Radical cations of phenyl-substituted aziridines: What are the conditions for ring opening?

Article abstract of DOI:10.1002/chem.200400557

Radical cations were generated from different phenyl-substituted aziridines by pulse radiolysis in aqueous solution containing T1OH⁺, N₃ or SO₄^- as oxidants or in n-butyl chloride, by ⁶⁰Co γ radiolysis in Freon matrices at 77 K, and in some cases by flash photolysis in aqueous solution. Depending on the substitution pattern of the aziridines, two different types of radical cations are formed: if the N atom carries a phenyl ring, the aziridine appears to retain its structure after oxidation and the resulting radical cation shows an intense band at 440-480 nm, similar to that of the radical cation of dimethylaniline. Conversely, if the N atom carries an alkyl substituent while a phenyl ring is attached to a C-atom of the aziridine, oxidation results in spontaneous ring opening to yield azomethine ylide radical cations which have broad absorptions in the 500-800 nm range. In aqueous solution the two types of radical cations are quenched by O₂ with different rates, whereas in n-butyl chloride, the ring-closed aziridine radical cations are not quenchable by O₂. The results of quantum chemical calculations confirm the assignment of these species and allow to rationalize the different effects that phenyl rings have if they are attached in different positions of aziridines. In the pulse radiolysis experiments in aqueous solution, the primary oxidants can also be observed, whereas in n-butyl chloride a transient at 325 nm remains unidentified. In the laser flash experiments, both types of radical cations were also observed.

Full text of DOI:10.1002/chem.200400557

Radical Cations of Phenyl-Substituted Aziridines: What Are the Conditions
for Ring Opening?
Carsten Gaebert,[a, d] Jochen Mattay,*[a] Marion Toubartz,[b] Steen Steenken,[b]
Beat Mꢀller,[c, e] and Thomas Bally*[c]
Abstract: Radical cations were gener-
ated from different phenyl-substituted
aziridines by pulse radiolysis in aque-
ous solution containing TlOHC+, N3C or
SO4Cꢀ as oxidants or in n-butyl chloride,
by 60Co g radiolysis in Freon matrices
at 77 K, and in some cases by flash
photolysis in aqueous solution. De-
pending on the substitution pattern of
the aziridines, two different types of
radical cations are formed: if the N
atom carries a phenyl ring, the aziri-
dine appears to retain its structure
after oxidation and the resulting radical
cation shows an intense band at 440–
480 nm, similar to that of the radical
cation of dimethylaniline. Conversely,
if the N atom carries an alkyl substitu-
ent while a phenyl ring is attached to a
C-atom of the aziridine, oxidation re-
sults in spontaneous ring opening to
yield azomethine ylide radical cations
which have broad absorptions in the
500–800 nm range. In aqueous solution
the two types of radical cations are
quenched by O2 with different rates,
whereas in n-butyl chloride, the ring-
closed aziridine radical cations are not
quenchable by O2. The results of quan-
tum chemical calculations confirm the
assignment of these species and allow
to rationalize the different effects that
phenyl rings have if they are attached
in different positions of aziridines. In
the pulse radiolysis experiments in
aqueous solution, the primary oxidants
can also be observed, whereas in n-
butyl chloride a transient at 325 nm re-
mains unidentified. In the laser flash
experiments, both types of radical cati-
ons were also observed.
Keywords: azomethine ylides · den-
sity functional calculations · matrix
isolation · radical ions · reaction
mechanisms
Introduction
radiation or on thermal activation aziridines undergo ring
opening to the corresponding azomethine ylides which can
be trapped in [3+2] cycloadditions with various dipolaro-
philes, to form nitrogen containing five-membered heterocy-
cles.[3–5] Under photoinduced electron transfer (PET) condi-
tions aziridines are oxidized to the corresponding radical
cations, which can react in a similar manner.[6–10] In the
course of our investigations aimed at the applications of
aziridines in organic synthesis[10,11] we became interested in
the reactive intermediates of the [3+2] cycloadditions, that
is, azomethine ylides and radical cations of aziridines, to elu-
cidate the mechanisms of the reactions. A few years ago we
published the results of studies in which azomethine ylides
were generated by laser flash photolysis.[12,13] In the present
study we investigate the radical cations that are obtained on
pulse radiolysis or 60Co g radiolysis of different phenylaziri-
dines.
