Magnetic Field Effects on the Photodissociation Reaction of Triarylphosphine in Nonviscous Homogeneous Solutions

Article abstract of DOI:10.1021/jp030965f

The magnetic field effects (MFEs) on the photodecomposition reactions of triphenylphosphine and its halogen and methyl derivatives are investigated in fluid solutions. The yield of diarylphosphinyl radicals decreased with increasing magnetic field from 0.1 to 5 T but was stationary below 0.1 T and above 5 T. The MFE becomes larger by the substitution of halogen atoms and the 3- or 4-methyl group. In cyclohexane, the yields of the escaped diarylphosphinyl radicals at 1 T are reduced to 0.69, 0.55, 0.59, and 0.56 of those at 0 T for triphenylphosphine and its tris(4-chloro), tris(3-methyl), and tris(4-methyl) derivatives, respectively. This magnetic field dependence was ascribed to originate from the deactivation process of the excited triplet state, which is a variant of the d-type triplet mechanism originally proposed by Steiner. The interaction between the closely lying nπ* and ππ* states makes their solvent dependence complicated.

Full text of DOI:10.1021/jp030965f

J. Phys. Chem. A 2004, 108, 3421-3429
3421
Magnetic Field Effects on the Photodissociation Reaction of Triarylphosphine in Nonviscous
Homogeneous Solutions
Yoshio Sakaguchi†,* and Hisaharu Hayashi
Molecular Photochemistry Laboratory, RIKEN (The Institute of Physical and Chemical Research),
Wako, Saitama 351-0198, Japan
ReceiVed: August 12, 2003; In Final Form: February 16, 2004
The magnetic field effects (MFEs) on the photodecomposition reactions of triphenylphosphine and its halogen
and methyl derivatives are investigated in fluid solutions. The yield of diarylphosphinyl radicals decreased
with increasing magnetic field from 0.1 to 5 T but was stationary below 0.1 T and above 5 T. The MFE
becomes larger by the substitution of halogen atoms and the 3- or 4-methyl group. In cyclohexane, the yields
of the escaped diarylphosphinyl radicals at 1 T are reduced to 0.69, 0.55, 0.59, and 0.56 of those at 0 T for
triphenylphosphine and its tris(4-chloro), tris(3-methyl), and tris(4-methyl) derivatives, respectively. This
magnetic field dependence was ascribed to originate from the deactivation process of the excited triplet state,
which is a variant of the d-type triplet mechanism originally proposed by Steiner. The interaction between
the closely lying nπ* and ππ* states makes their solvent dependence complicated.
1. Introduction
Most of the magnetic field effects (MFEs) of chemical
reactions in the liquid phase have successfully been explained
in terms of the radical pair mechanism (RPM).1-3 A radical
pair (RP) consists of two radicals surrounded by solvent
molecules. The component radicals are interacting weakly with
each other. Figure 1 shows the reaction scheme of an RP
generated photochemically by bond fission. Here, triarylphos-
phine (Ar3P) was taken as an example. This reaction scheme is
confirmed in section 3.1.
In this scheme, four possibilities (A-D) of the magnetic field-
dependent processes are shown. The first two (A, B) are the
intersystem crossing (ISC) processes to depopulate and to
populate the excited triplet state. To discriminate between them,
we use the words depopulation and population type (d- and
p-type) mechanisms for processes A and B.4 Usually, processes
A and B are not considered in the MFEs on chemical reactions
in the liquid phase. In a contact RP, which appears immediately
after the fission or just before the collision of an RP, the energy
separation between its singlet and triplet states becomes larger
than the magnetic interactions. Therefore, the spin conversion
between the singlet and the triplet states is blocked, where the
recombination of the pair proceeds through the singlet state.
The spin-forbidden recombination process, C, is sometimes
considered in the RP including heavy atoms, and this process
is usually considered to be independent of a magnetic field.
Although this does not mean that processes A-C are not
possible origins of the MFEs, almost all efforts have been
devoted to process D, the ISC of separated RPs.
Figure 1. Four possible pathways (A-D) of the magnetic field-
dependent processes in the photodecomposition reaction of triplet
3
triarylphosphine. Ar2P•, Ar•, and S* stand for the diarylphosphinyl
radical, the aryl radical, and a triplet sensitizer. Path X is excluded in
the real reaction scheme.
coupling mechanism (HFCM), the ∆g mechanism (∆gM), the
relaxation mechanism (RM), and the level-crossing mechanism
(LCM) are variants of the RPM. The lifetime of the RP and
hence the magnitude of its MFE are largely dependent on the
confinement of the RP (i.e., the “cage effect”) of its solvent.
Therefore, the MFE becomes smaller if the solution becomes
more fluid.
In 1979, Steiner reported large MFEs in the electron-transfer
reactions between thiazine dyes and electron donors containing
heavy atoms.5 Such large MFEs and the field dependence of
the effects could not be ascribed to the HFCM, ∆gM, RM, and
LCM. He ascribed the origin of their MFEs to the spin-forbidden
back-electron-transfer process of triplet exciplexes, which
competes with the escape to the free radicals.6,7 Although Figure
1 is presented for neutral RPs, the magnetic field-dependent
decay of these triplet exciplexes corresponds to process C
because each triplet exciplex is a contact RP-type intermediate
formed from the excited triplet state of the dye.6 This mechanism
has been called the d-type triplet mechanism (TM), where spin-
orbit coupling (SOC) plays an important role. Because the
recovery of the ground-state molecule from each exciplex must
compete with the diffusive separation, the reaction rate of the
backward electron transfer in the triplet exciplexes must be very
large. Steiner’s group reported and analyzed several MFEs due
to this d-type TM.8,9 Tero-Kubota’s group expanded this
Because the energy separation between the singlet and triplet
states of a separated RP is much smaller than that of contact
RPs, the magnetic interaction can induce a sizable effect on
separated RPs. Thus, most MFEs on chemical reactions through
RPs have been explained in terms of the RPM. The hyperfine
* Corresponding author. E-mail: ysakaguc@postman.riken.jp. Tel: +81-
48-467-9395. Fax: +81-48-462-4664.
† Present address: Surface Chemistry Laboratory, RIKEN.
10.1021/jp030965f CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/23/2004

3422 J. Phys. Chem. A, Vol. 108, No. 16, 2004
Sakaguchi and Hayashi
mechanism to chemically induced dynamic electron spin
polarization (CIDEP).6,7,10,11 As far as we know, however, there
has been no example of the MFE in process C for a contact RP
of neutral radicals. By investigating the CIDEP of the photolysis
of 2,2′-azobisbutyronitrile in solution, Paul’s group found a
d-type TM polarization of molecular origin.12,13 Recently, they
found MFEs and CIDEP for the sensitized photodecomposition
reaction of azocumene in fluid solutions and explained them in
terms of the d-type TM operating in the photoreactive triplet
state14 as anticipated by Steiner.6 The origin of this MFE
corresponds to process A in Figure 1.
