Spin chemical approach towards long-lived charge-separated states generated by photoinduced intramolecular electron transfer in a donor-bridge-acceptor system

Article abstract of DOI:10.1246/bcsj.74.293

Photo-induced intramolecular electron transfer (ET) and subsequent back ET have been studied for phenothiazine-bridge-pyromellitdiimide (D-B-A) and phenothiazine-bridge-pyromellitdiimide-nitroxide radical (D-B-A-R·) in various solvents by nano-second transient absorption spectroscopy. Upon excitation of the phenothiazine (D) moiety in these compounds, ET from ¹D* to A took place at a rate of ca. 5 × 10⁹ s^-1 to give charge-separated (CS) states, ¹[D^·+-B-A^·-] or ²[D^·+-B-A^·--R·], in high quantum yields. The decay dynamics of the CS states was determined by the interplay between the spin conversion and the back ET to the ground state. The efficiency of the interconversion between states with different spin multiplicities was strongly affected by the presence of R· and external magnetic fields. A faster spin conversion of [D^·+-B-A^·--R·] resulted in a slower initial decay of the CS state generated from the excited singlet state of D compared with [D^·+-B-A^·-]. Such a spin effect manifested most markedly in dioxane. On the other hand, the back ET rate (k_BET) varied by more than two orders of magnitude with the solvent polarity. In dimethyl sulfoxide, the back ET was predicted to be almost barrierless, k_BET≥ 10⁹ s^-1, while in benzene, this process fell in the inverted region, k_BET = 6.5 ×10⁶ s^-1.

Full text of DOI:10.1246/bcsj.74.293

© 2001 The Chemical Society of Japan
293
Bull. Chem. Soc. Jpn., 74, 293–304 (2001)
Spin Chemical Approach towards Long-Lived Charge-Separated States
Generated by Photoinduced Intramolecular Electron Transfer in a
Donor–Bridge–Acceptor System
Yukie Mori, Yoshio Sakaguchi, and Hisaharu Hayashi*
Molecular Photochemistry Laboratory, RIKEN (The Institute of Physical and Chemical Research),
Hirosawa, Wako, Saitama 351-0198
(Received September 1, 2000)
Photo-induced intramolecular electron transfer (ET) and subsequent back ET have been studied for phenothiaz-
ine−bridge−pyromellitdiimide (D−B−A) and phenothiazine−bridge−pyromellitdiimide−nitroxide radical (D−B−A−R•)
in various solvents by nano-second transient absorption spectroscopy. Upon excitation of the phenothiazine (D) moiety in
these compounds, ET from D to A took place at a rate of ca. 5 × 109 s–1 to give charge-separated (CS) states,
1
*
1
+
–
2
+
–
•
•
•
•
[D −B−A ] or [D −B−A −R•], in high quantum yields. The decay dynamics of the CS states was determined by the
interplay between the spin conversion and the back ET to the ground state. The efficiency of the interconversion between
states with different spin multiplicities was strongly affected by the presence of R• and external magnetic fields. A faster
+
–
•
•
spin conversion of [D −B−A −R•] resulted in a slower initial decay of the CS state generated from the excited singlet
+
state of D compared with [D −B−A – ]. Such a spin effect manifested most markedly in dioxane. On the other hand, the
back ET rate (kBET) varied by more than two orders of magnitude with the solvent polarity. In dimethyl sulfoxide, the
back ET was predicted to be almost barrierless, kBET ≥ 109 s–1, while in benzene, this process fell in the inverted region,
kBET = 6.5 × 106 s–1.
•
•
Photo-induced electron transfer (ET) reactions have been
extensively investigated for a variety of donor(D)−acceptor(A)
dyads as well as D−D′−A or D−A−A′ triads. These studies
have been aimed at mimicking the highly efficient charge sepa-
ration in the natural photosynthetic reaction center (RC).1–3
Many efforts have also been made to elucidate the mecha-
nisms, dynamics, and factors governing the rates for ET reac-
tions in chemically and biologically important systems.4 In
most of the photo-induced ET systems reported so far, includ-
ing the RCs, the excited singlet states of the light-absorbers un-
dergo charge separation to give the radical ion pair states
(RIPs); therefore, the fast spin-allowed charge recombination
to give the ground states inevitably shortens the lifetimes of the
RIP intermediates. On the other hand, for the modified RCs5
and synthetic model compounds,6 the long-lived triplet RIPs
were characterized by time-resolved EPR studies. The charge
recombination rates of such long-lived RIPs could be reduced
not only by the large separation distance, but also by the spin
forbiddeness for the triplet states. Thus, spin evolution can play
an important role in a sequence of ET processes, such as charge
separation, charge shift, and charge recombination. For typical
RIPs consisting of two organic ion radicals, spin conversion
from the singlet to the triplet states occurs at a rate of, at most,
~108 s–1, while ET reactions can take place much faster.
cals and that the spin dynamics of such three-spin systems was
modulated by an external magnetic field.7,8 The variations in
the spin dynamics of the intermediates were well reflected in
their lifetimes and/or the product yields under appropriate con-
ditions. These results prompted us to study the photoinduced
ET of a donor (D)−bridge (B)−acceptor (A)−radical (R•) sys-
tem toward the formation of long-lived charge-separated (CS)
states by a new approach based on spin dynamics. Elucidation
of the role of an additional radical center is also important in
the field of spin chemistry, because such an unreactive radical
center has been utilized as an “observer” of the spin dynamics
of the radical pair (RP) in EPR studies,9 while an ability as a
“catalyst” for the spin conversion processes of RP has been
proposed.10,11
In the present study, we have investigated the reaction path-
ways and dynamics of photo-induced charge separation in
D−B−A−R• and of the charge recombination in the resultant
CS state by means of a nano-second laser flash technique in
solvents of various polarity. Chart 1 shows the compounds
(1−5) studied in this work. Here, we used phenothiazine and
1,2:4,5-pyromellitdiimide as D and A, respectively. The bridg-
ing part was a semi-rigid biphenyl-4,4′-bis(methylene) unit.
The stable radical R• used was 2,2,6,6-tetramethylpiperidin-1-
oxyl (TEMPO), which was directly linked to A (compound 3 in
Chart 1). To clarify the effect of R• on the decay dynamics of
the CS state, we also examined the corresponding two-spin
Recently, we have found that spin conversion processes in
RIPs and biradicals were accelerated by the presence of an un-
reactive radical in the proximity of one of the component radi-
+
–
•
•
system (D −B−A ), which was generated upon the excita-

294 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
to the electronic structure of the S1 state of the phenothiazine
chromophore. The D−B−A and D−B−A−R• compounds, 2 and
3 respectively, exhibited fluorescence spectra which were simi-
lar to each other, but their intensities were reduced to about
one-tenth or less that of 1, as shown in Table 1. Because asym-
metric diimide 2′ (Chart 1) could not be obtained in a pure state
(see Experimental), symmetric diimide 2 was used as the
D−B−A compound. The similarity in the fluorescence spectral
shape among 1−3 indicated that the two D−B parts in 2 showed
no significant interaction in the S1 state. The fluorescence
quenching observed for 2 and 3 suggested that intramolecular
1
*
ET took place from D to A to give the CS state, which was
confirmed by the nano-second laser flash photolysis.
Figure 1 shows the transient absorption spectra for a ben-
zene solution of 3 recorded at various delay times after excita-
tion with a 355 nm laser pulse. The intense peaks at 720 and
660 nm are assigned to absorption due to the radical anion of
–
the 1,2 : 4,5-pyromellitdiimide moiety (A ),7b while the weak-
•
er broad band around 530 nm is assigned to the radical cation
+
of the phenothiazine part (D ).12,13 These absorption bands ap-
•
peared instantaneously after excitation and decayed to the zero
level in ca. 2 µs with no change in the spectral shape. Upon the
excitation of 2 in toluene, a quite similar transient spectrum
was observed. The transient spectrum for 1 showed a broad
band at 470 nm, which was assigned to the absorption due to
the lowest triplet excited (T1) state of the phenothiazine part
3
*
( D ) from the reported T−T absorption spectrum of MPTZ
(λmax = 465 nm, εmax = 2.3 × 104 M–1 cm–1).12 In the transient
spectra for 2 and 3, no absorption band assignable to the T−T
a)
Table 1. Integrated Fluorescence Intensities (IFL
)
of 2–
4 Relative to That of 1 and the ET Rate (kET) in Various
Solvents at 293 K on Excitation at 355 nm
Chart 1.
Solvent IFL(2)/IFL(1) kET /109 s−1 b) IFL(3)/IFL(1) IFL(4)/IFL(1)
tion of 2. The major photochemical process was ET from the
Benzene
Dioxane
THF
0.105
0.096
0.097
4.3
4.7
4.7
0.092
0.072
0.074
0.86
1
*
lowest singlet excited state of the donor part ( D ) to A for both
c)
—
2 and 3. The presence of R• accelerated the conversion from
the initially populated doublet spin states to the chemically in-
active quartet ones, retarding the decay of the CS state through
back ET. The solvent polarity was also found to be an impor-
tant factor in determining the lifetimes of the CS states derived
from 2 and 3 through a large variation in the spin-allowed back
ET rate.
c)
—
a) Integrated over 390–600 nm. b) Calculated by Eq. 3. c) Not
measured.
Results
1. Photo-induced Charge Separation. To investigate the
1
*
intramolecular quenching of D by A, we measured the fluo-
rescence spectra of D−B (1), D−B−A (2), and D−B−A−R• (3).
Compound 1 showed broad fluorescence emission with an ill-
resolved vibronic structure around 440 nm in both polar and
non-polar solvents at 293 K. The spectral shape of 1 was quite
similar to that of the fluorescence (λmax = 447 nm in MeCN) of
10-methylphenothiazine (MPTZ). The fluorescence quantum
yield of 1 in MeCN was ca. 0.01, and the ratio to that of MPTZ
was 0.96. These results indicate that the (biphenyl-4-yl)methyl
substituent at the 10-position gives no significant perturbation
Fig. 1. Transient absorption spectra of 3 in benzene at
various delay times after excitation.