Aziridines and their reactions are of great interest due to
their synthetic and pharmacological importance.[1,2] Upon ir-
[a] Dr. C. Gaebert, Prof. J. Mattay
Fakultꢀt fꢁr Chemie, Universitꢀt Bielefeld
33501 Bielefeld (Germany)
E-mail: mattay@uni-bielefeld.de
[b] M. Toubartz, Prof. S. Steenken
Max-Planck-Institut fꢁr Bioanorganische Chemie
45470 Mꢁlheim/Ruhr (Germany)
[c] Dr. B. Mꢁller, Prof. T. Bally
Departement de Chimie, Universitꢂ de Fribourg
1700 Fribourg (Switzerland)
Fax : (+41)26-300-9737
E-mail: thomas.bally@unifr.ch
[d] Dr. C. Gaebert
Present address:
Reports on radical cations of aziridines are scarce. In 1976
Holmes and Terlouw[14] reported that the metastable peaks
and kinetic energy release of C2H5NC+ obtained from dime-
thylamine, piperidine, or pyrrolidine were indistinguishable
from those obtained from aziridine. This led them to pro-
Consortium fꢁr elektrochemische Industrie GmbH
Zielstattstrasse 20, 81379 Mꢁnchen (Germany)
[e] Dr. B. Mꢁller
Present address: Bundesamt fꢁr Umwelt, Wald und Landschaft
3003 Bern (Switzerland)
1294
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200400557
Chem. Eur. J. 2005, 11, 1294 – 1304

FULL PAPER
pose for the first time that aziridine undergoes ring opening
to azomethine ylide radical cation on ionization. Conversely,
three years later Maquestiau et al. reported that C2H5NC+
generated from aziridine and pyrrolidine have distinct colli-
sion induced dissociation (CID) spectra, which led them to
claim that the aziridine radical cation retains its ring-closed
structure.[15] In 1984 Lien and Hopkinson carried out UHF/
4-31G calculations which predicted that the ring opening of
the arziridine radical cation is exothermic by over 26.5 kcal-
molꢀ1.[16]
Around the same time Schaap et al. investigated the pho-
tooxygenation of various substituted aziridines.[17–19] They
explained the observed reaction products in terms of ring-
opened intermediates. Finally, in 1986 Qin and Williams
measured the ESR spectra of parent aziridine subjected to g
radiolysis in Freon glasses at 77 K. The spectra showed the
unmistakable signature of an allylic radical which led these
authors to propose that the aziridine radical cation under-
goes ring opening, even under cryogenic conditions.[20]
The substrates examined experimentally in the present
study are 1-phenylaziridine 1, 1-butyl-phenylaziridine 2a,
the trans-2,3-diphenylaziridines 3a and 3b, the 1,2-dipheny-
laziridines 4a and 4b, and 1,2,3-triphenylaziridine 5. Compu-
tational model studies were done on compounds 2b and 3c,
instead of the n-butyl derivatives, 2a and 3b (see Scheme 1).
Figure 1. Transient absorption spectrum obtained after pulse radiolysis of
a 0.1 mm aqueous solution of 1, saturated with N2O and containing 1 mm
~
*
15.3 ms after the electron pulse.
Tl2SO4 at pH 6.6; & 0.9 ms,
5.3 ms,
5
ꢀ1
&
*
,
The insets shows the kinetics at 360 ( , k = 1.4ꢄ10 s ) and 440 nm (
k = 1.2ꢄ105 sꢀ1).
sorptions of TlOHC+ prevented a quantitative assessment of
the kinetics for the formation and decay of this species
which might be an adduct radical of 1 and OHC (see below).
When oxidation was effected by SO4Cꢀ (Figure 2) the
440 nm transient arose with
a rate constant of 9.2ꢄ
109 mꢀ1 sꢀ1. Its quenching with oxygen occurred with a rate
constant (k=6.0ꢄ107 mꢀ1 sꢀ1) in good agreement with the
+
one found after oxidation of 1 with TlOHC . Pulse radiolysis
of 4 mm of aziridine 1 in n-butyl chloride, and 60Co g radiol-
ysis in a Freon matrix at 77 K (lower part of Figure 2) gener-
ated the same species with lmax =440 nm but in n-butyl chlo-
ride this transient is not quenchable with oxygen.
