Synthesis, electrochemical investigation and EPR spectroscopy of a reversible four-stage redox system based on mesoionic 5,5'-azinobis(1,3-diphenyltetrazole) and related mesoionic compounds

Article abstract of DOI:10.1039/a809386e

Mesoionic 5,5'-azinobis(1,3-diphenyltetrazole) 1 was prepared, and its chemical oxidation gave stable crystals of the corresponding radical cation 1^.+ and dication 1²⁺, which reversibly gave back azine 1 on reduction with zinc. Electrochemical investigations of 1 using cyclovoltammetry and differential pulse voltammetry in pyridine (Py) or dichloromethane (DCM) also revealed the two reversible successive one-electron oxidations leading to dication 1²⁺ via radical cation 1^.+, both of which can be reduced to the neutral state 1. In the cathodic process, 1 was reduced by two consecutive one-electron transfers at only slightly different potentials up to the corresponding diamon 1^2- which could be re-oxidized to the neutral state; thus constituting a reversible four-stage redox system. Radical cation 1^.+ and anion 1^.- were characterized by EPR spectroscopy. In order to get more insights into the spin-density distribution of 1^.+, the bis- and tetra-¹⁵N-labelled species 1a^.+, 1b^.+ and 1c^.+ were synthesized and investigated by EPR and ¹⁵N as well as ¹⁴N ENDOR spectroscopy, revealing that the largest N hyperfine coupling constants are due to the nitrogen atoms of the central bridge. According to ¹H ENDOR there seems to be a small coupling with the protons of both phenyl rings which cannot be resolved in the EPR spectrum. The electrochemical properties of the related mesoionic compounds 5-10 were also investigated in Py or DCM solutions. In the cathodic process, a reduction peak of 9 and 10 was observed due to their reversible one-electron reduction to the corresponding radical anions. The radical obtained on reduction of 10 was characterized by EPR spectroscopy. On the other hand, 5-8 can be reduced by a formal two-electron transfer up to the corresponding dianions which are re-oxidizable to the neutral state. In the anodic process, 9/10 undergo irreversible one-electron oxidations whereas 5/6 (in DCM) and 7/8 (in Py) experience reversible or irreversible step by step two-electron oxidations leading to the dications. In Py the oxidation products of 5/6 react to further species revealing two more oxidation and several rereduction peaks. On the other hand, the oxidation products of 7/8 are instable in DCM (one main oxidation and rereduction peak). The electrochemical data are discussed in terms of delocalization in the cations and conjugation in the dications.

Full text of DOI:10.1039/a809386e

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis, electrochemical investigation and EPR spectroscopy
of a reversible four-stage redox system based on mesoionic
5,5Ј-azinobis(1,3-diphenyltetrazole) and related mesoionic
compounds†
Shuki Araki,*a Kaori Yamamoto,a Tomoko Inoue,a Koji Fujimoto,a Hatsuo Yamamura,a
Masao Kawai,a Yasuo Butsugan,a Jinkui Zhou,b Emerich Eichhorn,b Anton Rieker*b and
Martina Huberc‡
a Department of Applied Chemistry, Nagoya Institute of Technology, Gokiso-cho, Showa-ku,
Nagoya 466-8555, Japan
b Institute of Organic Chemistry, University of Tübingen, Auf der Morgenstelle 18,
D-72076 Tübingen, Germany
c Institute of Organic Chemistry, Free University of Berlin, Takustr. 3, D-14195 Berlin,
Germany
Received (in Cambridge) 1st December 1998, Accepted 17th February 1999
Mesoionic 5,5Ј-azinobis(1,3-diphenyltetrazole) 1 was prepared, and its chemical oxidation gave stable crystals of
the corresponding radical cation 1 ϩ and dication 12ϩ, which reversibly gave back azine 1 on reduction with zinc.
᝽
Electrochemical investigations of 1 using cyclovoltammetry and differential pulse voltammetry in pyridine (Py) or
dichloromethane (DCM) also revealed the two reversible successive one-electron oxidations leading to dication 12ϩ
via radical cation 1 ϩ, both of which can be reduced to the neutral state 1. In the cathodic process, 1 was reduced by
᝽
two consecutive one-electron transfers at only slightly different potentials up to the corresponding dianion 12Ϫ which
ϩ
could be re-oxidized to the neutral state; thus constituting a reversible four-stage redox system. Radical cation 1
᝽
and anion 1 Ϫ were characterized by EPR spectroscopy. In order to get more insights into the spin-density
᝽
distribution of 1 ϩ, the bis- and tetra-15N-labelled species 1a ϩ, 1b ϩ and 1c ϩ were synthesized and investigated by
᝽
᝽
᝽
᝽
EPR and 15N as well as 14N ENDOR spectroscopy, revealing that the largest N hyperfine coupling constants are due
to the nitrogen atoms of the central bridge. According to 1H ENDOR there seems to be a small coupling with the
protons of both phenyl rings which cannot be resolved in the EPR spectrum.
The electrochemical properties of the related mesoionic compounds 5–10 were also investigated in Py or DCM
solutions. In the cathodic process, a reduction peak of 9 and 10 was observed due to their reversible one-electron
reduction to the corresponding radical anions. The radical obtained on reduction of 10 was characterized by EPR
spectroscopy. On the other hand, 5–8 can be reduced by a formal two-electron transfer up to the corresponding
dianions which are re-oxidizable to the neutral state. In the anodic process, 9/10 undergo irreversible one-electron
oxidations whereas 5/6 (in DCM) and 7/8 (in Py) experience reversible or irreversible step by step two-electron
oxidations leading to the dications. In Py the oxidation products of 5/6 react to further species revealing two more
oxidation and several rereduction peaks. On the other hand, the oxidation products of 7/8 are instable in DCM (one
main oxidation and rereduction peak). The electrochemical data are discussed in terms of delocalization in the
cations and conjugation in the dications.
electrons constituting a novel four-stage redox system. Radical
Introduction
ϩ
cation 1 and dication 12ϩ were isolated in crystalline form,
᝽
Mesoionic compounds are an interesting family of hetero-
cyclic compounds. Their unique properties, such as reaction
behaviour, structural characteristics, and biological activities,
and the radical cation and radical anion of 1 were characterized
by EPR spectroscopy. The phenylogue-type analogues 5–8 of
mesoionic azine 1 were also prepared and their chemical and
electrochemical redox reactions were investigated, in com-
parison with mesoionic azine 1, by means of cyclovoltammetry
have attracted much attention, and a wide variety of derivatives
have so far been studied.1 However, their electrochemical prop-
erties have gathered relatively little attention. In this paper is
described the synthesis of mesoionic 5,5Ј-azinobis(1,3-diphenyl-
(CV) and differential pulse voltammetry (DPV).
tetrazole) 1 which was found to undergo reversible transfer of
Results and discussion
Synthesis and chemical reactions
† Supplementary data are available (SUPPL. NO. 57507, pp. 6) from
the British Library. For details of the Supplementary Publications
Scheme, see ‘Instructions for Authors’, J. Chem. Soc., Perkin Trans. 2,
available via the RSC web pages (http://www.rsc.org/authors). For
direct electronic access see http://www.rsc.org/suppdata/p2/1999/985.
‡ Present address: Laboratory of Astronomy and Physics, MAT
Group, Leiden University, PO Box 9504, 2300 RA Leiden, The
Netherlands.
We have recently found that a 5-chloro-1,3-diphenyltetrazolium
salt readily undergoes substitution reactions with nitrogen-
nucleophiles.2,3 By mixing with hydrazine hydrate, the chloro-
tetrazolium ion readily gave purple mesoionic azine 1 in good
1
yield [eqn. (1)]. H and 13C NMR chemical shifts of the 1,3-
diphenyltetrazolium ring protons and carbons of 1 are similar
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
985

View Article Online
ϩ
acetate gave the radical cation 1 as greenish-brown crystals.
Ph
᝽
Ph
ϩ
Radical cation 1 is extremely stable; it could be purified by
N
᝽
N
N
N
H2NNH2•H2O
N
N
Ph
column chromatography on silica gel and stored under air for
several months without appreciable decomposition. Further
N
N
N
Cl
2
N
N
(1)
N
N
-2HCl, -H2O
-2HBF4
N
Ph
oxidation of 1 ϩ with conc. nitric acid gave the orange dication
Ph
᝽
BF4
12ϩ. The longest-wavelength absorption of 1 in MeCN lies at
Ph
1
542 nm which shifts bathochromically to 625 nm (shoulder) on
oxidation to 1 ϩ, whereas further oxidation to 12ϩ results in a
to those of the previously prepared mesoionic tetrazolium
systems,2,3 thus supporting the tetrapolar structure of azine 1.