Y. Mori et al.
Bull. Chem. Soc. Jpn., 74, No. 2 (2001) 295
absorption was observed.
As can be seen in Table 1, the fluorescence yield of 3 was
somewhat lower than that of 2 in each solvent, suggesting that
there should be another process responsible for the fluores-
cence quenching of 3. It has been reported that nitroxide radi-
cals, such as TEMPO, can quench the excited singlet states
through energy transfer and/or enhanced intersystem crossing
(ISC).14 To examine the possibility of the fluorescence quench-
ing by R• in 3, we also measured the relative fluorescence in-
tensity for D−B−R• model compound 4,15 in which the diimide
moiety does not act as an electron acceptor. The fluorescence
intensity of 4 was somewhat lower than that of 1. This partial
quenching is likely to be due to excitation energy transfer via
the Förster mechanism, because the fluorescence spectrum of
1
*
D is largely overlapped with the absorption band of TEMPO
(λmax = 470 nm in benzene). These results indicated that the
1
*
D in 3 is quenched through energy transfer to R• as well as
ET. The reaction pathways for 2 and 3 are represented in
Schemes 1a and 1b, respectively.
2. Effects of Spin Multiplicity and Magnetic Fields on
Decay Dynamics of the CS States in Dioxane. To investi-
gate the effects of the presence of R• on the decay dynamics of
the CS states, we measured the time profiles of the transient ab-
sorbance, A(t) curves, due to the CS states by irradiating of 2
and 3 with a 355-nm laser pulse in several solvents. Although
the decay dynamics of the CS states derived from 2 and 3 were
different from each other in either solvent, the most prominent
spin effect was observed in dioxane. Figure 2(a) shows the A(t)
curves monitored at 650 nm for 3 in dioxane under various
magnetic fields (B’s). The A(t) curve under zero field can be fit-
ted to a mono-exponential decay with a rate constant of 6.87 ×
106 s–1. Because this observed rate constant corresponds to a
quarter of the kBET value (See Section 2a in Discussion), kBET in
dioxane was determined to be 2.74 × 107 s–1.
Fig. 2. A(t) curves monitored at 650 nm for 3 on exci-
tation at 355 nm under various magnetic fields (a) in
dioxane and (b) in pentyl acetate.
dioxane under conditions similar to those for Fig. 2a .The A(t)
curve for 2 under zero field was strongly biphasic. The decay of
Figure 3a shows the A(t) curves monitored at 715 nm for 2 in
Scheme 1. Reaction pathways for (a) 2 and (b) 3 on excitation of the phenothiazine chromophore. The k’s denote the rate
constants for the processes: kgr, radiative and non-radiative decay to the ground state; kISC (kIꢀSC), intersystem crossing from
∗
∗
∗
∗
∗
1D to 3D ; kEnT, excitation energy transfer from 1D to R•; kET, ET from 1D to A; kETT, ET from 3D to A; kST, spin conversion
from 1CS to 3CS; kTS, spin conversion from 3CS to 1CS; kBET, spin-allowed back ET reaction from A•− to D•+ within the CS
state.

296 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
decay. At an early stage (0 < t < 200 ns), the A(t) value de-
creased slightly with increasing B, while at later times (t > 200
ns) it marginally increased with increasing B. More distinctive
+
–
•
•
MFEs on the decay of [D −B−A −R•] were observed in pen-
tyl acetate (PenOAc), as shown in Fig. 2b. In this solvent, the
A(t) curve under zero field consisted of a major fast decaying
component and a minor slow decaying one. The former com-
ponent decayed much faster than the A(t) curve observed in di-
oxane. With increasing B, the A(t) value at an early time region
(t < 50 ns) decreased while that at later times (t > 80 ns) in-
creased. This tendency of the MFEs observed in PenOAc was
the same as that observed in dioxane (Fig. 2a), but the magni-
tudes of the MFEs in PenOAc were larger than those in diox-
ane. The A(t) curves in PenOAc under different B’s intersected
at a much earlier time than those in dioxane. The magnetically
induced changes in PenOAc were monotonous from 0 to 0.2 T,
but no further change was observed in the field range of
0.2−1.7 T.
The decay of [D −B−A – ] in dioxane exhibited complicat-
ed MFEs in the field range of 0−0.2 T, as shown in Fig. 3a. The
A(t) curves under different B’s intersect at 400−600 ns, indicat-
ing that the observed MFEs before this intersection have an op-
posite tendency to those after the intersection. To represent this
peculiar feature of MFEs, we integrated the A(t) curves in time
windows of 10−300 and 450−3350 ns and plotted the relative
integration values, I10(B)/I10(0) and I450(B)/I450(0), against B in
Fig. 3b. As shown in this figure, the I10(B) value steeply in-
creased with increasing B from 0 to 7.5 mT, decreased from 7.5
mT to 0.2 T, and finally slightly decreased from 0.2 to 1.7 T. On
the other hand, the I450(B) value showed opposite B-depen-
dence to the I10(B) one. Similar biphasic decay kinetics and
MFEs were reported for the CS states derived from fixed-dis-
tance triads consisting of porphyrins and an electron acceptor
in THF solution.16 These MFEs can be explained by combining
the level-crossing mechanism (LCM)17 and the relaxation
mechanism (RM)18 for singlet-born radical pairs, as discussed
later (See Section 2c in Discussion).
+
•
•
Fig. 3. (a) A(t) curves monitored at 715 nm for 2 on exci-
tation at 355 nm under various magnetic fields in diox-
ane. The broken line shows the normalized A(t) curve
for 3 in dioxane under zero field. (b) Ratios of the inte-
grated absorbance in the presence and absence of each
magnetic field in the time windows of 10–300 ns (cir-
cle) and 450–3350 ns (triangle) after excitation.
the shorter-lived component was much faster than that of the
+
–
•
•
CS state derived from 3, [D −B−A −R•] (broken line in Fig.
3a). This A(t) curve for 2 can be fitted to a double exponential
function with rate constants of 3.27 × 107 s–1 (k1) and 1.72 × 106
3. Solvent Effects on the Decay of the CS States. To in-
vestigate solvent effects on the kBET value, we measured the
A(t) curves for 3 in solvents of different polarity. Figure 4
shows the normalized A(t) curves observed for 3 under zero
1
+
–
s–1 (k2). As shown in Scheme 1a, [D −B−A ] (|S〉) decays
•
•
3
+
–
•
•
+
–
•
•
through the back ET and spin conversion to [D −B−A ]
(|T〉). Under zero field, each of the three sublevels of |T〉 is con-
verted to |S〉 at a rate of kTS, and the reverse conversion from
|T〉 to |S〉 takes place at a rate of kST = 3kTS. Under such condi-
tions, k1 and k2 are related to kTS and kBET by
field. The lifetime of [D −B−A −R•] decreased along with an
increase in the solvent polarity. The A(t) curves observed in tol-
uene and benzene under zero field exponentially decayed with
rate constants smaller than that observed in dioxane. The ob-
tained kBET values are listed in Table 2. In the presence of B’s
below 1.7 T, the A(t) curves in toluene and benzene exhibited
the same type of small deviation from a mono-exponential de-
cay as in dioxane.
2
k1,2 = 1/2[4kISC + kBET (16kTS2 + 4kTS
kBET + kBET
)
1/2]. (1)
The solution of Eq. 1 gave kTS and kBET values of 2.2 × 106 and
2.56 × 107 s–1, respectively. The obtained kBET value is in fairly
good agreement with the value (2.74 × 107 s–1) determined for
In THF, MeCN, and DMSO,19 the A(t) curves showed a bi-
phasic nature under zero field, and the initial decay was faster
than that observed in PenOAc. No difference in the initial de-
cay rate was detected among the THF, MeCN, and DMSO so-
lutions. However, when each of the sample solutions with the
same optical density at 355 nm was excited under the same
conditions, the maximal A(t) value, Amax, at 715 nm in MeCN
+
–
•
•
[D −B−A −R•].
Due to the presence of R•, MFEs on the decay of [D −B−
+
•
–
+
–
•
•
•
A −R•] were also different from those of [D −B−A ]. As
can be seen in Fig. 2a, the A(t) curves observed for 3 under B ≥
0.05 T exhibited a small deviation from a mono-exponential

