exhibits drastic solvent-polarity dependence and has been concluded to originate from a charge-
transfer species incorporating excited-state intramolecular charge transfer from ditolylamine to carbonyl oxygen.
In dipolar solvents, competitive rates between ESIPT and solvent relaxation were observed, and the solvated
charge-transfer state is thermodynamically more favorable, so that the T* f N* reverse proton transfer takes
places. Supplementary support was provided by the corresponding experiments for the methoxy derivative of
Ia as well as other relevant analogues. The results shed light on detailed proton/charge transfer coupled
dynamics as well as the associated solvent-relaxation dynamics at an early time domain.
1. Introduction
SCHEME 1: Energetics as a Function of Solvent
Polarizationa
Proton transfer1 represents one of the most fundamental
processes involved in chemical reactions as well as in living
systems.2 Vast numbers of research projects have been published
regarding various types of proton transfer, among which the
excited-state intramolecular proton transfer (ESIPT) has received
much attention.3 Numerous ESIPT molecules have been ex-
plored and investigated to shed light on their corresponding
spectroscopy and dynamics. Most ESIPT molecules possess
either six- or five-membered ring types of strong intramolecular
hydrogen bond between O-H (or N-H) and CdO (or pyridinic
nitrogen) groups, in which the intrinsic proton transfer in the
1ππ* state4 is essentially barrierless in nonpolar solvents5 and
may proceed either during the periods of low frequency, large-
amplitude vibrational motions associated with the hydrogen
bond6a or ballistically, as recently proposed by Lochbrunner.6b
In polar solvents, solvent static/dynamic perturbation may
play an important role in manipulating ESIPT pathways. On
one hand, for ESIPT molecules with a relatively weak intra-
molecular hydrogen bond, perturbation from solvent polarity,
especially protic environments, may modify the ESIPT dynamics
via the formation of a solute/solvent hydrogen-bonded complex.
As a result, ESIPT may either be prohibited within the excited-
state life span or proceed with a prerequisite of (protic) solvent
diffusive reorganization.7 On the other hand, considering the
possibly large differences in dipole moment (in terms of
orientation and/or magnitude) between excited normal (µ*N) and
tautomer species (µ*T), the normal and tautomer equilibrium
polarizations might be far separated. Since the large dipolar
change in solution is normally coupled to solvent polarization
effects, the relative energetics between normal and tautomer is
expected to be a function of solvent polarization coordinate.
a Considering the possibly large differences in dipole moment
between excited normal (µ*N) and tautomer species (µ*T), the normal
and tautomer equilibrium polarizations might be far separated. Ac-
cordingly, a solvent induced barrier (∆G+) may channel into the proton-
transfer reaction. Note that the orientation of the dipole moment for
both normal and tautomer species is arbitrarily picked.
Accordingly, as depicted in Scheme 1, the long-range polariza-
tion interactions may result in a solvent-induced barrier chan-
neling into the proton-transfer reaction.
* Address correspondence to these authors. E-mail: chop@ntu.edu.tw
and chenct@ntu.edu.tw.
10.1021/jp048415k CCC: $27.50 © 2004 American Chemical Society
Published on Web 07/03/2004

6488 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
SCHEME 2: Molecular Structures of Various
Compounds Studied
these goals, 2-hydroxy-4-(di-p-tolylamino)benzaldehyde (Ia, see
Scheme 2) was designed and synthesized. Ia is superior to the
aforementioned systems, i.e., I and II, based on several
advantages. First of all, unlike 3-hydroxyflavone derivatives (i.e.
system II), which are subject to the photolysis reaction on the
pyrone ring,12 Ia and corresponding analogues (Ib-d) are much
more photoresistant (vide infra). This is particularly important
in the fluorescence upconversion measurement, where the
degradation commonly takes place through multiphoton events,
resulting in experimental difficulties and/or artifacts. Second,
the existence of different hydrogen bonding conformers (e.g.
O-H- - -OdC and O-H- - -OCH3 sites) for system I in the
ground state complicates the spectral and dynamic assignments.
Conversely, the lack of a -OCH3 functional group ensures the
existence of only one HB (O-H- - -OdC) conformer in Ia,
simplifying both spectroscopic and dynamical approaches.
Moreover, in contrast to system I, introducing the di-p-tolyl
amino substituent to Ia pushes the spectral red shift of the S0
f S1 (ππ*) transition toward 380-420 nm, which is in the
optimal output range of our current femtosecond laser system.
The following sections are organized according to a sequence
of steps in which after the experimental expression, we first
performed detailed solvent-dependent absorption and emission
experiments to determine their steady-state spectral properties.
Subsequently, the dynamics of reaction/relaxation of Ia and its
corresponding derivatives are investigated and discussed on the
basis of femtosecond time-resolved studies under various solvent
polarities. Finally, a plausible proton/charge-transfer mechanism
is proposed to describe the overall reaction dynamics through
a combination of solvent polarity and proton-transfer reaction
coordinates.
However, cases regarding solvent dipolar relaxation coupled
ESIPT are rare because the dipolar change during ESIPT is, in
general, not as significant as that in the case of the electron
transfer, unless the Franck-Condon excited state undergoes
gigantic dipolar changes due to, for example, a charge-transfer
effect. In one approach, the design of such a system can be
accomplished by incorporating an electron-donating or -accept-
ing moiety into the chromophore, while the intramolecular
hydrogen bond remains intact, so that a electron/proton transfer
coupled reaction may take place in the excited state. Seminal
studies on molecules exhibiting remarkable solvent polarization
dependent ESIPT reaction dynamics should be ascribed to two
prototypical systems, namely p-N,N-(dimethylamino)methyl-
salicylate (I)8 and 4′-N,N-dialkylamino-3-hydroxyflavones (II)
(see Scheme 2).9-11 In polar, aprotic solvents, in contrast to a
dominant tautomer emission resulting from ultrafast ESIPT in
their corresponding parent molecules, i.e., methylsalicylate and
3-hydroxyflavone, multiple and dual emissions were observed
in systems I and II, respectively. For system I, steady state and
picosecond transient absorption measurements aimed at the
characterization of multiple emissions have been reported,8b
while detailed solvation coupled ESIPT dynamics have not yet
been explored. For system II, applying the Marcus electron-
transfer model and picosecond time-resolved study, Kelley and
co-workers10 were able to describe the overall reaction dynamics
through a combination of solvent polarity and proton-transfer
reaction coordinates, in which the proton-tunneling rate is
concluded to be relatively much slower than the rate of solvent
relaxation. Thus, a nonadiabatic type of proton transfer takes
place essentially after solvent equilibration is established in the
excited state.
2. Experimental Section
2.1. General Information. All reactions were performed
under nitrogen. Solvents were distilled from appropriate drying
agents prior to use. Commercially available reagents were used
without further purification unless otherwise stated. All reactions
were monitored by TLC with Merck precoated glass plates (0.20
mm with fluorescent indicator UV254) and were visualized with
UV light irradiation at 254 nm. Flash column chromatography
was carried out with use of silica gel from Merck (230-400
mesh). Mass spectra were obtained on a JEOL SX-102A
instrument operating in electron impact (EI) or fast atom
bombardment (FAB) mode. The 1H and 13C NMR spectra were
obtained on Bruker spectrometers operating at frequencies as
indicated for each compound. Chemical shifts were reported
relative to CDCl3 (δH 7.24) and at 77.0 ppm in CDCl3 [δC
(central line of t)]. FT-IR spectra were recorded on a Nicolet
magna-IR 550 series II. Melting points were uncorrected.
2.2. Materials. 9-Formyl-8-hydroxyjulolidine (Ic, Aldrich)
was twice recrystallized from CH2Cl2. The purity was checked
For both systems, limited by the temporal resolution, any
faster components of <10 ps were not resolvable in previous
studies. It is thus crucial to extend the investigation of relevant
systems toward the faster time domain to gain further insights
into the solvent polarity coupled ESIPT dynamics. Two key
issues to be explained include (1) the intrinsic ESIPT dynamics
for those charge/proton-transfer coupled systems in nonpolar
solvents such as cyclohexane, where the perturbation is mini-
mized, and (2) more rigorous investigation into the solvation
dynamics coupled ESIPT reaction, such as the differentiation
among solvent relaxation, proton-transfer dynamics, vibrational
cooling process, etc. on these systems. We believe that these
issues can be probed through femtosecond dynamic approaches
with the assistance of steady-state measurements. To achieve
by corresponding 1
H NMR and fluorescence excitation spectra.