Methylation of 1 with methyl iodide gave the N-methylated
mesoionic cation 2 [eqn. (2)]. Even with an excess of methyl
᝽
ϩ
hypsochromic shift to 446 nm. Reduction of 1 and 12ϩ with
᝽
zinc powder gave back 1; thus, the azine constitutes a reversible
ϩ
redox system, and the three oxidation states (1, 1 and 12ϩ
)
᝽
were isolated in crystalline form (Scheme 2). It should be noted
Ph
Ph
Ph
1. MeI
2. NaBF4
Me
N
N
N
N
Ph
1. HNO3
2. NaBF4
N
N
N
N
N
N
N
N
N
N
N
1
N
N
Ph
Ph
(2)
N
N
N
N
N
N
N
Ph
N
N
N
N
N
N
Zn
Ph
Ph
BF4
Ph
2BF4
2
Ph
Ph
1
12+
iodide, the dimethylated compound was not formed. On the
other hand, azine 1 was protonated step by step; thus, with a
limited amount of acetic acid brownish orange crystals of
monocation 3 were formed, and with trifluoroacetic acid the
colourless dication 4 was obtained. Compounds 3 and 4
regenerated the conjugate base 1 reversibly on treatment with a
base like NaOH (Scheme 1). Oxidation of 1 with lead tetra-
1. Pb(OAc)4
2. NaBF4
1. HNO3
2. NaBF4
Zn
Zn
Ph
N
N
N
N
Ph
N
N
N
N
BF4
Ph
N
Ph
N
N
H
N
Ph
NaOH
N
Ph
N
N
1
N
N
N
N
N
1•+
Ph
1. HOAc
2. NaBF4
BF4
Scheme 2
Ph
CF3COOH
NaOH
3
that 12ϩ was gradually converted to 1 ϩ merely by dissolving in
᝽
organic solvents such as acetonitrile and ethanol, indicating the
Ph
large stability of radical cation 1 ϩ. Although 1 undergoes a
᝽
N
N
H
N
Ph
reversible electrochemical reduction (see below), the chemical
reduction of 1 with zinc powder gave a complex mixture, from
which no defined products could be isolated.
N
N
N
N
N
N
N
H
Ph
The benzene-ring inserted analogues 5–8 were similarly syn-
thesized from the 5-chloro-1,3-diphenyltetrazolium ion and
appropriate aromatic amines [eqn. (3)]. (1,3-Diphenyltetra-
zolium-5-yl)-anilide2 (9) and -amide3 (10) were used as reference
Ph
2CF3CO2
4
Scheme 1
NH2
H2N
Ph
1.
or
NH2
N
NH2
N
Cl
2
N
N
2. NaOH
Ph
BF4
Ph
Ph
N
N
N
N
N
N
Ph
N
Ph
Ph
N
N
N
N
Ph
N
N
N
N
N
Ph
N
Ph
N
N
N
(3)
N
N
N
Ph
N
Ph
N
N
N
N
N
N
N
N
N
Ph
Ph
N
N
Ph
N
Ph
N
N
N
N
Ph
Ph
7
8
5
6
986
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995

View Article Online
Table 1 Formal and peak potentials of the electrochemical processes of the mesoionic compounds 1 and 5–10a
Compound
Solvent
E1r/mV
E2r/mV
E1o/mV
E2o/mV
E3o/mV
E4o/mV
1
1
5
5
6
6
7
7
8
Py
DCM
Py
DCM
Py
DCM
Py
Ϫ1853b
Ϫ1856b
Ϫ1805b
Ϫ1809b
Ϫ1755b
Ϫ1738b
Ϫ1792b
Ϫ1809b
Ϫ1750b
Ϫ1730c
Ϫ1756b
Ϫ1736c
Ϫ1775b
Ϫ1876b
Ϫ1967b
Ϫ2006b
Ϫ243b
Ϫ136b
3b
519b
610b
227b
340b
414c
414b
620b
598b
624c
870b
858c
61b
266c
333b
350d
323c
588d
544c
646d
573c
565c
1020d
1027c
1042d
974c
DCM
Py
771d
705c
8
DCM
9
10
Py
Py
10
DCM
Ϫ1885b
a Potentials are given vs. Ag/AgClO4 (0.01 M in MeCN–0.1 M Bu4NPF6). b Er and Eo were obtained by averaging of the corresponding anodic and
cathodic peak potentials. The potential separations ∆E between reduction and re-oxidation (or oxidation and re-reduction) peaks lie between 65 and
110 mV (in Py at v = 0.05 V sϪ1). c Er and Eo were obtained by DPV. d Eo are peak potentials of irreversible ET processes.
Ph
Ph
N
N
N
N
N
N
N
N
N
Ph
N
H
Ph
Ph
10
9
compounds. In contrast to the easy oxidation of 1, the
phenylogue 5 resists chemical oxidation with various oxidants,
such as conc. nitric acid, lead dioxide, and alkaline potassium
ferricyanate. On the other hand, reduction of 5 with zinc went
smoothly and deep-red crystals of 11 were obtained. Spectro-
scopic data show that, contrary to the reversible electro-
chemical reduction (see below), one of the tetrazolium rings of
5 was reductively cleaved to N-substituted NЈ-phenylguanidine
[eqn. (4)]. Similarly, zinc reduction of 9 gave N,NЈ-diphenyl-
guanidine in good yield.
Ph
N
N
N
Ph
N
N
Zn
5
(4)
H2N
NH
NPh
11
Fig. 1 CVs of cathodic reduction (a)–(d) and anodic oxidation (e)–(h)
of 1, 5, 7 and 10 in Py solution containing 0.1 M Bu4NPF6; potential vs.
Ag/AgClO4 (0.01 M in acetonitrile–0.1 M Bu4NPF6); (a) 1, (b) 5, (c) 7,
(d) 10, (e) 1, (f) 5, (g) 7 and (h) 10; scan rate, 100 mV sϪ1 for (a), (b), (d),
(e), (f) and (h), and 50 mV sϪ1 for (c) and (g).
Electrochemical investigations
The voltammetric investigations of 1 and 5–10 were performed
in pyridine (Py) or dichloromethane (DCM) solutions contain-
ing 0.1 M Bu4NPF6, at room temperature under an argon
atmosphere. Py was chosen as solvent to prevent protonation
and disproportionation reactions in the cathodic voltammetric
process. The anodic voltammograms, however, generally are
more reversible in DCM, presumably, because DCM unlike Py
does not act as a strong nucleophile towards cations and
dications. The cyclovoltammograms (at scan rates of 50 or 100
mV sϪ1) and differential pulse voltammograms of 1, 5, 7 and 10
for cathodic and anodic processes are shown in Fig. 1–4. The
potential data of 1 and 5–10 are summarized in Table 1. The
number of electrons involved in the electrochemical processes
was confirmed by controlled-potential electrolysis (CPE).
Let us first discuss the reductions. The cyclovoltammogram
of 1 [Fig. 1(a)] reveals two (quasireversible) couples Pr1/Po1
and Pr2/Po2 with only small potential separation between Pr1 and
Pr2. The corresponding DPV curves [Fig. 2(a)] exhibit well-
separated peaks Pr1 and Pr2 on the first scan and Po1 and Po2 on
the second scan (not shown). Obviously, there are two succes-
sive one-electron transfers to form the dianion via the mono-
anion (for EPR measurements see below). Presumably, the
electrons are entering both tetrazolium rings, step by step.
These rings are largely independent of each other, therefore,
their reduction potentials are close together. The anions can be
reoxidized to the neutral state (Scheme 3).
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
987

View Article Online
Fig. 3 CVs of the anodic oxidation (a)–(c) and their corresponding
DPVs (d)–(f) of the compounds 1, 5 and 7 in DCM solution, containing
0.1 M Bu4NPF6, potential vs. Ag/AgClO4 (0.01 M in acetonitrile–0.1 M
Bu4NPF6). (a) and (d) 1, (b) and (e) 5, (c) and (f) 7, for CVs: scan rate 50
mV sϪ1; for DPVs: scan rate 20 mV sϪ1, pulse amplitude 50 mV, pulse
width 50 ms, pulse period 200 ms.
Fig. 2 DPVs of cathodic reduction (a)–(d) and anodic oxidation (e)–
(h) of 1, 5, 7 and 10 in Py solution containing 0.1 M Bu4NPF6; potential
vs. Ag/AgClO4 (0.01 M in acetonitrile–0.1 M Bu4NPF6); scan rate, 20
mV sϪ1; pulse amplitude, 50 mV; pulse width, 50 ms; pulse period, 200
ms; (a) 1, (b) 5, (c) 7, (d) 10, (e) 1, (f) 5, (g) 7 and (h) 10.
-e
-e
-e
-e
1•+
12–
12+
1•–
1
oxidations only in DCM [Fig. 3(a) for CV and 3(d) for DPV]
leading to the dication 12ϩ via the mono-cation 1 ϩ, in agree-
+e
+e
+e
+e
᝽
ment with the chemical oxidation (for EPR measurements see
below). In Py, however, the second electron-transfer is no more
totally reversible, as can be seen from the smaller reduction
peak [Fig. 1(e) for CV and 2(e) for DPV]. Indeed, if the second
scan in pyridine is extended until Ϫ2.2 V [Fig. 4(a)] two small
rereduction peaks a, b occur between Ϫ1.0 and Ϫ1.5 V, at
potentials less negative than the above mentioned reduction
couples which appear between Ϫ1.8 and Ϫ2.0 V. In the third
(oxidative) scan, no reoxidation peaks corresponding to the
small reduction peaks a and b can be detected. If the anodic
sweep is stopped at about ϩ0.2 V (i.e. after the first but before
the second oxidation peak; Figure not shown), the reduction
peaks a and b of the second scan do not show up any more.