Y. Mori et al.
Bull. Chem. Soc. Jpn., 74, No. 2 (2001) 297
or DMSO was lower than that observed in THF (in Fig. 4, each
of the A(t) curves was normalized relative to its Amax). This dif-
ference in the Amax value suggests that the decay through back
2
+
–
•
•
ET of [D −B−A −R•] in MeCN and DMSO should be faster
than that in THF. Thus, the kBET value was found to increase in
the order benzene < toluene < dioxane << PenOAc < THF <
MeCN ~ DMSO, as the solvent polarity increases.
+
–
•
•
The decay dynamics of [D −B−A ] also depended on the
solvent polarity. Figure 5a shows the A(t) curves upon the exci-
tation of 2 in toluene under various B’s. Under zero field, the
+
–
•
•
decay rate was larger and smaller than that of [D −B−A −R•]
(broken line in Fig. 5a) in early (t ≤ 400 ns) and later (t ≥ 600
ns) time regions, respectively. This feature is similar to that ob-
served in dioxane (Fig. 3a), although the overall decay of
+
–
•
•
[D −B−A ] in toluene was much slower than that observed in
dioxane. The A(t) curves observed in THF under zero field also
showed biphasic decay, as shown in Fig. 5b. In DMSO, the ini-
tial decay part was too fast to observe with our nano-second ap-
paratus. Thus, the decay rate at early times (t < 100 ns) in-
creased in the order toluene < dioxane < THF < DMSO. The
acceleration of the initial decay in polar solvents reflects the in-
crease in the kBET value with increasing solvent polarity, as was
+
–
•
•
the case for [D −B−A −R•].
Fig. 5. A(t) curves for 2 under various magnetic fields
in (a) toluene and (b) THF. The broken line shows the
A650(t) curve for 3 in toluene under zero field.
As can be seen in Fig. 5a, the A(t) value in the monitored
time window (t < 1800 ns) for the toluene solution of 2 de-
creased from 0 to 0.01 T, increased from 0.01 to 0.35 T, and
slightly decreased from 0.35 to 1.7 T. This B-dependence is
quite similar to that observed for the A(t) curves in the time
window of 10−300 ns in dioxane (Fig. 3b), and explained by
the same mechanisms as the MFEs observed in dioxane. On the
other hand, a different type of MFE was observed in THF. As
shown in Fig. 5b, the decay rate in microsecond time region
markedly decreased with an increase in B from 0 to 0.1 T. With
a further increase in B, although no change in the A(t) curves
Fig. 4. A(t) curves for 3 under zero field in (A) toluene,
(B) dioxane, (C) pentyl acetate, and (D) THF. The mon-
itored wavelength is 650 nm for (A) and (B) or 715 nm
for (C) and (D). Each of the A(t) curves are normalized
with respect to the maximal value.
Table 2. Properties of Solvents,a) Free Energy Change (∆GBET), Total Reorganization Energy (λ), and Rate Constant (kBET) of
Back ET for the CS State Derived from 3 at 293 K
Solvent
εr
1/n2 − 1/εr
∆GBETb)/eV
λb)/eV
k
BET/s−1
Benzene
Toluene
Dioxane
Pentyl acetate
THF
2.2825
2.379
2.2189
4.79
7.52
47.24
36.64
0.0057
0.0259
0.0436
0.2994
0.3720
0.4358
0.5262
(−2.3)c)
N.A.e)
(−2.15)c)
(−2.0)
−1.81
(0.7)d)
N.A.e)
(0.85)f
(1.1)
1.3
6.5 × 106
7.4 × 106
2.7 × 107 [2.6 × 107]g)
≥ 108
≥ 108
≥ 109
DMSO
MeCN
−1.51
1.5
1.7
≥ 109
−1.52
a) Ref. 40. b) Values in parentheses are approximate. c) See text. d) λsol is assumed to be 0.4 eV. e) Not available. f) λsol is assumed to be
0.55 eV. g) Obtained for 2.

298 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
was observed from 0.1 to 0.5 T, the decay became slightly fast-
er from 0.5 to 1.7 T. In DMSO, the A(t) curve showed MFEs
similar to those observed in THF. These MFEs can be ex-
tems whose structures are believed to be rather rigid.25 Al-
though their calculations were carried out for ET reactions in
the gas phase, such structural deformation may be induced to
plained by the RM for triplet-born radical pairs. This result in-
some extent in solvents with low dielectric constants. Thus, the
3
1
*
•
*
dicated that D −B−A also underwent intramolecular ET to
intramolecular ET from D to A is exothermic by at least 0.28
3
+
–
•
give [D −B−A ] in polar solvents, such as THF and DMSO.
eV or more in all of the solvents examined.
1
*
No MFE due to the LCM was observed in THF or DMSO.
The unquenched D mainly undergoes ISC to the T1 state,
because the quantum yield of formation of the T1 state was re-
ported to be close to unity for MPTZ.12 A possible explanation
Discussion
1. Photo-induced Charge Separation. (a) Energy Lev-
for the lack of the T–T absorption band in the transient spectra
3
*
els of the CS States. As demonstrated by fluorescence and
of 2 and 3 is that D may rapidly decay through ET. The free-
1
3
*
*
transient absorption studies, the intramolecular ET from D to
energy change of ET from D to A in polar solvents can be es-
A in 2 and 3 took place sufficiently fast to compete with ISC to
timated from Eq. 2 by using the triplet excitation energy (ET)
reported for MPTZ (2.64 eV12) instead of ES. The calculated
values are –0.83, –1.13, and –1.12 eV in THF, DMSO, and
MeCN, respectively, suggesting that ET in the triplet manifold
can take place rapidly in these solvents.
3
*
D
in various solvents. The free-energy change (∆GET) ac-
companying this ET process can be estimated by Eq. 2.20
∆GET = (EODX − ERAED
)
MeCN − e2/(4πε0εrR)
+ e2/(8πε0)(1/rD + 1/rA)(1/εr − 1/εMeCN) − ES (2a)
(b) ET Rate Constants. The experimental rate constant
1
*
(kET) of ET from D to A for 2 was obtained from the relative
= −∆GBET − ES
(2b)
intensities of the fluorescence by
Here, EDOX and EARED are the oxidation potential of D and the re-
duction potential of A, respectively, measured in MeCN. Be-
cause the solubility in MeCN of 2 and 3 was too low for elec-
trochemical studies, EDOX = +0.345 V and EARED = –1.21 V (vs.
ferrocene) were used, which were measured for 1 and 5, re-
spectively. εr and εMeCN are the relative permittivities of the sol-
kET = τ0(1)−1[(IFL(1)/IFL(2)) − 1],
(3)
where IFL(1) and IFL(2) represent the integrated fluorescence
intensities for 1 and 2, respectively. τ0(1) is the lifetime of the
S1 state of 1, which is assumed to be the same as that of MPTZ
(2 ns in ethyl acetate,26 MeCN,26 and MeCN−water23). As
shown in Table 1, the kET values were similar to each other in
benzene, dioxane, and THF.
vent used for the ET reaction and of MeCN, respectively. The
+
ionic radii of D and A – are denoted by rD and rA, respective-
ly, where rD = rA = 3.7 Å were used. R is the center-to-center
distance between D+ and A–, which was assumed to be 12 Å on
the basis of the PM321-optimized structure for one of the stable
conformers in the neutral ground state.22 As the singlet excita-
tion energy of D (ES), the reported value for MPTZ (2.9 eV23)
•
•
According to the Marcus theory,27 kET for nonadiabatic ET is
expressed by
2
kET = 2π|V| (π/λkBT)1/2 exp [−(∆GET + λ)2/4λkBT], (4)
where V is the matrix element of electronic coupling,28 and λ is
the total reorganization energy, which consists of the contribu-
tions from solvent mode (λsol) and internal vibrations (λvib). Al-
though the exact value of ∆GET is unknown for each solvent, it
probably decreases in the order of benzene > dioxane > THF.
On the other hand, λsol increases in the order of benzene < diox-
ane < THF with an increase in polarity.27 As a result, the
Franck−Condon factors in Eq. 4 may be similar to each other
for these solvents, which accounts for the small solvent depen-
dence of kET. A similar insensitivity to the solvent polarity of
the ET rate was reported for the photo-induced intramolecular
charge separation in some D−B−A systems.24
was used. ∆GBET represents the free-energy change accompa-
–
nying back ET from A to D + within the CS state to give the
ground state. The ∆GET values are estimated to be –0.28, –0.24,
–1.09, –1.39, and –1.38 eV in benzene, dioxane, THF, DMSO,
and MeCN, respectively. However, Eq. 2 probably underesti-
mates the driving force for charge separation in benzene and
dioxane.
•
•
According to Warman et al.,24 the energy levels of the CS
states for donor–acceptor compounds separated by σ-bonds
were lower by 0.4 and 0.56 eV in benzene and dioxane, respec-
tively, than in saturated hydrocarbon solvents such as decalin.
They attributed this stabilization to quadrupolar interactions
between the solvent molecules and charged centers. Although
the quite low solubility of 2 and 3 to saturated hydrocarbons
precluded us from investigating in such solvents, the CS states
derived from these compounds are likely to be stabilized in
benzene and dioxane by the same type of solvent−solute inter-
actions. In addition, the R value may be smaller than that esti-
mated for the parent neutral species to cause a larger Coulom-
(c) Efficiencies of the Charge Separation.
Based on
1
+
–
•
•
Scheme 1a, the quantum yields of formation of [D −B−A ]
3
+
–
•
•
and [D −B−A ] from 2 are expressed by Eqs. 5a and 5b, re-
spectively:
ΦCSS(2) = kET/(kgr + kISC + kET) = kET/(τ0−1 + kET),
(5a)
ΦCTS(2) = kETT/(τT−1 + kETT) × kISC/(τ0−1 + kET
≤ kISC/(τ0−1 + kET) ≈ 1 − ΦCSS(2),
)
+
–
•
•
bic attraction between D and A . With respect to this point,
Shephard and Paddon-Row predicted based on the ab initio
calculations that the molecular structures can be significantly
deformed upon the charge-separation process in D–B–A sys-
(5b)
–1
–1
1
*
where τ0 and τT are the decay rates of D −B−A and
3
*
D −B−A to the ground state, respectively. With the kET values

