Chemiluminescence of 1,1′-biisoquinolinium and 2,2′-biquinolinium salts. Reactions of electron-rich olefins with molecular oxygen

Article abstract of DOI:10.1039/a702289a

2,2′-Dialkyl-1,1′-biisoquinolylidenes (BIQ), having been prepared by two-electron reduction of the corresponding diquaternary salts BIQ²⁺(X^-)₂, react with triplet dioxygen to produce chemiluminescence (CL). In aprotic solvents, the kinetics of the reaction are first-order with respect to the concentrations of the substrate and of oxygen. The second-order rate constants at 25°C increase in the order DMF < MeCN < DMSO, which correlate with the free energy change for an electron-transfer (ET) from BIQ to O₂, although this process is endothermic in all three solvents. In non-polar solvents such as benzene, the reaction proceeds much more slowly. From these results, a reaction mechanism is proposed as follows. An ET or partial charge-transfer gives a radical ion pair {BIQ^.+ ... O₂^.-} or a CT complex, in which intersystem crossing takes place from the triplet to the singlet state, and then radical coupling at the 1-position followed by cyclisation yields a 1,2-dioxetane, the postulated intermediate of the CL reaction. An electron-rich olefin having a closely related structure, 1,1′-dimethyl-2,2′-bisquinolylidene (BQ), shows a similar redox behaviour to that of BIQ. Reaction of BQ with O₂ produces CL, and the emitter has been identified as the singlet excited state of 1-methyl-2(1H)-quinolinone 8a. The effects of substituents in the quinoline ring on the autoxidation rate have been investigated in MeCN and DMSO. An olefin having the more negative oxidation potential shows the higher reactivity to O₂, suggesting that the autoxidation of BQ should take place via a similar reaction pathway as that of BIQ.

Full text of DOI:10.1039/a702289a

View Online / Journal Homepage / Table of Contents for this issue
Chemiluminescence of 1,1Ј-biisoquinolinium and 2,2Ј-biquinolinium
salts. Reactions of electron-rich olefins with molecular oxygen
Yukie Mori,* Kumiko Isozaki and Koko Maeda
Department of Chemistry, Faculty of Science, Ochanomizu University, Otsuka, Bunkyo-ku,
Tokyo 112, Japan
2
,2Ј-D ialkyl-1,1Ј-biisoquinolylidenes (BIQ), having been prepared by two-electron reduction of the
2ϩ Ϫ
corresponding diquaternary salts BIQ (X ) , react with triplet dioxygen to produce chemiluminescence
2
(
CL). In aprotic solvents, the kinetics of the reaction are first-order with respect to the concentrations of
the substrate and of oxygen. The second-order rate constants at 25 ЊC increase in the order
D M F < M eCN < D M SO, which correlate with the free energy change for an electron-transfer (ET) from
BIQ to O , although this process is endothermic in all three solvents. In non-polar solvents such as
2
benzene, the reaction proceeds much more slowly. From these results, a reaction mechanism is proposed as
ϩ
Ϫ
follows. An ET or partial charge-transfer gives a radical ion pair {BIQ ؒ ؒ ؒ O2 } or a CT complex, in
᝽
᝽
which intersystem crossing takes place from the triplet to the singlet state, and then radical coupling at the
-position followed by cyclisation yields a 1,2-dioxetane, the postulated intermediate of the CL reaction.
1
An electron-rich olefin having a closely related structure, 1,1Ј-dimethyl-2,2Ј-bisquinolylidene (BQ), shows
a similar redox behaviour to that of BIQ. Reaction of BQ with O produces CL, and the emitter has been
2
identified as the singlet excited state of 1-methyl-2(1H )-quinolinone 8a. The effects of substituents in the
quinoline ring on the autoxidation rate have been investigated in M eCN and D M SO. An olefin having the
more negative oxidation potential shows the higher reactivity to O , suggesting that the autoxidation of
2
BQ should take place via a similar reaction pathway as that of BIQ.
It was reported that 2,2Ј-ethylene-1,1Ј-biisoquinolinium
2
ϩ
Ϫ
5
8
4
dibromide 1 (Br ) produced a chemiluminescence (CL) on
2
6
7
3
N
N
addition of base and hydrogen peroxide in hydroxylic solv-
2
R
N+
N
N
1
–3
2ϩ
ents. Recently, we reinvestigated the CL reaction of 1 in
aprotic solvents and proposed that the two-electron reduced
species 1 reacted with molecular oxygen to give a 1,2-dioxetane
intermediate, thermal decomposition of which generated the
R
k2
O
O
R
1
2e–
R' +O2
+
R'
R'
N+
excited singlet state of 2,2Ј-ethylene-bis[1(2H)-isoquinolinone]
4
4
2
as the emitter (Scheme 1). Trimethylene-bridged analogue
BIQ2+
R,R' = -(CH2)2-
R,R' = -(CH ) -
BIQ
1, 2, 3
2
ϩ
Ϫ
2ϩ
Ϫ
(Br ) and an unbridged system 3 (I ) also exhibited CL
under similar conditions, although the intensity was lower than
that for 1 . The crucial factors controlling reactivity to O and
2+
2
2
1
2
3
2
+
2
3
2
ϩ
2+
R = R' = Me
2
from 3
from 1 or 2
CL quantum yield for this system have not yet been clarified.
The dihedral angles between the two isoquinoline planes are
2
ϩ
2ϩ 5,6
different among 1 –3 , which probably affects the electronic
structure and the reactivity of the reduced species.
Addition of triplet dioxygen to a C᎐C double bond in the
singlet state olefin is spin-forbidden. In the reaction of an
᎐
N
N
N
(CH2)n
Me
electron-rich olefin with O , a spin inversion process can take
2
O
4 n = 2, 5 n = 3
O
O
7
place via a charge-transfer (CT) complex. Alternatively, an
active oxygen species, such as O2 , ؒOOH or O , may be formed
Ϫ
1
6
+ hν
+ hν
2
as an intermediate. Thummel et al. reported cycloaddition of
Scheme 1
molecular oxygen to a C᎐C bond in diimidazolinylidenes, but
no reaction mechanism was described. In order to clarify
a detailed mechanism of the reaction of such electron-rich
᎐
8
ϩ
BQ᝽ , plausible intermediates of the CL reaction, are also
discussed on the basis of EPR and NMR spectra and semi-
empirical MO calculations.
olefins with O , a more systematic, quantitative investigation is
2
required. In the present study, we have investigated the kinetics
of oxygenation of electron-rich olefin 1 in various solvents and
proposed involvement of an electron transfer (ET) or a charge
transfer (CT) process in the rate determining step. As a closely
related system, the redox properties and CL reaction of 2,2Ј-
Results and discussion
2
؉
CL efficiency of BIQ salts ϩ
2
In the CL reaction of BIQ salts with alkaline hydrogen per-
2
ϩ
2ϩ
biquinolinium (BQ ) salts have been examined. The structures
oxide in MeOH–H O, 1 produced the most intense light
2
ϩ
and spin density distribution of radical cations BIQ᝽ and
among the three. The ratio of the integrated CL intensity was
2
ϩ
2ϩ
2ϩ
2ϩ
2ϩ
2ϩ
1
:2 :3 = 1:0.1:0.02. The order of 1 > 2 > 3 was in
1
agreement with the results reported by Mason and Roberts.
The difference in the CL efficiency, however, does not necessar-
ily reflect the difference in the yield of the singlet excited state.
*
Present address: Molecular Photochemistry Laboratory, The Institute
of Physical and Chemical Research (RIKEN), Wako, Saitama, 351-01,
Japan.
E-Mail ymori@postman.riken.go.jp
2ϩ
Ϫ
The chemical yield of 4 from 1 (Br ) was ca. 90%, while
2
J. Chem. Soc., Perkin Trans. 2, 1997
1969

View Online
Table 2 Second-order rate constants for reaction of 1 with O at 25 ЊC
ϩ
a
Table 1 Redox potentials EЊЈ (V vs. Ag/Ag ) in various solvents
2
Ϫ3
Ϫ1
3
Ϫ1 Ϫ1
Compd.