(3-Methoxyphenyl)di-p-tolylamine (1). A mixture of 3-meth-
oxyaniline (1.24 g, 10 mmol), 4-iodotoluene (5.45 g, 25 mmol),
copper bronze powder (1.92 g, 30 mmol), 18-crown-6 (0.52 g,
2 mmol), anhydrous potassium carbonate (4.83 g, 35 mmol),
and 1,2-dichlorobenzene (30 mL) was stirred and refluxed for
96 h. After cooling, the insoluble inorganic material was filtered
off and rinsed with CH2Cl2 a few times. The combined filtrates
were concentrated under reduced pressure, and the resulting
residue was purified by flash column chromatography on silica
gel eluted with hexane/CH2Cl2 (10:1) to yield the desired
product as a blonde oil (2.33 g, 77%). TLC (CH2Cl2/hexane,

Proton Transfer in p-N,N-Ditolylaminosalicylaldehydes
J. Phys. Chem. A, Vol. 108, No. 31, 2004 6489
1
1:10, v/v) Rf 0.3; H NMR (400 MHz, CDCl3) δ 7.09 (t, J )
residue was then extracted with CH2Cl2. The combined organic
layers were dried with MgSO4 and filtered, and the solvent was
removed by a rotary evaporator. The residue was purified by
flash column chromatography eluted with EtOAc/hexane (1:10
v/v), yielding the desired product (1.9 g, 89%). 1H NMR
(CDCl3, 400 MHz) δ 9.98 (1 H, s), 7.32 (1 H, s), 3.8 (3H, s),
3.26 (t, J ) 7.6 Hz, 4H), 2.72 (4 H, m), 1.93 (qu, J ) 6.0 Hz,
4 H); FAB-MS m/z (rel intensity) 231 (M+, 100%).
8.1 Hz, 1H), 7.05 (d, J ) 8.5 Hz, 4H), 6.99 (d, J ) 8.5 Hz,
4H), 6.60-6.56 (m, 2H), 6.48 (dd, J ) 8.1,2.4 Hz, 1H), 3.69
(s, 3H), 2.30 (s, 6H); 13C NMR (100 MHz, CDCl3) δ 160.33,
149.55, 145.26, 132.47, 129.79, 129.55, 124.72, 115.21, 108.48,
106.90, 55.16, 20.77; FAB-MS m/z 303.2 (M+), FAB-HRMS
for C21H21NO calcd 303.1623, found 303.1631; FT-IR (neat)
2998.9, 2919.3, 1597.8, 1448.6 cm-1
.
2-Methoxy-4-(di-p-tolylamino)benzaldehyde (Ib). To a solu-
tion of POCl3 (0.3 mL, 3.3 mmol) in 2 mL of DMF, which had
been stirred for 30 min at 0 °C for 30 min, was added compound
1 (0.91 g, 3.0 mmol) dissolved in 5 mL of dry DMF dropwise
at 0 °C. After being stirred an hour at 0 °C, the reaction was
slowly warmed to room temperature and then heated to 70 °C
for 6 h. The reaction was cooled to room temperature and
quenched with water. The mixture was extracted with CH2Cl2.
Combined organic layers were dried over MgSO4, filtered, and
concentrated under reduced pressure. The residue was purified
by flash column chromatography on silica gel eluted with
EtOAc/hexane (1:10) to give a yellowish solid Ib (0.70 g, 70%).
2.3. Photospectroscopic Measurements. Steady-state ab-
sorption and emission spectra were recorded by a Hitachi (U-
3310) spectrophotometer and an Edinburgh (FS920) fluorimeter,
respectively. The various solvents were of spectragrade quality
(Merck Inc.) and were used right after being received. Benzene
and acetonitrile showed traces of fluorescence impurities and
were fractionally distilled prior to use.
Nanosecond lifetime studies were performed with an Edin-
burgh FL 900 photon-counting system with a hydrogen-filled/
or a nitrogen lamp as the excitation source. The emission decays
were analyzed by the sum of exponential functions, which
allows partial removal of the instrument time broadening and
consequently renders a temporal resolution of ∼200 ps. The
setup of picosecond dynamical measurements consists of a
femtosecond Ti-Sapphire oscillator (82 MHz, Spectra Phys-
ics).13 The fundamental train of pulses was pulse-selected (Neos,
model N17389) to reduce its repetition rate to typically 0.8-8
MHz, and then used to produce second harmonics (375-425
nm) as an excitation light source. A polarizer was placed in the
emission path to ensure that the polarization of the fluorescence
was set at the magic angle (54.7°) with respect to that of the
pump laser to eliminate the fluorescence anisotropy. An
Edinburgh OB 900-L time-correlated single photon counting
system was used as a detecting system, rendering a temporal
resolution of ∼15 ps. The fluorescence upconversion measure-
ments were performed with a femtosecond optically gated
system (FOG-100, CDP). The fundamental of a Ti:sapphire laser
(Spectra Physics) at 750-850 nm with an average power of
0.5 W and a repetition rate of 82 MHz was used to produce
second harmonics (SH) at 375-425 nm by focusing onto a 0.5
mm thick BBO type-I crystal. The SH was then separated from
the fundamental pulses with a dichroic mirror and used as pump
pulses. The pump pulses were focused onto a rotating cell, and
the optical pathlength was 1.0 mm. The resulting fluorescence
was collected by an achromatic lens and then focused on another
BBO type-I crystal (0.5 mm). The optical delayed remaining
fundamental pulses were also focused on the BBO crystal and
used as gate pulses for the sum-frequency generation. A Berek’s
variable waveplate was placed in the pump beam path to ensure
that the polarization of the pump laser was set at the magic
angle (54.7°) with respect to that of the probe laser to eliminate
the fluorescence anisotropy. The upconverted signal was then
separated by an F/4.9 (f ) 380 mm) single monochromator
(CDP2022) and detected via a photon counting PMT (R1527P,
Hamamatsu). The cross correlation between SH and the
fundamental had a full width at half-maximum (fwhm) of ∼150
fs, which was chosen as a response function of the system.
1
Mp 103-105 °C; TLC (EtOAc/hexane, 1:10, v/v) Rf 0.3; H
NMR (400 MHz, CDCl3) δ 10.17 (s, 1H), 7.60 (d, J ) 8.7 Hz,
1H), 7.12 (d, J ) 8.4 Hz, 4H), 7.05 (d, J ) 8.4 Hz, 4H), 6.43
(dd, J ) 8.7,2.0 Hz, 1H), 6.36 (d, J ) 2.0 Hz, 1H), 3.66 (s,
3H), 2.32 (s, 6H); 13C NMR (100 MHz, CDCl3) δ 187.67,
163.14, 155.14, 143.36, 135.04, 130.24, 129.84, 126.48, 117.46,
111.15, 100.42, 55.33, 20.93; FAB-MS m/z 331.1 (M+), FAB-
HRMS for C22H21NO2 calcd 331.1572, found 331.1569; FT-IR
(neat) 2860.5, 2765.9, 1672.8, 1617.8, 1338.8 cm-1. Anal. Calcd
for C22H21NO2: C, 79.73; H, 6.39; N, 4.23. Found: C, 79.69;
H, 6.48; N, 4.05.
2-Hydroxy-4-(di-p-tolylamino)benzaldehyde (Ia). A solution
of boron tribromide (0.2 mL, 2.1 mmol) in 5 mL of dry CH2-
Cl2 was added dropwise to a solution of compound Ib (0.33 g,
1.0 mmol) in 20 mL of dry CH2Cl2 at -78 °C over 20 min.