In the spectrum recorded after oxidation with N3C the
band at 440 nm occurred only as a minor constituent that
arose with a rate constant of 8.0ꢄ108 mꢀ1 sꢀ1. The major tran-
sient, which was formed with a rate constant of 3.3ꢄ
108 mꢀ1 sꢀ1 showed the same 335 nm band that was observed
also after pulse radiolysis of aziridine 1 with Tl2SO4. Under
an atmosphere of N2O:O2 4:1, this 335 nm absorption was
completely quenched, so this transient is very sensitive to
oxygen. We propose that this species is an adduct radical
formed by attack of OHC onto a phenyl group of 1. In the
spectrum obtained with SO4Cꢀ as oxidant there is no absorp-
tion at 335 nm because the tert-butanol that was added in
this experiment immediately traps all OH radicals. This, and
the fact that adducts of aromatic systems with OH radicals
generally show absorptions in this region,[22–24] supports the
assignment of the 335 nm transient to an adduct radical. In
order to confirm this assignment we subjected a 0.1 mm so-
lution of 1 in N2O-saturated water at pH 8 to pulse radioly-
sis. Thereby we detected also a major transient at 335 nm
(Figure 3) while the 440 nm absorption appeared as a minor
secondary transient by OHꢀ elimination from the adduct
radical. The same OH adduct radical may also be formed in
the experiments with Tl2SO4 described above.
Scheme 1.
Results and Discussion
Pulse radiolysis and 60Co g irradiation
N-Phenylaziridine (1): The UV spectrum recorded 0.9 ms
after irradiating an N2O saturated solution of 0.1 mm 1-phe-
nylaziridine (1) containing 1 mm Tl2SO4 at pH 6.6 with an
electron pulse (Figure 1) shows the 360 nm band of the oxi-
dant, TlOHC+.[21] This absorption decreases with a rate con-
stant of k=1.4ꢄ109 mꢀ1 sꢀ1, while a new band arises with the
same rate constant at 440 nm (its maximum intensity is
reached after 15.3 ms). The presence of an isosbestic point at
420 nm suggests that the observed transformation occurs in
a single step. The 440 nm transient is quenchable with
oxygen (k=5.5ꢄ107 mꢀ1 sꢀ1).
Another transient was formed which has an absorption
peak near 335 nm the maximum of which is reached 5.3 ms
after the electron pulse (i.e., about three times more rapidly
than that of the 440 nm transient). Unfortunately, the ab-
Chem. Eur. J. 2005, 11, 1294 – 1304
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1295

J. Mattay, T. Bally et al.
um peroxodisulfate a new strong absorption occurred at
380 nm and a flat, weak one at 550 nm (Figure 4). These
bands rise concurrently, with a similar rate as that observed
for the decay of the sulfate radical anion at 450 nm[26,27] (k=
9.0ꢄ108 mꢀ1 sꢀ1), and they are also quenchable by oxygen
(k=1.3ꢄ108 mꢀ1 sꢀ1). A similar pair of bands (lmax =575 and
390 nm) was also observed on g radiolysis of aziridine 2a in
a Freon matrix at 77 K (Figure 4, solid line). On pulse irradi-
ation of 9 mm 2a in n-butyl chloride, a different transient
with a band at 325 nm was formed, which can also be
quenched by oxygen but we have been unable to assess the
identity of this species.
Figure 2. Top: Transient absorption spectrum obtained after pulse radiol-
ysis of a 0.2 mm aqueous solution of 1 containing 1 mm K2S2O8 and 0.1m
~
*
tert-butanol at pH 6.6;
0.7 ms, & 7.1 ms,
765 ms after the electron
=
pulse. Inset shows the kinetics at 440 nm (k
1.8ꢄ105 sꢀ1); bottom:
Transient absorption spectrum recorded after pulse radiolysis of 4 mm of
1 in n-butyl chloride ( ) and after 60Co g radiolysis in a Freon matrix at
~
Figure 4. Transient absorption spectrum observed 15 ms after pulse radiol-
ysis of a 0.2 mm aqueous solution of 2a containing 2 mmK2S2O8 and 0.1m
tert-butanol at pH 6 (&), and on 60Co g radiolysis in a Freon matrix at
77 K (c).
Evidently, these spectra are quite different from those
found for the radical cations of N-phenylaziridines, which
should come as no surprise because 2a lacks the N,N-dialky-
lanilino chromophore which was found to be responsible for
the 440–490 nm bands that are characteristic for the corre-
sponding radical cations. However, calculations predict that
C-phenylazirdines undergo facile ring opening to azome-
thine ylide radical cations on ionization (see below), so the
chromophore that is responsible for the spectra in Figure 4
is of an altogether different nature in this case. Very similar
spectra were obtained also for the corresponding N-methyl
as well as for the N-H derivative on g radiolysis in Freon
glasses.[28]
Figure 3. Transient absorption spectrum of 0.1 mm 1 observed on pulse
~
radiolysis of an N2O-saturated aqueous solution at pH 8;
0.8 ms, &
*
7.3 ms, 785 ms after the electron pulse.
77 K (c).
By comparison with the absorption spectrum of the radi-
cal cation of N,N-dimethyl aniline (DMAC+) which shows an
intense band at 465 nm,[22,25] we assigned the 440 nm band of
ionized 1 to the (ring closed) radical cation of N-phenylaziri-
dine (a discussion of the electronic structure of 1C will be
provided in Section on Calculations).
trans-2,3-Diphenylaziridines 3a and 3b: 60Co g radiolysis of
3a and 3b in a Freon matrix at 77 K yields nearly indistin-
guishable spectra (shown in Figure 5), so we can assume
that similar cations are formed. As will be shown below, 2,3-
diphenylaziridines undergo spontaneous ring opening upon
ionization to yield azomethine ylide radical cations of the
same type as 2a, cations which are characterized by their
broad bands in the 500–800 nm range.
+
1-Butyl-2-phenylaziridine (2a): In the spectrum of 0.2 mm
aziridine 2a in an aqueous solution containing 2 mm potassi-
1296
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1294 – 1304

Aziridines
FULL PAPER
Freon. It shows four absorptions at 280, 320, 380 and
630 nm that appear with the same rate constant (k=6.6ꢄ
109 mꢀ1 sꢀ1) and are quenchable with oxygen (k=5.8ꢄ
108 mꢀ1 sꢀ1, the highest such rate constant observed in this
study!) in the peroxodisulfate experiments. We also note
that the major absorption at 380 nm is very intense (e=
21100mꢀ1 cmꢀ1) compared with those of the oxidation prod-
ucts of the other aziridines (e=2000–4000mꢀ1 cmꢀ1) that
were identified in this study.
In order to identify the transient observed upon pulse ra-
diolysis of 3a we measured the conductivity of an aqueous
peroxodisulfate solution of this compound and compared it
with a similar solution containing aziridine 3b (note that
these solutions have a pH of 4.9 so the aziridines are present
mostly in their protonated forms, 3H+). After the pulse the
conductivity was found to be much higher in the case of 3a
than if the solution contained 3b. This finding leads us to
postulate that 3aC+ undergoes spontaneous deprotonation to
yield radical 6 (Scheme 2), a decay pathway that is not avail-
+
able to 3bC .
Figure 5. Transient absorption spectra recorded after pulse radiolysis of
8 mm of 3a (&, 2.5 ms) and 9 mm of 3b (&, 2 ms after the electron pulse)
in n-butyl chloride and after 60Co g radiolysis of the same two 2,3-diphe-
nylaziridines in a Freon matrix at 77 K (c).
Scheme 2.
Conversely, the spectra obtained on pulse radiolysis of
8 mm 3a and 9 mm 3b in n-butyl chloride are quite different
in appearance (Figure 5). That obtained from 3b shows
bands with peaks at 430 and about 700 nm which arise and
decay with the same rate. The same bands are also observed
if SO4Cꢀ is used as an oxidant whereby they also build up
with the same rate (k=9.0ꢄ108 mꢀ1 sꢀ1) and are both
quenched by oxygen (k=7.9ꢄ107 mꢀ1 sꢀ1). Since the two
bands coincide with those in the spectra obtained after g ra-
diolysis in Freon, we conclude that the same radical cation
is formed in the two sets of experiments. In contrast to the
radical cations of the N-arylaziridines 1, 4a, and 4b (see
below), this one is, however, also quenchable with oxygen in
n-butyl chloride. Similar to the case of 2a we cannot explain
the oxygen quenchable transient absorbing at 330 nm after
pulse radiolysis in butyl chloride.