MeCN
DMF
DMSO
Solvent
Method
[O ]/mol dm
kobs/s
k /dm mol
s
2
2
2
2
2
ϩ
ϩ
ϩ
2
a
Ϫ3
3
1
2
3
7
7
7
Ϫ0.42, Ϫ0.80
Ϫ0.645, Ϫ0.945
Ϫ0.82, Ϫ0.93
Ϫ0.57, Ϫ0.72
Ϫ0.62, Ϫ0.78
Ϫ0.68, Ϫ0.79
Ϫ1.20
Ϫ0.51, Ϫ0.81
Ϫ0.73, Ϫ0.96
Ϫ0.93
Ϫ0.66, Ϫ0.72
Ϫ0.705, Ϫ0.78
Ϫ0.77
Ϫ0.57, Ϫ0.86
Ϫ0.79, Ϫ1.00
Ϫ0.98
Ϫ0.735
Ϫ0.785
DMSO
MeCN
MeCN
MeCN
DMF
CV
2.1 × 10
2.63
2.43
0.51
0.32
0.452
0.090
1.25 × 10
2.96 × 10
Ϫ3
2
CV
CVb
CL
8.21 × 10
1.72 × 10
1.72 × 10
4.72 × 10
Ϫ3
2
2.94 × 10
ϩ
ϩ
Ϫ3
2
a
b
c
O2
1.9 × 10
2
Ϫ3
CV
CL
96
91
3.4 × 10
2
ϩ
Ϫ4
Ϫ0.81
Ϫ1.16
DMF
9.9 × 10
9.1 × 10
c
Ϫ3
Ϫ4
Ϫ2
Ϫ1.25
Benzene UV–VIS
3.1 × 10
a
Ϫ3
Ϫ1
a
b
c
0
.1 mol dm Bu NBF was contained. The scan rate was 20 mV s
.
Cyclic voltammetry. Time course of CL intensity. Decay of the
4
4
absorption of 1.
2
ϩ
Ϫ
3
1
2
(I ) gave a number of by-products other than 2-methyl-
2
lists the observed rate constants of the disappearance of 1
under pseudo-first-order conditions, kobs. As for the electro-
chemical studies, the kobs values were determined from the
(2H )-isoquinolinone 6. Furthermore, the actual emitter for
2
ϩ
2ϩ
or 3 was not the S state of the isoquinolinone but the
1
4
excited state of another compound formed via energy transfer.
ϩ
2
ϩ
cathodic peak potentials of the 1᝽ /1 couple at various scan
A reaction of 1 with KO in MeCN also produced a CL
2
rates, using a working curve for an EC reaction investigated by
emission and gave 4 in a high yield (>99%). A trace amount
1
4
Nicholson and Shain. The presence of O did not affect the
2
(
<1%) of the intramolecular [2 ϩ 2] cycloadduct of 4 was also
2
ϩ
ϩ
1
redox wave of the 1 /1᝽ couple. The reaction rate of the dis-
detected by H NMR spectroscopy. The CL quantum yield was
determined to be (6 ± 1) × 10 . This value was comparable
Ϫ3
appearance of 1 was found to be first order with respect to the
concentration of O , and therefore it is expressed by eqn. (2).
2
with that of 9,10-dicyanoanthracene–KO reaction in THF
2
Ϫ3
9
(
2 × 10 ) reported by Breslin and Fox. For direct (not via
Ϫd[1]/dt = k [1][O ]
(2)
2
2
energy transfer to a secondary emitter) CL reactions, ΦCL is
expressed as eqn. (1) where Φ , Φ * and Φ are the chemical
R
S
F L
15
Using the reported data for solubility of O2, bimolecular
rate constants k were obtained. The k values determined by
2
2
ΦCL = Φ Φ * Φ
(1)
R
S
F L
spectrophotometric and electrochemical methods were in
agreement with each other. In nonpolar solvents such as ben-
zene, the disappearance of 1 was much slower than that in polar
reaction yield, fraction of the S state and fluorescence quan-
1
tum yield of the emitter, respectively. In the present case,
because no alternative pathway was observed to compete with
the CL process, ΦR is equal to unity. Using ΦF L value of 0.097
for 4 in aerated MeCN,† eqn. (1) gives Φ * of 0.06, which is
much higher than that for decomposition of unsubstituted or
solvents, suggesting that the 1–O reaction should involve a
2
polar intermediate in the rate determining step. As is seen in
Table 2, however, the k value in DMSO was more than ten
2
S
times larger than that in DMF in spite of a similar polarity.
Ϫ6
Ϫ3 10
Inspection of Tables 1 and 2 reveals that the rate constant k
correlated with the free energy change for an electron-transfer
(ET) from 1 to O , although this ET process was endothermic
∆GЊ = 29–42.5 kJ mol ) in all the solvents studied. The more
2
methyl-substituted 1,2-dioxetanes (3 × 10 –2.5 × 10 ). As
for some dioxetanes with an electron-rich substituent, intra-
1
1
12
13
2
molecular CIEEL or a CT mechanism has been proposed,
Ϫ1
(
but this process is unlikely to occur for neutral symmetrical
dioxetanes. Intermolecularly, aromatic compounds having a
low oxidation potential such as 1 (Table 1) could play the role
negative the redox potential, the larger the k2 value was
observed.
1
1,12
2ϩ
Eberlein and Bruice investigated autoxidation of 1,10-
ethano-5-ethyl-1,5-dihydrolumiflavin (Flred) in an aqueous solu-
tion and proposed that Fl transferred an electron to O to
of an activator in CL reactions.
In the reaction of 1 , how-
ever, it was not 1* but 4* that emitted light, and therefore, this
possibility was ruled out. Under these conditions the CL inten-
sity of 2 or 3 was lower than that of 1 . In the case of 2 ,
the major product was 5 and the intramolecular cycloadduct of
was also formed (ca. 10% yield). Since the cycloadducts had
red
2
ϩ
Ϫ
2
ϩ
2ϩ
2ϩ
2ϩ
yield a radical pair {Fl᝽ ؒ ؒ ؒ O ᝽ } followed by radical coupling
2
leading to a zwitterionic intermediate, although this ET reac-
Ϫ1 16
tion was rather endothermic (∆GЊ = 57 kJ mol ). Nanni et al.
5
2
ϩ
4
reported that radical cation of methyl viologen (MV ) reacted
17,18
been obtained by triplet-sensitised photolysis of 4 or 5, these
products may be formed from the chemically generated T state
with O in a DMF solution.
In the presence of a large excess
ϩ
2
1
ϩ
2
ϩ
Ϫ
of MV᝽ , O
2
Ϫ
was reduced via one ET from MV᝽ , and the
in the CL reaction. On addition of KO in MeCN, 3 (I ) gave
2
2
ϩ
resultant O
reacted with another molecule of MV to give a
2
᝽
᝽
a complex mixture including 6.
radical-coupling product. They suggested that the coupling
Ϫ
ϩ
18
would be fast, although O2 can reduce MV to MV. In
᝽
᝽
Reaction mechanism and kinetics
We focus on the CL reaction of 1 and its reduced species in
aprotic solvents for the following reasons: (i) the chemical yield
ϩ
Ϫ
2
ϩ
order to examine whether 1᝽ reacts with O
2
᝽ to give 4* in
aprotic solvents or not, KO was added to an MeCN solution of
2
ϩ
1
. A bright emission was instantaneously observed, and the
CL spectrum coincided with the fluorescence spectrum of 4.
᝽
of 4 was high, (ii) the emitter is the S state of 4 which is formed
1
Ϫ
directly via thermal decomposition of the dioxetane, (iii), O2
is stable against disproportionation on the timescales of our
᝽
From these facts, a possible reaction mechanism is proposed.
ϩ
Ϫ
2
One ET from 1 to O
2
yields a radical ion pair {1᝽ ؒ ؒ ؒ O
2
᝽ },
experiments (< 10 s) in aprotic media, (iv) the system does not
Ϫ
Ϫ
and the coupling of the component radicals gives a 1,2-
contain OH or MeO , which may cause nucleophilic attack on
2
ϩ
ϩ
dioxetane via a zwitterion intermediate (Scheme 2). Under con-
1
and/or 1 . In order to get information on the rate deter-
mining step in the reaction of 1 with O , the kinetics of CL
᝽
2
emission and disappearance of 1 were examined at 25 ЊC under
air or an oxygen atmosphere. Time courses of the CL intensity
were recorded at 390 nm in DMF and MeCN. After a time lag
of a few seconds, the decay curve could be fitted to a single
exponential function. The time lag may be an artifact because
the absorption and CL spectra partially overlapped. Table 2
N
ET
1
+ O2
{
1•+••• O2•–
}
–O
O
dioxetane
BET
N+
†
The CL quantum yield was also determined under air.