In 1995, we reported large MFEs for the photodecomposition
reaction of triphenylphosphine (Ph3P) in fluid solutions below
1.34 T.15 The magnitude of the MFEs was anomalously larger
than usual MFEs of neutral RPs in fluid solutions. Tentatively,
we ascribed the MFEs of Ph3P to ∆gM. In 1997, we found that
the MFEs at higher fields (1.5-10 T) did not obey the field
dependence due to ∆gM. Therefore, we started to apply the
d-type TM to our reactions.16 The MFEs found by Paul’s group
encouraged our analyses.
Figure 2. (A) Transient absorption spectra of 4-ClPh3P in n-dodecane
in the absence (b) and presence (O) of a magnetic field of 1 T. (B)
MFE spectrum, the difference in the above spectra.
In this article, we report on the reconsideration of the MFEs
of Ph3P using its halogen and methyl derivatives. Our conclusion
for the origin of the MFEs of the photodecomposition reactions
of Ar3P is the d-type TM in the photoreactive molecular triplet
state. This is analogous to the mechanism put forth by Paul
and co-workers. However, we considered all four alternatives
(A-D) for MFEs and consequently eliminated three of them.
In each of our reactions, the solute is dissolved in a fluid solvent
at low concentration. These are the simplest reaction conditions
for the observation of MFEs in the condensed phases and for
practical applications.
were generated by an Oxford 37057 superconducting magnet
with a PS120-10 power supply. The experiments under the
lowest magnetic field (B < 0.2 mT for the electromagnet, B <
0.3 mT for the superconducting magnet) are denoted as those
in the absence of a magnetic field.
In the present experiments, the time profiles of the transient
absorbance, A(t, B) curves, were measured in the absence and
presence of a magnetic field. Because the magnetic field
modified the immediate A(t ) 0 ns, B) values, we could not
use them as an internal standard to calibrate the fluctuation of
laser power. To reduce the fluctuation of the power of each
laser shot, we averaged the data over 10 shots. To avoid long-
term variation of the laser power, we measured the A(t, 0 T)
curves frequently during the measurements of A(t, B) ones.
Using a second- or third-order polynomial fit for all A(t, 0 T)
measurements, we estimated the long-term variation of the laser
power. Then we normalized all data assuming that the laser
power was changing smoothly during the measurements at other
fields between the measurements at 0 T.
2. Experimental Section
Triphenylphosphine (Ph3P, elemental analysis grade, Merck),
tris(4-chlorophenyl)phosphine (4-ClPh3P, Aldrich) and its 3-an-
alogue (3-ClPh3P, Aldrich), 4-bromophenyldiphenylphosphine
(4-BrPhPh2P, Aldrich), tris(2-methylphenyl)phosphine (2-
MePh3P, Aldrich) and its 3- and 4-analogues (3, and 4-MePh3P,
Aldrich), and tetrafluoro-1,4-dicyanobenzene (F4DCNB, Ald-
rich) were applied without further purification. 1,4-Dicyanoben-
zene (DCNB, TCI) was recrystallized from an ethanol-benzene
mixture. 2-Propanol, cyclohexane (high-performance liquid
chromatography grades, Cica-Merck), n-hexane (high-perfor-
mance liquid chromatography grade or UV analysis grade, Cica-
Merck), n-decane, n-dodecane, n-tetradecane, and n-hexadecane
(99+% grades, Aldrich) were used as solvents without further
purification.
Phosphine compounds are rather sensitive to oxygen in air.
Although triarylphosphines are not pyrophoric, the weighing,
the dissolution in solvents, and the preservation during the
measurement were carried out in a glovebox filled with nitrogen
gas. All solvents were degassed by sonication before the
dissolution under nitrogen. The sample solution was supplied
to the quartz flow cell by a Teflon tube surrounded by a rubber
tube. Nitrogen gas flowed in the rubber tube to prevent the
permeation of oxygen to the Teflon tube. The irradiated solution
was not reused. No Tygon tube was applicable to the present
experiments.
3. Results and Discussion
3.1. Reaction Scheme. The A(t, B) curves were measured
for Ph3P, 4-ClPh3P, and 4-BrPhPh2P in nonviscous homoge-
neous solutions upon laser excitation (λ ) 266 nm) at 293 K.
Here, we concentrate on those of 4-ClPh3P because we described
those of Ph3P in the preliminary report.15 The transient absorp-
tion spectra of 4-ClPh3P (1 × 10-3 mol dm-3) observed in
n-dodecane 200 ns after excitation in the absence and presence
of a magnetic field of 1 T are shown in Figure 2A. Each
spectrum shows a sharp peak at 340 nm and a broad one around
480 nm. The shape of each spectrum shown in Figure 2A was
not changed during t ) 0 ns and ∼4 µs. However, the magnetic
field dependence of the peak at 340 nm is different from that at
480 nm. In Figure 2B, we plotted the MFE spectrum,18 which
is derived from the difference A(t, 0 T) - A(t, 1 T). This
spectrum corresponds to the species whose yield is affected by
the magnetic field. In this MFE spectrum, the peak at around
480 nm is missing. The spectrum shown in Figure 2B is very
similar to the spectrum of the diphenylphosphinyl radical (Ph2P•
) observed in the reaction of Ph3P,15,19 although the peak position
is shifted from 320 to 340 nm. Thus, we can definitely conclude
Laser flash photolysis experiments were performed at 293
K. They were basically the same as described previously.17 The
sample solutions were excited with the fourth (266 nm)
harmonic of a Quanta-Ray GCR-1 or a GCR-103 Nd:YAG laser.
The external magnetic fields (B) of 0-1.75 T were generated
by a Tokin SEE-10W electromagnet. Those of B ) 0-10 T

Photodissociation of Triarylphosphine
J. Phys. Chem. A, Vol. 108, No. 16, 2004 3423
In the previous report, we proposed the observed MFEs to
be due to the triplet state from the polarization of the CIDEP
spectra observed in the 2-propanol solution of Ph3P. We can
further confirm this conclusion using F4DCNB in the sensitiza-
tion experiments. Figure 3C shows the A(t) curves observed at
340 nm for the 2-propanol solutions of 4-ClPh3P (1 × 10-3
mol dm-3) and F4DCNB (1 × 10-3 mol dm-3). The reaction
dynamics changed drastically from that without F4DCNB (cf.
Figure 3B). Nevertheless, the observed transient absorption
spectrum was not changed from that without F4DCNB. Each
of the A(t) curves in Figure 3C consists of an immediate rise
and a gradual rise that continues up to ∼5 µs. The immediate
rise can be ascribed to the reaction by the direct excitation of
4-ClPh3P. However, the slow rise was observed only in the
presence of F4DCNB. Because the lifetime of the singlet excited
state of DCNB is reported to be merely 8.8 ns in acetonitrile,20
the slow rise is not attributed to the sensitized decomposition
of 4-ClPh3P by the excited singlet state of F4DCNB but to that
of the excited triplet state. Similar behavior was observed in
the reaction of Ph3P with F4DCNB or DCNB in 2-propanol,
but the slow rise terminated within 1.5 µs. Therefore, we can
draw the reaction scheme shown in Figure 1. Here, path X,
which is the dissociation from the singlet excited state, is
3
3
3
excluded, and S* indicates F4DCNB* or DCNB*.