Y. Mori et al.
Bull. Chem. Soc. Jpn., 74, No. 2 (2001) 299
listed in Table 1 and an assumption that τ0–1 should be the same
as that of MPTZ (5 × 108 s–1), ΦCSS(2) was estimated to be 0.9 in
benzene, dioxane, and THF. Because τT–1 (~ 105 s–1)32 is proba-
bly much lower than kTET, the ΦTCS(2) value could be close to 1–
ΦCSS(2), which equals to 0.1. This upper limit of ΦTCS(2) was still
bination of the LCM at B ~ 7.5 mT and the RM in the higher
field region. As shown in Fig. 6a, the |S〉 and |T〉 states are ener-
getically separated by |2J| / gβ (= ca. 7.5 mT) under zero field.
Owing to this energy gap larger than the B1/2 value (= 2.8 mT),
the efficiency of the HFC-induced |S〉 → |T〉 conversion under
zero field is lower than that of a biradical with |2J| / gβ = 0 mT.
As a result, most of [D −B−A ] born in the |S〉 state decays
through back ET. With increasing B, the energy of the |T+1〉 (or
|T–1〉) sublevel is raised (or lowered). Under B = |2J| / gβ, where
the energy level of |T+1〉 (or |T–1〉) matches that of |S〉, HFC-in-
duced |S〉 → |T+1〉 (or |S〉 → |T–1〉) conversion takes place in
competition with back ET (Fig. 6b). Due to this enhancement
of the |S〉 → |T〉 conversion, the apparent decay at the early
stage (t < 300 ns) under B = 7.5 mT is retarded compared with
that under zero field, as can be seen in Fig. 3a. Under fields
much smaller than ΦCSS(2).
1
+
–
•
•
*
In the case of 3, the quantum yield of ET from D to A is
represented by
Φ
CS(3) = kET/(kgr + kIꢀSC + kEnT + kET) = kET × τ(3),
(6)
2
1
*
where τ(3) is the lifetime of [ D −B−A−R•], which can be
evaluated as τ0 × IFL(3) / IFL(1). Assuming that the kET value of 3
is the same as that of 2 in each solvent, we obtained ΦCS(3) val-
ues of 0.8, 0.7, and 0.7 in benzene, dioxane, and THF, respec-
2
1
*
tively. Thus, the major process in the decay of [ D −B−A−R•]
was the ET reaction with minor contribution from the energy
transfer. This is consistent with the result that the fluorescence
higher than 7.5 mT, the |T 〉 sublevels decayed only through
1
the |T 〉 → |S〉 and |T 〉 → |T0〉 relaxation, and the rates of
1
1
these relaxation processes are much smaller than the |S〉 ↔ |T0〉
quenching for 4 was much less effective than that observed for
3
conversion rate and kBET (Fig. 6c). The populations of the |T 〉
*
1
2 or 3, as shown in Table 1. ET from D to A may also take
+
–
•
•
sublevels through the |S〉 → |T 〉 relaxation may become negli-
1
place to give [D −B−A −R•] in either the doublet (|D〉) or
gibly small, but these sublevels can be populated through ET of
quartet (|Q〉) sublevels depending on the total spin multiplicity
3
3
of the precursor, 4,2[ D −B−A−R•] (Scheme 1b). It was report-
*
the triplet precursor, D −B−A (Scheme 1a). The observed
*
MFE on I450(B) in a field region of 7.5 mT << B ≤ 1.7 T, which
ed that the T1 state of MPTZ was intermolecularly quenched by
TEMPO through energy transfer (T−D quenching).6 This
quenching process requires a close contact of the triplet excited
molecule and the radical.33 In the case of 3, the contribution of
reflects the B-dependence of the decay rate of the |T 〉 sublev-
1
els, can be explained by the RM. The |S〉 ↔ |T0〉 interconver-
sion rate somewhat increases with increasing B owing to the
difference in the isotropic g-factors between D + (g = 2.00527b)
•
intramolecular T−D quenching is negligible because of the
–
and A (g = 2.00417b) (∆g mechanism18a). However, this en-
•
3
*
large separation distance between D and R•.
hancement is not so significant because the magnitude of this
interaction (∆gB ) is 1.9 mT under B = 1.7 T, which is smaller
than |2J| / gβ and B1/2.
2. Spin Conversion Processes of the CS States. (a) Sin-
+
glet–Triplet Conversion of [D –B–A – ]. First, we discuss
•
•
the |S〉 ↔ |T〉 conversion of the two-spin intermediate,
+
–
•
•
+
–
•
•
(b) Spin Conversion of [D –B–A –R•] under Zero
[D −B−A ], and its B-dependence, which can be explained
by the well-established mechanisms of MFEs for radical pairs.
From an analysis of the A(t) curve in dioxane under zero field,
+
–
•
•
Field. The three-spin intermediate, [D −B−A −R•], can ex-
ist in either of four quartet (|Q〉) and four doublet (|D〉, |D′〉)
+
–
•
•
spin sublevels. In this expression, the |D′〉 states have the total
the |T〉 → |S〉 conversion rate (kTS ) of [D −B−A ] was deter-
mined to be 2.2 × 106 s–1 (See Section 2 in Results). If the |S〉
+
spin of singlet with respect to D −A – pair, while the |D〉 states
•
•
have the triplet one.10f The energy levels of the |D〉, |D′〉 and |Q〉
states are determined by J(D , A ), J(D , R•), and J(A ,
R•), where J(i, j) is the exchange interaction between the radi-
cal centers, i and j, under zero field.10 Upon ET of the doublet
precursor [ D −B−A−R•], the |D′〉 states are populated as
shown in Scheme 1b. A mono-exponential decay of
+
–
•
•
and |T〉 states of [D −B−A ] are degenerate, the HFC-induced
|S〉 ↔ |T〉 conversion can occur at a rate of kHFC, which is esti-
mated to be 1.2 × 108 s–1 by Eq. 7;34
+
–
+
–
•
•
•
•
kHFC = πgβB1/2/h,
(7a)
2
1
*
2
2
B1/2 = 2[|AHFC(D•+)| + |AHFC(A•−)| ]
+
–
•
•
[D −B−A −R•] observed in dioxane (Fig. 2a) can be ex-
plained by one of the three cases: Case (i) where all the eight
sublevels (|D′〉, |D〉, and |Q〉) are in a thermal equilibrium, Case
(ii) where the |D′〉 and |D〉 states are equilibrated with no popu-
lation of the |Q〉 states, or Case (iii) where only the |D′〉 states
were populated. According to Buchachenko et al., the |D′〉 ↔
|D〉 conversion is induced by the difference in the exchange in-
/[|AHFC(D•+)| + |AHFC(A•−)|].
(7b)
The effective HFC interactions for D and A – were obtained
by the relation |AHFC| = Σj[Ij(Ij + 1)aj2]1/2, where Ij and aj are the
spin quantum number and the isotropic HFC constant of the j-
+
•
•
th nucleus. Here, the reported aj values of MPTZ + and the rad-
•
ical anion of N,N′-dihexyl-1,2 : 4,5-pyromellitdiimide13,16
+
–
teraction, ∆J(D , A ) ≡ |J(D , R•) – J(A , R•)|.10 To obtain
experimental evidence for this spin catalysis theory, Turro et al.
investigated whether the spin conversion of RPs was affected
by the presence of the TEMPO radical, which was introduced
in the reaction systems either intermolecularly or intramolecu-
larly (i.e. covalently connected to one of the component radi-
cals of the RPs).10a,b,e They observed an enhancement of the
+
–
+
–
•
•
•
•
•
•
were used for D andA , respectively. The observed kTS value
+
–
•
•
of [D −B−A ] was much smaller than the calculated one.
This reduction of the spin conversion rate is attributed to its ex-
change interaction, the absolute value (|J|) of which is larger
than the magnitude of the effective HFC interactions.
+
–
•
•
The MFEs observed for the decay of [D −B−A ] in diox-
ane (Fig. 3) and in toluene (Fig. 5a) can be explained by a com-