This means that the dication reacts with a nucleophile, presum-
ably pyridine, in a fast reversible way, or its tetrazolium ring
is cleaved into many fragments. The same CV- and DPV-
responses are observed when the chemically produced and iso-
Scheme 3
In contrast to 1, compounds 5–10 exhibit only one couple
of reduction and reoxidation peaks (Pr and Po) each in their
voltammograms [Fig. 1(b) for 5, 1(c) for 7, 1(d) for 10], but the
DPV curves of 5–8 [Fig. 2(b) for 5, 2(c) for 7] show much
broader peaks than those of 9 and 10 [Fig. 2(d) for 10]. Obvi-
ously, there are two nearly simultaneous one-electron reduc-
tions of 5–8 being a consequence of the larger separation of the
tetrazolium rings in 5–8 as compared to 1, whereas the peak
couples of 9 and 10 correspond to one-electron transfers each.
Thus, in the first step in all cases a one-electron transfer occurs
to form the radical anions. A second one-electron transfer to
give the dianions follows in the case of 1 and 5–8, however, only
in the case of 1 the second electron-transfer occurs at noticeably
larger potential.
As we can see from Table 1, the (first) reduction potential
does not much vary with the type of compounds or with the
solvent (Py, DCM). This indicates that the tetrazolium system
acts as the pitfall for the electrons. However, Py seems to relieve
the electron-transfer to a small extent. Also, insertion of
aromatic systems at the exo-nitrogen lowers the potential, e.g.
1→5, 1→7, 10→9. Interestingly, a biphenylene bridge is more
effective than a phenylene bridge (5→6, 7→8), and there is
practically no difference between a p- and a m-arylene bridge.
This may be a consequence of the larger distance between the
centres of negative charge in the latter cases (6 and 8) resulting
in a lower energy of the corresponding dianions.
lated cations 1 ϩ or 12ϩ (see above) are subjected to the electro-
᝽
chemical investigation. Thus, 1 was characterized as a new
compound exhibiting four-step redox behaviour (Scheme 3)
capable of acting both as an electron acceptor and as an elec-
tron donor, depending on the ionization potential and electron
affinity of a suitable partner.4
The p-phenylene bridged species 5 undergoes also two revers-
ible one-electron oxidations in DCM [Fig. 3(b) for CV, Fig. 3(e)
for DPV]. As can be seen from Fig. 3 and Table 1, the
peak-couples are less separated for 5 than for 1; this may be a
consequence of the larger distance (and hence greater
independence) of the two tetrazolium rings in 5. It further turns
out that the first electron-transfer is more facile (∆E = 197 mV)
In the oxidation process, the species show stronger diversities.
Compound 1 exhibits two successive reversible one-electron
988
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995

View Article Online
(∆E = 80 mV), which is again a consequence of the larger dis-
tance and hence stronger independence of the active tetrazo-
lium centres in 6. The first oxidation peak, and to a lesser degree
also the second, is shifted to a more positive potential (see Table
1), which may be explained by the following arguments: In the
ϩ
radical cation 6 the delocalization of the odd electron is
᝽
hampered by the twisting (atropisomerism) of the biphenyl
ϩ
ϩ
system leading to a higher energy of 6 with respect to 5
.
᝽
᝽
Although the “dication effect” is further relieved in 62ϩ as com-
pared to 52ϩ, the biphenoquinone-bisimine structure (C) seems
Ph
N
N
N
N
Ph
N
N
N
Ph
N
N
N
C
Ph
to be less stable than the quinone-bisimine structure B; we even
have to consider an equilibrium with a biradical state of 62ϩ 5
.
Fig. 4 Multi-sweep CVs of compounds 1, 5, 7 and 10 (a) 1 in Py, (b) 5
in Py, (c) 7 in DCM and (d) 10 in Py. Potential vs. Ag/AgClO4 (0.01 M in
The m-phenylene bridged compound 7 gives rise to a CV
[Fig. 1(g)] in Py, which shows two oxidation peaks being
irreversible (I) or partially reversible (II) as well as a rereduction
peak a. The DPV [Fig. 2(g)] confirms the two-peak oxidation.
To our surprise, however, in DCM only one oxidation peak can
be observed, which is followed by several small peaks at higher
potentials, and two rereduction peaks a and b [Fig. 3(c)]. This
feature with one dominant oxidation peak is also evident from
the DPV [Fig. 3( f )]. The results indicate a slow decomposition
of 7 in the oxidation process to give a complicated mixture
especially in DCM. In Fig. 4(c) the return potential was laid
close behind peak I to narrow the sweep-range in the positive
part of the potential scale. As a result, the rereduction peak a in
the second scan has decreased in intensity but the reduction
peak b remains nearly unaffected. Therefore, the peak a must
belong to a product which is formed from the oxidized species
of 7 more slowly than the product which gives the rereduction
peak b. In the third scan, it turns out that b has a corresponding
oxidation peak bЈ forming a reversible peak couple with b,
whereas a shows no corresponding oxidation peak. The product
X responsible for b/bЈ is apparently so stable that on oxidation
of 7 in peak I it is produced at or near the working electrode
surface in a considerable amount. Therefore, if we do not stir
the solution after CV measurement and run a DPV closely
afterwards, we observe a peak at the position bЈ in the positive
DPV scan (not shown). We are currently endeavouring to
isolate and characterize X after large scale electrolysis.
acetonitrile–0.1 M Bu4NPF6), scan rate 50 mV sϪ1
.
from 1, presumably because of the better delocalization of the
ϩ
odd electron over both sides in the cation radical 1 as com-
᝽
pared to the situation with 5 ϩ. The difference in the second
᝽
oxidation potentials (∆E = 270 mV) is most striking. Again, the
two above mentioned effects seem to be responsible: the
dication 12ϩ apparently adopts a structure A of the π-system, in
Ph
N
Ph
N
N
N
N
N
N
Ph
N
N
N
Ph
N
N
N
N
N
Ph
N
Ph
N
N
N
Ph
N
A
B
Ph
which the positive charges are localized mainly in the tetrazo-
lium rings. This is, of course, true also for 52ϩ, however, here the
Coulomb repulsion of the two charges, the so-called “dication
effect”, is relieved owing to the larger distance between the
tetrazolium rings (B), which stabilizes the dication.
In Py 5 behaves completely differently: there are apparently
four successive quasi-reversible or irreversible oxidation peaks
in the CV [Fig. 1( f )]. According to the DPV [Fig. 2( f )] it looks
as if a successive four-step one-electron oxidation up to the
tetracations would be possible. The number of transferred elec-
trons as determined by coulometry, however, varies between 4.4
and 8.3, depending on the conditions. It is quite conceivable
that also in Py no higher oxidation than up to the dication
occurs, which then reacts with Py to give products which are
further oxidized in peaks III and IV. Indeed, there are several
rereduction peaks, e.g. those denoted by a and b, if the reverse
scan is continued until Ϫ2.2 V [Fig. 4(b)]. The third scan
does not show any reoxidation peaks of a and b (cf. the situ-
ation with 1, above). If the switch potential is varied (Fig. S1,
Supplementary Material), it turns out that the first oxidation
peak is reversible on the time-scale of CV also in Py, and no
rereduction peak a or b occurs, when the potential is reversed
between the first and second oxidation peak.
The CV features of 8 are similar to those of 7, (Fig. S4,
Supplementary Material) especially in Py, whereas in DCM
only one rereduction peak a is observed. The oxidation poten-
tials of 8 are even higher than those of 7, which means that now
an interaction between the two tetrazolium rings is practically
no longer possible. And, indeed, the first oxidation potential is
about the same (Table 1) as that of 9, a mono-tetrazolium
system with one phenyl ring at the exocyclic nitrogen (see
below). Thus, 8 behaves like two units of 9, arbitrarily con-
nected by a bond.
The mono-tetrazolium compounds 9 and 10, finally, undergo
irreversible one-electron oxidations [cf. 10: Fig. 1(h) for CV,
Fig. 2(h) for DPV]. A three-scan multicycle CV [Py as solvent,
Fig. 4(d)] shows two rereduction peaks a and b in the second
scan, similarly to the situation with 7 in DCM. The third scan
reveals that the rereduction b is partially reversible. We have to
assume fragmentation of the tetrazolium part following the
oxidation. The oxidation potentials of 10 are the highest in the
whole series, nearly 1.2 V higher than that for 1. In 10 there is
The biphenylenyl compound 6, on oxidation, gives CV and
DPV curves revealing the same features as those of 5, in Py as
well as in DCM (Fig. S2 and S3, Supplementary Material).