In addition, there are pronounced shoulders at 480 (3a)
and 500 nm (3b), respectively, the kinetics of which differ
from that for the formation and decay of the 430 and
700 nm bands. Previous flash photolysis experiments in ace-
tonitrile and methanol had given rise to bands at 480 (3a)
or 500 nm (3b)[12] which were assigned to the neutral azome-
thine ylides that are formed by ring opening of aziridines.[29]
Perhaps the same species are formed as minor products on
pulse radiolysis in BuCl or g radiolysis in Freon.[30]
We propose a ring-opened structure for the 6 because
a) calculations predict that 3aC+ undergoes spontaneous ring
opening (see below) and b) because the allylic radical 6
enjoys considerably more stabilization by the two terminal
phenyl rings than an aziridinyl radical. Of course in n-butyl
+
chloride, spontaneous deprotonation of 3aC is not expected
to occur because the proton cannot be solvated. We can see
two possibilities for the formation of 6 under these condi-
tions: either 3aC+ is deprotonated by neutral 3a, or 3a un-
dergoes hydrogen atom loss on radiolysis in n-butyl chloride.
A similar process was observed by Qin and Williams to
occur on g irradiation of parent aziridine in a CF2ClCCl2F
matrix at 114 K which led, next to the ring-opened azome-
thine ylide radical cation, to the aziridinyl radical.[20]
+
+
In aqueous solution 2aC and 3bC are trapped faster by
oxygen than the radical cations of the 1-arylaziridines and,
in contrast to the latter, 3bC+ is quenchable with oxygen
even in n-butyl chloride. The fact that 3b was found to un-
dergo [3+2] cycloadditions with various dipolarophiles
under PET conditions[6,7,10–13] also speaks for a ring-opened
structure of the radical cation.
1-Aryl-2-phenylaziridines 4a and 4b: Upon pulse radiolysis
of an aqueous solution of 0.2 mm 1,2-diphenylaziridine (4a)
containing peroxodisulfate a transient absorbing at 440 nm
builds up with a rate constant of 6.0ꢄ108 mꢀ1 sꢀ1. This transi-
ent is quenchable with oxygen (k=5.5ꢄ105 mꢀ1 sꢀ1). Pulse ra-
diolysis of a 9 mm solution of 4a in n-butyl chloride also af-
In contrast, the spectrum obtained on pulse radiolysis of
3a, either in n-butyl chloride or in aqueous peroxodisulfate
solution, differs strongly from that found after g radiolysis in
Chem. Eur. J. 2005, 11, 1294 – 1304
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1297

J. Mattay, T. Bally et al.
forded a broad absorption with a maximum at 440 nm which
is, however, not quenchable by oxygen. Additional evidence
for the identification of this species is obtained by the re-
sults of g radiolysis in a Freon matrix at 77 K which leads to
a spectrum with an intense band peaking at 445 nm (solid
line in Figure 6). In addition, a broad band with lmax
ꢁ
650 nm, similar to that observed after ionization of 1-butyl-
2-phenyldiaziridine 2a, was observed. However, this latter
band can be bleached separately by irradiation through a
590 nm cutoff filter, so it does not belong to the same spe-
cies as the intense UV band. Apparently this second, un-
identified species was not formed on pulse radiolysis in so-
lution.
Figure 7. Transient absorption spectra observed after pulse radiolysis of a
0.1 mm aqueous solution of 4b containing 10 mm NaN3 at pH 8, 7.1 ms
), a 9 mm solution of 4b in n-butyl chloride, 2.3 ms (&), and on 60Co g
~
(
radiolysis of 4b in a Freon matrix at 77 K (c).