Scheme 2
1
970
J. Chem. Soc., Perkin Trans. 2, 1997

View Online
ditions used for the 1–O reaction, a large excess of O exists in
2
2
ϩ
the solution. If 1 were trapped by O , a peroxyl radical would
be formed, which might cause a radical chain reaction. In fact,
᝽
2
ϩ
the second-order rate constant for the reaction of 1 with O2
᝽
Ϫ2
3
Ϫ1 Ϫ1
was 1.5 × 10 dm mol
s
in MeCN at 25 ЊC, which was
negligible compared with the k value of the 1–O reaction. This
2
2
ϩ
low reactivity of 1 to O was consistent with the result that
᝽
2
2
ϩ
ϩ
the redox wave of the 1 /1 couple was not affected by the
᝽
2
ϩ
presence of O . In the reaction of 1 with an excess amount of
KO , 1 is readily reduced with O2 and the resultant 1
combines with another molecule of O2 as shown in eqn. (3).
2
2
ϩ
Ϫ
ϩ
2
᝽
᝽
Ϫ
᝽
2
ϩ
Ϫ
ϩ
1
ϩ O2᝽
→ 1 ϩ O2
(3a)
(3b)
᝽
ϩ
Ϫ
1᝽
ϩ O2᝽
→ Dioxetane
ϩ
Ϫ
2ϩ
ϩ
Further reduction of 1᝽ with O2᝽
[eqn. (4)] can also take
Fig. 1 Absorption spectra of 7a (——), 7a (– – –) and 7a (···)
᝽
2ϩ Ϫ5 Ϫ3
in MeCN. [7a ] = 5.9 × 10 mol dm
.
ϩ
Ϫ
1᝽
ϩ O2᝽
→ 1 ϩ O2
(4)
redox potential similar to that of BIQ may undergo the same
type of oxygenation reaction. We have synthesized 2,2Ј-
place, but 1 thus formed will react with O . In the mechanism
shown in Scheme 2, the coupling of 1 and O2 should be fast
enough to compete with back electron-transfer (BET). It may
be possible if the coupling reaction takes place within the
geminate radical ion pair {1 ؒ ؒ ؒ O2 }. Recently Yamaguchi
and co-workers investigated the mechanism of autoxidation of
phenol and indole derivatives. According to their calculations,
one ET from phenolate anion to O can occur with assistance
by hydrogen bonding between O2 and protic solvents, while in
aprotic media formation of a CT complex followed by spin
inversion to a CT state is energetically favourable. If a similar
CT complex is formed from 1 and O , then Scheme 2 should be
modified as shown in Scheme 3. Formation of a C᎐O bond
2
ϩ
Ϫ
2
ϩ
᝽
᝽
biquinolinium (BQ ) salts , which have a closely related struc-
2ϩ
ture with BIQ , and investigated their redox properties and CL
reactions.
ϩ
Ϫ
2
ϩ
2ϩ
᝽
᝽
Cyclic voltammograms of 7a –7c in MeCN showed two
reversible one-electron redox waves in each case. The peak sep-
aration between cathodic and anodic scans was 55–70 mV at
1
9
Ϫ1
2
the scan rate of 10–100 mVs . As is shown in Table 1, the redox
Ϫ
ϩ
ϩ
/1.
᝽
potential of BQ
/BQ is a little less negative than that of 1᝽
᝽
3
In a solvent having a high donor number, the first reduction
1
2
ϩ
potential of BQ is shifted towards the more negative value.
On the other hand, the second redox potential is largely
unaffected, and as a result only one redox wave is observed in
DMSO. Substitution by methyl groups shifted reduction poten-
tials to the negative direction, and a 4-methyl group had a larger
effect than one at the 6-position. Because introduction of a
methyl group at the 4- or 6-position does not cause further
steric interaction between the two quinoline rings, the observed
effect is attributed to the electronic nature of the substituent.
2
[
zwitter ion
3{1δ+•••O2δ–}
1{1δ+•••O2δ–}
dioxetane
1
+ O2
or biradical]
Scheme 3
within the singlet CT complex will give a zwitterion (or less
likely biradical) intermediate, which will then cyclise to a 1,2-
dioxetane.
2
0
2
2
ϩ
PM3 calculations for 7a indicated that the spin density
᝽
Rate constants for reaction of 2 or 3 with O could not be
at the 4-position is higher than that of the other CH carbons
(vide infra), and therefore, an electron-donating substituent at
the 4-position seems to make the radical less unstable.
2
determined because of experimental difficulties. The absorption
spectrum of the starting olefin and the CL spectrum overlapped
with each other. Electrochemical methods seemed to be better,
but the second reduction potential of 2 or 3 was close to
that of O in MeCN or DMSO. Qualitative analyses of the
cyclic voltammograms in aerated DMF indicated that the dis-
appearance of 2 or 3 was faster than that of 1. As is seen in
Table 1, the free energy change for the ET process from 2 or 3 to
2
ϩ
Reduction of BQ with Na/Hg was carried out using the
2
ϩ
2ϩ
2ϩ 4
same procedure as that reported for BIQ . Variation of the
2
ϩ
absorption spectra on reduction of 7a in MeCN is shown in
Fig. 1. A transient orange species with a structured absorption
band in 400–580 nm showed an EPR signal, and the signal
disappeared on further reduction to a reddish violet form.
These results confirmed that 7a was reduced to radical cation
7a᝽ and then to diamagnetic neutral species 7a (Scheme 4).
2
2
ϩ
O was smaller than that from 1, and this correlation supported
2
ϩ
the proposed mechanism involving an ET or CT process.
Although a single ET cannot be discriminated from a partial
CT for BIQ–O system, the relatively sharp dependence of k on
GЊ (ET) suggests a significant contribution of a state where an
electron has been transferred to O from the olefin.
In solvents with low polarity, oxygenation of 1 was rather
slow. On addition of a drop of MeOH to the solution, however,
a bright CL emission was instantaneously observed. Previously,
Heller et al. reported that an alcoholic solvent worked as
catalyst for oxygenation of electron-rich olefins such as 1 and
tetrakis(dimethylamino)ethylene. A plausible reaction path-
way for this catalytic process is proposed. In the presence of
MeOH, oxygen will accept one electron from 1 more easily,
since superoxide is stabilised by hydrogen bonding with MeOH
R2
4
2
2
5
∆
R1
Me
3
2
6
2
N+
7
N
+e–
+
1
7a•+–7c•+
8
–
e–
Me
R1
R2
–e–
+e–
3
,21
a2+–7c2+
7
1
2
7a R = R = H
R2
1
2
7b R = Me, R = H
R1
1
2
Me
N
7
c R = H, R = Me
or by protonation to give a hydroperoxyl radical ؒOOH. Such
a reactive radical thus formed rapidly combined with 1 , and
eventually 4* is formed via the 1,2-dioxetane.
ϩ
N
᝽
Me
R1
2؉
R2
Redox behaviour of BQ salts
If an ET or CT process generally takes place between an
electron-rich olefin and molecular oxygen, a substrate with a
7a–c
Scheme 4
J. Chem. Soc., Perkin Trans. 2, 1997
1971

View Online
2
ϩ
ϩ
Table 3 Absorption maxima (nm) of BQ , BQ and BQ in MeCN
᝽
2
ϩ
ϩ a
BQ
BQ᝽
BQ
7
7
7
a
b
c
329
336
327
404
406
396
506
516
505
a
ϩ
The longest wave absorption of BQ᝽
was observed in the near IR
region.
2
ϩ
Ϫ
Fig. 2 (a) CL spectrum of the reaction of 7a (PF6 )2 with KO2 in
MeCN–DMSO. (b) Fluorescence spectrum of 8a in MeCN (excitation
at 332 nm).
2
ϩ
2ϩ
2ϩ
Fig. 3 Cyclic voltammograms of (a) 7a , (b) 7b and (c) 7c in
Ϫ1
DMSO at 25 ЊC at a scan rate of 30 mV s . —— : under Ar. – – – :
2
ϩ
Ϫ4
Ϫ3
under O . [7 ] = ca. 1.6 × 10 mol dm
.