However, the peak-couples of 6 in DCM move closer together
no stabilization of the primarily formed cation-radical 10 ϩ by
᝽
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
989

View Article Online
10 nitrogen and 20 hydrogen atoms, if it is delocalized over the
whole system. Therefore, we have to expect that simulations will
only give rough coupling constants. One of them with a com-
ponent linewidth (∆H) of 0.16 mT gave the hyperfine splitting
constants (hfc) aN (2) = 0.527, aN (4) = 0.195 mT (the values
in parentheses denote the number of equivalent nuclei). This
ϩ
means, that some of the nitrogen atoms of 1 as well as the
᝽
hydrogen atoms seem to couple only insignificantly with the
free electron. As judged from the molecular structure we should
expect a detachment of the electron on oxidation of 1 from the
exocyclic nitrogen atom Nexo. The free spin density should
therefore be highest and the hfc largest for Nexo (resonance
structure D). Via allylic resonance, spin density should also be
transferred to N-4 (resonance structure E). The moiety Ph–
Ph
Ph
1
R1
2
3
N
N
N
N
N
N
N
β
Nexo
N
R2
N
α
N
N
Ph
N
N
Ph
4
Ph
F
D
E
(N-3)–(N-4) in E corresponds to the fragment F present in
linear (DPPH) or cyclic hydrazyls (verdazyls, tetrazolyls). These
compounds develop a free spin density also at the adjacent Nβ
atom, being nearly as large as that on Nα,7 whereas the coupling
with the hydrogen atoms of the phenyl ring is small and about
the same for ortho- and para-protons.7 By analogy, two hfc’s aN
ϩ
(2) (Nexo) and aN (4) (N-3, N-4) would be reasonable for 1
,
᝽
implying that the electron is equally distributed over equivalent
positions in both moieties of the molecule.
Ph
N
15N
Ph
ϩ
ϩ
Fig. 5 EPR spectra of (a) 1 ϩ, (b) 1a ϩ, (c) 1b and (d) 1c in
N
N
N
᝽
᝽
᝽
᝽
N
N
BF4
N
DCM.
N
Ph
N15
any substituent at the exocyclic nitrogen. In agreement with this
neither in Py nor in DCM any reversibility can be observed.
The number of electrons transferred, as determined by coulom-
etry, is ca. 0.8 which points to a one-electron oxidation of 10 in
the oxidation peak. However, n-values determined by CPE in a
completely irreversible peak are surely not very accurate.
Ph
1a
Ph
N
N
N
Ph
N15
N15
N
N
BF4
N
N
N
Spectroscopic and magnetic investigations
Ph
The redox processes of the tetrazolium compounds produce
radical anions or cations as the primary species, which we tried
to detect and characterize by UV/Vis, EPR, and ENDOR
spectroscopy, and susceptibility measurements. The majority of
these investigations were performed on the chemically prepared
Ph
1b
Ph
cation-radical 1 ϩ, which is very stable and can easily be
᝽
N
15N
N
handled (see above). In situ generation by electrochemical
Ph
N15
N15
N
N
N
BF4
methods was also applied in some cases.
N
Ph
N15
UV Spectra. As already mentioned in the synthesis section,
Ph
ϩ
the radical cation 1 shows a longest-wavelength absorption
᝽
(625 nm) which is bathochromic with respect to the neutral
species 1, as well as to the dication 12ϩ, being typical for such
systems.6
1c
In order to facilitate a determination of the nitrogen coup-
lings experimentally, we prepared three types of 15N-labelled
EPR Spectra of 1 ϩ. Radical cation 1 ϩ, generated by the
analogues (1 ϩa–c) of 1 with 15N atoms in the positions
ϩ
᝽
᝽
᝽
᝽
electrochemical oxidation of 1 in pyridine solution at room
shown in the respective formulae. Their EPR spectra in DCM
temperature, showed the same EPR spectrum as that of the
[Fig. 5(b)–(d), respectively] reveal an interesting phenomenon:
chemically prepared species 1 ϩ BF4Ϫ in DCM [Fig. 5(a)]; aside
The spectrum of 1a ϩ (15N in positions 2 and 2Ј) does not show
᝽
᝽
from the number of low-intensity lines observable at the wings
of the spectrum. The g-value was found to be 2.00391. Both
spectra are only moderately resolved, which is quite under-
standable because the odd electron has a chance to couple with
any change as compared to that of 1 ϩ, whereas the spectrum
᝽
of 1b (15N in both Nexo positions) is distinctly different. The
ϩ
᝽
spectrum of the tetra-labelled species 1c (15N in 2-, 2Ј- and
ϩ
᝽
both exo-positions), as expected from the just described
990
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995

View Article Online
ϩ
Fig. 6 ENDOR spectra of (a) 1a ϩ and (b) 1b ϩ. The spectrum of 1
᝽
᝽
᝽
(not shown) is similar to (a). Nitrogen ENDOR lines are centred about
1/2aiso, the isotropic hfc, and spaced by 2νn, the Larmor frequency of
the free nucleus (14N, 15N or H). Experimental conditions: solvent:
DCM; Prf (radiofrequency power): 30 W, Pmw (microwave power): 24
mW, rf frequency modulation: 30 kHz, with 100 kHz deviation, broad
baseline features were subtracted. (a): Temperature T = 280 K, 45 min
total measuring time; (b): T = 270 K, 34 min.
Fig. 7 EPR spectra obtained on cathodic reduction of (a) 1 and (b) 10
in Py.
lar not in the frequency region where ENDOR resonances
would be expected if any of the 14N nuclei contributing to a2
were replaced by 15N. This again suggests that the hfc of 15N-2
ϩ
features of 1a and 1b ϩ, completely coincides with that of
᝽
᝽
᝽
1b ϩ. As a consequence, a coupling of the free electron with
N-2 and N-2Ј cannot be detected, i.e. aN-2 and aN-2Ј should be
smaller than the line-width (<0.09 mT). The EPR spectra of
1b ϩ (and 1c ϩ), on a first order treatment, show a large triplet
in 1c ϩ does not contribute to a2, and that the hfc of the nitro-
᝽
gens at that position was too small to be observed by ENDOR
(a < 1.5 MHz; 0.05 mT).§ Proton ENDOR lines (0.50 MHz;
0.018 mT) were observed in the region around νH, the Larmor
frequency of the free proton, which are assigned to the only
protons in the radical, i.e. those of the phenyl rings. No attempt
was made to resolve the hfc’s expected for the different protons
of the phenyl rings.
᝽
᝽
structure (1:2:1) with each triplet line being further split into
a nonet [the multiplicity is close to the theoretical one
(1:4:10:16:19:16:10:4:1)], the three nonets being partially
overlapped. The triplet splitting results from two equivalent
15N-nuclei, the nonet splitting from four equivalent 14N-nuclei.
ϩ
The a-values, directly extracted from the EPR spectrum of 1b
are: a (2) = 0.75 mT and a (4) = 0.185 mT. The value of
(2) corresponds to a (2) = 0.53 mT in 1 (a N = a
0.7129 from the gyromagnetic ratio). The results agree well with
᝽
15N
14N
EPR Spectra on cathodic reductions. The EPR spectrum
ϩ
15N
14N
N
14
15
a
×
᝽
obtained on cathodic reduction of 1 is shown in Fig. 7(a). Con-
ϩ
trary to the radical-cation 1 only a broad line could be
᝽
ϩ
the simulation of the spectrum of 1 mentioned above and
᝽
observed; the g-value was 2.00294. The poor resolution might
prove (i) that the largest hfc belongs to Nexo, (ii) that both Nexo
are equivalent, i.e. the free electron is symmetrically delocalized
over both moieties of the molecule. In conformity with earlier
considerations (see above), the smaller a N-value should be
attributed to N-3, N-3Ј, N-4 and N-4Ј. Strictly speaking, aN-3
result from an interaction of the odd electron with all ten nitro-
gen nuclei in a radical-anion 1 Ϫ. However, since the two reduc-
᝽
tion potentials of 1 lie close together, and a potential control
was not possible in our intra-muros electrochemical cell, a
reduction to the dianion 12Ϫ might have occurred. Contrary to
the dication 12ϩ (see below), this dianion could well be in equi-
librium with a diradical whose dipolar coupling would then be
responsible for the bad resolution. This possibility is currently
being tested.
14
and aN-4 are not exactly equivalent and, indeed, a refined simu-
ϩ
᝽
lation (∆H = 0.0837 mT) of the spectrum of 1 (S5, Sup-
plementary Material) reveals two inequivalent 14N-couplings,
14
i.e. 0.185 and 0.199 mT, besides a N = 0.530 mT. Presumably,
the larger one (0.199 mT) belongs to N-3 and N-3Ј, the smaller
one (0.185 mT) to N-4 and N-4Ј, quite in analogy to the results
obtained with DPPH, verdazyls and tetrazolyls.7
Compound 10, however, can be reduced electrochemically by
one-electron transfer to a radical anion 10 Ϫ. We observed an
᝽
EPR spectrum [Fig. 7(b)] with g = 2.00397. The resolution is
much better than in all other cases discussed in this paper.