Figure 6. Transient absorption spectra recorded after pulse radiolysis of a
0.2 mm aqueous solution of 4a containing 1 mmK2S2O8 and 0.1m tert-bu-
*
tanol at pH 6, 62 ms after the electron pulse ( ), on pulse radiolysis of a
9 mm solution of 4a in n-butyl chloride, 2.5 ms after the electron pulse
(&), and on 60Co g radiolysis in a Freon matrix at 77 K (c).
Aqueous peroxodisulfate spontaneously oxidized 1-(p-me-
thoxyphenyl)-2-phenylaziridine (4b), even at room tempera-
ture, to yield a purple solution with a concomitant decrease
of the pH. Therefore 4b cannot be subjected to pulse radiol-
ysis to generate its radical cation under controlled condi-
tions. However, if N3C is used as oxidant we observed a tran-
sient absorption at 470 nm which arose with a rate constant
of 2.9ꢄ109 mꢀ1 sꢀ1 and was quenched by oxygen (k=1.5ꢄ
107 mꢀ1 sꢀ1). Once again, pulse radiolysis of a 9 mm solution
of aziridine 4b in n-butyl chloride resulted in a similar ab-
sorption which was, however, not quenchable with oxygen.
A 480 nm band with a tail that extended to over 700 nm was
observed after g radiolysis of 4b in a Freon matrix at 77 K
(Figure 7).
The similarity in the kinetic and spectroscopic data of the
oxidized 1-phenyl and the 1-aryl-2-phenylaziridines indicat-
ed that the resulting radical cations retain a ring-closed
structure in all cases. This conclusion was confirmed by
preparative studies which yielded no products from reac-
tions with dipolarophiles, products that would have indicat-
ed that ring opening to azomethine ylide radical cations had
occurred, under PET conditions.[11]
Figure 8. Top: difference spectrum observed after g radiolysis of 1,2,3-tri-
phenylaziridine (5) in Freon at 77 K; center: difference spectrum for the
subsequent photolysis at >715 nm; bottom: difference spectrum for the
bleaching of the azomethine ylide radical cation at 475 nm.
1,2,3-Triphenylaziridine (5): This compound was only inves-
tigated by g radiolysis, but the results proved to be quite in-
teresting: After radiolysis, a spectrum with a band peaking
at 451 nm, that is, in the region where N-phenylaziridine
radical cations absorb, was observed in addition to a broad,
weak absorption between 500 and 700 nm (Figure 8a). After
only three minutes of photolysis through a 715 nm cutoff
filter, a much more intense spectrum arose (Figure 8b)
which could in turn be bleached completely by irradiation
at >475 nm (Figure 8c). This latter spectrum was similar to
1298
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1294 – 1304

Aziridines
FULL PAPER
those observed after radiolysis of the 2,3-diphenylaziridines
3a and 3b which underwent spontaneous ring opening upon
oxidation.
Thus, 5 appeared to retain its ring-closed structure after
ionization at 77 K but can be converted to its ring-opened
azomethine ylide radical cation by subsequent photolysis, a
unique behaviour in the series of phenyl substituted aziri-
dines investigated in this study.
concerted conrotatory fashion. According to B3LYP, the
transition state for this reaction lies 33 kcalmolꢀ1 above 1C+,
but the process leading to the 2-phenylazomethine ylide rad-
+
ical cation 7C (see Scheme 3) is
nearly thermoneutral. This con-
trasts with the parent aziridine
radical cation where the same
level of theory predicts a barri-
er of only 10.1 and an exother-
Calculations
micity of 33.6 kcalmolꢀ1.[28] To
Scheme 3.
understand the influence that
In an effort to substantiate the conclusions we had reached
from the experiments described above, and to understand
the reasons for the different fate of the ionized aziridines
and the electronic structure of the resulting transients, we
carried out a comprehensive set of quantum chemical calcu-
lations the results of which are presented and discussed in
this Section.
the N-phenyl ring has on the
thermochemistry and the kinetics of this process we calcu-
lated the barriers for rotation of this phenyl ring in 1C , at
+
the transition state, and in 7C+. Whereas in 1C+ it is 26.3 kcal-
molꢀ1—thus testifying to the substantial resonance stabiliza-
tion of this cation by the phenyl ring—it falls to
9.6 kcalmolꢀ1 at the transition state (i.e., 16.7 kcalmolꢀ1 of
resonance energy is lost at this point). In 7C+ the phenyl ring
is already twisted by 538 and bringing it into a perpendicular
position equires only 0.6 kcalmolꢀ1. This comes as no sur-
prise, because the phenyl ring is attached to a nodal position
of the allylic HOMO of 7C+ where it cannot exert a stabiliz-
ing influence. Thus, our conclusion that 1 retains its ring-
closed structure on ionization is fully confirmed by the cal-
culations.