2
The absorption spectrum of 7a in benzene (λmax 505 nm) was
quite similar to that recorded in MeCN. As is seen in Table 3,
substitution by methyl groups caused only a small shift of the
absorption maxima. In the case of the BIQ series studied, the
absorption spectra showed a large variation with the bridge on
the nitrogen atoms, reflecting the difference in the degree of
1
13
was not determined, the H and C NMR spectra indicated
inequivalence of the two quinoline groups, and it was likely to
be an OH adduct. Addition of aqueous H
brought about no CL, indicating that it was not involved in the
Ϫ
O
to this product
2
2
4
2ϩ
conjugation between the two isoquinoline rings. It is note-
CL reaction. Reaction of 7a with KO
2
or 7a with O in MeCN
2
worthy that BQ shows a longer wavelength absorption than
BIQ (400–450 nm), which is probably due to the more planar
structure of the former π system. The H NMR spectrum of 7a
also gave 8a with light emission. Fig. 2(a) shows the CL spec-
trum recorded on addition of KO to a solution of 7a (PF )
2 6 2
in MeCN. The spectrum was similar to the fluorescence of 8a in
MeCN [Fig. 2(b)], and the emitter of this CL was determined to
2
ϩ
Ϫ
1
in C D showed that it was a single isomer, probably (E)-form.
6
6
PM3 calculations showed that (E)-7a with C symmetry has
be the singlet excited state of 8a. Reaction of 7b and 7c with O
2
i
a lower energy than the (Z )-isomer. In the case of olefin 3, a
mixture of (E)- and (Z )-forms was obtained on reduction of
in MeCN showed blue emission and gave the corresponding 1-
methyl-2(1H )-quinolinones 8b and 8c, respectively as the major
2
ϩ
Ϫ
4
3
(I ) with aqueous Na S O . In the PM3 optimized struc-
product. Compared with the reaction of 1 with O , oxygenation
2
2
2
2
4
ture of either isomer, the central C᎐C double bond is highly
twisted. The torsion angles of N(2)᎐C(1)᎐C(1Ј)᎐N(2Ј) are 142
of 7a under the same conditions was slower, and the CL effi-
ciency (ΦCL) was lower by two orders of magnitude. The lower
᎐
and 34Њ for the (E)- and (Z)-forms, respectively.
ΦCL value is partially attributed to the lower fluorescence quan-
tum yield of 8a (0.017). Furthermore, in the CL reactions of
2ϩ
2ϩ
CL reaction of BQ salts
BQ–O or BQ –KO , several unidentified minor products were
2
2
On addition of aqueous H O –NaOH to an aqueous solution
also formed, which also resulted in a decrease in the CL
efficiency.
In the BQ series studied, the steric interaction in the form-
ation of the corresponding dioxetane is almost the same among
the three compounds, and therefore it seems suitable to investi-
2
2
2
ϩ
Ϫ
of 7a (MeSO4 )2, a bluish violet CL was observed. The major
product of this reaction was identified to be 1-methyl-2(1H )-
1
quinolinone (8a) based on a comparison of the H NMR spec-
2
ϩ
trum with that of the authentic sample. At high pH 7a was
2
3
unstable. On addition of an excess amount of NaOD to a
gate effects on the reactivity of O stemming from the redox
2
2
ϩ
Ϫ
1
ϩ
solution of 7a (MeSO4 ) in D O, the variation of the H
potentials and/or the electronic structure of BQ and BQ . As
for the 7a–O reaction in MeCN, because the CL intensity was
2
2
᝽
2
ϩ
NMR spectrum showed that 7a was completely transformed
2
to a new compound. Although the structure of this product
very low, the disappearance of the absorption due to 7a was
1
972
J. Chem. Soc., Perkin Trans. 2, 1997

View Online
ϩ
᝽
monitored at 505 nm. The absorbance vs. time curve was
The EPR spectrum of 1 in MeCN with a modulation width
fitted to first-order kinetics to give a rate constant of
of 1 G consisted of 11 lines with an averaged line separation of
Ϫ2
Ϫ1
1
(
(
1.2 ± 0.1) × 10 s , and using the concentration of O
Ϫ1 Ϫ1
4.1 G, which was similar to that recorded in MeOH. At a small
2
Ϫ3
Ϫ3 15
3
1.72 × 10 mol dm ), a k value of 7.0 ± 0.6 dm mol
s
modulation (0.05 G) about 150 lines were observed over 45 G,
but low- or high-field wings could not be analysed due to a poor
S/N ratio, while in the central part several signals seemed to
overlap owing to coincidental degeneracy. Determination of the
hyperfine coupling constants of nitrogen and each proton
seemed difficult from the EPR spectra alone. The relative
magnitudes of proton hyperfine couplings can be estimated
by NMR line broadening resulting from an electron exchange
between related diamagnetic and paramagnetic species.
According to Johnson’s treatment using density matrix form-
alism, the linewidth contribution from the self-exchange is
given by eqn. (5) where k is the second-order rate constant, a is
2
was obtained at 25 ЊC. For 7b and 7c, the CL intensity per unit
time was higher than that of 7a, and the decay was faster under
the same conditions. The observed CL intensity vs. time curve
could be fitted to an exponential decay with a second-order rate
3
Ϫ1 Ϫ1
3
Ϫ1 Ϫ1
constant of 26 dm mol
for 7c). The obtained rate constants for autoxidation of 7a–7c
were lower than that of 1.
Electrochemical methods were also attempted. In MeCN, the
s
(for 7b) or 25 dm mol
s
(
2
4
2
ϩ
ϩ
two reduction peaks partially overlapped and the 7 /7 redox
wave was affected by the presence of O , indicating that 7 also
᝽
ϩ
25
2
᝽
reacted with O on a 1 s timescale. Fig. 3 shows cyclic voltam-
2
2
ϩ
2ϩ
mograms of 7a –7c in deaerated or O -saturated DMSO at a
2
Ϫ1
Ϫ1
2
2
2
scan rate of 30 mV s . Although only one redox wave was
∆T2 = k[P]/(1 ϩ 4k [D] /a )
(5)
observed in each case, the ratio of the cathodic and the anodic
peak currents was almost unity under anaerobic conditions,
indicating that both of the two redox reactions were electro-
the hyperfine coupling constant in angular frequency units, [D]
and [P] are the concentrations of the diamagnetic and para-
magnetic partners. In fast-exchange limit or intermediate cases,
the line becomes broader with an increase in the a value. In the
chemically reversible. In O -saturated solution, the reoxidation
2
peak was significantly diminished (for 7a and 7b) or disap-
peared (for 7c), and the order of reaction rate was 7c > 7b > 7a.
Thus the relative reactivity of 7a–7c to oxygenation clearly cor-
relates with free energy change for an ET from these olefins to
1
2ϩ
H NMR spectrum of 1 in the presence of a small amount of
ϩ
1
᝽
, the observed line broadening showed that hyperfine coup-
H
ling constants |a | increase in the order 8-H < 6-H ~ 5-H < 3-
O (Table 1). From these results and an analogy with the BIQ–
2
H < 7-H < NCH < 4-H. As for the NCH protons, there are
2
2
O system, reaction of BQ with O is considered to be initiated
two nonequivalent sets of positions, pseudo-axial and pseudo-
equatorial, which are interconvertible. In the absence of elec-
tron exchange, the NMR signals due to the methylene protons
were observed as an AAЈBBЈ system even at 70 ЊC, indicating
that the inversion of the ethylene bridge was frozen on
2
2
by an ET process or formation of a CT complex, and coupling
at the 2-position leads to a zwitterion intermediate, which
eventually cyclises to a 1,2-dioxetane. Thermal decomposition
of the dioxetane will generate the S state of 8 as the emitter
1
(
Scheme 5).