The solution was stirred at -78 °C for an additional 2 h and
then at room temperature. After the mixture was stirred for 20
h at room temperature, water was added carefully to consume
excess boron tribromide. The acidic aqueous solution was
extracted with CH2Cl2 (50 mL × 3), and the combined organic
layers were dried over MgSO4 and concentrated under reduced
pressure. The greenish residue was purified by flash column
chromatography eluted with EtOAc/hexane (1:10) to yield a
yellowish solid (0.24 g, 75%). Mp 133-135 °C; TLC (EtOAc/
1
hexane, 1:10) Rf 0.40 (EtOAc/hexane, 1:10); H NMR (400
MHz, CDCl3) δ 11.40 (s, 1H), 9.54 (s, 1H), 7.22 (d, J ) 8.7
Hz, 1H), 7.14 (d, J ) 8.4 Hz, 4H), 7.06 (d, J ) 8.4 Hz, 4H),
6.39 (dd, J ) 8.7, 2.2 Hz, 1H), 6.26 (d, J ) 2.2 Hz, 1H), 2.33
(s, 6H); 13C NMR (100 MHz, CDCl3) δ 192.89, 163.61, 155.71,
142.78, 135.68, 134.58, 130.35, 126.92, 113.86, 110.01, 103.71,
20.95; FAB-MS m/z 317.1 (M+), FAB-HRMS for C21H19NO2
calcd 317.1416, found 317.1385; FT-IR (neat) 3025.4 (br),
2919.3, 1631.3, 1609.4, 1498.9 cm-1. Anal. Calcd for C21H19-
NO2: C, 79.47; H, 6.03; N, 4.41. Found: C, 79.21; H, 6.02; N,
4.40.
9-Methoxy-2,3,6,7-tetrahydro-1H,5H-pyrido[3,2,1-ij]quino-
line-8-carbaldehyde (Id). To a suspension of NaH (60% in oil,
176 mg, 2.6 mmol) in THF (20 mL) was added slowly Ic (500
mg, 2.2 mmol) at 0 °C. After the mixture was stirred for 30
min, methyl iodide (369 mg, 2.6 mmol) under nitrogen (25 °C)
was added and the solution was subjected to heat at 60 °C for
12 h. The mixture was then cooled to room temperature followed
by the addition of water to consume unreacted NaH. The organic
solvent was removed via reduced pressure and the resulting
2.4. Computational Approaches. The Gaussian 98 pro-
gram14 was used to perform the ab initio calculation on the
molecular structure. Geometry optimizations for all structures
were carried out with the 6-31G(d′,p′) basis set at the Hartree-
Fock (HF) level. Hessians and hence vibrational frequencies
were also performed to check whether the optimized geometrical
structure for those molecules is at an energy minimum, transition
state, or higher order saddle point.

6490 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
Figure 1. Static absorption and fluorescence spectra of Ia in (A)
cyclohexane, (B) benzene, (C) dichloromethane, and (D) acetonitrile
at 298 K. Dashed curves express the spectra fit for Ia in cyclohexane.
Figure 2. Static absorption and fluorescence spectra of Ib in (A)
cyclohexane, (B) benzene, (C) dichloromethane, and (D) acetonitrile
at 298 K.
3. Results
emission maximized at ∼435 nm in cyclohexane, while the 530-
nm band apparently disappeared (see Figure 2A). The difference
in structures between Ia and Ib mainly lies in the lack of
intramolecular HB in Ib. Furthermore, 2HBA, a well-known
ESIPT model, exhibits a unique 520-nm proton-transfer tautomer
emission.15 Thus, it is reasonable to ascribe the 530-nm band
in Ia to the tautomer emission resulting from ESIPT. The 450-
nm emission was accordingly assigned to the normal (i.e.
nonproton-transfer) emission. Similarly, the entire emission of
Ia in benzene originating from the same Franck Condon excited
state is ascertained by the identical excitation spectra (not shown
here). However, since the emission full width at half-maximum
(fwhm) is wider than that of Ib, the existence of dual emissions
in Ia is also possible (see Figures 1B and 2B for comparison),
which unfortunately could not be resolved due to the strong
spectral overlap (vide infra).
3.1. Steady-State Approach. In cyclohexane, Ia exhibits the
lowest singlet S0 f S1 (ππ*) absorption band maximum at 365
nm (see Figure 1A, ꢀ365 ∼ 1.1104 M-1 cm-1). The spectral
feature resembles that of Ib (see Figure 2A), generally treated
as a nonintramolecular hydrogen bond (HB) and hence a non-
ESIPT model due to the lack of the O-H proton. In contrast to
2-hydroxybenzaldehyde (2HBA), in which the absorption
maximum could be bathochromically shifted to as large as 15
nm with respect to 2-methoxybenzaldehyde due to the intramo-
lecular HB enhancing π electrons delocalization,15 the spectral
shift on the S0 f S1 peak wavelength between Ia and Ib was
negligibly small (<3 nm, see Figures 1 and 2 for comparison).
Upon increasing the solvent polarity, a blue shift of the
absorption maximum was observed for Ia from 365 (cyclohex-
ane) to 358 nm (CH3CN, see Figure 1D). In view of the low
ionization energy for the di-p-tolyl amino substituents, these
results can plausibly be rationalized by the manifestation of the
S0 f S1 absorption band in Ia and Ib by a charge-transfer
character incorporating ditolylamine (electron donor) and car-
bonyl oxygen (electron acceptor). The dominant charge-transfer
property suppresses the π electrons delocalization induced by
HB, resulting in a negligible difference in absorption peak
wavelength between Ia and Ib. Franck-Condon excitation
causes significant dipolar changes on Ia (or Ib), and hence the
unfavorable solvated configuration in the excited state, rational-
izing the hypsochromic shifts of the absorption spectrum upon
increasing solvent polarity.
The emission of Ia in cyclohexane consists of dual bands
maximized at 450 and ∼530 nm (shoulder, see Figure 1A). The
ratio for the 450 nm versus 530 nm emission intensity reveals
concentration independence, eliminating the emission associated
with any high-order aggregation. The entire dual emission
originating from a common ground-state species is ascertained
by the same fluorescence excitation spectra throughout the
monitored wavelengths of 420-650 nm, which are also ef-
fectively identical with the absorption spectrum, indicating that
the entire emission results from a common Franck-Condon
excited state. In a comparative study, Ib exhibited a single
In contrast to the occurrence of ultrafast ESIPT in 2HBA,16
giving rise to an exclusive ∼520 nm tautomer emission, the
appearance of dual emissions for Ia in nonpolar solvents, in
which the normal fluorescence dominates in cyclohexane, is
remarkable. Because there are no other HB conformers existing
for Ia in the ground state, the steady state result simply implies
the existence of either an appreciable barrier or an excited-state
equilibrium associated with ESIPT. Further insights into the
reaction dynamics will be gained via the femtosecond fluores-
cence upconversion study.
As shown in Figure 1, in contrast to the appearance of dual
emissions in Ia in cyclohexane, the steady-state approach
revealed a broad, single-fluorescence band in all polar, aprotic
solvents studied. Furthermore, despite the blue shift in the
absorption profile, the fluorescence peak frequency exhibited
drastic solvent dependence, being decreased as the solvent
polarity increased. The results support the operation of excited-
state charge transfer, yielding a gigantic change in the dipole
moment. Perhaps the strongest support for this viewpoint is
given by the nonproton-transfer model Ib, which revealed
similar solvent-polarity dependence in the peak maximum, being
red shifted from 435 nm in cyclohexane to 580 nm in CH3CN
(see Figure 2). Further structural verification of the charge-

Proton Transfer in p-N,N-Ditolylaminosalicylaldehydes
J. Phys. Chem. A, Vol. 108, No. 31, 2004 6491
Figure 3. Static absorption and fluorescence spectra of Ic (s) and Id
(- - -) in (A) cyclohexane, (B) benzene, (C) dichloromethane, and (D)
acetonitrile at 298 K.