300 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
Fig. 6. Energy diagrams and interconversion routes of the spin sublevels for the CS state derived from 2 in the cases of (A)
positive J or (B) negative J. The bold (dotted) arrows present processes faster (slower) than those presented by the solid thin
arrows.
radical coupling within the RPs by the presence of the TEMPO
radical in non-viscous solvents, such as benzene and dioxane,
where the lifetimes of the RPs were probably short (≤ 10–9 s).
(PI–R•, see Chart 1) in 2-propanol (2-PrOH).7b The observed
MFEs on the free-ion yields from the three-spin intermediate,
4,2
+
–
•
•
[MPTZ PI −R•], were explained by the relaxation mecha-
nism for the |D〉 ↔ |Q〉 conversion.36 We suggested that the |D〉
+
–
•
•
In the present case, where |J(D , R•)| << |J(A , R•)|, a large
+
–
+
–
•
•
•
•
∆J(D , A ) can cause efficient |D′〉 ↔ |D〉 conversion of
↔ |Q〉 conversion of [MPTZ PI −R•] should be induced by
+
–
[D −B−A −R•] within its lifetime, which is governed by the
kBET value (2.74 × 107 s–1). From these facts, it can be consid-
ered that thermal equilibrium is established between the |D′〉
mainly a dipolar interaction between PI – and R•. The magni-
•
•
•
–
•
tude of this dipolar interaction (|DZFS|) of PI −R• was estimat-
ed to be 4.35 mT using a point-dipole approximation with a
+
–
–
•
•
•
and |D〉 states during the decay of [D −B−A −R•].
distance of 8.6 Å between the center of PI moiety and the
midpoint of the N−O bond in R•.7b The observation of large
MFEs indicates that this interaction can effectively induce the
|D〉 ↔ |Q〉 conversion within the lifetime of the ion-pair in 2-
PrOH (~10–8 s).
It is expected that information on the |D〉 ↔ |Q〉 conversion
process can be obtained from MFEs on the decay dynamics of
+
–
•
•
[D −B−A −R•], because the efficiency of the |D〉 ↔ |Q〉 con-
version, but not of the |D′〉 ↔ |D〉 one, is affected by external
magnetic fields.35,36 As shown in Fig. 2, the decay of
–
+
–
•
•
•
It is likely that the A −R• part in [D −B−A −R•] has al-
+
–
–
•
•
•
[D −B−A −R•] at an early time region in dioxane and Pe-
nOAc increased with increasing B. The observed MFEs indi-
cate that the |D〉 states were efficiently converted to the |Q〉 ones
under zero field and that this |D〉 → |Q〉 conversion was partial-
ly inhibited in the presence of B’s. The mechanism of the
MFEs on |D〉 ↔ |Q〉 conversion is discussed below (in Section
2(c)) in more detail. From the consideration on the |D′〉 ↔ |D〉
and |D〉 ↔ |Q〉 conversion processes, the mono-exponential de-
cay of the A(t) curve observed in dioxane under zero field can
be interpreted as first-order decay of an equilibrium mixture of
all the eight sublevels, that is Case (i). In such a situation, the
observed decay rate corresponds to a quarter of the spin-al-
lowed back ET rate (kBET), because the singlet character with
most the same |DZFS| value as PI −R•. Therefore, the MFEs
observed for 3 can be explained by the same mechanism as the
previous case.7 Although the estimated |DZFS| value (4.35 mT)
+
–
•
•
of [D −B−A −R•] is only 1.6-fold of the calculated B1/2 value
+
–
•
•
(2.8 mT) of [D −B−A ], the observed differences in the de-
cay dynamics indicate that the |D〉 ↔ |Q〉 conversion of
+
–
•
•
[D −B−A −R•] should be much faster than the |S〉 ↔ |T〉 con-
version of [D −B−A ] under zero field, as was mentioned
above. The |D〉−|Q〉 energy gap of [D −B−A −R•] can be ei-
ther larger or smaller than the |S〉−|T〉 gap of [D −B−A ], de-
pending on the signs and magnitudes of the three J (i, j) val-
ues.10 These J values, however, could not be obtained in the
present study because no MFE due to the LCM was observed.
+
–
•
•
+
–
•
•
+
–
•
•
+
–
+
–
+
–
•
•
•
•
•
•
respect to D −A pair is one-fourth for each sublevel of
The J(D , A ) value in [D −B−A −R•] is likely to be simi-
+ – +
+
–
•
•
•
•
•
[D −B−A −R•].
lar to that in [D −B−A ], whereas |J(D , R•)| is negligibly
+
–
+
•
•
•
(c)MFEsonthe|D〉↔|Q〉Conversionof[D –B–A –R•].
Previously we studied photo-induced ET reaction of MPTZ
with N-hexyl-N′-(TEMPO-4-yl)-1,2 : 4,5-pyromellitdiimide
small due to the large separation of D and R•. Although the
+
–
•
•
|D〉−|Q〉 energy gap of [D −B−A −R•] could not be obtained
–
•
because of the lack of the knowledge on the J(A , R•) value,

Y. Mori et al.
Bull. Chem. Soc. Jpn., 74, No. 2 (2001) 301
the occurrence of efficient |D〉 ↔ |Q〉 conversion under zero
field suggests that this energy gap may be smaller than the
|S〉−|T〉 gap of [D −B−A ].
from Eq. 4. Although the decrease in λ causes an increase in
the pre-exponential factor, the decrease in the exponential term
is much more dominant. For some other D−B−A compounds, a
large solvent dependence of the back ET rate was also report-
ed.24
+
–
•
•
In Fig. 2b, the decay observed at the early time region,
which mainly represents the decay of the CS states generated
2
1
*
from [ D −B−A−R•], becomes faster with increasing B be-
cause the spin conversion to the unreactive |Q〉 states becomes
less efficient. On the other hand, the A(t) curves at the later time
region predominantly reflect the decay of the |Q 3/2〉 sublevels.
These sublevels may be populated through ET of
(b) Solvent Effect on J. The solvent effects on the decay
+
–
+
–
•
•
•
•
dynamics of both [D −B−A −R•] and [D −B−A ] can be
attributed to the solvent dependence of kBET, as discussed
+
–
•
•
above. However, the J(D , A ) value is also affected by the
solvent polarity, and the change in J(D , A ) can alter the |S〉
↔ |T〉 conversion rate of [D −B−A ]. According to the per-
turbation theory, the exchange interaction can be estimated
by38
+
–
•
•
4
3
+
–
•
•
*
[ D −B−A−R•]. The decay of the |Q 3/2〉 sublevels is governed
by the spin relaxation to |Q 1/2〉 and |D 1/2〉, which becomes
slower with increasing B.36
A comparison of the photo-induced processes for 3 and 2
demonstrates that the introduction of a nitroxide radical (R•) to
the D−B−A system markedly accelerates the spin conversion
of the CS states, while the efficiency of the charge separation
(ΦCS) and the spin-allowed back ET rate (kBET) were not largely
affected by the presence of R•. An enhancement of the spin-
2
J ≈ − |V| /(∆GBET + λ).
(9)
Equation 9 predicts that the J value can vary with solvents
through variations in ∆GBET and λ. When ∆GBET + λ is close to
zero, this theory cannot be applied, because the right-hand side
of Eq. 9 diverges. More rigorous theoretical treatments with an
equilibrium distribution for nuclear configurations predict that
J becomes zero at ∆GBET + λ = 0.38a,d As can be seen in Table 2,
conversion rate gives two advantageous features to the photo-
1
*
chemical reaction of D−B−A. First, when ET occurs from D ,
+
–
•
•
the conversion from the initially generated CS states to the un-
reactive states with different spin multiplicity decelerates the
decay through the back ET. As a result, the CS states can un-
dergo successive reactions other than back ET, such as a charge
shift to a secondary donor or acceptor, if available. Secondly,
rapid equilibration of all the accessible spin sublevels of the in-
termediate simplifies its decay kinetics compared with a spin-
selective reaction accompanied by slow reversible spin conver-
sion, which is favorable for determining of the reaction rate
constants.
because ∆GBET + λ for [D −B−A ] in DMSO is nearly zero,
the |S〉 and |T〉 states are nearly degenerate. In such a case, |S〉
↔ |T〉 conversion is induced by the HFC interaction with a rate
of kHFC (= 1.2 × 108 s–1 from Eq. 7, vide supra) under zero field.
+
–
•
•
The lack of observation of [D −B−A ] born in the |S〉 state
suggests that kBET in DMSO should be larger than kHFC
.
Conclusion
1
*
(1) The S1 state of phenothiazine moiety ( D ) in the present
D−B−A(−R•) systems was effectively quenched through the
intramolecular ET to give the CS state in high quantum yields,
even in solvents of low polarity, such as benzene and toluene.
The TEMPO moiety (R•) in 3 also caused excitation energy
transfer as another quenching pathway, but the quenching
3. Solvent Effects on the Decay Dynamics of the CS
States. (a) Solvent Effect on kBET
. Table 2 shows that the
kBET value dramatically increases with increasing solvent po-
larity. This solvent dependence can be explained as follows. In
the limit of the dielectric continuum model, the solvent reorga-
nization energy (λsol ) can be expressed by Eq. 8,27
through ET was much faster than the energy transfer. In polar
solvents, such as MeCN and DMSO, ET from D toA also oc-
3
*
curred to give the CS state with a different spin multiplicity
λsol = e2/(8πε0)(1/rD + 1/rA − 2/R)(1/n2 − 1/εr),
(8)
1
*
from that formed from D .
where n is the refractive index of the solvent. As for the internal
reorganization energy, we assume that λvib = 0.3 eV, which was
(2) The decay dynamics of the CS state was strongly affect-
ed by the presence of R•, especially in dioxane. Without R•,
+
1
+
–
•
•
•
estimated for an RIP consisting of MPTZ and radical anion
of an electron acceptor with one benzene ring in a previous
work.7b The ∆GBET and λ values estimated by Eqs. 2 and 8, re-
spectively, are listed in Table 2. Approximate ∆GBET values for
benzene and dioxane were obtained by subtracting of the re-
ported additional solvation energies (0.4 and 0.55 eV24) from
∆GBET calculated for trans-decalin by Eq. 2. As the λsol values
for benzene and dioxane, the corresponding values reported by
Warman et al. for similar systems were used.24 Although more
accurate ∆GBET and λ values for solvents of low or medium po-
larity are necessary for a quantitative discussion,37 Table 2
shows that the back ET reactions vary from a deeply inverted
region to the normal region along with an increase in the sol-
vent polarity. The observation that kBET decreases with a de-
crease in the solvent polarity is consistent with the prediction
most part of [D −B−A ] decayed through the spin-allowed
3 + –
•
•
back ET, because the conversion to [D −B−A ] was much
slower due to an exchange interaction larger than B1/2. In the
presence of R•, |D〉 ↔ |Q〉 conversion was effectively induced
by a dipole−dipole interaction between A – and R•, and the ef-
•
ficient |D′〉 ↔ |D〉 conversion resulted from the difference be-
–
+
•
•
tween J(A , R•) and J(D , R•). As a result of these fast spin
+
–
•
•
conversions, the decay of the [D −B−A −R•] born in the |D′〉
+
–
•
•
state was markedly retarded compared with [D −B−A ] born
in the |S〉 state. This point is advantageous for efficient chemi-
cal transformations initiated by photo-induced ET and/or de-
termination of back ET rates.
(3) The back ET rate constant (kBET) of the CS state to give
the ground state varied by more than two orders of magnitude
with the solvent property. The kBET value decreased with de-