the NMR timescale. It seems reasonable to assume that the
ϩ
inversion rate in 1 should be similar or a little larger than
᝽
2
ϩ
R2
R2
that in 1 from an analogy with the cases of the 9,10-
2
6
R1
R1
dihydrophenanthrene radical anion and the diquat radical
cation. Accordingly, the signals due to the axial and equa-
2
7
+
O2
7
a–c
torial protons should show not an averaged but respective
N
N
Ϫ1
∆
T2 values. Unfortunately, because of small differences in the
O
O
1 1
chemical shift and a large H– H coupling constant, these two
signals overlapped with each other, and contribution to the line-
width from each signal could not be separated. In order to
avoid the complexity resulting from the H– H coupling, the H
NMR spectrum of [ H ]ethylene-bridged compound [ H ]1
was recorded in the presence of [ H ]1 . The ∆T2 values
for the axial and equatorial deuterons were the same within
experimental error, indicating that either position showed a
similar hyperfine coupling. The EPR spectrum of [ H
sisted of seven lines with an averaged separation of 4.4 G. Such
spectral change on deuterium substitution can be explained by
Me
Me
1
1
2
2
2
2ϩ
4
4
R2
2
ϩ
Ϫ1
4
᝽
R1
2
ϩ
N
O
4
]1᝽ con-
Me
+
hν
H
coupling with four protons having an equal a value of 4.0–4.3
8
a–c
G. The angular dependence of a hyperfine coupling constant
1
2
H
2
8
8
8
a R = R = H
for a β-proton is most simply expressed as aβ = Bcos θ where θ
is the angle between the p orbital on the nitrogen and the C ᎐H
1
2
b R = Me, R = H
z
β
1
2
c R = H, R = Me
27
bond. The θ values for axial and equatorial protons in a PM3-
ϩ
Scheme 5
optimized structure for 1 are 30 and 116Њ, respectively, which
᝽
ax
eq
predicts an aβ /aβ ratio of 3.9:1. A discrepancy in the results
ax
eq
Electronic structure of radical cations
of the NMR experiments (aβ /aβ = ca. 1) cannot be explained
at the present stage. A large twisting angle between the two
isoquinoline rings (37Њ) might affect the hyperfine coupling
constants, and the above formula for the angular dependence
may be over simplified.
As was described above, the chemical yield of the emitter was
not so high except for the 1–O reaction. In alkaline aqueous
2
Ϫ
Ϫ
2ϩ
solution, nucleophilic attack of OH or OOH to BIQ or
2
ϩ
BQ caused undesired side reactions. On the other hand, in the
coupling between BIQ or BQ and O2 , regioselectivity of
ϩ
ϩ
Ϫ
2ϩ
2ϩ
Similar NMR experiments were attempted for 3 and 7a .
2ϩ
᝽
᝽
᝽
radical attack is important. If C᎐O bond formation occurs
In the case of 3 , the signals due to 8-H and 6-H were less
ϩ
H
exclusively at the 1-position in BIQ (or the 2-position in
broadened than the other signals, indicating that the a values
᝽
ϩ
BQ ), the 1,2-dioxetane will be formed quantitatively, result-
at these positions were small. However, the linewidths of all the
᝽
Ϫ
2ϩ
ing in a high yield of the emitter. Alternatively, O2 may attack
another position with a high spin density to give another
product. In this regard the spin density distributions in 1
other signals were almost the same. As for 7a , all the signals
᝽
were equally broadened. This result means that the self
ϩ
,
2ϩ
ϩ
exchange for 7a /7a was so slow that the situation fell in a
᝽
᝽
ϩ
ϩ
3
and 7a were examined by means of EPR, NMR and
slow-exchange limit even at a relatively high concentration of
the diamagnetic partner (0.06 mol dm ).
᝽
᝽
Ϫ3
semiempirical MO calculations.
J. Chem. Soc., Perkin Trans. 2, 1997
1973

View Online
Table 4 Calculated spin densities for the most stable conformer of
radical cations
Shimadzu UV2200 spectrophotometer and a Shimadzu RF510
a
fluorophotometer equipped with a quantum counter, respect-
ively. Fluorescence quantum yields were determined at 25 ЊC
using anthracene in EtOH as standard. The CL spectra were
recorded with a Hamamatsu Photonics PMA-10 spectro-
C-1 or
b
Compd. C-2
N
C-3
C-4
C-5
C-6
C-7
C-8
ϩ
1
1
3
7
0.149 0.152 0.013 0.062 0.006 0.021 0.012 0.013
0.161 0.151 0.009 0.061 0.009 0.022 0.013 0.018
0.146 0.163 0.019 0.063 0.007 0.021 0.011 0.015
photometric multichannel analyser. H (270 or 400 MHz) and
᝽
ϩ
2
᝽
H (61.4 MHz) were obtained with a JEOL JNM-GSX400 or
ϩ
a
᝽
JNM-GSX270 spectrophotometer. J values are given in Hz.
The residual protons in the solvent or sodium 2,2-dimethyl-2-
a
ϩ
The structure was optimized by the ROHF (for 1 ) or UHF (for
᝽
ϩ
ϩ
b
ϩ
ϩ
silapentane-5-sulfonate (in D O) were used as internal refer-
2
3
and 7a ) method using PM3 parameters. C-1 for 1
and 3᝽
,
᝽
᝽
᝽
ϩ
ence. EPR spectra (X-band) were recorded on a JEOL JES-
FE2XG spectrometer equipped with a temperature control unit
ES-DVT1 with 100 kHz modulation. IR spectra were obtained
with a Perkin-Elmer Spectrum 2000 FTIR spectrometer as KBr
disks. Melting points were determined on a Yanaco micro
apparatus and are uncorrected.
C-2 for 7a
᝽
In order to investigate the structure of the radical cations,
ϩ
ϩ
ϩ
PM3 calculations were carried out for 1 , 3 and 7a . In the
᝽
᝽
᝽
optimized structure, the central C᎐C bond distance was short-
ened by 0.04–0.05 Å compared with that of the corresponding
ϩ
dication. In the case of 1 , other structural parameters includ-
᝽
ing the torsion angle of N(2)᎐C(1)᎐C(1Ј)᎐N(2Ј) were similar to
Materials
2
ϩ
ϩ
the corresponding values in 1 . On the other hand, 3 and
Solvents used for the measurements of the fluorescence spectra
were of spectroscopic grade. DMSO and DMF were distilled
᝽
ϩ
7
a
showed two local minima with different twisting angles. In
the conformer with the lowest energy, the torsion angle was 115
and 140Њ for 3 and 7a , respectively, while the dications
showed a single minimum at ca. 100Њ along the rotation of the
᝽
from CaH under reduced pressure just before use. tert-Butyl
2
ϩ
ϩ
alcohol was dried over activated molecular sieves 4 Å. Potas-
sium tert-butoxide was dried in vacuo at room temp. before use.
᝽
᝽
central bond. The large geometry change on reduction from
CD CN (Isotec, 99.8% D) was dried over molecular sieves 3 Å
3
2
ϩ
ϩ
7
a
to 7a probably caused a high intramolecular reorganiz-
and stored under vacuum. Anthracene was recrystallised from
EtOH. The purity of luminol (Wako) was checked with the UV
spectrum and used without further purification. Other reagents
were used as received.
᝽
ation energy, and the self-exchange rate decreased, as was men-
tioned above. Shortening of the central bond and an increase in
coplanarity of the two aromatic rings suggest that conjugation
between the two halves of the molecule should stabilise the
radical.
2,2Ј-Ethylene-1,1Ј-biisoquinolinium
hexafluorophosphate
2
؉
2ϩ
Ϫ
1 (PF
mmol), aq. KPF
ant yellow precipitate was filtered, washed with water and
؊
)
. To an aqueous solution of 1 (Br )
(189 mg, 0.42
6
2
2
Table 4 lists calculated RHF spin densities of the most stable
conformer for each radical. The three radicals showed a similar
6
(200 mg, 1.09 mmol) was added. The result-
2
ϩ
Ϫ
6 2
spin density distribution. The unpaired electron delocalises on
recrystallised from aqueous MeCN to give 1 (PF
yellow crystals (124 mg) [Found: C, 42.3; H, 2.8; N, 4.95.
(PF requires C, 41.8; H, 2.8; N, 4.9%]. δ (CD CN)
5.19 (2H, m, NCH ax), 5.30 (2H, m, NCH eq), 7.83 (2H, d, J
) as fine
ϩ
the whole molecule, but the C-1 (or C-2 in 7a ) has the highest
᝽
spin density among the carbon atoms, which is consistent with
C
20
H
16
N
2
6
)
2
H
3
Ϫ
the result that O2 attacked this position of the radical cation.
2
2
᝽
ϩ
The calculated spin densities for 1 predicts that the hyperfine
8.8, 5,5Ј-H), 7.92 (2H, ddd, J 8.8, 7 and 1.2, 6,6Ј-H), 8.33 (2H,
ddd, J 8.5, 7 and 1.2, 7,7Ј-H), 8.51 (2H, d, J 8.5, 8,8Ј-H), 8.89
(2H, d, J 6.5, 4,4Ј-H) and 8.91 (2H, d, J 6.5, 3,3Ј-H).