Figure 4. The plot of fluorescence maximum frequencies for (A) Ia,
(B) Ib, (C) Ic, and (D) Id as a function of solvent polarity parameter
f(ꢀ): O, data for cyclohexane, ethyl ether, ethyl acetate, dichloro-
methane, and acetonitrile; 2, data for benzene.
transfer emission is given by Ic and Id (see Scheme 2), in which
the nitrogen lone pair electrons are locked, to a certain extent,
in an unfavorable orientation so that the charge transfer
efficiency may be largely inhibited. Figure 3 depicts the
emission of Id in various solvents, in which the emission peak
is only slightly red shifted from 375 nm in cyclohexane to 400
nm in CH3CN, supporting the proposed charge-transfer mech-
anism incorporated diarylamino nitrogen in Ib. Conversely, as
shown in Figure 3, Ic revealed a large Stokes shifted, solvent-
polarity independent ∼520 nm emission band. Obviously, under
the diminution of excited-state charge transfer, ESIPT is
decoupled from the solvent perturbation. Accordingly, Ic is
expected to exhibit similar photophysical properties as 2HBA,
i.e., fast ESIPT reaction, resulting in a proton-transfer tautomer
emission. To simplify a further approach, the emission spectra
of Ib and Ic can thus be used to simulate the normal (i.e. charge
transfer) and tautomer emission profiles, respectively, of Ia. As
shown in the dotted-line curve of Figure 1, the overall emission
spectrum of Ia in cyclohexane is well fitted by the combination
of Ib and Ic, further supporting the dual emission properties
for Ia in cyclohexane.
vac
where v˜f and v˜f in eq I are the spectral position (in terms of
wavenumber) of the solvation equilibrated fluorescence maxima
and the value extrapolated to the diluted gas phase, respectively,
bµg and bµe are the dipole moment vectors of the ground and
excited states, and ∆f is the solvent polarity parameter function
and is generally expressed as ∆f ) (ꢀ - 1)/(2ꢀ + 1), where ꢀ
denotes the static dielectric constant of the solvent. The plot of
fluorescence peak frequency as a function of ∆f is shown in
Figure 4, parts A and B, for Ia and Ib, respectively. As predicted
by eq I, a linear relationship is found from ethyl ether to
acetonitrile, and slopes as steep as -2.44 104 and -2.78 104
cm-1 were obtained for Ia and Ib, respectively, consistent with
its assignment of the charge-transfer emission. a0 in eq I was
estimated to be 6.55 Å for both Ia and Ib via the Hartree-
Fock method with 6-31G(d′,p′) basis sets. Accordingly, the
change in dipole moment between ground and excited states
was deduced to be as large as 27 and 28 D for Ia and Ib,
respectively. Moreover, the proton-transfer tautomer emission
of Ic can be clearly distinguished by its rather small polarity
dependence in the Lippert’s plot (see Figure 4C).
The solvent-polarity dependent emission property can be
specified quantitatively according to the theory derived from
dielectric polarization, specifying that the spectral shifts of the
fluorescence upon increasing the solvent polarity depend on the
difference in permanent dipole moments between ground and
excited states. The magnitude of the excited-state dipole
moments can thus be estimated by a method incorporating the
fluorescence solvatochromic shift.17-21 If the dipole moments
of the solute are approximated by a point dipole in the center
of a spherical cavity with a radius a0, on the basis of small
solvent dependent absorption properties and negligence of the
solute polarizability, one obtains
3.2. Femtosecond Time-Resolved Spectroscopy. 3.2.1.
Femtosecond Dynamics of Nonproton-Transfer Cases. To make
a clear differentiation between charge transfer and proton
transfer, we would first like to present the relaxation dynamics
of the nonproton-transfer case Ib. The time-resolved fluores-
cence of Ib in cyclohexane is straightforward, and it was
established by a response-limited (<130 fs) rise and a single-
exponential decay component (τf ∼ 70 ps, see Figure 5) at any
monitored wavelengths. The fast decay rate correlates well with
its rather small quantum yield of ∼1.2 × 10-3 measured in
cyclohexane. Due to the appreciable triplet-triplet absorption
(λmax at 480 and 560 nm) and intensive phosphorescence in a
77 K methylcyclohexane glass (not shown here), it is plausible
that similar to o-methoxybenzaldehyde,3a the lowest singlet
2(bµe - bµg)2
vac
ν˜f ) ν˜f
-
∆f
(I)
3
1
hca0
excited state of Ib in cyclohexane is in an nπ* configuration.

6492 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
Figure 7. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ib in acetonitrile. The solid lines express the
corresponding best-fitted curves.
Figure 5. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ib in cyclohexane. The solid lines express the
corresponding best-fitted curves.
Figure 8. The temporal emission spectra of Ib in CH2Cl2 acquired at
a delay time of (b) 0.25, ([) 0.5, (]) 1.0, and (O) 3.0 ps. The dashed
line indicates the steady-state fluorescence. The solid lines express the
fitted curves. Insert: The plot of C(t) versus the delay time in CH2Cl2.
C(t) is defined in the text.
Figure 6. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ib in dichloromethane. The solid lines express
the corresponding best-fitted curves.
The proximity in energetics between Sππ* and Snπ* manifolds
may cause significant state mixing, resulting in the fluorescence
relaxation dynamics dominated by the rate of nπ* f ππ*
intersystem crossing.
the best-fitted time constants for the fast decay (or rise)
components vary by the probed emission wavelengths. Second,
as shown in Figure 8, the temporal spectral evolution of the
emission in, e.g., CH2Cl2 reveals a continuous spectral shift in
fluorescence peak maxima. If the large Stokes shift results from
the interconversion of two electronically excited states, i.e., a
locally excited and a charge transfer species, two distinct
fluorescence bands should appear during the time-dependent
spectral evolution, and the time constants should be independent
of the monitored wavelengths. We thus ascribe the early time
evolution of Ib in dipolar solvents to a continuously evolving
solvent relaxation rather than an interconversion between two
excited species. On the basis of the time-resolved Stokes shifted
model,22 the solvation relaxation time τs can be extracted from
the experimental function expressed as
1
3
Figures 6 and 7 show the time-resolved fluorescence spectra
of Ib in dichloromethane and acetonitrile, respectively. Similar
trends were observed in both solvents, in which the emission
dynamics are nonsingly exponential. Detailed fitting parameters
for both early-time relaxation dynamics and population decay
rate are listed in Table 1. Upon monitoring at shorter emission
wavelengths, the relaxation dynamics consist of an ultrafast rise
(<130 fs), a fast, resolvable (∼100 fs to a few ps) decay, and
rather long (>1 ns) decay components. The decay components
require at least two, and for some wavelengths even three,
exponential terms to achieve good convoluted fits for Ib. The
>1-ns decay component, measured by the pico-nanosecond
single photon counting technique, was resolved to be 3.7 and
1.1 ns in CH2Cl2 and CH3CN, respectively, and was then
attributed to the population decay (radiative plus nonradiative)
of the solvent-equilibrated S1 state. Upon increasing the
monitored wavelength, while the population decay remains
unchanged, the decay time of the early (fast) components
gradually increases, accompanied by a decrease in the magni-
tude, and finally becomes a rise component at the emission tail.