ENDOR Spectra. These results were confirmed by ENDOR
Unfortunately, all attempts to simulate it in terms of the 14N
ϩ
experiments (Fig. 6). Two 14N hfc’s were observed for 1a
:
Ϫ
and 1H nuclei of 10 failed. Therefore, the EPR spectrum
᝽
᝽
a1 = 15.1 MHz (0.539 mT) and a2 = 5.9 MHz (0.21 mT). The
larger hfc a1 can again unambiguously be assigned to the exo-
cyclic nitrogens by comparison with the selectively 15N labelled
could belong to a mixture of 10 Ϫ and a cleavage product there-
᝽
from, the more so, since the analogue 9 could be cleaved by
reduction with zinc (see above). Unfortunately, these radical
anions are not persistent enough for the ENDOR measurement
ϩ
compound 1b ϩ, because in the ENDOR spectra of 1b the
᝽
᝽
lines due to a1 in 1a ϩ are absent. Instead, two lines correspond-
᝽
ing to a = 21.2 MHz (0.756 mT) are observed, as expected. The
ENDOR lineshapes for the lines corresponding to a2 show a
small splitting, indicating that two inequivalent groups of
§ Although this is a plausible suggestion, which qualitatively agrees
with the results of the simulation of the EPR spectra, the difficulties
associated with ENDOR at these low frequencies (see Experi-
mental section), which are compounded by the unfavourable
relaxation properties of nuclei like 14N and 15N in solution, limits
the degree of confidence in arguments based on the absence of
ENDOR signals.
nitrogens contribute, one with a2a = 5.6 MHz (0.198 mT), the
ϩ
other with a2b = 6.0 MHz (0.214 mT). Comparison with 1c
᝽
reveals no changes in the position and in the shape of these
lines. Furthermore, no additional lines are observed, in particu-
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
991

View Article Online
ϩ
᝽
to get a final proof for their structure and further insight into
the odd electron distribution.
on the other hand, give cation-radicals 7 ϩ, 8 and dications
᝽
(presumably existing as diradicals) 72ϩ, 82ϩ which are much less
stabilized, because the meta-junction, aside from an additional
degree of twisting, neither allows a delocalization of the odd
electron in 7 ϩ, 8 ϩ over the whole system in terms of canonical
Susceptibility measurements. The dication 12ϩ is diamagnetic
and no triplet system, as can be seen from its well-resolved
NMR spectrum and the temperature-independent magnetic
susceptibility (Ϫ2.4 × 10Ϫ4 emu molϪ1) observed in a SQUID
magnetometer for a powdered sample, if care is taken that no
monocation 1 ϩ is present as impurity. The monocation 1 ϩ, on
᝽
᝽
structures nor a conjugated quinonoid π-system in 72ϩ, 82ϩ
.
This is nicely reflected by the higher oxidation (peak) potentials
of 7 and 8 relative to those of 5 and 6 and the small variation of
these values between Py and DCM as solvents. The instability
of the oxidized species of 7 and 8 is thus an intrinsic one and
not caused by the nucleophilic attack of Py; therefore, DCM
does not give any advantage over Py as solvent in the oxidation
of these species.
᝽
᝽
the other hand, reveals temperature-dependent paramagnetism
(ca. 1.42 Bohr magnetons without diamagnetic correction) with
a normal Curie–Weiss behaviour between 50–300 K. At lower
temperature, there is a deviation to higher magnetic moments,
which is not yet understood.
In the case of reduction, the formation of the anion-radicals
and dianions occurs at very similar potentials, which either
means that the anion-radical state is more difficult or the
dianion state more easy to reach from the neutral state as com-
pared to the corresponding cationic species. Since the potentials
correspond to those of the monovalent reduction of the mono-
tetrazolium systems 9 and 10, the first suggestion seems to be
true, which means that resonance stabilization of the anion-
radicals should be less effective than that of the cation-radicals;
unfortunately, the EPR spectra obtained on reduction of 1 and
10 do not give any information hereto. At the same time, the
dianions might exist as diamagnetic conjugated species or as
diradicals with negligible interaction behaving like two separate
mono-radicals. This would explain why the potential of the
second electron transfer is about the same as that of the first
one and why for all compounds 1, 5–10 the reduction potentials
do not vary appreciably. Altogether, the mesoionic species pre-
sented here show several interesting features,5 which could be of
importance for materials science, whose understanding however
requires further investigations.
Conclusions and summary
A series of mesoionic compounds 1 and 5–10 are now easily
accessible, even in the solid state. These compounds are very
stable and can be oxidized or reduced chemically or electro-
chemically; in some cases the processes are reversible and occur
stepwise via ion-radicals to diions (cations and anions). It
turned out that, in general, the oxidation products are more
stable in DCM, the reduction products in Py. This is reasonable
because DCM is a weak electrophile reacting with anions,
whereas Py acts as a strong nucleophile, attacking cations.
Aside from these solvent effects, the stability of the substrate
ions seems to be favoured largely by two effects: (i) a strong
delocalization of the free electron in the ion-radical with a
preferential residence on the exocyclic link atoms, thus allowing
for a more or less localization of the charges in the tetrazolium
rings, (ii) a π-conjugation of the ionic centres in the diions,
which tends to separate the charges as far as possible. Both
effects will be strongly influenced by the distance of the tetra-
zolium rings. The larger the distance, the less the Coulomb
repulsion between the charges (diion effect)—but also the
delocalization of the free electron is less effective.
Experimental
General
The effects (i) and (ii) are well demonstrated in the case of 1
Melting points were measured with a hot-stage apparatus and
are uncorrected. Elemental analyses were carried out at the
Elemental Analysis Centre of Kyoto University. IR spectra
were taken as KBr discs on a JASCO A-102 spectrometer.
Electronic spectra were measured on a Hitachi 124 or U-3500
spectrophotometer. NMR spectra were obtained using a
which is the most thoroughly investigated species of the present
series. It can be oxidized stepwise to the cation-radical 1 ϩ and
᝽
the dication 12ϩ, both of which could be isolated as stable BF4
Ϫ
salts. A comprehensive EPR/ENDOR study, using also specific
15N-labelling, shows that in the cation-radical 1 ϩ, indeed, the
᝽
free electron is delocalized over both moieties of the molecule,
although not uniformly over all N-atoms of the tetrazolium
ring; only two ring N-atoms reveal a detectable coupling and
hence indicate a free spin density at the corresponding ring
positions.¶ In the sense of the above statement (i), however, the
predominant part of the free spin density is experimentally
found at the two exocyclic (hydrazo) nitrogens. This may, there-
fore, be the reason for the stability and the very low first oxid-
ation potential of 1. The likewise remarkable stability of the
diamagnetic dication 12ϩ, on the other hand, may be a con-
sequence of (ii), i.e. the formation of an azo-bridge between the
tetrazolium cationic centres.
1
Hitachi 90 (90 MHz for H), a Varian XL-200, Varian Gemini
1
200 (200 MHz and 50 MHz, for H and 13C, respectively) or a
1
Bruker AMX 400 (400 MHz and 100 MHz, for H and 13C,
respectively). Chemical shifts are recorded in ppm downfield
from tetramethylsilane. J values are given in Hz. Assignment of
signals was made based on the chemical shifts, peak intensities
and INEPT techniques. Mass spectra were taken with a Hitachi
M-2000S spectrometer (EI, 70 eV). SQUID measurements were
carried out with a Quantum Design, MPMS system.
15N-sodium nitrite (99% 15N) and 15N2-hydrazine sulfate
(99% 15N) were purchased from Isotec Inc.
Although no problems were encountered, it should be noted
that polyaza compounds as well as the perchlorates used for the
electrochemical procedures in general may be explosive and
should be handled with due care.
In the case of 5–10, we note a fundamental difference
between the para-bridged bis-tetrazolium systems 5 and 6 on
the one hand and the meta-bridged species 7 and 8 on the other.
On oxidation, the first mentioned molecules produce cation-
radicals (5 ϩ, 6 ϩ) and dications (52ϩ, 62ϩ) successively, which
᝽
᝽
are stabilized by resonance (5 ϩ, 6 ϩ) or by formation of a con-
Electrochemical and EPR/ENDOR measurements
᝽
᝽
jugated quinonoid system (52ϩ, 62ϩ). The difference in Py and
Electrochemical experiments were carried out on a Windows-
driven BAS 50W electrochemical analyzer (Bioanalytical
Systems, Inc., West Lafayette, IN 47906, USA). For CV and
DPV a 3 mm platinum or glassy carbon electrode (Metrohm)
was used as working electrode. The auxiliary electrode con-
sisted of a Pt wire or disk. Ag/AgClO4 (0.01 M in MeCN/0.1 M
Bu4NPF6) was used as reference electrode and all potentials
given relate to this electrode. The formal potentials of the
ferrocene/ferrocenium electrode against this reference electrode
DCM as solvents results only from the ability of Py to exert a
nucleophilic attack on 5 ϩ, 6 and, especially, on 52ϩ, 62ϩ
.
ϩ
᝽
᝽
Since the positive charges in the dications 52ϩ and 62ϩ are far
more separated than in 12ϩ, the second oxidation potentials of 5
and 6 are much lower than that of 1. The meta-bridged species,
¶ Therefore, strictly speaking, the ring symbols in the tetrazolium rings
used in the formulae 1a ϩ, 1b ϩ and 1c ϩ are only of a formal nature.