Transient optical absorption spectra and their kinetic behavior
for 2-MePh3P, 3-MePh3P, and 4-MePh3P in cyclohexane solu-
tions were also measured upon laser excitation (λ ) 266 nm)
at 293 K. Spectra similar to that of Ph3P in cyclohexane were
observed. The peak positions were 325, 325, and 330 nm for
2-, 3-, and 4-derivatives, respectively. The reaction scheme
shown in Figure 1 is also applicable to these compounds.
3.2. Magnetic Field Dependence. In Figure 4A, the magnetic
field dependence of the ratio R(B) ) A(t, B)/A(t, 0 T) observed
for Ph3P in 2-propanol, cyclohexane, and n-hexane solutions is
plotted against log(B/T). Here, the observation time, t, is taken
to be 200 ns after laser excitation. This figure also shows the
dependence of R(B) observed at t ) 5 µs for Ph3P with F4DCNB
(1 × 10-1 mol dm-3) in 2-propanol. The results in n-hexane
and cyclohexane were obtained at 320 nm. The others were
obtained at 330 nm. The selection of each observation wave-
length was determined by respective experimental conditions.
These R(B) values correspond to the magnetic field dependence
of the yield of the escaped Ph2P•. The smooth lines are the best-
fit curves based on the d-type molecular triplet mechanism,
which is discussed later.
From Figure 4A, we observed the following peculiar MFEs
for Ph3P:
(a) The R(B) values are almost constant between 0 and 0.1
T.
(b) Above 0.1 T, they start to decrease with increasing B.
(c) Their decrements become smaller above 1 T and are
almost stationary above 5 T.
(d) The yields of the escaped Ph2P• at 10 T become as small
as 0.68, 0.62, and 0.75 of those at 0 T in 2-propanol,
cyclohexane, and n-hexane, respectively.
Figure 3. A(t) curves observed at 340 nm for (A) 4-ClPh3P in
n-dodecane, (B) 4-ClPh3P in 2-propanol, and (C) 4-ClPh3P with
F4DCNB in 2-propanol.
that the peak at 340 nm is due to the bis(4-chlorophenyl)-
phosphinyl radical (4-ClPh2P• ). Because the absorption around
480 nm shows no MFE and it was not observed in the reactions
of Ph3P and other derivatives, we shall not discuss it further.
The transient absorption spectrum of 4-BrPhPh2P (1 × 10-3
mol dm-3) in cyclohexane observed 200 ns after the excitation
showed a sharp peak at 330 nm and a weak absorption up to
600 nm (data not shown). The resemblance and clear shift of
the spectrum from that of Ph2P• implies that the observed sharp
peak can be assigned to the 4-bromophenylphenylphosphinyl
radical (4-BrPhPhP• ).
Figure 3A shows the typical A(t, B) curves observed in the
dodecane solution of 4-ClPh3P at 340 nm in the absence and
presence of magnetic fields. Figure 3B shows those observed
in a 2-propanol solution. Each of these A(t) curves consists of
a very fast rise and a slow decay. The slow decay is ascribed to
the simple disappearance process of escaped radicals, a bimo-
lecular process. Consequently, we can conclude that all pro-
cesses prior to the escape of the RPs are completed before the
decay of A(t) curves, and we can safely use the A(t) values later
than 50 ns as the measure of the yield of the escaped Ar2P•
radical.
In the study of Ph3P in 2-propanol, we reported a sharp rise
and decay immediately after the excitation together with a slow
decay.15 Preliminary picosecond laser photolysis revealed that
the lifetime of the excited Ph3P in solution was much shorter
than the pulse width of our nanosecond laser. This means that
the sharp rise observed in the 2-propanol solution is the
integrated one of the singlet and/or triplet excited states during
our nanosecond-laser pulse. However, we could not find such
an absorption for 4-ClPh3P on the nanosecond scale, which
indicates that the lifetime of its precursor excited state should
be shorter than that of Ph3P.
(e) If we take the R(B) values at 10 T as limiting values,
then the half fields of the effects, B1/2, in 2-propanol, cyclo-
hexane, and n-hexane are 0.4, 0.5, and 0.7 T, respectively.
(f) The sensitized reaction with F4DCNB showed the same
magnetic field dependence as that without F4DCNB within
experimental error.
In Figure 4B, the magnetic field dependence of the R(B)
values observed for 4-ClPh3P in 2-propanol (t ) 200 ns),
cyclohexane (t ) 100 ns), and n-hexane (t ) 100 ns) solutions

3424 J. Phys. Chem. A, Vol. 108, No. 16, 2004
Sakaguchi and Hayashi
is plotted. This figure also shows the R(B) values observed at
t ) 4 µs for 4-ClPh3P and F4DCNB (2 × 10-1 mol dm-3) in
2-propanol. The observation wavelength is 340 nm except for
the low-field data in cyclohexane (335 nm). As is the case for
Ph3P, the R(B) values are almost constant between 0 and 0.05
T. Above 0.05 T, they start to decrease with increasing B. Their
decrements become smaller above 1 T and are almost stationary
above 5 T. The R(10 T) values for the escaped 4-ClPh2P•
become as small as 0.46 (B1/2 ) 0.4 T) and 0.62 (B1/2 ) 0.5 T)
in cyclohexane and n-hexane, respectively. The R(1.75 T) value
in 2-propanol is 0.5, and its B1/2 is estimated to be 0.3 T. The
sensitized reaction with F4DCNB gave the same magnetic field
dependence as the direct one within experimental error.
In Figure 4C and D, we compared the magnetic field
dependence of the R(B) values observed for Ph3P, 3-ClPh3P,
4-ClPh3P, and 4-BrPhPh2P in cyclohexane and that of Ph3P and
2-, 3-, and 4-MePh3P in cyclohexane, respectively. The observa-
tion wavelengths are 320, 330, 325, 325, and 330 nm for
3-ClPh3P, 4-BrPhPh2P, 2-MePh3P, 3-MePh3P, and 4-MePh3P,
respectively. The data of Ph3P and 4-ClPh3P are the same as
those presented in Figure 4A and B, respectively. The B1/2 values
for 3-ClPh3P, 4-BrPhPh2P, 2-MePh3P, 3-MePh3P, and 4-MePh3P
in cyclohexane are estimated to be 0.4, 0.4, 0.4, 0.4, and 0.3 T,
respectively.
The above-mentioned MFEs are very unique as compared
with the ordinary MFEs observed in fluid solutions. The present
MFEs are characterized as follows:
(i) The magnitudes of the present MFEs (1-R(B)) are much
larger than the conventional ones. Even in the most fluid solvent
examined, n-hexane, the R(10 T) values attained such small
values as 0.75 and 0.62 for Ph3P and 4-ClPh3P, respectively.