᝽
coupling will increase in the order 5-H < 7-H < 8-H ~ 3-H < 6-
H < 4-H. Although this order is slightly different from the
result of NMR experiments, the spin density at the 4,4Ј-
1,1Ј-Dimethyl-2,2Ј-biquinolinium
hexafluorophosphate
Ϫ
2؉
؊
2ϩ
Ϫ 23
position has been proved to be high. If the O2 attacks this
7a (PF ) . To an aqueous solution 7a (MeSO ) , aq. KPF
6
2
4
2
6
᝽
position, a dioxetane bridged at the 3,4-positions will be formed
and the cleavage of its four-membered ring will result in
destruction of the isoquinoline ring. As for the autoxidation of
BIQ and BQ, factors controlling the chemical yield of the emit-
ter as well as an origin of the inconsistency between experi-
mental and calculated spin densities is still an open question.
Further investigation of the electronic structures of these rad-
icals is in progress.
(2.5 equiv.) was added. The resultant white precipitate was
filtered, washed with water and dried. Recrystallisation from
EtOH–MeCN gave faint pink, plate-like crystals [Found: C,
42.15; H, 3.3; N, 5.8. C20
42.3; H, 3.3; N, 5.9%]; νmax/cm 3096, 1627, 1590, 1521, 1442,
1392, 1351, 842, 779, 755 and 559; δ (CD CN) 4.41 (6H, s,
H N (PF ) ؒ0.5(CH CN) requires C,
18 2 6 2 3
Ϫ1
H
3
Me), 8.21 (2H, d, J 8.5, 3,3Ј-H), 8.25 (2H, t, J 8, 6,6Ј-H), 8.51
(2H, ddd, J 8, 7 and 1.5, 7,7Ј-H), 8.57 (2H, d, J 8, 5,5Ј-H), 8.59
(2H, d, J 7, 8,8Ј-H) and 9.43 (2H, d, J 8.5, 4,4Ј-H).
1
,1Ј,6,6Ј-Tetramethyl-2,2Ј-biquinolinium hexafluorophosphate
Conclusion
2
؉
؊
7
b (PF ) . The corresponding biquinoline was prepared by
6 2
28 3
Autoxidation of electron-rich olefins BIQ and BQ is not a rad-
ical chain process but a bimolecular reaction of the substrate
and triplet dioxygen. In polar aprotic solvents the reaction rate
correlated with the free energy change for an ET from the sub-
strate to O , while in non-polar solvents the reaction is very
slow. These results strongly suggest that the reaction should be
initiated by an ET or CT process to give a radical ion pair, and
then intersystem crossing takes place from the triplet to the
singlet state followed by formation of a C᎐O bond. All the
olefins studied produce light emission through the same type of
reaction pathway, but the CL efficiency and the chemical yield
of the emitter varies from compound to compound, probably
reflecting the difference in the electronic structure and the steric
interactions during bond formation.
the methods of Tiecco et al. A DMF solution (10 cm ) of
NiCl ؒ6H O (547 mg, 2.3 mmol) and triphenylphosphine
2
2
(
2.41 g, 9.2 mmol) was stirred under nitrogen at 50 ЊC, and zinc
powder (290 mg, 4.4 mmol) was added to the solution. To the
resultant reddish brown solution, a DMF solution (5 cm ) of 2-
chloro-6-methylquinoline (350 mg, 2.0 mmol) was added
dropwise. The mixture was stirred at 50 ЊC for 3 h under nitro-
gen, cooled, poured into 5% aq. NH (70 cm ) and extracted
with CHCl . The organic layer was washed with water, dried
over Na SO and concentrated. The resultant precipitate was
filtered and recrystallised from CHCl to give 6,6Ј-dimethyl-
,2Ј-biquinoline as fine, pale-yellow needles (176 mg, 43%). Mp
60–261 ЊC; δ (CDCl ) 2.57 (3H, s, Me), 7.58 (1H, dd, J 8.5 and
.7), 7.64 (1H, br s), 8.11 (1H, d, J 8.5), 8.22 (1H, d, J 8.5) and
.77 (1H, d, J 8.5).
3
2
3
29
3
3
3
2
4
3
2
2
1
8
H
3
Experimental
Apparatus
A mixture of 6,6Ј-dimethyl-2,2Ј-biquinoline (80 mg, 0.28
mmol) and dimethylsulfate (1.5 cm ) was heated to 170 ЊC for
3
Absorption and fluorescence spectra were recorded on a
80 min. The mixture was cooled, water and diethyl ether were
1
974
J. Chem. Soc., Perkin Trans. 2, 1997

View Online
2
ϩ
Ϫ
3
added and the aqueous layer was separated, to which aq. KPF6
186 mg, 1.0 mmol) was added. The resultant white precipitate
was filtered and recrystallised from aqueous MeCN to give
20 min. To this mixture, 7a (MeSO4 ) (4 mg) in buffer (2 cm )
2
(
was added with a syringe. The mixture was bubbled with Ar at
room temp. for 45 min, then the wine-red species was com-
pletely transferred to the organic layer and the aqueous phase
became almost colourless. A part (ca. 0.6 cm ) of the C D
solution was transferred to an NMR tube under an Ar atmos-
2
ϩ
Ϫ
7
b (PF )2 as pale-yellow plates (106 mg, 59%) [Found: C,
6
3
4
5.3; H, 3.8; N, 6.3. C H N (PF ) ؒ(CH CN) requires C, 44.7;
2
2
22
Ϫ1
2
6
2
3
6
6
H, 3.9; N, 6.5%]. νmax/cm 3095, 1629, 1589, 1513, 1401, 1351,
8
21, 715 and 559; δ (CD CN) 2.73 (6H, s, 6,6Ј-Me), 4.35 (6H,
phere with a syringe, the tube was stoppered with a Teflon cap
H
3
1
s, 1,1Ј-Me), 8.12 (2H, d, J 8.5, 3,3Ј-H), 8.3 (4H, m), 8.44 (2H, d,
J 9.5) and 9.26 (2H, d, J 8.5, 4,4Ј-H).
and the H NMR spectrum was recorded at room temp. δ 2.71
H
(6H, s, Me), 6.03 (2H, d, J 10, 3,3Ј-H), 6.47 (2H, d, J 8, 8,8Ј-H),
6.58 (2H, d, J 10, 4,4Ј-H), 6.82 (2H, t, J 7.5), 6.88 (2H, d, J 7,
5,5Ј-H) and 7.08 (2H, t, J 7.5).
1
,1Ј,4,4Ј-Tetramethyl-2,2Ј-biquinolinium hexafluorophosphate
2؉
؊
7
c (PF ) . This compound was synthesised from 2-chloro-4-
6 2
2ϩ Ϫ
methylquinoline by the same procedure as that for 7b (PF ) .
Pale-violet needles [Found C, 44.7; H, 3.8; N, 6.4.
C H N (PF ) ؒ(CH CN) requires C, 44.7; H, 3.9; N, 6.5%].
6
2
2
؉
Identification of the main products in CL reactions of BQ salts
in MeCN
2
2
22
2
Ϫ1
6
2
3
2ϩ
Ϫ
Ϫ4
Ϫ3
νmax/cm 3090, 1622, 1596, 1522, 1370, 836, 766 and 558; δH -
A solution of BQ (PF6 )2 (2–3 × 10 mol dm ) was reduced
with 0.5% Na/Hg under vacuum by procedures previously
reported for reduction of BIQ . The progress of the reaction
was monitored with UV–VIS spectrometry to ensure complete
two-electron reduction. The solution of BQ thus prepared was
exposed to air and shaken. After light emission had ceased, the
(CD CN) 3.12 (6H, s, 4,4Ј-Me), 4.32 (6H, s, 1,1Ј-Me), 8.02 (2H,
3
2
ϩ 4
s, 3,3Ј-H), 8.21 (2H, ddd, J 8, 7 and 1), 8.43 (2H, ddd, J 9, 7 and
1
), 8.52 (2H, d, J 9) and 8.65 (2H, dd, J 8 and 1).