302 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
1
creasing solvent polarity. This solvent effect can be explained
by the fact that the back ET process falls in the deeply inverted
region due to the smaller λsol and the more negative ∆GBET in
less-polar solvents.
798, 758, 748 cm–1. H NMR δ 3.88 (2H, s, CH2NH2), 5.13
(2H, s, NCH2), 6.73 (2H, d, J = 8.0 Hz), 6.89 (2H, t, J = 7.5
Hz), 7.03 (2H, ddd, J = 8.0, 7.5, and 1.3 Hz), 7.11 (2H, dd, J =
7.5 and 1.3 Hz), 7.1−7.4 (4H, m), 7.56 (4H, m).
N,N′-Bis{[4′-(10-Phenothiazinylmethyl)biphenyl-4-yl]me-
thyl}-1,2:4,5-pyromellitdiimide (2) and N-[4′-(10-Phenoth-
iazinylmethyl)biphenyl-4-yl]methyl-N′-(2,2,6,6-tetramethylpi-
peridin-1-oxyl-4-yl)-1,2 : 4,5-pyromellitdiimide (3). A mix-
ture of 7 (138 mg, 0.35 mmol), 1,2 : 4,5-pyromellitic dianhydride
(120 mg, 0.55 mmol), and 4-amino-2,2,6,6-tetramethylpiperidin-
1-oxyl(4-amino-TEMPO) (123 mg, 0.72 mmol) in 5 cm3 of DMF
was heated under reflux overnight. The mixture was cooled, poured
into water, and extracted with benzene. The organic extracts were
washed with water and brine, and concentrated. The residue was
subjected to flash chromatography with benzene−ethyl acetate
(20 : 1 to 10 : 1 v/v) as the eluent affording 2 and 3 as the first and
second fractions, respectively. When 7 and 1,2 : 4,5-pyromellitic
dianhydride were reacted in a molar ratio of 2 : 1, 2 was obtained in
a higher yield. 2: IR 3060, 3030, 2930, 2850, 1775 (νC=O), 1709
(νC=O), 1593, 1572, 1499, 1464, 1453, 1394, 1260, 1101, 936, 800,
791, 754, 748 cm–1. 1H NMR δ 4.91 (4H, s), 5.12 (4H, s), 6.70 (4H,
d, J = 7.5 Hz), 6.87 (4H, td, J = 7.5 and 1.6 Hz), 7.00 (4H, td, J = 7.5
and 1.6 Hz), 7.09 (4H, dd, J = 7.5 and 1.6 Hz), 7.38 (4H, d, J = 8
Hz), 7.5 (8H, AA′BB′), 7.55 (4H, d, J = 8 Hz), 8.26 (2H, s). 3 : IR
3050, 2975, 2938, 1773 (νC=O), 1713 (νC=O), 1595, 1572, 1499,
1466, 1442, 1379, 1364, 1342, 1096, 856, 810, 748 cm–1. 1H NMR
δ 4.94 (2H, s), 5.13 (2H, s), 6.71 (2H, d, J = 7.8 Hz), 6.88 (2H, t, J
= 7.5 Hz), 7.00 (2H, t, J = 7 Hz), 7.10 (2H, d, J = 7 Hz), 7.4 (2H, d,
J = 8 Hz), 7.5−7.6 (6H, m), 8.29 (2H, br s).
Experimental
For syntheses, anhydrous THF (stabilizer-free), hexane, tolu-
ene, and DMF (Organics, for organic syntheses) were used as re-
ceived.All of the reactions were carried out under a N2 atmosphere.
Flash chromatography was carried out on silica gel (Kanto Chemi-
cals, Silica gel 60, spherical, 270−325 mesh). IR spectra were ob-
tained as KBr disks. 1H NMR spectra (270 MHz) were measured in
CD2Cl2 with the signal due to residual 1H of the solvent (δ 5.32) as
an internal standard. EPR spectra were recorded at room tempera-
ture with 100-kHz modulation on an X-band EPR spectrometer
(JEOL, JES-RE1X). The magnetic field and the microwave fre-
quency were determined with an NMR field meter (Echo Electron-
ics, EFM-2000AX) and a microwave counter (Echo Electronics,
EMC-14), respectively. A microwave power of 0.8 mW was used.
For optical measurements, THF and MeCN were of spectroanalyt-
ical grade and dried over Molecular Sieves 3A before use. Benzene
and toluene of spectroanalytical grade were used as received. Diox-
ane was distilled from CaH2 under a N2 atmosphere. DMSO of
spectroanalytical grade was distilled from CaH2 under reduced
pressure.
10-[(4′-Bromomethylbiphenyl-4-yl)methyl]phenothiazine
(6). To a suspension of NaH (5.0 mmol, washed with anhydrous
hexane) in THF (6 cm3), a solution of phenothiazine (0.8 g, 4.0
mmol) in THF was added with stirring. The mixture was stirred un-
der N2 at room temperature until the evolution of H2 gas ceased
(about 1 h) and at 40 °C for an additional 1 h. The resultant red-
brown solution was slowly transferred by a cannula to a 100 cm3
three-necked flask containing a suspension of 4,4′-bis(bromometh-
yl)biphenyl (1.70 g, 5.0 mmol) in 30 cm3 of THF. This mixture was
stirred at 60 °C for 3 h, cooled down to room temperature, and fil-
tered. The filtrate was concentrated, diluted with benzene and wa-
ter. The organic layer was separated, washed with water and then
brine, dried over MgSO4, and evaporated. The residue was subject-
ed to flash chromatography with benzene−hexane (1 : 2 to 1 : 1).
The obtained bromide 6 was contaminated with a small amount of
the starting dibromide, but used for the next step without further
purification.
10-[(4′-Aminomethylbiphenyl-4-yl)methyl]phenothiazine
(7). A mixture of 6 (500 mg, 1.1 mmol), potassium phthalim-
ide (278 mg, 1.5 mmol) and 18-crown-6 (40 mg, 0.15 mmol) in
10 cm3 of toluene was stirred at 90 °C for 12 h.39 The resultant
mixture was cooled down to room temperature and diluted
with benzene and water. The organic layer was separated,
washed with brine, dried over ΜgSO4, and concentrated. Flash
chromatography of the residue eluting with benzene−ethyl ace-
tate (20 : 1 v/v) gave the phthalimide (440 mg). This phthalim-
ide was suspended to EtOH (25 cm3) and a 6-fold excess of hy-
drazine monohydrate was added. The resultant mixture was re-
fluxed for 3 h. After cooling, an aqueous solution (10 cm3) of
Na2CO3 (320 mg) was added, and concentrated until colorless
precipitate was separated. This precipitate was extracted with
benzene, and the organic layer was washed with water and
brine, dried over MgSO4, and evaporated up to give amine 7 in
58 % yield. IR 3400 (νNH), 3370 (νNH), 3050, 2920, 2850,
1591, 1570, 1499, 1485, 1462, 1399, 1371, 1258, 1220, 858,
Attempted Synthesis of N-Hexyl-N′-[4′-(10-phenothiazinyl-
methyl)biphenyl-4-yl]methyl-1,2:4,5-pyromellitdiimide (2′).
A
mixture of 7 (60 mg, 0.15 mmol), 1,2 : 4,5-pyromellitic dianhy-
dride (65 mg, 0.30 mmol), and hexylamine (46 mg, 0.45 mmol) in
3 cm3 of DMF was heated overnight under reflux. A work-up pro-
cedure similar to the case of 2 and 3 gave a mixture of 2′ and N,N′-
dihexyl-1,2 : 4,5-pyromellitdiimide in a ratio of ca. 1 : 1.5 (based
on the integration ratio of the 1H NMR signals), which could not be
separated from each other after repeated flash chromatography.
N-[4′-(10-Phenothiazinylmethyl)biphenyl-4-yl]methyl-N′-
(2,2,6,6-tetramethylpiperidin-1-oxyl-4-yl)bicyclo[2.2.2]oct-7-
ene-2,3 : 5,6-bis(dicarboximide) (4). A mixture of amine 7 (59
mg, 0.15 mmol), 4-amino-TEMPO (43 mg, 0.25 mmol), and bicy-
clo[2.2.2]oct-7-ene-2,3:5,6-tetracarboxylic dianhydride (50 mg,
0.20 mmol) in DMF (3 cm3) was stirred at 130 °C overnight. After
cooling to room temperature, the mixture was poured into water
and extracted with benzene. The organic extracts were washed with
brine, dried over MgSO4, and concentrated. Flash chromatography
of the residue with benzene−ethyl acetate (20 : 1 to 4 : 1 v/v) as
eluent gave 4 (25 mg, 16%). IR 3060, 3030, 2970, 2930, 1773
(νC=O), 1701 (νC=O), 1595, 1575, 1500, 1464, 1443, 1377, 1255,
1213, 804, 781, 752 cm–1. 1H NMR δ 2.91 (2H, br s), 3.02 (2H, s),
3.72 (2H, br s), 4.58 (2H, s, imide−CH2), 5.13 (2H, s, phenothiaz-
ine−CH2), 6.04 (2H, br s, olefinic H), 6.71 (2H, d, J = 7.3 Hz), 6.88
(2H, t, J = 7.3 Hz), 7.01 (2H, t, J = 7.3 Hz), 7.10 (2H, d, J = 7.3 Hz),
7.31 (2H, d, J = 7.6 Hz), 7.39 (2H, d, J = 7.6 Hz), 7.52 (2H, d, J =
7.6 Hz), 7.54 (2H, d, J = 7.6 Hz).
N-(Biphenyl-4-yl)-N′-(2,2,6,6-tetramethylpiperidin-1-oxyl-
4-yl)-1,2:4,5- pyromellitdiimide (5). A mixture of 4-phenyl-
benzylamine (366 mg, 2.0 mmol), 4-amino-TEMPO (1.0 g, 5.8
mmol), and 1,2 : 4,5-pyromellitic dianhydride (763 mg, 3.5 mmol)
in DMF (14 cm3) was heated under reflux overnight. After cooling