The magnitudes of the MFEs (25 and 38% decreases) are much
larger than those of conventional MFEs in fluid solutions
(usually less than 10%).
(ii) There is no detectable MFE below 0.1 T for Ph3P and
2-MePh3P and below 0.05 T for 3-ClPh3P, 4-ClPh3P, 4-BrPhPh2P,
3-MePh3P, and 4-MePh3P, respectively. However, the effects
converge below 10 T for all Ar3P species investigated. A
magnetic field dependence such as that observed in the present
study is completely different from those expected from the
conventional mechanisms, as explained later.
(iii) The solvent dependence of the magnitude of the MFEs
for Ph3P and 4-ClPh3P (n-hexane < 2-propanol < cyclohexane)
does not correlate with the viscosity (0.313, 2.379, and 0.9751
cP, respectively). Furthermore, the effects of 2-propanol com-
pared to those of cyclohexane for Ph3P and 4-ClPh3P are very
different from each other.
(iv) The magnitude of the MFEs becomes larger by the
substitution with heavy atoms. The magnitude of the MFEs of
halogen-containing Ar3P species is larger than that of Ph3P, and
the effect of one para-bromine substitution in 4-BrPhPh2P is
almost the same as that of three para-chlorine substitutions in
4-ClPh3P. This heavy-atom effect is also position-dependent.
The para substitution of chlorine atoms exhibited a larger effect
than the meta substitution.
(v) The 3- and 4-methyl substitutions also enhanced the MFEs
in the same way that heavy atoms did. However, the 2-methyl
substitution reduced the magnitude of the MFE to be smaller
than that of Ph3P.
Previously, we reported results on MFEs of neutral RPs in
fluid solutions. In the photochemical electron-transfer reactions
between triplet xanthone and N,N-diethylaniline in 2-propanol,
we observed R(1.5 T) ) 1.13 for the escaped aniline cation
radical.21 However, no MFE was observed beyond experimental
Figure 4. R(B) values versus log(B/T) of (A) Ph3P in n-hexane and
cyclohexane and in 2-propanol in the absence and presence of F4DCNB,
(B) 4-ClPh3P in n-hexane and cyclohexane and those in 2-propanol in
the absence and presence of F4DCNB, (C) Ph3P, 3-ClPh3P, 4-BrPhPh2P,
and 4-ClPh3P in cyclohexane, and (D) Ph3P, 2-MePh3P, 3-MePh3P, and
4-MePh3P in cyclohexane.

Photodissociation of Triarylphosphine
J. Phys. Chem. A, Vol. 108, No. 16, 2004 3425
error for the hydrogen-abstraction reaction between triplet
xanthone and xanthene in 2-propanol.21 The absence of an MFE
in the last reaction is ascribed to the lack of a Coulombic
attraction between the component radicals in a pair. In a very
viscous solution of glycerin, the hydrogen-abstraction reaction
between triplet benzophenone and phenol forming a neutral RP
showed an increase of up to 1.28 at 0.34 T.22
In micellar solutions, we could observe larger MFEs than in
fluid solutions. In the photochemical hydrogen-abstraction
reactions of triplet 1,4-naphthoquinone in a sodium dodecyl
sulfate (SDS) micelle, R(10 T) for the escaped semiquinone
radical became 2.26.23 In the photodecomposition reaction of
triplet (2,4,6-trimethylbenzoyl)diphenylphosphine oxide, a phos-
phorus-containing compound, in an SDS micelle, R(1.2 T) for
the escaped diphenylphosphonyl radical was 1.12.24 It is clear
that the present MFEs observed for Ar3P species in fluid
solutions are almost as large as those observed in micellar
solutions. This must be related to the origin of the present MFEs.
3.3. Mechanism of the Magnetic Field Effects. The most
probable origin of the MFE of chemical reactions in the liquid
phase is the RP, process D in Figure 1, which should be
considered at first. The MFEs in this process have been
explained by HFCM, ∆gM, RM, and LCM.1,3,27 The absence
of an MFE in the present reactions at fields lower than 0.1 or
0.05 T is interesting. The mechanism to induce MFEs in such
low fields is ascribed to the HFCM. According to our CIDEP
results, the hyperfine coupling (hfc) constant of 31P in Ph2P• is
8.73 mT.15 The proton hfc constants of Ph• are reported to be
1.743 mT for the 2 and 6 positions, 0.625 mT for the 3 and 5
positions, and 0.204 mT for the 4 position.25 We can estimate
B1/2 due to the HFCM using the following equation:26
Figure 5. (A) Level diagram and kinetic constants to describe the
d-type triplet mechanism proposed by Steiner. (B) Modified diagram
and kinetic constants to describe the d-type molecular triplet mechanism.
seems to be too long to invoke a complete mixing between the
S and T0 states even at 10 T. Furthermore, the observed largest
decrease down to 0.46 for 4-ClPh3P in cyclohexane is beyond
the limit of ∆gM because its most optimistic estimation cannot
exceed 0.67 (2/3). Therefore, we conclude that the magnetic field
dependence of Ar3P cannot be explained by ∆gM, which was
tentatively proposed in the previous study.15
The relaxation mechanism (RM) is the third one to describe
the MFEs of RPs.28 The estimated effect due to the RM for the
photodecomposition reactions of Ar3P in fluid solutions should
be at least 1000 times smaller than the observed one15 even if
Ph2P• has a large anisotropy in its g tensor.29 MFEs due to the
LCM are sometimes observed in reactions of RPs, which have
nonzero values for the exchange interaction.27 The expected field
dependence due to the LCM is, however, completely different
from that observed in the present study. Consequently, the MFEs
of Ar3P cannot be interpreted by the conventional mechanisms,
process D in Figure 1.
2
•
2(BPh P 2 + BPh
)
•
2
B1/2(hfc) )
(1)
(BPh P• + BPh
)
•
2
where
B radical ) (ΣjIj(Ij + 1)aj2)1/2
Using the above hfc constants, we obtained a B1/2(hfc) value of
12.7 mT for an RP consisting of Ph2P• and Ph•. Because the
largest hfc constant in Ar2P• is the one for 31P and other
contributions such as that of protons at the 4 position of Ar•
are negligible, this value is applicable to all compounds in the
present study. Consequently, the absence of an MFE below 50
mT means the absence of a contribution from the HFCM beyond
experimental error.
Next we consider the d-type triplet mechanism (TM) proposed
by Steiner8 for a contact RP, process C in Figure 1. In Figure
5A, we applied the d-type TM to the present reaction scheme.