᝽
᝽
᝽
992
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995

View Article Online
are 0.19 and 0.21 V in Py and DCM solutions, respectively. It
was separated by two glass frits from the Haber-Luggin capil-
lary. The compartment between reference electrode and H.-L.
capillary was filled with 0.1 M Bu4NPF6–acetonitrile solution.
The H.-L. capillary itself always contained the same solution as
the cell for the measurements. For CPE cylindrical Pt (10%
iridium) gauze working and auxiliary electrodes, separated by
two porous glass frits, were used. To the compartment contain-
ing the auxiliary electrode, quinhydrone in 10-fold excess over
the investigated compound was added to ease the electron
transfer. High quality commercial pyridine (Fluka) stored over
4 Å molecular sieves was used without further treatment.
Alternatively, commercial pyridine was purified by distilling it
three times under argon after drying over KOH for several
weeks, and then kept over 4 Å molecular sieves. DCM was
refluxed over P2O5 for several hours and afterwards distilled
twice. Tetrabutylammonium hexafluorophosphate (Bu4NPF6)
as supporting electrolyte was prepared from tetrabutyl-
ammonium bromide and ammonium hexafluorophosphate or
purchased (Fluka), recrystallised four times from EtOH and
dried in vacuo at 110 ЊC for 48 h. The concentration of the
substrates was 0.3–2 mM.
EPR spectra were obtained with a Bruker ESP 300E
spectrometer. For g-factor measurements, the field gradients
were corrected by replacing the sample with a reference com-
pound (2,6-di-tert-butyl-4-tert-butoxyphenoxyl in benzene,
g = 2.00463). All ENDOR spectra were measured with a
laboratory developed spectrometer described elsewhere8 on
samples prepared on a high vacuum line, by dissolving the
solid radical in DCM, and degassed by repeated freeze–pump–
thaw cycles. DCM was dried over calcium hydride. As baseline
artifacts due to cross talking of microwave (mw) and radio-
frequency (rf) fields are common in the low frequency region of
the ENDOR spectrum, and since the partially resolved EPR
spectrum of the sample causes baseline artifacts which can
appear as lines, the following control experiments were
performed to discriminate artifacts from ENDOR lines: by
changing the magnitude of the static magnetic field B0 and
employing B0 modulation, lines due to cross talking of mw and
rf fields should shift or be suppressed, thus ENDOR signals
were identified by not being susceptible to these changes. The
temperature dependence of the relaxation rates governing
liquid solution ENDOR intensities9 causes pronounced inten-
sity changes of ENDOR lines with temperature, which is not
the case for the spurious background signals. Thus, the
ENDOR signals were further identified by having temperature
dependent intensities.
methyl iodide (0.34 g, 2.4 mmol) was added and the mixture
was further refluxed for 3 h. NaBF4 (1.0 g) in water (10 ml) was
added and the organic layer was separated, dried (Na2SO4), and
evaporated. The residue was chromatographed on silica gel
(DCM–acetone = 1:1) to give 2 as brass-yellow crystals (41 mg,
56%); mp 221–222 ЊC (from EtOH) (Found: C, 56.4; H, 4.2; N,
24.2. Calc. for C27H23BF4N10: C, 56.5; H, 4.0; N, 24.4%); νmax
/
cmϪ1 1658, 1606, 1580, 1492, 1472, 1462, 1378, 1340, 1300,
1284, 1222, 1180, 1060, 990, 774 and 688; δH (90 MHz; CDCl3)
3.64 (3H, s, Me), 7.00–7.78 (16H, m, Ph) and 7.98–8.32 (4H, m,
Ph); δC (50 MHz; CDCl3) 40.2 (Me), 120.9 (C-o), 121.5 (two
overlapped signals) (C-o), 126.6 (C-o), 129.1 (C-m), 129.4 (C-p),
129.7 (C-m), 130.6 (two overlapped signals) (C-m), 131.6 (C-p),
132.8 (C-p), 133.6 (C-p), 133.8 (C-i), 134.2 (C-i), 135.7 (C-i),
135.8 (C-i), 157.8 (Cϩ) and 163.4 (Cϩ).
Protonation of 1. Azine 1 (60 mg, 0.13 mmol) was dissolved
in DCM (15 ml) and acetic acid (50 µl) was added. The mixture
was stirred for 5 min and poured into aqueous sodium
tetrafluoroborate (2.7 g in 30 ml of water). The organic layer
was separated, dried (Na2SO4), and the solvent was evaporated
to give monoprotonated compound 3 (63 mg, 85%) as brownish
orange crystals; mp 130 ЊC (from EtOH–hexane) (Found: C,
55.1; H, 4.0; N, 24.3. Calc. for C26H21BF4N10ؒ0.5H2O: C, 54.9;
H, 3.9; N, 24.6%); νmax/cmϪ1 3345, 3040, 1696, 1614, 1586, 1490,
1460, 1380, 1338, 1292, 1234, 1196, 1060, 984, 924, 772 and 688;
δH (200 MHz; CDCl3) 7.62 (12H, br m, Ph) and 8.25 (8H, br m,
Ph).
Diprotonated compound (4). Azine 1 (44 mg, 0.093 mmol) was
dissolved in trifluoroacetic acid (0.50 ml) to give a colourless
solution. Excess trifluoroacetic acid was evaporated to dryness
to give colourless crystals of 4 in quantitative yield. An ana-
lytical sample was dried under vacuum at 50 ЊC (Found: C,
41.9; H, 2.4; N, 13.6. Calc. for C30H22F6N10O4ؒ3(CF3CO2H): C,
41.5; H, 2.4; N, 13.4%); νmax/cmϪ1 3105, 2900, 1780, 1652, 1626,
1496, 1464, 1436, 1376, 1324, 1298, 1212, 1170, 1142, 998, 928,
842, 818, 800, 786, 766, 730, 712, 702 and 680; δH (90 MHz;
CF3CO2D) 7.45–7.85 (16H, m, Ph) and 7.90–8.24 (4H, m, Ph);
δC (50 MHz; CF3CO2D) 122.5 (C-o), 126.5 (C-o), 132.3 (C-i),
132.6 (C-m), 133.2 (C-m), 135.9 (C-p), 136.5 (C-p), 136.8 (C-i)
and 159.9 (Cϩ).
Oxidation of 1: Synthesis of radical cation (1 ϩ). To a solu-
᝽
tion of 1 (0.47 g, 1.0 mmol) in DCM (20 ml) was added lead
tetraacetate (0.25 g, 0.50 mmol), and the mixture was stirred at
room temperature for 10 min. NaBF4 (1.0 g) in water (10 ml)
was added and the organic layer was separated. The layer was
washed with NaHCO3 (5%) and the solvent was evaporated
Synthesis
under vacuum. The residue was column-chromatographed on
ϩ
5,5Ј-Azinobis(1,3-diphenyltetrazole) (1). A mixture of 5-
chloro-1,3-diphenyltetrazolium tetrafluoroborate3 (0.35 g, 1.0
mmol) and hydrazine hydrate (80%, 0.32 ml, 5.0 mmol) in
acetonitrile (50 ml) was stirred at room temperature for 2 h. The
solvent was evaporated and the residue dissolved in DCM.
Insoluble materials were filtered off and the filtrate was evapor-
ated to dryness. The residual solid was recrystallised from
DCM–ether to give 1 as deep purple crystals (0.17 g, 72%);
mp > 300 ЊC (Found: C, 66.2; H, 4.2; N, 29.8. Calc. for
C26H20N10: C, 66.1; H, 4.3; N, 29.65%); νmax/cmϪ1 3090, 1614,
1580, 1490, 1470, 1372, 1322, 1290, 1214, 1156, 1070, 970, 760,
700 and 692; λmax (DCM)/nm 269 (sh, log ε 4.64), 280 (4.65) and
542 (3.73); δH (90 MHz; CDCl3) 7.12–7.76 (12H, m, Ph) and
7.94–8.40 (8H, m, Ph); δC (50 MHz; CDCl3) 120.4 (C-o), 120.9
(C-o), 127.1 (C-m), 129.7 (C-p), 129.9 (C-p), 131.5 (C-m), 136.5
(C-i), 136.9 (C-i) and 158.4 (Cϩ); m/z 472 (Mϩ, 100%).
silica gel (DCM–acetone = 4:1) to give 1 as greenish brown
᝽
crystals (0.17 g, 31%); mp 185–189 ЊC (from MeOH–Et2O)
(Found: C, 55.7; H, 3.3; N, 24.75. Calc. for C26H20BF4N10: C,
55.8; H, 3.6; N, 25.0%); νmax/cmϪ1 1592, 1492, 1450, 1358, 1210,
1054, 986, 762 and 670; λmax (MeCN)/nm 297 (log ε 4.58), 350
(sh, 4.29), 448 (4.03) and 625 (sh, 3.46).