Y. Mori et al.
Bull. Chem. Soc. Jpn., 74, No. 2 (2001) 303
to room temperature, the reaction mixture was poured into water,
and the precipitate was collected by filtration, washed with water,
and dried in air. The filtrate was extracted with CH2Cl2 and the or-
ganic extracts were washed with 1 M (1 M = 1 mol dm–3) HCl, wa-
ter and then brine, dried over MgSO4, and evaporated. The result-
ant residue and the collected precipitate were combined and puri-
fied by flash chromatography with CH2Cl2−ethyl acetate as eluent.
The first orange-colored band gave the desired product 5 (272 mg).
10-[(Biphenyl-4-yl)methyl]phenothiazine (1) was prepared
from phenothiazine and 4-bromomethylbiphenyl by the same pro-
cedures as those for 6. IR 3057, 2851, 1590, 1568, 1487, 1460,
1453, 1408, 1373, 1327, 1284, 1255, 856, 828, 754, 731, 692 cm–1.
1H NMR δ 5.14 (2H, s), 6.73 (2H, d, J = 8 Hz), 6.88 (2H, m), 7.02
(2H, td, J = 7.3 and 1.7 Hz), 7.10 (2H, dd, J = 7.3 and 1.7 Hz),
7.30−7.45 (5H, m), 7.55−7.61 (4H, m).
Electrochemistry. Cyclic voltammograms of 1 and 5 were re-
corded with a BAS CV-1B voltammetric controller in MeCN con-
taining 0.1 M n-Bu4NBF4 as the supporting electrolyte under Ar at
ca. 298 K. Glassy carbon, Pt wire, and Ag/Ag+ (BAS RE-5) were
used as the working, counter, and reference electrodes, respective-
ly. The scan rate was 50 mV s–1. The ferrocene was used as the in-
ternal standard, which showed a reversible redox wave (the peak
separation was 60 mV between the cathodic and anodic scans) at
–0.365 V vs. SCE in our electrolytic solutions.
Fluorescence Spectra. Fluorescence spectra were recorded
on a Shimadzu RF510 spectrofluorometer equipped with a quan-
tum counter at 293 K. Each of the sample solutions (ca. 5 cm3) in a
long-necked quartz cell (10 × 10 × 45 mm) was bubbled withAr for
20 min before the measurements. The relative quantum yields of
fluorescence were determined based on the integrated intensity af-
ter subtracting of the background due to the Raman scattering by
the solvent.
Laser Flash Photolyses. All measurements were carried out
at 293 K. Each of the sample solutions (ca. 2.5 cm3) was placed in
a long-necked quartz cell (10 × 5 × 45 mm) and bubbled withAr for
20 min before the measurements. The magnetic fields were gener-
ated by a Tokin SEE-10W electromagnet. The third (355 nm) har-
monic of a Quanta-Ray GCR-103 Nd : YAG laser was used as ex-
citation light. The monitored wavelength of the A(t) curves was
chosen for each sample solution so that the maximal absorbance
change (Amax) on excitation was in a range of 0.12−0.22 (transmit-
tance of 60−75% relative to that before excitation). The concentra-
tion of 3 was (1−2) × 10–4 M, while for 2 saturated solutions were
used (≤ 1 × 10–4 M). The optical densities at 355 nm were
0.12−0.13 and 0.7−0.12 for 3 and 2, respectively, with a path length
of 1 cm.
cules. Part I and II,” ed by M. Bixon and J. Jortner, Wiley, New
York (1999).
5
a) For example, R. J. Hulsebosch, I. V. Borovykh, S. V.
Paschenko, P. Gast, and A. J. Hoff, J. Phys. Chem. B, 103, 6815
(1999). b) Recent review, K. Möbius, Chem. Soc. Rev., 2000, 129.
6
a) D. Carbonera, M. D.Valentin, C. Corvaja, G.Agostini, G.
Giacometti, P. A. Liddell, D. Kuciauskas, A. L. Moore, T. A.
Moore, and D. Gust, J. Am. Chem. Soc., 120, 4398 (1998). b) G.
Elger, H. Kurreck, A. Wiehe, E. Johnen, M. Fuhs, T. Prisner, and J.
Vrieze, Acta Chem. Scand., 51, 593 (1997). c) F. Lendzian, J.
Schlüpmann, J. von Gersdorff, K. Möbius, and H. Kurreck, Angew.
Chem., Int. Ed. Engl., 39, 1461 (1991).
7
a) Y. Mori, Y. Sakaguchi, and H. Hayashi, Chem. Phys.
Lett., 286, 446 (1998). b) Y. Mori, Y. Sakaguchi, and H. Hayashi, J.
Phys. Chem. A, 104, 4896 (2000).
8
Y. Mori, Y. Sakaguchi, and H. Hayashi, Chem. Phys. Lett.,
301, 365 (1999).
9
a) A. J. Hoff and P. J. Hore, Chem. Phys. Lett., 108, 104
(1984). b) K. M. Salikhov, A. J. van der Est, and D. Stehlik, Appl.
Magn. Reson., 16, 101 (1999).
10 a) E. N. Step, A. L. Buchachenko, and N. J. Turro, J. Am.
Chem. Soc., 116, 5462 (1994). b) A. L. Buchachenko, L. V. Ruban,
E. N. Step, and N. J. Turro, Chem. Phys. Lett., 233, 315 (1995). c)
A. L. Buchachenko and V. L. Berdinsky, Russ. Chem. Bull., 44,
1578 (1995); Translated from Izv. Akad. Nauk. Ser. Khim., 1995,
1646. d) A. L. Buchachenko and V. L. Berdinsky, J. Phys. Chem.,
100, 18292 (1996). e) A. L. Buchachenko, V. L. Berdinsky, and N.
J. Turro, Kinetics and Catalysis, 39, 301 (1998); Translated from
Kinetika i Kataliz, 39, 325 (1998). f) A. L. Buchachenko and V. L.
Berdinsky, Chem. Phys. Lett., 298, 279 (1998). In our case, the |D〉
and |D′〉 states in the presence of magnetic fields should be ex-
pressed as follows instead of Eqs. (2g) and (2h) in their paper:
|D+1/2〉 = 6–1/2 (2|αD +α β 〉 – |α +β α 〉 – |β +α α 〉); |D
〉
–
–
–
•
A R
•
•
•
R
•
•
•
•
•
A
D
A
R
D
–1/2
= 6–1/2 (2|βD +β α 〉 – |β +α β 〉 – |α +β β 〉); |D′+1/2〉 = 2–1/
–
–
–
•
A R
•
•
•
•
•
•
R
•
•
A
R
D
A
D
2
(|αD +β α 〉 – |β +α α 〉); |D′–1/2〉 = 2–1/2 (|αD +β β 〉 –
–
•
–
•
–
•
A R
•
•
•
•
R
•
•
A
R
D
A
–
|βD +α β 〉).
•
•
•
R
A
11 A. I. Ivavov, V. A. Mikhailova, and S. V. Feskov, Appl.
Magn. Reson., 16, 481 (1999).
12 Y. Moroi, A. M. Braun, and M. Grätzel, J. Am. Chem. Soc.,
101, 567 (1979).
13 Y. Sakaguchi and H. Hayashi, J. Phys. Chem. A, 101, 549
(1997).
14 a) M. Kaholek and P. Hrdlovic, J. Photochem. Photobiol. A:
Chem., 127, 45 (1999). b) N. V. Blough and D. J. Simpson, J. Am.
Chem. Soc., 110, 1915 (1988). c) S. A. Green, D. J. Simpson, G.
Zhou, P. S. Ho, and N. V. Blough, J. Am. Chem. Soc., 112, 7337
(1990). d) S. E. Herbelin and N. V. Blough, J. Phys. Chem. B, 102,
8170 (1998).
15 The D−R• distance in 4 is somewhat different from that in 3,
because the bicyclic diimide part in 4 is not planar as the 1,2 : 4,5-
pyromellitdiimide part in 3. However, we could not obtain any bet-
ter D−B−R• model compound than 4.
YM thanks the financial supports from The Nishida Re-
search Fund for Fundamental Organic Chemistry and from
Kenkyu-Shoureihi (Research Facilitation Grant) of RIKEN.
References
16 U. Werner, Y. Sakaguchi, H. Hayashi, G. Nohya, R.
Yoneshima, S. Nakajima, and A. Osuka, J. Phys. Chem., 99, 13930
(1995).
1
2
M. R. Wasielewski, Chem. Rev., 92, 435 (1992).
D. Gust, T. A. Moore, and A. L. Moore, Acc. Chem. Res.,
26, 198 (1993).
17 Y. Fujiwara, R. Nakagaki, and Y. Tanimoto, “Dynamic Spin
Chemistry,” ed by S. Nagakura, H. Hayashi, and T. Azumi, Kodan-
sha–Wiley, Tokyo and NewYork (1998), Chap. 3.
18 a) H. Hayashi, “Dynamic Spin Chemistry,” ed by S.
Nagakura, H. Hayashi, and T.Azumi, Kodansha–Wiley, Tokyo and
New York (1998), Chap. 2. b) H. Hayashi and S. Nagakura, Bull.
Chem. Soc. Jpn., 51, 2862 (1978).
3
a) C. Luo, D. M. Guldi, H. Imahori, K. Tamaki, and Y.
Sakata, J. Am. Chem. Soc., 122, 6535 (2000). b) H. Tsue, H.
Imahori, T. Kaneda, Y. Tanaka, T. Okada, K. Tamaki, and Y.
Sakata, J. Am. Chem. Soc., 122, 2279 (2000), and references there-
in.
4
“Electron Transfer : from Isolated Molecules to Biomole-