The key species in this mechanism is the triplet contact RP
(radical pair complex), 3(Ar2P• •Ar) in this figure. In the absence
of a magnetic field, each sublevel (x, y, z) of the triplet complex
decays to the starting material with a different rate constant (kisc.x
,
A convergence of the MFE at high fields was not observed
in the previous study where the maximum magnetic field was
1.34 T. In the previous study, we ascribed the origin of the
magnetic field dependence to ∆gM15 because the yield of the
escape radical from a triplet precursor should decrease because
of ∆gM.27 The photosensitization by F4DCNB again supports
the triplet precursor. However, the convergence of the depen-
dence below 5 T cannot be rationalized by ∆gM. We can
estimate the Larmor frequency difference from the g values of
Ph2P• and Ph•. In the previous study, we reported the g value
of Ph2P• to be 2.007, but a more accurate value taking into
account the second-order effect is 2.0065. The g value of Ph•
is reported to be 2.00227.25 Consequently, the ∆g value between
Ph2P• and Ph• is estimated to be 0.00423, which implies a
Larmor frequency difference of 59 kHz T-1. Thus, the S-T0
conversion takes 8.4 ns at 1 T or 0.84 ns at 10 T. Because the
lifetime of an RP in fluid solutions is 0.1-1 ns, this interval
k
isc.y, kisc.z). Under this anisotropic condition, the decay rate from
each sublevel is assumed to be as follows:
kisc.z ) z × kisc
(0 e z e 1)
(2)
(3)
kisc.x ) kisc.y ) kisc
The slower recombination from the z sublevel causes a larger
yield of the escaped radicals in the absence of a magnetic field.
This anisotropy in the ISC process levels off with increasing
B. In the limit of high field, the ISC process becomes isotropic
so that the yield of the escaped radicals may be reduced.
Therefore, the d-type TM for the contact RP predicts a decrease
of the escaped radicals due to the transition from the anisotropic
stage to the isotropic one. The yield of the escaped radicals,
Y(B), is expressed as follows:7-9

3426 J. Phys. Chem. A, Vol. 108, No. 16, 2004
Sakaguchi and Hayashi
1 + (1/3)k′isc + 6D′rot (B)
Y(B) )
(4)
(1 + k′isc)(1 + 4D′rot (B)) + 2D′rot (B)
Here, k′isc ) kisc/kesc, D′rot(B) ) Drot(B)/kesc, and the magnetic
field dependence of the rotational relaxation rate, Drot(B), is
described in the original article. kesc is the rate constant of escape
of the RP. In eq 4, kisc.z is assumed to be 0 (i.e., z ) 0). The
magnetic field dependence of Y(B) predicted by eq 4 is similar
to that shown in Figure 4, where the relative yield, Y(B)/Y(0 T)
is plotted.
Two observations support the d-type TM: (1) The absence
of the effect at low fields and the convergence at high fields
depicted in feature ii. (2) The heavy-atom effect described in
feature iv. However, we need to address several unique points
that may not be affirmative for the d-type TM. The first point
is that the key species is composed of two neutral radicals in
the present reaction. This invokes two difficulties. Because the
recombination process is spin-forbidden, the SOC interaction
between the component radicals becomes important, which
requires close interaction between the component radicals. A
neutral RP has no special interaction to hold the component
radicals as a relatively rigid complex.
Steiner investigated the photoexcited electron-transfer reac-
tions of thiazine dyes with electron donors and assumed triplet
exciplexes, each of which consists of the neutral dye radical
and cation radical.8,9 Therefore, such an exciplex has a charge
but no Coulombic interaction between them. The charge-transfer
interaction between the charged molecule and the counterpart
seems to support the rigid structure. Tero-Kubota et al.
investigated the CIDEP spectra of similar systems11 and found
a special polarization, when the MFE of the system obeys the
d-type TM. They found that no such polarization appeared when
neutral radicals are produced. These results mean that a strong
interaction, such as the charge-transfer interaction, is necessary
between the component radicals for the d-type TM for contact
RPs. No MFE due to this mechanism is expectable for neutral
RPs such as Ar2P• and Ar•.
Another important issue is the nature of the depopulation
process, which has to be competitive in rate with rotational
relaxation. In the systems investigated by the groups of Steiner
and of Tero-Kubota, the depopulation process corresponds to
back-electron transfer, which apparently can be fast enough to
compete with rotational relaxation. That σ-bond formation
between two neutral radicals should be comparably fast seems
at least doubtful. Furthermore, the escape of two neutral radicals
should be faster than that of molecules held together by charge-
transfer interaction. Faster escape, however, diminishes the
magnitude of MFE.
The second point is the solvent viscosity dependence of the
present reactions. This is the reason that we cast doubt on the
d-type TM for the contact RP in a previous study.15 According
to Steiner’s simulation, the magnitude of the MFE becomes
larger when the solvent viscosity becomes larger. As described
in feature iii, such dependences were not observed. This is
contradictory to the expectation of the d-type TM for a contact
RP. However, the strange effect of 2-propanol indicates that
the solvent dependence of the present reaction is not simple.
Apart from the discrepancy in the viscosity dependence,
several features observed in the present MFEs correspond well
to the expectation by the d-type TM. Consequently, we
investigated the viscosity dependence of the reaction of 4-ClPh3P
in detail. For that purpose, we chose n-alkanes as solvents. From
n-hexane to n-hexadecane, their viscosity changes by more than
a factor of 10, from η ) 0.313 to 3.53 cP at 20 °C. However,
Figure 6. Results and simulation of the solvent dependence of R(B)
values of 4-ClPh3P. (A) Viscosity dependence of Drot and kesc. (B)
Viscosity dependence of Drot only.
their dielectric constant changes by merely a factor of 1.11, from
1.8371 to 2.0460 at 20 °C. Consequently, we can neglect the
change in polarity that may be expected between cyclohexane
and 2-propanol, although its contribution to the neutral RP is
not clear.
In Figure 6A, the R(B) values obtained for 4-ClPh3P in these
solvents are shown. To obtain these values, we used the R(B)
at t ) 100 ns observed at 340 nm. The simulated R(B) curves
for the solvents are discussed later. The MFE becomes larger
from n-hexane to n-hexadecane, and we can definitely conclude
that this dependence originates from viscosity. The R(B) values
at 1.5 T range from 0.67 in n-hexane to 0.41 in n-hexadecane.
This means that the magnitude of the MFE becomes 1.6 times
larger when the viscosity increases by a factor of 11.
Steiner and Haas investigated the viscosity dependence of
the MFE in the photochemical electron-transfer reaction between
triplet methylene blue and p-iodoaniline in a mixture of methanol
and ethylene glycol.9 The mechanism of its MFE is interpreted
by the combined effects of the d-type TM and ∆gM. They
presented a simulation to extract the contribution of d-type TM.
The comparison with their results and ours indicates that the
viscosity dependence of methylene blue is much larger than
that of 4-ClPh3P. The enhancement of the MFE of methylene
blue observed in the increase of the viscosity by 3.5-fold seems
much larger than that of 4-ClPh3P, which was observed to have
an 11-fold viscosity increase. The shift of B1/2 values in the
increase of the viscosity by a factor of 3.5 is similar to ours
when the viscosity increases by a factor of 11. A quantitative
evaluation of the observed MFE may shed light on this
discrepancy.