Two remarks can be made according to the data analyses. First,
V(t) - V(∞)
V(0) - V(∞)
C(t) )
(II)
where V(0), V(∞), and V(t) denote frequency maxima of the
upconverted fluorescence probed at t ) 0, ∞, and intermediate
times t, respectively. The plot of C(t) versus delay time in CH2-
Cl2 shown in the insert of Figure 8 reveals a dual exponential
decay-like profile, and a mean τs23 was estimated to be 0.75

Proton Transfer in p-N,N-Ditolylaminosalicylaldehydes
J. Phys. Chem. A, Vol. 108, No. 31, 2004 6493
TABLE 1: The Photophysical Properties of Ia and Ib in Various Solvents
solvent
CHE
emission (Φ)
450 nm (1.6 × 10-2
early dynamicsa
population decayb
0.36 (0.042c)
Ia
)
)
450 nm [τ1: 0.37 (0.123)]
488 nm [τ1: 0.35 (0.096)]
BEN
DCM
ACN
550 nm (1.1 × 10-1
575 nm (5.6 × 10-3
620 nm (1.0 × 10-3
460 nm [τ1: 0.31 (0.02); τ2: 3.2 (4.5 × 10-3)]
485 nm [τ1: 2.81(0.08); τ2: 12.1 (0.13)]
525 nm [τ1: 4.11 (6 × 10-3); τ2: 17.9 (0.015)]
575 nm [τ1: 0.17 (-0.27); τ2: 14.6 (8 × 10-3)]
650 nm [τ1: 0.32 (-6.9 × 10-3)]
2.0
)
)
500 nm [τ1: 0.41 (6 × 10-3); τ2: 1.7 (8.5 × 10-3)]
565 nm [τ1: 0.91 (4 × 10-3); τ2: 3.2 (6.5 × 10-3)]
680 nm [τ1: 0.18 (-1.1 × 10-3); τ2: 4.1 (9 × 10-3)]
720 nm [τ1: 0.34 (-0.037); τ2: 4.9 (9.5 × 10-3)]
750 nm [τ1: 0.48 (-6 × 10-3); τ2: 3.9 (2.9 × 10-3)]
480 nm [τ1: 0.21 (4 × 10-3)]
0.37
540 nm [τ1: 0.51 (2.5 × 10-3)]
620 nm [τ1: 0.92 (2.8 × 10-3); τ2: 3.3 (4.5 × 10-3)]
670 nm [τ1: 1.22 (1.5 × 10-3); τ2: 8.3 (4.4 × 10-4)]
720 nm [τ1: 1.55 (3.6 × 10-4); τ2: 11.7 (2.2 × 10-4)]
0.11
0.07
Ib
CHE
BEN
435 nm (1.2 × 10-3
490 nm (1.1 × 10-1
)
)
440 nm [τ1: 2.51 (0.093)]
460 nm [τ1: 3.9 (0.35)]
488 nm [τ1: 5.05 (0.14)]
510 nm [τ1: 1.65 (-0.2); τ2: 4.7 (0.16)]
560 nm [τ1: 2.11 (-0.18)]
5.2
3.7
1.1
650 nm [τ1: 2.47 (-0.042)]
DCM
ACN
555 nm (2.2 × 10-1
580 nm (4.2 × 10-2
)
)
480 nm [τ1: 0.76 (0.14); τ2: 3.1 (0.32)]
525 nm [τ1: 0.75 (0.11); τ2: 3.8 (0.25)]
550 nm [τ1: 0.23 (-0.086); τ2: 3.0 (0.19)]
625 nm [τ1: 0.59 (-0.098); τ2: 2.9 (0.11)]
675 nm [τ1: 0.85 (-0.02)]
500 nm [τ1: 0.39 (4.5 × 10-3); τ2: 1.54 (5.3 × 10-3)]
540 nm [τ1: 0.41 (1.4 × 10-3); τ2: 1.80 (4.7 × 10-3)]
585 nm [τ1: 0.45 (3.6 × 10-4); τ2: 1.98 (2.1 × 10-3)]
640 nm [τ1: 0.43 (-2.1 × 10-3); τ2: 2.08 (2.3 × 10-3)]
700 nm [τ1: 0.55 (-2.6 × 10-3); τ2: 2.34 (1.1 × 10-3)]
a In ps. Data listed in the early dynamics are the rise (negative preexponential factor) and short decay components. The much longer population
decay components are not shown in this column. Data in parentheses are the fitted preexponential factors. b In ns. c Preexponential factor at 450 nm.
ps. Similar methods were applied for Ib in CH3CN, and a τs of
0.42 ps was then extracted. Within the experimental error, the
deduced τs of 0.75 in CH2Cl2 and 0.42 ps in CH3CN is
respectively in agreement with the average solvent relaxation
time of 0.56 and 0.26 ps reported previously.23
It should be noted that the static spectroscopy and relaxation
dynamics of Ib in benzene are quite different from those in
cyclohexane. Despite the zero net dipole moment of benzene,
the trend of time-dependent fluorescence in this solvent is more
or less the same as those in dipolar solvents, consisting of
wavelength-dependent short decay (or rise) components and a
rather long population decay of 5.2 ns (see Figure 9). The τf of
5.2 ns with a corresponding quantum efficiency of 0.11 is
drastically different from those in cyclohexane (τf ) 70 ps,
Φf ∼ 1.2 × 10-3). The anomalous photophysical behavior in
benzene is also revealed in the Lippert’s plot (see Figure 4), in
which the peak frequency versus ∆f(ꢀ) deviates considerably
from others. The continuum dielectric models used in the
Lippert’s plot predict a similar emission peak frequency in both
cyclohexane and benzene. Such models rely solely on the infinite
wavelength dielectric properties of the solvent (ꢀ0 or ꢀ(w)) and
thus cannot distinguish between truly nonpolar solvents (e.g.
cyclohexane) and nondipolar solvents such as benzene. Benzene
contains significantly easily polarized π electrons, which lead
to large higher order multipole moments (quadrupole, octapole,
etc.).23,24 Thus, both the statics and dynamics of solvation in
benzene are not qualitatively different from those in dipolar
solvents. Equation I fails to convey usable information in the
case of nondipolar solvents. However, just as in the case of the
energetics, we expect the dynamics of solvation in nondipolar
solvents to involve similar reorientational/translational solvent
Figure 9. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ib in benzene. The solid lines express the
corresponding best-fitted curves.
motions as in dipolar solvents, even if they are less readily
modeled.24 At this stage, simple theories to accurately model
the solvation dynamics in benzene are not available. In addition,
since the relevant approach is not a core issue in point, details
will be elaborated elsewhere. Nevertheless, note that the fast
decay monitored at, e.g., 440 nm (see Figure 9) is fitted to be

6494 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
SCHEME 3: The Proposed Free Energy of the Solvated
Ib as a Function of the Solvent Polarizationa
Figure 10. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ia in cyclohexane. The solid lines express the
corresponding best-fitted curves.
a PN and P*N are the equilibrium polarization for the normal species
in the ground and excited states, respectively.
700 nm. Note that although much of the normal and tautomer
emissions overlap, according to the spectral fit in Figure 1, there
is negligible overlap between normal (e.g. 420 nm) and tautomer
(e.g. 680 nm) emissions, indicating that the identical population
lifetime cannot be solely explained by the spectral overlap. The
results lead us to conclude the establishment of an excited-state
equilibrium between normal (i.e., charge transfer) and proton-
transfer tautomer states, in which the rates of forward and
reverse ESIPT are much faster than that of the population decay
rate (i.e., k1 and k-1 . kN + kT, see the following scheme) for
both states. The ESIPT reaction scheme and its associated
relaxation dynamics can be expressed as follows.
2.5 ps, which is in the same magnitude as the average solvation
time of 2.1 ps.23
In summary, upon electronic excitation, the dipolar change
in Ib is quite large and can thus be treated as a charge transfer
process. The coupling between locally excited and charge
transfer states must be fairly strong so that the solvent
polarization induced barrier is negligible. Under the absence of
a reaction barrier, a common adiabatic solvation energetics of
Ib is depicted in Scheme 3, where the nonequilibrium free
energy (G) of the solute/solvent system for the ground state
(S0) and excited state (S1) of Ib can be plotted as a function of
an instantaneous solvent polarization P. Accordingly, the
observed relaxation dynamics is mainly governed by the
solvation process. Relative to cyclohexane, the much longer
decay, and hence higher quantum yield, of Ib in benzene and
dipolar solvents can be tentatively rationalized by the solvent
stabilization of the Sππ* manifold, leading to the interconversion
between Sππ* and Snπ* states so that the deactivation pathway
dominated by the Snπ* f Tππ* intersystem crossing is drastically
diminished.
3.2.2. Dynamics of Charge/Proton-Transfer Coupled Reac-
tion. In comparison with Ib, the relaxation dynamics of Ia are
relatively complicated due to the addition of one more reaction
channel, namely the ESIPT pathway demonstrated by the steady-
state emission measurements. In cyclohexane, the temporal
resolution at the blue side, e.g. 450 nm, of the fluorescence
(see Figure 10) consists of a response limited (<130 fs) rise
component, a fast but resolvable decay of 0.37 ps, followed by
a much slower decay that can be treated as an offset constant
within an acquisition duration of 20 ps. While the decay time
of 0.37 ps remains unchanged, the intensity ratio for the fast
versus the slower decay component at t ∼ 0 decreases with
increases in the monitored emission wavelengths. The 0.37 ps
decay component eventually disappears when the emission
wavelength is monitored at, e.g., >560 nm, leading to a response
limited rise and approximately constant decay components
within 20 ps. Interestingly, despite the resolution of dual
emission bands from the steady-state approach, the lifetime of
the slow decay component, i.e., the population decay, acquired
by the picosecond single photon counting method, was resolved
to be 360 ps throughout the monitored dual emission of 420-
[N*]0
[N*] )
{(λ2 - X)e-λ t + (X - λ1)e-λ t
k1[N*]0
}
(1)
(2)
1
2
λ2 - λ1
1
2
[T*] )
{e-λ t - e-λ t
}
λ2 - λ1
kN + kTKeq
1 + Keq
λ1 ≈
,
λ2 ) k1 + k-1
Y ) kT + k-1
X ) kN + k1,
where k1 and k-1 denote the forward and reverse rates of ESIPT,
respectively. Keq represents the equilibrium constant between
N* and T* species. Accordingly, upon instantaneous population
of N*, the excited normal species should undergo two decay
pathways with time constants of λ1 and λ2, while the overall
dynamics of the tautomer species should theoretically reveal
rise and decay components with rate constants of λ2 and λ1,
respectively. As shown in the dual spectral fit of the overall
emission (see Figure 1A), the time-resolved normal emission
monitored at <450 nm should be free from the interference of
the tautomer emission. Thus, the resolved fast decay component

Proton Transfer in p-N,N-Ditolylaminosalicylaldehydes
J. Phys. Chem. A, Vol. 108, No. 31, 2004 6495
Figure 12. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ia in dichloromethane. The solid lines express
the corresponding best-fitted curves.