15N-labelled analogues (1 ϩa–c). The labelled compounds
᝽
1
᝽
ϩa–c were synthesized by the oxidation of 1a–c. Compound
1a was prepared from 2-15N-5-chloro-1,3-diphenyltetrazolium
tetrafluoroborate, which was synthesized according to the pub-
lished method3,10 by using 15N-sodium nitrite. Compound 1b
was obtained by the condensation of 5-chloro-1,3-diphenyl-
tetrazolium tetrafluoroborate with 15N2-hydrazine sulfate: To a
mixture of 5-chloro-1,3-diphenyltetrazolium tetrafluoroborate
(0.70 g, 2.0 mmol) and 15N2-hydrazine sulfate (0.13 g, 1.0 mmol)
in acetonitrile (100 ml) was added dropwise aqueous NaOH
(1 M, 20 ml). The mixture was stirred for 10 min. The resulting
precipitate was filtered, washed with water, and dried to give 1b
(0.25 g, 52%). In the same manner, 1c was obtained from
Methylation of 1. To a solution of azine 1 (60 mg, 0.13 mmol)
in DCM (10 ml) was added methyl iodide (0.34 g, 2.4 mmol),
and the mixture was refluxed for 12 h. An additional amount of
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
993

View Article Online
2-15N-5-chloro-1,3-diphenyltetrazolium tetrafluoroborate and
N,NЈ-Bis(1,3-diphenyltetrazolium-5-yl)biphenyl-4,4Ј-diamide
15N2-hydrazine sulfate.
(6). A mixture of 5-chloro-1,3-diphenyltetrazolium tetrafluoro-
borate (0.35 g, 1.0 mmol) and benzidine (0.18 g, 1.0 mmol) in
acetonitrile (3.0 ml) was stirred at room temperature for 24 h.
Water was added and the resulting precipitate was filtered and
washed with water. The product was suspended in DCM (30
ml) and aqueous NaOH (1 M, 10 ml) was added. The mixture
was vigorously stirred and the red DCM layer was separated,
dried over anhydrous sodium sulfate, and evaporated. A dark-
red solid (0.25 g, 79%) was obtained, which was recrystallised
from DCM (ca. 15 ml) to give the pure product (0.18 g, 55%);
mp > 300 ЊC (Found: C, 72.5; H, 4.6; N, 22.2. Calc. for
C38H28N10ؒ0.5H2O: C, 72.0; H, 4.6; N, 22.1%); νmax/cmϪ1 1620,
1580, 1488, 1370, 1334, 1286, 1254, 1210, 1174, 1110, 1070, 962,
826, 756, 680; λmax (CHCl3)/nm 296 (log ε 4.46) and 376 (sh,
3.18); δH (200 MHz; CDCl3) 7.42 (2H, t, J 8), 7.48–7.70 (18H,
m), 8.22 (4H, m) and 8.40 (4H, br d, J 8); δC (100 MHz, CDCl3)
120.2, 121.3, 122.9 (C-o and C-Ar); 126.7 (C-Ar), 127.9 (C-p),
129.1 (C-m), 129.6 (C-m), 131.2 (C-p); 134.0, 135.3, 136.4 (C-i
and Cq-Ar); 147.4 (Ar-C-N) and 154.3 (Cϩ); m/z (FD) 624
(Mϩ).
Dication (12؉). Crystals of 1 (0.30 g, 0.63 mmol) were added
in portions to concentrated nitric acid (10 ml), and the mixture
was stirred for 10 min. NaBF4 (11 g) in water (20 ml) was added
and the resulting orange crystals were filtered, washed with
water, and dried to give 12ϩ (0.36 g, 87%); mp 193–196 ЊC
(Found: C, 48.3; H, 3.2; N, 21.85. Calc. for C26H20B2F8N10:
C, 48.3; H, 3.1; N, 21.7%); νmax/cmϪ1 1492, 1460, 1374, 1308,
1290, 1246, 1180, 1060, 1000, 926, 786, 770, 710, 696 and 684;
λmax (MeCN)/nm 293 (log ε 4.49), 352 (sh, 4.15) and 446 (3.88);
δH (90 MHz; CF3CO2D) 7.76–8.14 (12H, m, Ph), 8.14–8.36 (4H,
m, Ph) and 8.36–8.70 (4H, m, Ph); δC (50 MHz, CF3CO2D)
123.5 (C-o), 127.6 (C-o), 132.9 (C-m), 133.0 (C-m), 133.3 (C-i),
136.6 (C-p), 137.4 (C-p), 137.1 (C-i) and 161.8 (Cϩ).
Oxidation of radical cation 1 ϩ to dication 12ϩ. Radical cation
᝽
ϩ
1
ؒBF4Ϫ (26 mg, 0.047 mmol) was added to concentrated nitric
᝽
acid (0.50 ml). The mixture was ultrasonicated for 5 min and
worked up as above to give 12ϩ (29 mg, 95%).
N,NЈ-Bis(1,3-diphenyltetrazolium-5-yl)-m-phenylenediamide
(7). A mixture of 5-chloro-1,3-diphenyltetrazolium tetrafluoro-
borate (0.41 g, 1.2 mmol) and m-phenylenediamine (0.13 g, 1.2
mmol) in acetonitrile (3.0 ml) was stirred at room temperature
for 24 h. Water was added and the resulting precipitate was
filtered, washed with water, and dried to give crystals of the
conjugate acid (0.22 g, 51%); mp 112–120 ЊC. The acid (0.13 g,
0.18 mmol) was dissolved in acetonitrile (2.0 ml) and aqueous
NaOH (1 M, 6.0 ml) was added. A reddish-purple precipitate
was formed immediately, which was filtered, washed with water,
and dried to give N,NЈ-bis(1,3-diphenyltetrazolium-5-yl)-m-
phenylenediamide (89 mg, 92%); recrystallisation from ethanol
gave a pure sample (28 mg, 27%); mp 83 ЊC (Found: C, 66.3; H,
4.25; N, 24.1. Calc. for C32H24N10ؒ1.5H2O: C, 66.8; H, 4.7; N,
24.3%); νmax/cmϪ1 1620, 1556, 1484, 1466, 1364, 1322, 1278,
1238, 1202, 1148, 1104, 1064, 952, 904, 872, 816, 750 and 678;
λmax (CHCl3)/nm 272 (log ε 4.50) and 448 (sh, 3.20); δH (200
MHz; [2H6]DMSO) 6.93 (2H, d, J 8), 7.10 (1H, t, J 8), 7.46–7.68
(13H, m), 8.15 (4H, br d, J 8, Ph) and 8.37 (4H, br d, J 8, Ph);
δC (100 MHz; CDCl3) 116.0 (C-Ar), 117.6 (C-Ar), 120.3 (C-o),
121.3 (C-o), 127.7 (C-p), 128.6 (C-Ar), 129.1 (C-m), 129.5
(C-m), 131.1 (C-p), 135.7 (C-i), 136.5 (C-i), 149.3 (Ar-C-N) and
154.3 (Cϩ); m/z 548 (Mϩ, 100%), 353 (34), 328 (34), 245 (8), 169
(48), 105 (PhN2, 8) and 77 (Ph, 61).
ϩ
Reduction of radical cation 1 to azine 1. A mixture of
᝽
radical cation 1 ϩ (39 mg, 0.070 mmol) and zinc powder (14 mg,
᝽
0.21 mmol) in acetonitrile (10 ml) was stirred for 10 min under
argon. The solvent was removed under reduced pressure and
the residue was dissolved in DCM. The mixture was shaken
with aqueous NaHCO3 (5%). The organic layer was separated,
dried (Na2SO4), and evaporated. The residue was column-
chromatographed on silica gel (DCM–acetone = 4:1) to give 1
(25 mg, 76%).
Reduction of dication 12ϩ to azine 1. A mixture of dication 12ϩ
(42 mg, 0.065 mmol) and zinc powder (71 mg, 1.1 mmol) in
acetonitrile (20 ml) was stirred for 20 h. The solvent was
removed under reduced pressure and the residue was dissolved
in DCM. The solution was shaken with aqueous NaHCO3
(5%). The organic layer was separated, dried (Na2SO4), and
evaporated. The residue was column-chromatographed on
alumina (DCM–acetone = 4:1) to give 1 (20 mg, 65%).