Using eq 4, we tried to fit the relatiVe yields observed in the
present reactions by calculating R(B) ) Y(B)/Y(0). In these
fittings, kisc, kesc, Drot are the variables to be optimized. Because
the three independent parameters are still abundant, it is difficult

Photodissociation of Triarylphosphine
J. Phys. Chem. A, Vol. 108, No. 16, 2004 3427
to evaluate the validity of the obtained values for the data shown
in Figure 4. As for the results shown in Figure 6A, the
parameters are restricted by the viscosity. kesc and Drot are
considered to be dependent on solvent viscosity. We made a
global nonlinear least-squares fit of the whole data set shown
in Figure 6A by adding the following limitations:
For a contact RP, the independence of kesc from solvent
viscosity seems hardly acceptable, but the scheme becomes
sensible by reassigning the origin of the present MFE to process
A in Figure 1. This means that the excited triplet state of Ar3P
is affected by external magnetic fields, as is the case for Paul
and co-workers.14 This model is shown in Figure 5B. For
convenience, we would like to call this mechanism the d-type
molecular TM. In this scheme, the kesc value is regarded as the
dissociation rate, kdis, of triplet Ar3P. If this is the case, then
we can consider the kdis value not to be strongly dependent on
solvent viscosity. At the same time, we can avoid the situation
in which the obtained kesc value (1.90 × 1010 s-1) is much larger
than the possible escaping rate of RPs (4.8 × 108 s-1).
(1) kisc is a common constant in all alkane solvents.
(2) kesc and Drot are inversely dependent on solvent viscosity.
The second assumption is well established, and theoretical
expressions are given.30,31 Instead of substituting such theoretical
values, we fit a set of kesc and Drot values for n-hexadecane,
and others are fixed to the values inversely dependent on their
viscosity. With this assumption, we could fit the data shown in
Figure 6A with the following values: kisc ) 3.25 × 1011 s-1
,
For Drot, this scheme also gives a reasonable value. According
kesc ) 1.80 × 105 s-1, and Drot ) 3.22 × 109 s-1 for
n-hexadecane. The kesc and Drot values for other solvents are
obtained by multiplying by the ratio η(n-hexadecane)/η(solvent).
The fitted curves are shown in Figure 6A. We note the large
discrepancies on the low- and high-viscosity sides indicating
that the viscosity dependence of the MFE of 4-ClPh3P is smaller
than expected for the d-type TM for a contact RP. The numerical
simulations indicate that this size of the change can be explained
by the viscosity increase by a factor of about 2.5, not 11.
Furthermore, the fitted value of kesc is 106 times smaller than
that of kisc, which means that there are almost no escaped
radicals. Although we did not measure the quantum yields of
the present reactions, it is very unlikely that we could observe
radical formation with a quantum yield smaller than 10-6. This
result is, of course, due to the limitation of the fitting process
to optimize the relative yield, R(B) ) Y(B)/Y(0), instead of the
real yield, Y(B). Similarly, we could have observed neither a
kinetic process having a rate constant of 1.80 × 105 s-1 in
n-hexadecane nor a process having a rate constant of 2.03 ×
106 s-1 in n-hexane.
to Debye’s derivation, the following relationship is expected:30
kT
8πR3η
Drot
)
(6)
Here, k is the Boltzmann constant, and R is the effective
hydrodynamic radius of the rotating species. By substituting
the Drot and η(n-hexadecane) values, we get the R value of
4-ClPh3P to be 4.0 Å for spherical geometry. Steiner and Haas
obtained 3.6 Å for methylene blue/aniline complexes.9 Consid-
ering the molecular structures of these compounds, we consider
the present R value to be reasonable. The obtained rate constants
based on this scheme show that the dynamic process that induces
the present MFEs terminates within 1 ns after excitation, which
is consistent with our observation. Therefore, theoretical
calculations and experimental observations support the d-type
molecular TM.
The model can further explain the strange solvent effect of
2-propanol indicated in feature iii. As mentioned in the previous
study, Ph3P has closely lying nπ* and ππ* excited states.32
Consequently, the locations of these two states are dependent
on the solvent polarity and on the substituents, as is the case
for aromatic carbonyl compounds. The change in mixing of
these two states induces a variation in the character of their
triplet sublevels and in the dissociation rate. This may be the
reason for the difference between Ph3P and 4-ClPh3P in
2-propanol. Preliminary picosecond laser photolysis revealed
that the lifetime of the excited singlet or triplet state of Ph3P in
2-propanol is longer than that in cyclohexane. Although the
character of this state has not yet been defined, this is clear
evidence for the solvent dependence of the excited states.
The lifetime, τ, of a neutral RP is estimated to be τ ) 4r2/
D.31 Here, r is the radius of the component radical, and D is
the sum of the diffusion coefficients of each radical. If we derive
D from the Stokes-Einstein relation, then kesc () 1/τ) is
described as follows:
kT
12πr3η
kesc
)
(5)
Here, k is the Boltzmann constant, and T is the temperature. If
we assume the radius of a radical to be 4.0 Å (this may be an
overestimate, see below), then kesc in n-hexadecane at 20 °C is
estimated to be 4.8 × 108 s-1. This rate is 2.7 × 103 times
larger than the one derived by the optimization. These discrep-
ancies seem to imply the invalidity of the model. Consequently,
the experimental and theoretical arguments require us to discard
the physical model of the original d-type TM for our reaction.
Therefore, we can exclude process C in Figure 1.
3.4. d-Type Triplet Mechanism in the Photoreactive
Molecular Triplet State. We obtained much better fitting
curves shown in Figure 6B by introducing new limitations as
follows:
(1) kisc is a common constant in all alkane solvents.
(2) kesc is also a common constant in all alkane solvents.
(3) Drot is inversely dependent on the solvent viscosity.
With these assumptions, we obtained the following values:
kisc ) 1.87 × 1011 s-1, kesc ) 1.90 × 1010 s-1, and Drot ) 7.02
× 108 s-1 for n-hexadecane. The Drot values for other solvents
are obtained by multiplying by the ratio η(n-hexadecane)/
η(solvent).
The next evidence to support the d-type molecular TM is
the MFEs of methyl-substituted compounds. As described in
feature v, the observed MFEs of 3- and 4-MePh3P are almost
the same as that of 4-BrPhPh2P. Because methyl substitution is
not expected to have large SOC interaction, the large MFE
without heavy atoms can be ascribed to the modification of the
nπ* and ππ* excited states as in the case of the solvent polarity.
The absence of the difference between 3- and 4-substitution of
the methyl group is clearly different from that of the chlorine
atom as described in feature iv. This may also come from the
difference in the mechanism (i.e., nπ*-ππ* modification and a
direct SOC contribution). However, large effects of methyl
substitution in the RP are not expected. In the case of 2-MePh3P,
its small MFE may be ascribed to its steric hindrance if the
triplet molecule itself is the origin of the MFE. For the RP
formed in this case, its steric hindrance seems to be significantly
reduced in comparison with that of the original molecule.