Figure 11. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ia in benzene. The solid lines express the
corresponding best-fitted curves.
with a rate constant of 370 fs-1 can be ascribed to the λ2
()k1 + k-1), and the decay rate of 360 ps-1 corresponds to the
equilibrated population decay, i.e., λ1. The preexponential factor
for eq 1 can be further derived to
X - λ1
λ2 - λ1 k1 + k-1
k1
≈
(3)
(4)
λ2 - X k-1
λ2 - λ1 k1 + k-1
≈
Thus, the ratio for (3) versus (4) is equivalent to k1/k2 ) Keq,
which is also equal to the ratio of two exponential factors
extracted experimentally (see Table 1). Accordingly, Keq was
deduced to be 2.94. Since k1 + k-1 was measured to be 2.7 ×
1012 s-1, k1 and k-1 were then calculated to be 2.0 × 1012 and
6.8 × 1011 s-1, respectively.
Figure 13. Time-resolved sum frequency signal of fluorescence and
gate pulse (800 nm) for Ia in acetonitrile. The solid lines express the
corresponding best-fitted curves.
According to the precursor-successor type of kinetics
described above, the upconverted signal monitored at the
tautomer emission region should give rise to a rise and a decay
component. In contrast, however, only a population decay rate
constant λ1 of 0.36 ns could be resolved upon monitoring at
>550 nm. The results can be tentatively rationalized by the fact
that the tautomer emission spectrum overlaps considerably with
the normal species (see Figure 1A). As a result, the magnitude
of the rise component of the tautomer emission is largely
canceled out by the decay component of the normal emission.
We have also made attempts to extract the thermodynamics
between N* and T* species. In a temperature-dependent steady-
state approach for Ia in methylcyclohexane, the overall emission
intensity increased upon a decrease in the temperature. However,
the intensity ratio for the 530 nm versus 450 nm band remained
nearly unchanged from 330 to 160 K. Assuming a negligible
ESIPT barrier and a similar temperature-dependent nonradiative
decay rate, we thus tentatively propose an accidental degeneracy
between N* and T* for the case of Ia in cyclohexane. Due to
its charge-transfer property, any subsequent increase of the
solvent polarity should further stabilize the normal (i.e., charge-
transfer) state, resulting in a breakdown of the degeneracy
described as follows.
tailed fitting parameters for early-time relaxation dynamics and
population decay rate are listed in Table 1. Although both Ia
and Ib seem to reveal similar steady-state spectral features in
these solvents (see Figures 1 and 2 for comparison), drastically
different relaxation dynamics between Ia and Ib were resolved.
Because a similar trend in the differences was observed in these
solvents, the results shown in Figures 11-13 can be discussed
concurrently. In comparison with the relaxation dynamics of
Ib in the same solvent, two remarks can be promptly pointed
out for Ia. First of all, under identical experimental conditions,
i.e., the same excitation frequency, intensity, and detecting
parameters, the decay components of Ia monitored at short
wavelength regions are always faster than those measured in
Ib (see Figures 6, 7, and 9 for comparison). For example, the
fast decay component at 500 nm was fitted to be 410 fs for Ia
in CH2Cl2, which is nearly twice as fast as that of 760 fs
measured in Ib (see Table 1). Second, when monitoring at longer
wavelengths of >550 nm in, e.g., CH2Cl2, the rise dynamics in
Ib, found to be resolvable and correlated with the solvent
relaxation time, could no longer be resolved (i.e., <150 fs) in
the case of Ia. The results cannot be rationalized by a model
simply incorporating only solvent relaxation dynamics. Instead,
an additional deactivation pathway incorporating ESIPT in Ia
Figures 11-13 depict the time-resolved spectra of Ia in
benzene, dichloromethane, and acetonitrile, respectively. De-

6496 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
SCHEME 4: The Proposed Relaxation Dynamics of Ia
as a Function of Solvent Polaritiesa
Figure 14. The temporal spectral evolution of Ia in dichloromathane.
The solid lines express the corresponding fitted curves.
must play a key role for the observed dynamics in these solvents.
Support of this viewpoint is rendered by extracting the spectral
temporal evolution for Ia. As shown in Figure 14, the time-
dependent spectral evolution acquired at an earlier time domain
of <1.0 ps revealed two distinct bands maximized at 500 and
∼530 nm in benzene, followed by a continuously time-
dependent spectral red-shift toward a steady-state peak wave-
length of ∼560 nm in a period of ∼1 ps. The result is in sharp
contrast to the continuous, time-dependent spectral evolution
solely due to the solvent relaxation in Ib (see Figure 9 for
comparison), supporting the viewpoint in that a precursor-
successor ESIPT must be involved in the deactivation processes
for Ia.
a PT denotes the solvent polarity perturbed proton-transfer process.
P*N and P*T are the equilibrium polarization for N* and T*,
respectively. P+ represents the critical solvation configuration, in which
the N* and P* are in the same energy. Note that the ground-state
tautomer (T) is not shown in this scheme.
component of Ia, which is dominated by rates of ESIPT and
solvation process, is on the order of a few hundred femtosec-
onds, the rate of ESIPT must be resolvable. For a simplified
approach, the decay at a blue region of ∼480 nm was fitted to
be 780 and 410 fs for Ib and Ia, respectively, in CH2Cl2. Taking
the solvent relaxation rate for both Ia and Ib to be the same,
the rate of ESIPT is thus estimated to be ∼860 fs-1. The results
are in contrast to the system response limited rate of ESIPT in
the case of 2-hydroxylbenzaldehyde (2HBA),25 indicating that
the charge-transfer property in Ia leads to different distribution
of the charge density from that of 2HBA, so that the driving
force (i.e. photoacidity and basicity, etc.) for ESIPT is ap-
preciably reduced in Ia.
After reaching a critical solvation configuration, such as point
P+ depicted in Scheme 4, ESIPT becomes energetically uphill
due to the unfavorable solvent polarization, so the deactivation
subsequently should be dominated by the solvent relaxation,
finally achieving a solvent-stabilized, equilibrated charge transfer
state. As supported by the steady-state approach, in that the
charge-transfer emission is predominant in benzene and dipolar
solvents, the more stable one in the equilibrated excited state
should be ascribed to the charge-transfer manifold. Thus, ESIPT
from N* f T* after equilibrium population is energetically
prohibited. Conversely, T* f N* reverse proton transfer, in
which T* is populated originally from the early time domain,
is expected to be thermodynamically allowed. Because of the
different equilibrium polarization configuration between N* and
T*, the solvent polarity effect is expected to incorporate into
the ESIPT reaction pathway, resulting in a nonnegligible barrier.
If the coupling is small, the solvent-induced barrier in ESIPT
may be described semiquantitatively by the Marcus theory,26-29
commonly applied in the photoinduced electron transfer reaction,
which simply predicts that the solvent-induced barrier ∆G+ for
the T* f N* proton-transfer reaction is related to the reaction
exothermicity ∆G (see Scheme 1) and the solvent reorganization
4. Discussion
To view the above results on a fundamental basis, we have
performed theoretical approaches based on the 6-31G(d′,p′) basis
sets at a Hartree-Fock level. As a result, dipole moments for
the geometry optimized Ia normal and tautomer species in the
ground state were calculated to be 3.2 and 4.2 D, respectively,
and the angle between these two dipole vectors was less than
15°. Conversely, as supported by the nearly solvent-polarity
independent tautomer emission in Ic (see Figure 3), it is also
reasonable to assume a similar dipole moment in terms of both
magnitude and orientation between ground and excited states
of the tautomer species. Accordingly, we can securely conclude
that the excited proton-transfer tautomer possesses a similar
dipolar character with respect to that of the normal ground state.