304 Bull. Chem. Soc. Jpn., 74, No. 2 (2001)
Spin Effect on Charge-Recombination Dynamics
19 In MeCN and DMSO, the A(t) curves did not decay to the
zero level within 1 µs, but a small fraction of non-decaying compo-
nent (ca. 5% of Amax) was observed. This component probably cor-
responded to the absorption due to the singly charged ions which
were formed by the following intermolecular charge shift reaction.
lene in DMF was reported to be –1.98V vs. mercury pool. If the re-
dox potentials are the same in DMF and MeCN, this value corre-
sponds to –2.29 V vs. SCE.30d d) G. J. Kavarnos and N. J. Turro,
Chem. Rev., 86, 401 (1986).
31 a) M. Bixon and J. Jortner, in Ref. 4, pp. 35–202. b) H. M.
McConnel, J. Chem. Phys., 35, 508 (1961).
D•+ B A•− R• + D B A R• → D B A•− R• + D+• B A R•
20 A. Z. Weller, Z. Phys. Chem., 1982, 133.
32 The decay rate of the T1 state of MPTZ (1.5 × 105 s–1 in 2-
PrOH at 293 K under N2 atmosphere) determined at a higher con-
centration (1 × 10–3 M).7b
21 J. J. P. Stewart, J. Comp. Chem., 10, 209 and 221 (1989).
22 a) The structure optimization was carried out for 2′ by the
PM3 methods with the program package MOPAC.22b Two stable
conformers with almost the same energy were obtained , and their
R values were 12 and 14 Å. For the CS state, the shorter value was
used, because the larger Coulombic stabilization was obtained. b)
WinMOPAC Ver. 2.0, Fujitsu Ltd., Japan, 1997.
33 a) G. -H. Goudsmit, H. Paul, and A. I. Shushin, J. Phys.
Chem., 97, 13243 (1993). b) A. I. Shushin, Z. Phys. Chem., 182, 9
(1993). c) A. I. Shushin, Chem. Phys. Lett., 208, 173 (1993). d) A.
Kawai, T. Okutsu, and K. Obi, J. Phys. Chem., 95, 9130 (1991). e)
A. Kawai and K. Obi, J. Phys. Chem., 96, 5701 (1992). f) Y.
Kobori, K. Takeda, K. Tsuji, A. Kawai, and K. Obi, J. Phys. Chem.
A, 102, 5160 (1998).
34 (a) A. Weller, K. Nolting, and H. Staerk, Chem. Phys. Lett.,
96, 24 (1983). b) K. Schulten and P. G. J. Wolynes, Chem. Phys.,
68, 3292 (1978).
35 a) L. R. Faulkner, H. Tachikawa, and A. J. Bard, J. Am.
Chem. Soc., 94, 691 (1972). b) H. Tachikawa and A. J. Bard, Chem.
Phys. Lett., 26, 10 and 246 (1974). c) K. Razi Naqvi, H. Staerk, and
T. Gillbro, Chem. Phys. Lett., 49, 160 (1977). d) Y. Iwasaki, K.
Maeda, and H. Murai, The Sixth International Symposium on
Magnetic Field and Spin Effects in Chemistry and Related Phe-
nomena, Emmetten, Switzerland, Book of Abstracts, p. 67 (1999).
36 a) R. C. Johnson and R. E. Merrifield, Phys. Rev. B, 1, 896
(1970). b) D. R. Kearns and A. J. Stone, J. Chem. Phys., 55, 3383
(1971). c) P. W.Atkins and G. T. Evans, Mol. Phys., 29, 921 (1975).
d) K. Schulten, J. Chem. Phys., 80, 3668 (1984).
37 Photoexcitation of some D−B−A systems to a charge-sepa-
rated excited state in benzene and dioxane required excess reorga-
nization energies of 0.4 and 0.55 eV, respectively, compared with
the same process in trans-decalin.24 PenOAc has a relatively low
macroscopic dielectric constant, but contains an ester group, which
could have strong interactions with charged solutes.
23 Y. Kawanishi, N. Kitamura, and S. Tazuke, J. Phys. Chem.,
90, 2469 (1986).
24 a) J. M. Warman, M. P. de Haas, J. W. Verhoeven, and M. N.
Paddon-Row, in Ref. 4, pp. 571–601. b) J. M. Warman, K. J. Smit,
M. P. de Haas, S. A. Jonker, M. N. Paddon-Row, A. M. Oliver, J.
Kroon, H. Oevering, and J. W. Verhoeven, J. Phys. Chem., 95, 1979
(1991).
25 M. J. Shephard and M. N. Paddon-Row, J. Phys. Chem. A,
103, 3347 (1999).
26 S. Nath, H. Pal, D. K. Palit, A. V. Sapre, and J. P. Mittal, J.
Phys. Chem. A, 102, 5822 (1998).
27 R. A. Marcus, J. Chem. Phys., 24, 966 and 979 (1956).
28 The magnitude of V largely depends on the ET mechanism.
In some D−B−A systems, two-step sequential ET involving an in-
termediate in which the bridge part bears a charge (D+−B–−A or
D−B+−A–) has been postulated.29 In the present case, however, the
+
–
•
•
sequential mechanism involving D −B –A can be ruled out, be-
cause no charge separation occurred in 1 even in solvents of high
polarity such as MeCN. The ∆GET value for the hypothetical ET re-
1
+
action of D –B to give D –B – was estimated to be ca. +0.2 eV
from Eq. 2, with ERED of biphenyls (–2.36 and –2.55 V vs. SCE for
unsubstituted and 4,4′-di-t-butyl substituted compounds),30 EOX of
1 (+ 0.71V vs. SCE), R ≈ 6 Å, and ES = 2.9 eV. Thus, this hypothet-
•
•
*
38 a) Y. Kobori, K. Akiyama, and S. Tero–Kubota, J. Chem.
Phys., 113, 465 (2000). b) Y. Kobori, S. Sekiguchi, K. Akiyama,
and S. Tero–Kubota, J. Phys. Chem. A, 103, 5416 (1999). c) S.
Sekiguchi, Y. Kobori, K. Akiyama, and S. Tero–Kubota, J. Am.
Chem. Soc., 120, 1325 (1998). d) M. Volk, T. Häberle, R. Feick, A.
Ogrodnik, and M.-E. Michel-Beyerle, J. Phys. Chem., 97, 9831
(1993). e) M. Y. Okamura, R. A. Isaacson, and G. Feher, Biochim.
Biophys. Acta, 546, 394 (1979).
ical ET is energetically unfavorable. It is likely that the single-step
1
*
long-distance ET from D to A takes place by the super-exchange
mechanism.31
29 a) R.A. Marcus, Chem. Phys. Lett., 133, 471 (1987). b) S. F.
Fischer and P. O. J. Scherer, Chem. Phys., 115, 151 (1987). c) M.
-E. Michel-Beyerle, M. Bixon, and J. Jortner, Chem. Phys. Lett.,
151, 188 (1988). d) M. Volk, G. Aumeier, T. Langerbacher, R.
Feick, A. Ogrodnik, and M. -E. Michel-Beyerle, J. Phys. Chem. B,
102, 735 (1998).
39 Advantage of addition of a catalytic amount of crown ether
was reported. K. Soai, A. Ookawa, and K. Kato, Bull. Chem. Soc.
Jpn., 55, 1671 (1982).
30 a) P. K. Freeman and L. L. Hutchinson, J. Org. Chem., 45,
1924 (1980). b) M. D. Curtis and A. L. Allred, J. Am. Chem. Soc.,
87, 2554 (1965). c) In Ref. 30a, the reduction potential of naphtha-
40 “Handbook of Chemistry and Physics,” 77th edition, ed by
D. R. Lide, CRC Press, FL (1996), pp. 6−151.