30
1
,6-Dimethyl-2(1H )-quinolinone (8b) and 1,4-dimethyl-
31
2
(1H )-quinolinone (8c) were synthesised according to the
2
reported procedures. [ H ]2,2Ј-Ethylene-1,1Ј-biisoquinolinium
dibromide [ H ]1 (Br ) was prepared with the same pro-
cedures as 1 (Br ) using [ H ]1,2-dibromoethane (99% D).
reaction mixture was evaporated under reduced pressure, and
4
2
2ϩ
Ϫ
1
the H NMR spectrum of the residue was recorded in CDCl .
4
2
3
2
ϩ
Ϫ
2
3
In benzene. To a mixture of C H (4 cm ) and Na S O in
phosphate buffer (pH 7.0, 3 cm ), a solution of 7a (MeSO4 )2
2
4
6
6
2
2
4
3
2ϩ
Ϫ
Ϫ3
Ϫ3
3
Cyclic voltammetry
Cyclic voltammograms were obtained on a BAS 100B electro-
chemical workstation at a scan rate of 10–100 mV s using
glassy carbon, Pt wire and Ag/Ag as working, counter and
reference electrodes, respectively. The sample concentration was
ca. 2 × 10 mol dm . Deaeration was done by bubbling with
argon until a reduction wave of oxygen was no longer observed
in the buffer (1.8 × 10 mol dm , 2 cm ) was added under an
Ar atmosphere. The mixture was continuously bubbled with Ar
until the wine-red reduced species moved to the organic phase.
The organic layer was transferred to another flask to be
exposed to air, and a few drops of MeOH were added as a
Ϫ1
ϩ
Ϫ4
Ϫ3
catalyst for the CL reaction. The mixture was evaporated and
1
the H NMR spectrum of the residue was recorded in CDCl .
3
2
ϩ
Ϫ
2ϩ
Ϫ
(20 min). As for experiments under an oxygen atmosphere, the
In the cases of 7b (PF ) and 7c (PF )2, because of the low
6 2 6
solution was bubbled with O for 20 min. The background cur-
solubility in water, a suspension of the salt in buffer and C H
6 6
2
rent was corrected. For determination of rate constants, the
temperature was maintained at 25 ЊC by circulation of water to
a water jacketed cell from a Shimadzu TB-85 thermostatted
bath.
were put into a reaction vessel, to which Na S O in buffer was
2 2 4
added under Ar. The mixture was agitated by vigorous bubbling
with Ar, and similar work-up procedures were done.
2
ϩ
Ϫ
Reaction with KO . To a solution of BQ (PF6 )2 in MeCN, a
2
suspension of KO in MeCN was added and the mixture was
2
Measurements of relative intensity and quantum yield for CL of
BIQ
stirred until the CL had ceased. The supernatant was concen-
2؉
trated and water and CHCl were added. The organic layer was
3
The CL intensity was recorded on a Niti-on lumicounter 1000
connected with a personal computer through an analogue to
digital (A/D) converter. In the case of H O reaction in alkaline
separated, dried over Na SO and evaporated.
2
4
2
؉
Product analysis of BIQ –KO reaction
2
ϩ
2
2
2
Ϫ
Ϫ5
2ϩ
Ϫ
2
aqueous methanol, a solution of BIQ (X ) (1.3 × 10 mol
To a solution of BIQ (X ) in [ H ]DMSO, KO was added,
2
2 6 2
Ϫ3
dm , 300µl)and 4%aq. hydrogen peroxide(50µl)wereplaced in
and the mixture was stirred until the CL emission ceased. The
1
a Pyrex cell which was set in the optical unit of the lumicounter.
H NMR spectrum of the supernatant was recorded.
Ϫ3
An aqueous solution of NaOH (0.1 mol dm , 50 µl) was added
as a starter and the light intensity was collected for 1 min.
Determination of rate constants by emission or UV–VIS
spectrometry
As for the reaction with KO , solid KO (ca. 10 mg) was placed
2
2
2
ϩ
Ϫ
Ϫ4
Ϫ3
in a dried cell, in which a solution of BIQ (X ) in MeCN
1 × 10 mol dm , 150 µl) was injected. CL quantum yields
were determined against a luminol standard in DMSO. Five
or six runs were performed for each experiment.
CL spectrum of 1 –KO reaction. A solution of 1 in MeCN
A sample solution of 1 in MeCN (2 × 10 mol dm ) or DMF
Ϫ4 Ϫ3
2
Ϫ6
Ϫ3
(
(1 × 10 mol dm ) was prepared in sealed glassware equipped
3
2
2ϩ
Ϫ
with a 1 cm quartz cuvette with Na/Hg reduction of 1 (Br ) .
2
Immediately after the glassware was opened and shaken to
expose the solution to air, the time course of CL was recorded
on a Shimadzu RF510 fluorophotometer at 390 nm (bandwidth
10 nm). During the experiments water at 25 ЊC was circulated to
the cell holder from a Shimadzu TB-85 thermostatted bath. An
absorbance or CL intensity vs. time curve was fitted to a single
exponential decay to obtain the rate constant under pseudo-
first-order conditions. Because the absorption of 1 and the CL
emission spectra were partially overlapped, the data collected
for the initial 5 s were not included for curve fitting. In the cases
of 7b and 7c, similar procedures were carried out in MeCN.
ϩ
ϩ
᝽
2
᝽
2
ϩ
Ϫ
was prepared with Na/Hg reduction of 1 (Br ) in sealed
2
glassware equipped with a 1 cm quartz cuvette. As soon as the
cell was cut off, a small amount of solid KO was added to the
2
solution and the CL spectrum was recorded on the spectro-
photometric multichannel analyser.
2
؉
2ϩ
Ϫ
CL spectrum of 7a –KO reaction. A solution of 7a (PF6 )2
2
Ϫ3
Ϫ3
3
in dry MeCN (1.7 × 10 mol dm , ca. 1.5 cm ) was placed in a
cm quartz cuvette, which was set to the multichannel analyser.
Immediately after addition of a solution (0.5 cm ) of saturated
1
3
Ϫ3
2ϩ
KO in DMSO containing dibenzo-18-crown-6 (0.01 mol dm
)
As for a benzene solution, after 1 was reduced to 1 in
2
to the cell, the light emission was accumulated for 5 s.
MeCN using glassware equipped with a vacuum-tight stop-
cock, the glassware was again connected to the vacuum line to
remove the solvent and degassed benzene (3.5 cm ) was
vacuum-transferred onto the residue. The glassware was sealed
off from the vacuum line and the reduced species was dissolved
in benzene. The cell was opened and connected with a reservoir
1
3
H NMR measurement of 7a
3
A mixture of C D (2 cm ) and a solution of Na S O in phos-
6
6
2
2
4
3
phate buffer (pH 7.0, 4 cm ) was put into a test tube stoppered
with a rubber septum and bubbled with Ar through a needle for
J. Chem. Soc., Perkin Trans. 2, 1997
1975

View Online
containing 1 atm oxygen. After shaking, variation of the UV–
VIS spectrum was recorded every 3 min up to 72 min, during
which isosbestic points were observed at 350 and 316 nm. From
stable conformer was obtained with a single point ROHF calcu-
lation using the UHF-optimized geometry. As for neutral spe-
cies 3 and 7a, both (E)- and (Z )-isomers were optimized with
C or C symmetry.
the plot of the logarithm of the absorbance at 440 nm vs. time,
2
i
ϩ
the rate constant was obtained. The reaction rate of 1 with O2
᝽
was determined by similar procedures. In the sample prepar-
2
ϩ
Acknowledgements
ation, the reduction was stopped when ca. 90% of 1 was con-
ϩ
verted to 1 to avoid concomitance of 1. The absorbance at
We wish to thank Professor Yuichi Masuda and Ms Haruko
᝽
2
5
00 nm was monitored for 6 h.
Hosoi, Ochanomizu University, for their help with H NMR
measurements and valuable discussions. We are grateful to Pro-
fessor Mamoru Ohashi and Dr Takashi Hirano, The University
of Electro-Communications, for measuring CL spectra. Thanks
are also due to Professor Hajime Nagano, Ochanomizu Univer-
sity, for his discussion and encouragement and to Professor
Yutaka Fukuda for his kindness in putting his electrochemical
workstation at our disposal. This work was supported by a
Grant-in-Aid for Scientific Research from the Ministry of Edu-
cation, Science and Culture, Japan (No. 08554026).