The results, in combination with experimental data, lead us to
propose a mechanism incorporating competitive proton-transfer
versus solvent relaxation for Ia, depicted in Scheme 4.
Upon Franck-Condon excitation, Ia undergoes an instant,
gigantic dipolar change due to the charge redistribution.
Subsequently, the unfavorable polarization configuration is
subject to a rapid solvent relaxation/reorientation to reach an
energetically more favorable configuration. However, due to
the identical polarization environments in the normal ground
state and excited proton-transfer tautomer state, the solvent
polarization effect should be decoupled from the ESIPT pathway
at early times after the Franck-Condon excitation. If ESIPT is
energetically allowed, proton transfer free of the solvent-polarity
perturbation is expected. The rate of intrinsic ESIPT, which is
normally dominated by the proton tunneling process, may be
on the same magnitude as the rate of solvent relaxation, resulting
in a competitive deactivation pathway. Because the fast decay

Proton Transfer in p-N,N-Ditolylaminosalicylaldehydes
J. Phys. Chem. A, Vol. 108, No. 31, 2004 6497
energy λ, expressed as ∆G+ ) (λ/4)(1 + ∆G/λ)2. λ may be
electronvolt.20,21 In addition, a feature reflecting the absence of
bond breaking/making in outer-sphere electron transfer is
essentially present in the proton-transfer reaction. Recently,
detailed discussion of the modified theory suited to acid-base
ionization proton-transfer reaction in polar media has been
reported by Kiefer and Hynes.32,33 Since a proton-transfer
hypersurface is much more complicated in the excited state,
further details are not discussed in this study.
obtained from an Onsager cavity model described as
(µN - µT)2 ꢀ0 - 1
ꢀ∞ - 1
λ )
-
3
(
)
2ꢀ0 + 1 2ꢀ∞ + 1
a0
where ꢀ∞ is the high-frequency dielectric constant, which is
equivalent to the square of the refraction index n. ꢀ0 is the static
dielectric constant and can be experimentally calculated from
the Classius-Mossotti equation.30 Consequently, a three-
dimensional plot of free energy for ESIPT as a function of the
solvent polarization coordinate P and the proton-transfer reaction
coordinate qH at solvation coordinates of excited tautomer (P*T),
intermediate (P+), and normal (P*N) is depicted in Scheme 4.
The reaction coordinate qH may turn out to be complicated due
to numerous internal motions that may modulate the height and
width of the proton-transfer barrier. The population with energy
being equal to the solvent-induced barrier ∆G+ is given by the
Boltzman factor under the assumption that solvent relaxation
in the reactant well is quite fast. For proton transfer to occur,
this population must advance along the solvent coordinate P
and hop from the reactant to the product surface. The latter
process normally corresponds to tunneling through a barrier
along the proton motion coordinate qH. Two extreme limits can
be considered to treat the overall reaction dynamics. Consider
the first case, where tunneling through the qH coordinate at P
) P+ is assumed to be relatively slow. In this case, crossing
from the normal to the proton-transfer tautomer surface is quite
unlikely. Since the coupling between the reactant and product
coupling enters directly into the kinetic model, this expression
is essentially analogous to the nonadiabatic electron transfer.31
On the other hand, in the case where tunneling along the qH
coordinate is very fast, proton transfer from the normal to the
tautomer species along the excited-state potential hypersurface
occurs essentially every time the P+ solvent configuration is
achieved. In this case, the preexponential A factor is ap-
proximately equal to the inverse of the solvent longitudinal
relaxation time (τL-1) and is independent of the reactant/product
coupling. This expression is analogous to adiabatic electron
transfer, and the rate of ESIPT must be fast.
Finally, a large number of different mechanisms have been
proposed to account for the p-N,N-dialkylaminobenzaldehyde
types of molecules exhibiting anomalous charge-transfer emis-
sion.34 Among these models, the most widely accepted one is
based on the mechanism incorporating twisted intramolecular
charge transfer (TICT). One main debatable controversy for
TICT lies in whether these molecules twist or not during the
charge-transfer process.35,36 Unfortunately, the femtosecond
relaxation dynamics presented here could not resolve any
dynamics associated with the rotational motion of Ia. On one
hand, the electronic coupling between ditolyl amino (donor) and
formaldehyde (acceptor) through the π conjugation might be
very strong, such that the instant electron rearrangement, i.e.,
optical transfer, takes place. On the other hand, it is notable
that the TICT model incorporating a rigid and perpendicular
geometry is only an idealized limiting case, and the reality
involves broad angular distributions centered around perpen-
dicular twist angles.35 In addition, there is a reservation if the
electron donor subgroup is completely decoupled from the
acceptor subgroup in the ICT state. According to an ab initio
approach (6-31G(d′,p′) basis sets at a Hartree-Fock level), the
geometrically optimized structures for both Ia and Ib possess
a large twist angle of ∼40° in the ground state. Due to the
pretwisted angle in the ground state, Ia and Ib may undergo
ultrafast charge transfer, i.e., charge redistribution, in the excited
sate, of which the dynamics of the partially decoupled species
are not resolvable in this work due to the system response of
150 fs. Focus on determining the possible twist motion coupled
charge-transfer process is currently in progress.
5. Conclusion
Because no slow rise (or decay) dynamics of <1012 s-1 could
be resolved, it is more likely that the ESIPT dynamics for Ia
can be attributed to an adiabatic type of proton transfer, in which
the solvent-induced barrier must be negligibly small. However,
as indicated by Figures 11-13, because no rise dynamics
attributed to the solvent relaxation can be resolved, we cannot
rule out another possibility, that the lack of solvent-induced
dynamics is due to the strong overlap between charge-transfer
and proton-transfer emission. Furthermore, one has to be wary
of applying the Marcus model in interpreting the proton-transfer
reaction. The Marcus theory was originally derived for outer-
sphere electron-transfer reactions in solution, which might not
be a suitable model for dealing with the proton-transfer reaction.
Particularly, proton transfer simultaneously involves bond
breaking and making as well as strong electrostatic solute-
surrounding solvent interaction. Nevertheless, several research-
ers17,18 have stressed the importance of understanding how an
equation whose structure is based on electron-transfer theory
could be successful in the proton-transfer context, where the
assumptions of electron-transfer theory do not apply. For
example, electron-transfer theory generally assumes that the
electronic coupling between reactant and product states is small
(e.g., 1 kcal/mol or even less),19 whereas for the proton transfer,
a typical electronic coupling value might be on the order of an
In conclusion, we report on the excitation behaviors for charge
versus proton transfer for Ia and its related derivatives (Ib-c).
In cyclohexane, near isoenergetics are established between
charge-transfer (normal) and proton-transfer states for Ia. The
excited charge-transfer state, possessing gigantic dipole moment
change with respect to the ground state, is further stabilized
via a solvent relaxation process. In benzene and dipolar solvents,
the rates of ESIPT and solvent relaxation are competitive, and
the populated charge-transfer state is thermodynamically favor-
able. Once the proton-transfer tautomer is populated, an
adiabatic, fast excited-state reverse proton transfer takes place.
The solvent polarity induced barrier is concluded to be small,
such that the reaction rate is approximately equal to the inverse
of the solvent longitudinal relaxation time and is independent
of the reactant/product coupling. The results are in sharp contrast
to system II, based on 3-hydroxyflavone derivatives, in which
the intrinsic proton-tunneling rate has been proposed to be
relatively much slower than the solvent relaxation dynamics and
in which the solvent coupling effect strongly affects the ESIPT
dynamics.10 It is thus also intriguing to examine the associated
solvent polarity coupled ESIPT dynamics for 3-hydroxyflavone
based system II analogues in the early time domain. Focus on
this issue is currently in progress.

6498 J. Phys. Chem. A, Vol. 108, No. 31, 2004
Chou et al.
(12) For examples, see: (a) Studer, S. L.; Brewer, W. E.; Martinez, M.
L.; Chou, P.-T. J. Am. Chem. Soc. 1989, 111, 7643. (b) Chou, P.-T.; Studer,
S. L.; Martinez, M. L.; Orton, E.; Young, M. Photochem. Photobiol. 1991,
53, 587.