N-Phenylaziridine (1): On ionization of 1 an electron is re-
moved from the HOMO depicted on the right-hand side of
Figure 9. Contrary to that of the iso-p-electronic benzyl
anion, this MO is antibonding between the phenyl ring and
the N atom, hence we expect this bond to be strengthened
in the radical cation of 1. B3LYP calculations confirmed
that upon ionization this bond assumes partial double bond
character which is also given by the fact that the aziridine
+
and the phenyl rings are nearly coplanar in 1C (Figure 9).
N-Methyl-2-phenylaziridine (2b): If a phenyl ring is attach-
ed to a C rather than the N atom of the aziridine, the situa-
tion changes completely compared with that in 1. For rea-
sons of computational economy calculations were carried
out on the 1-methyl rather than the 1-butyl derivative (the
cations that result from the two compounds have virtually
indistinguishable spectra[28]). Now the HOMO arises
through interaction of the benzene moiety with the symmet-
ric Walsh-MO in the three-membered ring (see Figure 10).
In contrast, the length of the bonds in the aziridine moiety
are hardly affected by ionization, which indicates that ring-
opening will not be much easier in the radical cation than in
the neutral aziridine.
Figure 9. Structural changes on ionization of 1 and shape of the singly oc-
cupied MO in the resulting radical cation.
The photoelectron spectrum of 1[31,32] shows two more
bands within 3 eV of the first, the second of which lies only
1 eV above the first. Hence, the observed absorption bands
of 1C+ at 440 nm (2.8 eV) must correspond to D0 ! D2 exci-
tation. Indeed, TD-B3LYP excited state calculations predict
another, very weak transition at 690 nm which corresponds
to HOMOꢀ1 ! HOMO excitation (the observed D0 ! D2
Figure 10. Shape of the singly occupied MO of the radical cation ob-
tained from 1-methyl-2-phenylaziridine 3c and structural changes on oxi-
dation.
The geometry changes on ionization are in accord with
expectations from the nodal properties of the MO from
ꢀ
transition, which involves HOMOꢀ2
!
HOMO and
which the electron is removed (Figure 10): The aziridine C
C bond lengthens by 0.12 ꢅ, which will facilitate the cleav-
HOMO ! LUMO excitation is predicted at 400 nm).
ꢀ
ꢀ
If enforced (e.g.. by lengthening the C C bond in the azir-
age of this bond, whereas the aziridine phenyl bond short-
ens by 0.04 ꢅ.
idine), ring opening of the radical cation proceeds in a non-
Chem. Eur. J. 2005, 11, 1294 – 1304
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1299

J. Mattay, T. Bally et al.
Indeed, the transition state for the ring opening, which
occurs again in a conrotatory fashion, was found to lie only
4 kcalmolꢀ1 above 2bC+, and the process is exothermic by
25 kcalmolꢀ1, that is, almost as much as in the parent com-
pound. An intrinsic reaction coordinate calculation from the
transition state leads to the exo-isomer of the C-phenyl-N-
+
methyl azomethine ylide radical cation 8C although a sepa-
rate calculation shows that the endo-conformer lies even
0.6 kcalmolꢀ1 lower in energy, in spite of the twisting of the
phenyl ring that results from steric interactions in this case.