Consequently, the smaller and strange viscosity dependence and

3428 J. Phys. Chem. A, Vol. 108, No. 16, 2004
Sakaguchi and Hayashi
TABLE 1: Observed R(B) Values at 1.5 T, Derived Rate Constants, and Estimated Yields of Escaped Radicals at 0 T
compound
4-ClPh3P
4-ClPh3P
4-ClPh3P
4-ClPh3P
4-ClPh3P
Ph3P
solvent
R(1.5 T)
kisc/1011s-1
k
dis/1010s-1
Drot/109s-1
Y(0)
n-hexadecane
n-tetradecane
n-dodecane
n-decane
n-hexane
n-hexane
cyclohexane
cyclohexane
cyclohexane
cyclohexane
cyclohexane
cyclohexane
cyclohexane
2-propanol
2-propanol
2-propanol
2-propanol
0.41
0.45
0.47
0.54
0.67
0.80
0.65
0.64
0.57
0.53
0.80
0.56
0.55
0.71
0.72
0.51
0.54
1.87
1.87
1.87
1.87
1.87
1.53
1.91
1.75
2.05
2.27
1.02
1.69
1.61
1.74
1.88
1.67
1.64
1.90
1.90
1.90
1.90
1.90
3.91
3.84
3.08
2.87
2.38
3.84
2.17
1.98
5.04
5.56
2.27
2.38
0.702
(1.07)
(1.64)
(2.67)
(7.92)
(7.92)
(2.54)
(2.54)
(2.54)
(2.54)
(2.54)
(2.54)
(2.54)
(1.04)
(1.04)
(1.04)
(1.04)
0.360
0.344
0.325
0.298
0.228
0.378
0.398
0.374
0.349
0.316
0.497
0.327
0.318
0.468
0.472
0.373
0.383
Ph3P
3-ClPh3P
4-BrPhPh2P
4-ClPh3P
2-MePh3P
3-MePh3P
4-MePh3P
Ph3P
Ph3P + F4DCNB
4-ClPh3P
4-ClPh3P + F4DCNB
the effect of methyl substitution seem to be enough for us to
conclude that the species sensitive to the external magnetic field
is the excited triplet state of Ar3P and not their RP.
3.5. Comparison among Compounds. In Figure 4, we
presented the simulation curves of the experimental results. They
are obtained under the following assumptions:
(1) The Drot values are derived from that of 4-ClPh3P in
n-hexadecane by multiplying by η(n-hexadecane)/η(solvent).
The size difference has been neglected.
However, if we accept the fact that the fate of the excited
triplet species is magnetic field-dependent, then we should be
aware of another process that may exhibit anisotropy in addition
to the depopulation process. That is the population process to
the triplet state, the rate of which is kpop, in Figure 5B. This
corresponds to the p-type TM, process B in Figure 1. The result
of the sensitization experiment allows us to assess the contribu-
tion of the p-type TM. If, in the intramolecular triplet population
process, the zero-field sublevels were populated at different
rates, then the triplet sensitization experiment should lead to
an MFE that is different in size from that observed on direct
excitation. As described above, the magnetic field dependence
of the sensitized reaction is the same as that in the absence of
sensitizer. Therefore we can certainly discard contributions from
the p-type TM. The dissociation of the triplet state is definitely
accomplished within the triplet manifold. There should be no
anisotropy in kdis of each triplet sublevel.
(2) The kisc and kdis values are the variables to be fit.
The R(B) values at 1.5 T, the derived rate constants, and the
estimated escaped radical yields at zero field, Y(0), are shown
in Table 1, where the data are arranged by solvents. The
reasonable values of Y(0) support the validity of the fitting
processes optimizing R(B) ) Y(B)/Y(0). Of course, these values
should be regarded as qualitative because the applied limitations
are too restricted to derive the real parameters, particularly in
the absence of recombination from the triplet z sublevel (the
assumption that the z value in eq 2 is zero.). Many numerical
simulations to interpret the effect of halogen or methyl substitu-
tion revealed that the most effective single parameter among
kisc, kdis, and z is z when the others are taken to be common
among the derivatives. Under such conditions, the optimized z
value is generally small (0.01-0.09, except 0.23 for 2-MePh3P),
as expected. In these optimizations, 4-ClPh3P had the smallest
z value, which is in line with the expectation from the SOC.
However, we have no choice but to set z ) 0 to obtain a reliable
Consequently, we can draw the following conclusions:
(1) Our results are explained in terms of the MFE of the
excited triplet molecules and not in terms of the succeeding
radical pairs.
(2) Our results are explained by the magnetic field dependence
of the depopulation kinetics of the excited triplet molecules and
not by the magnetic field dependence of the population or
dissociation kinetics.
Drot of 4-ClPh3P in n-hexadecane. Therefore, the interpretation
of Table 1 must be done in keeping with these uncertainties.
In Table 1, we can see the following trends:
(1) The kdis value of Ph3P is larger than those of other
compounds except for 2-MePh3P.
This means that our results can be interpreted by the d-type
TM in the photoreactive triplet state. In this paper, we exclude
processes B-D in Figure 1 from the candidates for the present
MFEs, and process A is ascribed to their origin. This elimination
of possible mechanisms has been a lengthy procedure, but it
was unavoidable to reach a reliable conclusion. As described
in the Introduction, Paul and co-workers assigned the origin of
the MFE of the sensitized photodecomposition of azocumene
to the same mechanism.14 Interestingly, the yield next to 0 T
(maybe 0.1 T) shown in Figure 6 is clearly smaller than the
yield at 0 T.14 Of course, the starting field of the isotropic
transition can be smaller than 0.1 T in the d-type TM; such
field dependence, which starts smoothly from 0 T, can also
originate from the RPM. The presented CIDEP spectrum of the
cumyl radical is a little bit weaker on the higher-field side, which
suggests a contribution of the RPM. Although the saturation of
MFE at high fields seems to support their assignment, they did
not explicitly exclude other possibilities.
(2) In the halogen-substituted compounds, the increase of
MFE (1-R(B)) mainly reflects the increase in kisc.
(3) In the methyl-substituted compounds, the increase of MFE
(1-R(B)) mainly reflects the decrease in kdis.
Observation 1 seems to indicate that any substitution tends
to stabilize the triplet states. This may be due to the lowering
of the energy levels of the excited triplet states by substitution.
The increase of kisc (observation 2) is in good agreement with
the heavy-atom effects enhancing intersystem crossing. To some
extent, the decrease of kdis (observation 3) is also responsible
for the increase of the MFE in the series of halogen-substituted
compounds, but the changes in kdis are smaller than those of
the methyl-substituted compounds. However, observation 2 is
not applicable to the methyl-substituted compounds. This
supports the discussion in section 3.4 that the mechanisms to
enhance the MFEs of halogen- and methyl-substituted com-
pounds are different. For further discussion of these reactions,

Photodissociation of Triarylphosphine
J. Phys. Chem. A, Vol. 108, No. 16, 2004 3429
the determination of the lifetime of the excited states by
picosecond photolysis system is needed, but this is beyond the
scope of the present study.