(13) Chou, P.-T.; Chen, Y.-C.; Yu, W.-S.; Chou, Y.-H.; Wei, C.-Y.;
Cheng, Y.-M. J. Phys. Chem. A 2001, 105, 1731.
(14) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.;
Head-Gordon, M.; Replogle, E. S.; Pople, J. A. GAUSSIAN 98, Revision
A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.
References and Notes
(1) In general, it is difficult to distinguish the mechanism between
proton transfer and hydrogen atom transfer reactions. In this review article
we simply adopted the conventional term namely “proton transfer”
throughout the text, which is particularly more suitable in describing the
solvent polarity coupled proton-transfer tautomerism.
(2) For example, see: Studies in Physical and Theoretical Chemistry;
Mu¨ller, A., Ratajack, H., Junge, W., Diemann, E., Eds.; Electron and Proton
Transfer in Chemistry and Biology, Vol. 78; Elsevier: Amsterdam, The
Netherlands, 1992.
(3) (a) Scheiner, S. J. Phys. Chem. A 2000, 104, 5898. (b) Waluk, J.
Conformational Aspects of Intra- and Intermolecular Excited-State Proton
Transfer. In Conformational Analysis of Molecules in Excited States; Waluk,
J., Ed.; Wiley-VCH: New York, 2000. (c) Chou, P. T. J. Chin. Chem. Soc.
2001, 48, 651. (d) Wu, K.-C.; Cheng, Y.-M.; Lin, Y.-S.; Yeh, Y.-S.; Pu,
S.-C.; Hu, Y.-H.; Yu, J.-K.; Chou, P.-T. Chem. Phys. Lett. 2004, 382, 203.
(e) Paterson, M. J.; Robb, M. A.; Blancafort, L.; DeBellis, A. D. J. Am.
Chem. Soc. 2004, 126, 2912. (f) Lochbrunner, S.; Wurzer, A. J.; Riedle, E.
J. Phys. Chem. A 2003, 107, 10580. (g) de Vivie-Riedle, R.; De Waele,
V.; Kurtz, L.; Riedle, E. J. Phys. Chem. A 2003, 107, 10591. (h) Shynkar,
V. V.; Mely, Y.; Duportail, G.; Piemont, E.; Klymchenko, A. S.;
Demchenko, A. P. J. Phys. Chem. A 2003, 107, 9522. (i) Cheng, C.-C.;
Chang, C.-P.; Yu, W.-S.; Hung, F.-T.; Liu, Y.-I.; Wu, G.-R.; Chou, P.-T.
J. Phys. Chem. A 2003, 107, 1459. (j) Lukeman, M.; Wan, P. J. Am. Chem.
Soc. 2003, 125, 1164.
(4) In some ESIPT molecules, the lowest excited singlet state is in an
nπ* configuration, in which ESIPT is prohibited. For a recent review, see
ref 3a.
(5) ESIPT incorporating phenol O-H to a â-carbon atom is not
included in this category, see: Lukeman, M.; Wan, P. J. Am. Chem. Soc.
2002, 124, 9458.
(6) For recent examples, see: (a) Chudoba, C.; Riedle, E.; Pfeiffer,
M.; Elsaesser, T. Chem. Phys. Lett. 1996, 263, 622. (b) Lochbrunner, S.;
Wurzer, A. J.; Riedle, E. J. Chem. Phys. 2000, 112, 10699. (c) Chou, P.-
T.; Chen, Y.-C.; Yu, W.-S.; Chou, Y.-H.; Wei, C.-Y.; Cheng, Y.-M. J.
Phys. Chem. A 2001, 105, 1731. (d) Ameer-Beg S.; Ormson, S. M.; Brown,
R. G.; Matousek, P.; Towrie, M.; Nibbering, E. T. J.; Foggi, P.; Neuwahl,
F. V. R. J. Phys. Chem. A 2001, 105, 3709. (e) Chou, P.-T.; Chen, Y.-C.;
Yu, W.-S.; Cheng, Y.-M. Chem. Phys. Lett. 2001, 340, 89.
(7) For examples, see: (a) McMorrow, D.; Kasha, M. J. Phys. Chem.
1984, 88, 2235. (b) Brucker, G. A.; Kelley, D. F. J. Phys. Chem. 1987, 91,
2856. (c) Brucker, G. A.; Kelley, D. F. J. Phys. Chem. 1987, 91, 2862. (d)
Strandjord, A. J. G.; Barbara, P. F. J. Phys. Chem. 1985, 89, 2355. (e)
Brucker, G. A.; Kelley, D. F.; Swinney, T. C. J. Phys. Chem. 1991, 95,
3190.
(15) (a) Nagaoka, S. I.; Hirota, N.; Sumitani, M.; Yoshihara, K. J. Am.
Chem. Soc. 1983, 105, 4220. (b) Toribio, F.; Catala´n, J.; Amat, F.; Acun˜a,
A. U. J. Phys. Chem. 1983, 87, 817.
(16) Stock, K.; Bizjak, T.; Lochbrunner, S. Chem. Phys. Lett. 2002, 354,
409.
(17) Lippert, E. Z. Naturforsch. 1955, 10a, 541.
(18) Mataga, N.; Kubota, T. Molecular Interactions and Electronic
Spectra; Marcel Dekker: New York, 1970; p 385.
(19) Bilot, L.; Kawski, A. Z. Naturforsch. 1962, 17a, 621.
(20) Liptay, W. Z. Naturforsch. 1965, 20a, 1441.
(21) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Plenum
Press: New York, 1983.
(22) (a) Nagarajan, V.; Brearley, A. M.; Kang, T. J.; Barbara, P. F. J.
Chem. Phys. 1987, 86, 3183. (b) Barbara, P. F.; Jarzeba, W. AdV. Photochem.
1990, 15, 1.
(23) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. J.
Phys. Chem. 1995, 99, 17311.
(24) Reynolds, L.; Gardecki, J. A.; Frankland, S. J. V.; Horng, M. L.;
Maroncelli, M. J. Phys. Chem. 1996, 100, 10337.
(25) Stock, K.; Bizjak, T.; Lochbrunner, S. Chem. Phys. Lett. 2002, 354,
409.
(26) Marcus, R. A. J. Chem. Phys. 1965, 43, 379.
(27) Marcus, R. A. Annu. ReV. Phys. Chem. 1964, 15, 155.
(28) Marcus, R. A. J. Chem. Phys. 1965, 43, 679.
(29) Marcus, R. A. ReV. Mod. Phys. 1993, 65, 599.
(30) Bordewijk, O. Theory of Electric Polarization; Elsevier: Amster-
dam, The Netherlands, 1978; Vol. 2.
(31) Calef, D. F.; Wolynes, P. G. J. Phys. Chem. 1983, 87, 3387.
(32) Kiefer, P. M.; Hynes, J. T. J. Phys. Chem. A 2002, 106, 1834.
(33) Kiefer, P. M.; Hynes, J. T. J. Phys. Chem. A 2002, 106, 1850.
(34) (a) Weisenborn, P. C. M.; Huizer, A. H.; Varma, C. A. G. O. J.
Chem. Soc., Faraday Trans. 1989, 85, 1895. (b) Blib, B.; Lommatzsch,
U.; Monte, C.; Rettig, W.; Brutschy, B. Chem. Phys. 2000, 254, 407.
(35) Rettig, W.; Zietz, B. Chem. Phys. Lett. 2000, 317, 187.
(36) Rappoport, D.; Furche, F. J. Am. Chem. Soc. 2004, 126, 1279.
(8) (a) Gormin, D.; Kasha, M. Chem. Phys. Lett. 1988, 153, 574. (b)
Heldt, J.; Gormin, D.; Kasha, M. Chem. Phys. 1989, 136, 321.
(9) (a) Chou, P.-T.; Martinez, M. L.; Clements, J. H. J. Phys. Chem.
1993, 97, 2618. (b) Chou, P.-T.; Martinez, M. L.; Clements, J. H. Chem.
Phys. Lett. 1993, 204, 395.
(10) (a) Swinney, T. C.; Kelley, D. F. J. Chem. Phys. 1993, 99, 211.
(b) Parsapour, F.; Kelley, D. F. J. Phys. Chem. 1996, 100, 2791.
(11) Ormson, S. M.; Brown, R. G.; Vollmer, F.; Rettig, W. J. Photochem.
Photobiol. A 1994, 81, 65.