Reduction of dication 12ϩ to radical cation 1 ϩ. A suspension
᝽
of dication 12ϩ (50 mg, 0.077 mmol) in ethanol (50 ml) was
stirred at room temperature for 24 h. The reaction mixture
turned from orange to greenish brown, and the crystals of 12ϩ
gradually went into solution. The solvent was evaporated and
the residue was chromatographed on silica gel (DCM–acetone)
N,NЈ-Bis(1,3-diphenyltetrazolium-5-yl)biphenyl-3,3Ј-diamide
(8). A mixture of 5-chloro-1,3-diphenyltetrazolium tetrafluoro-
borate (0.37 g, 1.1 mmol) and 3,3Ј-diaminobiphenyl11 (0.20 g,
1.1 mmol) in acetonitrile (5.0 ml) was stirred at room temper-
ature for 20 h. Aqueous NaOH (1 M, 30 ml) was added and the
product was extracted with DCM and dried (Na2SO4). Then,
the solvent was evaporated, giving a crude product (0.42 g),
which was recrystallised from a chloroform–ethanol mixture to
give the pure product (0.13 g, 38%) as a brick-red powder; mp
227 ЊC (dec.) (Found: C, 73.0; H, 4.5; N, 22.15. Calc. for
C38H28N10: C, 73.1; H, 4.5; N, 22.4%); νmax/cmϪ1 1622, 1564,
1490, 1468, 1368, 1330, 1288, 1206, 1160, 1112, 1070, 966, 758
and 680; λmax (MeCN)/nm 262 (log ε 3.70), 302 (3.72) and 442
(2.74); δH (400 MHz; CDCl3) 7.22 (2H, d, J 7.6), 7.32 (4H, m),
7.44 (12H, m), 7.78 (2H, t, J 1.8), 8.07 (4H, m) and 8.32 (4H, d,
J 7.9); δC (100 MHz; CDCl3) 119.9 (C-Ar), 120.0 (C-o), 121.1
(C-Ar), 121.2 (C-o), 122.1 (C-Ar), 127.8 (C-Ar or C-p), 128.6
(C-Ar or C-p), 129.1 (C-m), 129.5 (C-m), 131.1 (C-Ar or C-p),
135.4 (C-i), 136.3 (C-i), 142.4 (Cq-Ar), 149.1 (Ar-C-N) and
154.6 (Cϩ); m/z 624 (Mϩ, 0.3%), 404 (40) and 77 (Phϩ, 100).
ϩ
to give 1 ؒBF4Ϫ (21 mg, 49%).
᝽
N,NЈ-Bis(1,3-diphenyltetrazolium-5-yl)-p-phenylenediamide
(5). A mixture of 5-chloro-1,3-diphenyltetrazolium tetrafluoro-
borate (2.0 g, 5.8 mmol) and p-phenylenediamine (0.65 g, 6.0
mmol) in acetonitrile (17 ml) was stirred at room temperature
for 24 h. Water was added and the resulting precipitate was
filtered, washed with water, and dried to give crystals of the
conjugate acid (1.5 g, 70%); mp 260–262 ЊC. The acid (2.1 g, 2.9
mmol) was dissolved in acetonitrile (29 ml) and aqueous NaOH
(1 M, 87 ml) was added. A reddish-purple precipitate was
formed immediately, which was filtered, washed with water,
and dried to give N,NЈ-bis(1,3-diphenyltetrazolium-5-yl)-p-
phenylenediamide (1.6 g, 100%); mp 284–286 ЊC (Found: C,
69.2; H, 4.4; N, 25.3. Calc. for C32H24N10ؒ0.5H2O: C, 68.9; H,
4.5; N, 25.1%); νmax/cmϪ1 1614, 1590, 1484, 1464, 1368, 1322,
1288, 1252, 1196, 1170, 1148, 1112, 1090, 1064, 962, 750, 698
and 678; λmax (CHCl3)/nm 296 (log ε 4.96), 370 (sh, 4.04) and
443 (sh, 3.64); δH (200 MHz; CDCl3) 7.26 (4H, s, ArH), 7.39
(2H, t, J 8, Ph), 7.48–7.64 (10H, m, Ph), 8.21 (4H, m, Ph) and
8.43 (4H, br d, J 8, Ph); m/z 548 (Mϩ, 100%), 105 (PhN2, 15)
and 77 (Ph, 18).
Chemical reduction of 5 with zinc powder. A mixture of 5
(0.22 g, 0.39 mmol) and zinc powder (0.21 g, 3.2 mmol) was
994
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995

View Article Online
C. A. Ramsden, Tetrahedron, 1982, 38, 2965; W. D. Ollis, S. P.
ultrasonicated in DCM (6 ml) containing 15 drops of acetic
acid for 20 min at room temperature. Aqueous NaOH (1 M)
was added and the product was extracted with DCM. The
extracts were dried (Na2SO4) and concentrated. The residue was
purified by preparative TLC (alumina/DCM) to give 11 (79
mg, 46%) as deep-red crystals; mp 183–186 ЊC (Found: C, 70.0;
H, 4.8; N, 25.4. Calc. for C26H22N8: C, 69.9; H, 5.0; N, 25.1%);
νmax/cmϪ1 3420, 1620, 1572, 1488, 1400, 1360, 1324, 1280, 1230,
1200, 1160, 1100, 1064, 960, 750 and 680; λmax (MeCN)/nm 278
(log ε sh, 4.34), 309 (4.48) and 442 (3.38); δH (200 MHz; CDCl3)
4.62 (3H, br s, NH and NH2), 6.95–7.64 (15H, m, Ar and Ph),
8.16 (2H, m, Ph) and 8.34 (2H, br d, J 8, Ph); δC (50 MHz;
CDCl3) 120.3 (C-o), 121.7 (C-o), 124.0 (C-o), 124.1 (CH of Ar),
124.2 (C-p), 125.7 (CH of Ar), 128.4 (C-p), 129.4 (C-m), 129.8
(C-m), 130.0 (C-m), 131.2 (C-i), 131.6 (C-p), 135.3 (C-i), 136.4
(C-i), 143.2 (C-N), 147.3 (C-N), 152.3 (N-C-N) and 154.9 (Cϩ);
m/z 446 (Mϩ, 0.02%), 328 (100) and 77 (89).
Stanforth and C. A. Ramsden, Tetrahedron, 1985, 41, 2239.
2 S. Araki, Y. Wanibe, F. Uno, A. Morikawa, K. Yamamoto,
K. Chiba and Y. Butsugan, Chem. Ber., 1993, 126, 1149.
3 S. Araki, K. Yamamoto, M. Yagi, T. Inoue, H. Fukagawa,
H. Hattori, H. Yamamura, M. Kawai and Y. Butsugan, Eur. J. Org.
Chem., 1998, 121.
4 For reviews on multi-stage redox systems see: K. Deuchert and
S. Hünig, Angew. Chem., Int. Ed. Engl., 1978, 17, 875; S. Hünig and
H. Berneth, Top. Curr. Chem., 1980, 92, 1; H. Perlstein, Angew.
Chem., Int. Ed. Engl., 1977, 16, 519. For recent examples see: J. Zhou
and A. Rieker, J. Chem. Soc., Perkin Trans. 2, 1997, 931; J. Zhou,
M. Felderhoff, N. Smelkova, L. M. Gornostaev and A. Rieker,
J. Chem. Soc., Perkin Trans. 2, 1998, 343.
5 For the singlet–triplet problem of similar systems see: A. Rebmann,
P. Schuler, H. B. Stegmann and A. Rieker, J. Chem. Res. (S), 1996,
318; (M), 1996, 1765; A. Rebmann, J. Zhou, P. Schuler, A. Rieker
and H. B. Stegmann, J. Chem. Soc., Perkin Trans. 2, 1997, 1615.
6 See, e.g. G. Frenking, A. Rieker, J. Salbeck and B. Speiser,
Z. Naturforsch., 1996, 51b, 377.
7 See, e.g. A. R. Forrester and F. A. Neugebauer, in Landolt–
Börnstein, Numerical Data and Functional Relationships in Science
and Technology, ed. H. Fischer and K.-H. Hellwege, Springer-
Verlag, Berlin, Heidelberg, New York, 1979, New Series, Group II:
Atomic and Molecular Physics, Vol. 9, Magnetic Properties of Free
Radicals, Part c1, Organic N-Centered Radicals and Nitroxide
Radicals, pp. 51, 69, 75.
8 K. Möbius and R. Biehl, in Multiple Electron Spin Resonance
Spectroscopy, ed. M. M. Dario and J. H. Freed, Plenum Press,
New York, 1979, p. 475; K. Möbius, M. Plato and W. Lubitz,
Phys. Rep., 1982, 87, 171.
9 M. Plato, W. Lubitz and K. Möbius, J. Phys. Chem., 1981, 85, 1202.
10 R. N. Hanley, W. D. Ollis and C. A. Ramsden, J. Chem. Soc., Perkin
Trans. 1, 1979, 736.
11 C. C. Barker and F. D. Casson, J. Chem. Soc., 1953, 4184.
12 J. Macholdt-Erdniss, Chem. Ber., 1958, 91, 1992.
Chemical reduction of 9 with zinc powder. In a similar manner,
9 (0.10 g, 0.32 mmol) was reduced with zinc powder (0.10 g, 1.5
mmol) in DCM (5.0 ml) containing 15 drops of acetic acid.
Aqueous workup and preparative TLC (alumina/DCM) gave
N,NЈ-diphenylguanidine (53 mg, 78%) as colourless crystals;
mp 144–150 ЊC (lit.,12 150 ЊC).
Acknowledgements
The Tübingen authors wish to thank VW-Stiftung, Hannover,
for a grant (I/71009). We also have to thank Mr Paul Schuler for
running EPR experiments and simulations. M. H. would like
to thank Prof. Klaus Möbius for support, in particular for the
use of the spectrometers, and Dipl. Phys. Jacob Lopez for his
contribution to the ENDOR experiments.
References
1 For reviews on mesoionic compounds see: W. D. Ollis and C. A.
Ramsden, Adv. Heterocycl. Chem., 1976, 19, 1; C. G. Newton and
Paper 8/09386E
J. Chem. Soc., Perkin Trans. 2, 1999, 985–995
995