E. Steiner of Konstanz University for discussions and supplying
the computer program to evaluate the triplet mechanism for the
assignment of the characteristic parameters. The program was
completely rewritten for the numerical fitting to optimize the
parameters and for the d-type molecular triplet mechanism.
Therefore, any fault in the analyses of the present results is our
responsibility. We are also sincerely grateful to Professor
Tadashi Okada and Professor Yoshinori Hirata of Osaka
University for picosecond transient absorption measurements.
4. Conclusions
The MFEs of triarylphosphines were found to be observable
in dilute solutions, irrespective of the solvent polarity. The
concentration of the substrate is arbitrary for the present MFEs
because the reactions for the MFEs are unimolecular. Such
requirements for large MFEs cannot be fulfilled by the d-type
TM and the RPM of unchained RPs. From the MFE, it has to
be concluded that the reaction rates of the triarylphosphines are
very large. This means that it is very hard to quench these
reactions. However, MFEs of the same magnitude appeared in
the sensitized reactions. Last but not least, the magnitude of
the MFEs of triarylphosphines was very large. The above
conditions are very suitable to the practical application of the
present MFEs. If we use triarylphosphine as an initiator of
radical reactions, then we can almost freely settle the reaction
condition by considering only the reactant. Some of the reactions
that obey the d-type molecular TM may afford this option. The
large MFE without heavy atoms other than phosphorus is still
the excellent merit of triarylphosphines. Halogen atoms that
pollute the environment if applied on a large (technical) scale
are not required to induce the MFE for triarylphosphines.
The magnetic field effects on the photodecomposition reac-
tions of triarylphosphines in fluid solutions were investigated
in the present study. The yield of the diarylphosphinyl radicals
was invariant from 0 to 0.05 T. Then it started to decrease up
to 5 T and became invariant again above it. The magnetic field
effect became larger by halogen substitution and by 3- and
4-methyl substitution. The viscosity dependence of tris(4-
chlorophenyl)phosphine using n-alkanes revealed that the effect
became larger with increasing solvent viscosity. This depen-
dence was similar to that due to the d-type triplet mechanism
reported by Steiner but was much smaller than the expectation
of the standard mechanism for the triplet contact RPs. By
assigning the magnetic field-dependent process to the excited
triplet state as the precursor of the contact RP, the discrepancy
was removed. The closely lying 3nπ* and 3ππ* states of
triarylphosphine have been suggested to explain the difference
in the solvent effect of polar and nonpolar solvents and the
enhancement of the magnetic field effects of 3- and 4-methyl
substitutions.
References and Notes
(1) Nagakura, S.; Hayashi, H.; Azumi, T. Dynamic Spin Chemistry;
Kodansha: Tokyo, 1998.
(2) Kaptain, R. J. Am. Chem. Soc. 1972, 94, 6251.
(3) Steiner, U. E.; Ulrich, T. Chem. ReV. 1989, 89, 51.
(4) Katsuki, A.; Kobori, Y.; Tero-Kubota, S.; Milikisyants, S.; Paul,
H.; Steiner, U. E. Mol. Phys. 2002, 100, 1245.
(5) Steiner, U. Z. Naturforsch., A 1979, 34, 1093.
(6) Steiner, U. Chem. Phys. Lett. 1980, 74, 108.
(7) Steiner, U. Ber. Bunsen-Ges. Phys. Chem. 1981, 85, 228.
(8) Ulrich, T.; Steiner, U. E.; Fo¨ll, R. E. J. Phys. Chem. 1983, 87,
1873.
(9) Steiner, U. E.; Haas, W. J. Phys. Chem. 1991, 95, 1880.
(10) Katsuki, A.; Akiyama, K.; Tero-Kubota, S. J. Am. Chem. Soc. 1994,
116, 12065.
(11) Tero-Kubota, S.; Katsuki, A.; Kobori, Y. J. Photochem. Photobiol.,
C 2001, 2, 17 and references therein.
(12) Savitsky, A. N.; Batchelor, S. N.; Paul, H. Appl. Magn. Reson.
1997, 13, 285.
(13) Savitsky, A. N.; Paul, H.; Shushin, A. I. J. Phys. Chem. A 2000,
104, 9091.
(14) Milikisyants, S.; Katsuki, A.; Steiner, U.; Paul, H. Mol. Phys. 2002,
100, 1215.
(15) Sakaguchi, Y.; Hayashi, H. Chem. Phys. Lett. 1995, 245, 591.
(16) Preliminary reports are presented in New Aspects of Photochemistry
and Reaction Dynamics, Kisarazu, Japan, July 1997 and Symposium on
Photochemistry 1997 IA108, Sendai, Japan, September 1997.
(17) Sakaguchi, Y.; Hayashi, H. J. Phys. Chem. 1984, 88, 1437.
(18) Jeschke, G.; Wakasa, M.; Sakaguchi, Y.; Hayashi, H. J. Phys. Chem.
1994, 98, 4069.
(19) Otrebski, W.; Getoff, N.; Wilke, G. Radiat. Phys. Chem. 1984, 23,
691.
(20) Horn, K. A.; Whitenack, A. A. J. Chem. Phys. 1988, 92, 3875.
(21) Igarashi, M.; Sakaguchi, Y.; Hayashi, H. Chem. Phys. Lett. 1995,
243, 545.
(22) Levin, P. P.; Khudyakov, I. V.; Kuzmin, V. A. J. Phys. Chem.
1989, 93, 208.
(23) Nakamura, Y.; Igarashi, M.; Sakaguchi, Y.; Hayashi, H. Chem.
Phys. Lett. 1994, 217, 387.
(24) Hayashi, H.; Sakaguchi, Y.; Kamachi, M.; Schnabel, W. J. Phys.
Chem. 1987, 91, 3936.
(25) Zemel, H.; Fessenden, R. W. J. Phys. Chem. 1975, 79, 1419.
(26) Weller, A.; Nolting, F.; Steark, H. Chem. Phys. Lett. 1983, 96, 24.
(27) Sakaguchi, Y.; Hayashi, H.; Nagakura, S. Bull. Chem. Soc. Jpn.
1980, 53, 39.
(28) Hayashi, H.; Nagakura, S. Bull. Chem. Soc. Jpn. 1984, 57, 322.
(29) Geoffroy, M.; Lucken, E. A. C.; Mazeline, C. Mol. Phys. 1974,
28, 839.
(30) Debye, P. Polar Molecules; Dover Publications: New York, 1945.
(31) Salikhov K. M. Magnetic Isotope Effect in Radical Reactions;
Springer-Verlag: Wien, Germany, 1996.
Acknowledgment. Y.S. acknowledges support from a Grant-
in-Aid for Scientific Research (08640662) from the Ministry
of Education, Culture, Sports, Science and Technology, Japan,
and from the MR Science Research Project of RIKEN. H.H.
acknowledges support from a program entitled “Research for
the Future” of the Japan Society of the Promotion of Science
(RFTF:99P01201). We are sincerely grateful to Professor Ulrich
(32) Fife, D. J.; Morse, K. W.; Moore, W. M. J. Photochem. 1984, 24,
249.