EPR spectra of radical cations
A sample solution of a radical cation in MeCN (2 × 10 –
Ϫ4
Ϫ3
Ϫ3
1
× 10 mol dm ) was prepared by Na/Hg reduction of the
corresponding dication under vacuum in sealed glassware
equipped with an EPR tube. The variation of the UV–VIS spec-
trum was monitored, and when ca. 90% of the dication was
reduced, the solution was transferred to the EPR tube, which
was then mounted in an EPR cavity. The signal output for a
recorder was fed to an A/D converter, and the spectrum was
accumulated 8–32 times using a personal computer.
References
2
ϩ
NMR line broadening experiments for 1
1 S. F. Mason and D. R. Roberts, J. Chem. Soc., Chem. Commun.,
1967, 476.
2 R. A. Henry and C. A. Heller, J. Luminescence, 1971, 4, 105.
2
ϩ
Ϫ
3
A solution of 1 (Br ) (ca. 1 mg) in D O (0.9 cm ) was placed
2
2
2
ϩ
Ϫ
in glassware equipped with an NMR tube in which 1 (Br )
2
3
C. A. Heller, R. A. Henry and J. M. Fritsch, in Chemiluminescence
and Bioluminescence, ed. M. J. Cormier, D. M. Hercules and J. Lee,
Plenum Press, New York, 1973, p. 249.
(22 mg) was placed and a branch containing zinc powder. After
the solution was degassed by freeze–pump–thaw cycles, a por-
tion of the solution was transferred to the branch to obtain a
4 K. Maeda, Y. Matsuyama, K. Isozaki, S. Yamada and Y. Mori, J.
Chem. Soc., Perkin Trans. 2, 1996, 121.
ϩ
small amount of 1 , and then the whole solution was trans-
᝽
5
6
7
Y. Mori, Y. Matsuyama, S. Yamada and K. Maeda, Acta
Crystallogr., Sect. C, 1992, 48, 894.
Y. Mori, Y. Matsuyama, J. Suzuki, Y. Ishii, S. Yamada and K.
Maeda, Acta Crystallogr., Sect. C, 1993, 49, 1398.
ferred to the NMR tube. The tube was sealed off, the solid
2
2
ϩ
Ϫ
1
1
2
(Br ) was completely dissolved (The final concentration of
2
ϩ
Ϫ3
was ca. 0.055 mol dm ) and NMR spectra were recorded at
2
2–65 ЊC. As for H NMR measurements, the sample solution
was prepared by the same procedures using [ H ]1 (Br ) and
H O instead of 1 (Br ) and D O, respectively. The experi-
K. Yamaguchi, Y. Ikeda and T. Fueno, Tetrahedron, 1985, 41, 2099.
2
2ϩ
Ϫ
4
2
8 Z. Shi, V. Goulle and R. P. Thummel, Tetrahedron Lett., 1996, 37,
2357.
2
ϩ
Ϫ
2
2
2
9
D. T. Breslin and M. A. Fox, J. Am. Chem. Soc., 1993, 115, 11716.
mental spectra were fitted to the sum of two Lorentzian curves
to obtain their linewidths. The frequency difference and line-
widths without exchange were determined from the spectrum
of a sample containing no radical. The linewidth of a signal
due to natural abundance of HDO in the solvent showed no
1
1
1
1
0 W. Adam and W. J. Baader, J. Am. Chem. Soc., 1985, 107, 410.
1 G. B. Schuster, Acc. Chem. Res., 1979, 12, 366.
2 L. H. Catalani and T. Wilson, J. Am. Chem. Soc., 1989, 111, 2633.
3 For examples, W. Adam, R. Fell and M. H. Schulz, Tetrahedron,
1993, 49, 2227; F. McCapra, Tetrahedron Lett., 1993, 34, 6941.
1
difference between samples with and without radical. The H
14 R. S. Nicholson and I. Shain, Anal. Chem., 1964, 36, 706.
1
5 M. Tsushima, K. Tokuda and T. Ohsaka, Anal. Chem., 1994, 66,
551.
NMR signals were assigned based on NOESY experiments
4
(
270 MHz, 50 ЊC) for a sample without a radical. Cross peaks
were observed between the pseudo-equatorial methylene pro-
tons and 3,3Ј-H as well as between 4,4Ј-H and 5,5Ј-H. δ (D O,
1
1
6 G. Eberlein and T. C. Bruice, J. Am. Chem. Soc., 1983, 105, 6685.
7 D. T. Sawyer and E. J. Nanni, in Oxygen and oxy-radicals in
chemistry and biology, ed. M. A. Rodgers and E. L. Powers,
Academic Press, New York, 1982, pp. 15–44.
H
2
2
7
8
8
4 ЊC, 400 MHz) 5.35 (2H, m, CH ax), 5.54 (2H, m, CH eq),
2
2
.87 (2H, d, J 8.2, 5,5Ј-H), 7.92 (2H, dd, J 8.2 and 7, 6,6Ј-H),
.33 (2H, dd, J 8.2 and 7, 7,7Ј-H), 8.51 (2H, d, J 8.2, 8,8Ј-H),
.92 (2H, d, J 6.5, 4,4Ј-H) and 9.11 (2H, d, J 6.5, 3,3Ј-H).
18 E. J. Nanni, Jr., C. T. Angelis, J. Dickson and D. T. Sawyer, J. Am.
Chem. Soc., 1981, 103, 4268.
1
9 Y. Yoshioka, S. Yamanaka, S. Yamada, T. Kawakami, M. Nishino,
K. Yamaguchi and A. Nishinaga, Bull. Chem. Soc. Jpn., 1996, 69,
2
701.
1
2؉
2؉
H NMR line broadening experiments for 3 and 7a2
2
0 N. J. Turro, Tetrahedron Lett., 1985, 41, 2089.
21 C. A. Heller, Adv. Chem. Ser., 1968, p. 225.
22 J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209.
23 J. E. Dickeson and L. A. Summers, J. Heterocycl. Chem., 1970, 7,
ϩ
Ϫ
In a Pyrex NMR tube, NaBPh (7 mg) and 3 (PF )2 or
7
CD CN (0.8 cm ) was vacuum-transferred to the tube, which
was then sealed off from the vacuum line. The concentration of
the dication was ca. 0.06 mol dm . The sample solution was
irradiated with a 500 W xenon lamp for 20 s to obtain a small
amount of radical cation via photoinduced electron-transfer
from tetraphenylborate anion. The H NMR spectrum was
4
6
2
ϩ
Ϫ
a
(PF6 )2 were placed and degassed on a vacuum line.
3
3
4
01.
2
2
2
4 A. L. Rieger and P. H. Rieger, J. Phys. Chem., 1984, 88, 5845.
5 C. S. Johnson, Jr., Adv. Magn. Reson., 1965, 1, 33.
6 F. C. Adam, Can. J. Chem., 1971, 49, 3524.
Ϫ3
27 P. D. Sullivan and M. L. Williams, J. Am. Chem. Soc., 1976, 98,
1
1711.
2
2
3
8 M. Tiecco, L. Testaferri, M. Tingoli, D. Chianelli and M.
Montanucci, Synthesis, 1984, 736.
9 P. Desos, G. Schlewer and C. G. Wermuth, Heterocycles, 1989, 28,
recorded at room temp.
MO calculations
PM3 calculations were carried out with the program of
MOPAC version 6.01. As for 1 the structure of 1 deter-
mined by X-ray analysis was used as the initial model and
structure optimization was performed by ROHF methods with
C symmetry. In the cases of 3 and 7a , UHF structure
optimization was carried out with C symmetry from several
initial structures with various torsion angles of N᎐C᎐C᎐N to
1
085.
0 H. Balli and D. Schelz, Helv. Chim. Acta, 1970, 53, 1903.
31 C. E. Kaslow and D. J. Cook, J. Am. Chem. Soc., 1945, 67, 1969.
32 J. Lee and H. H. Seliger, Photochem. Photobiol., 1965, 4, 1015.
3 J. J. P. Stewart, MOPAC Ver. 6.0, QCPE No. 455; Revised as Ver.
6.01 by T. Hirano, Ochanomizu University.
3
3
ϩ
2ϩ
᝽
5
3
ϩ
ϩ
2
᝽
᝽
Paper 7/02289A
Received 3rd April 1997
Accepted 19th June 1997
2
give two local minima. The unpaired spin density of the more
1
976
J. Chem. Soc., Perkin Trans. 2, 1997