To gain insight into the role of the phenyl ring in 2b we
computed again the barriers for twisting this ring to a posi-
tion that is perpendicular to that which it assumes at the op-
+
Figure 11. Molecular orbitals of the 1-phenyl-azomethine ylide radical
timized geometries of 2bC , the transition state for ring
cation 8C+ that are involved in the electronic transitions listed in Table 1.
opening, and 8C+: this process requires only 7.8 kcalmolꢀ1 in
2bC+ and 8.0 kcalmolꢀ1 in 8C+,
but 18.7 kcalmolꢀ1 at the transi-
Table 1. Calculated electronic transitions of 8C+ by using the CASPT2 method.
tion state. Thus, in contrast to
Isomer
States
Experiment
[eV]
CASPT2
[eV]
CASSCF
1C+, the phenyl ring has
a
[nm]
[nm]
f[a]
Configurations[b]
11 kcalmolꢀ1 greater stabilizing
effect at the transition state
than in the reactant in the case
of 2bC+, which readily explains
why the barrier for ring open-
ing is so much lower in this
case. Apparently this barrier is
too low to even prevent ring
opening from occurring at 77 K.
Next we addressed the ques-
tion whether the spectra shown
in Figure 4 are compatible with
those expected for 8C+. In this
case, the TD-B3LYP method
provided quite unsatisfactory
results, so we turned to the
CASSCF/CASPT2 method to
model the excited states. Thereby we distinguished between
the two isomers which are expected to have slightly differ-
ent spectra. The results of these calculations are summarized
in Table 1 which shows that the most important transitions
in 8C+ involve the allylic azomethine ylide moiety (MOs 34,
36, and 37 in Figure 11). On the whole, the spectrum that is
predicted for the exo-isomer is in better accord with the ex-
perimental one than that computed for the endo-isomer. To-
gether with the IRC calculation mentioned above, this
seems to indicate that it is indeed the exo-isomer of 8C+
which is formed in the ring-opening of 2bC+.
exo
12 A’’
22 A’’
–
583
–
2.12
–
584
–
2.12
–
75% (36)1[c]
44% 36 ! 37
20% 34 ! 36
55% 35 ! 36
10% 34 ! 36
28% 34 ! 36
19% 36 ! 38
10% 36 ! 37
77% (36)1[c]
31% 34 ! 36
27% 36 ! 37
10% 36 ! 39
62% 35 ! 36
23% 34 ! 36
17% 36 ! 37
16% 36 ! 40
0.028
32 A’’
42 A’’
(440 sh)
398
(2.82)
3.12
484
396
2.56
3.13
0.002
0.252
endo
1
2
–
583
–
2.12
–
566
–
2.19
–
0.010
3
4
(440 sh)
398
2.82
3.12
485
407
2.56
3.05
0.002
0.129
[a] Oscillator strength for electronic transition. [b] Active space: nine electrons in five occupied + four virtual
MOs. The most important of those are shown in Figure 11. [c] The superscripted 1 indicates that orbital 36
(the HOMO) is occupied by a single electron in the ground state.
less than 1 kcalmolꢀ1 remains to effect full cleavage of this
bond to yield the endo–exo isomer of 9C+ where both phenyl
rings are twised by about 208 relative to the C-N-C plane of
the azomethine ylide moiety. Due to the presence of phenyl
rings on both aziridine carbon atoms, the SOMO is centered
even more strongly in the symmetric Walsh-MO of the
three-membered ring than in the 2-phenylaziridine
(Figure 12).
N-Methyl-2,3-diphenylaziridine (3c): In view of the above
conclusions with regard to 2-phenylaziridine, it certainly
comes as no surprise that 2,3-diphenylaziridines undergo
almost barrierless ring opening on oxidation. Actually it
turned out to be quite tricky to prevent the radical cation of
the model system 3c from relaxing spontaneously to the
azomethine ylide radical cation 9C+, and to find a transition
Figure 12. Shape of the singly occupied MO of the radical cation ob-
tained from 1-methyl-, 2,3-diphenylaziridine and structural changes on
oxidation.
ꢀ
state for this process. On ionization of 3c, the C C bond in
the aziridine moiety lengthens to 1.86 ꢅ and a barrier of
1300
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1294 – 1304

Aziridines
FULL PAPER
Unfortunately, 9C+ proved to be too large to be amenable
to CASSF/CASPT2 calculations of its excited states, so we
had to resort to the more economical TD-DFT method to
companied by an planarization
of the N atom (the angle be-
tween the plane of the aziridine
+
ꢀ
model the electronic spectrum of 9C . Although the resulting
ring and the C phenyl bond de-
predictions are not in such good quantitative accord with
the experimental spectra of ionized 3a and 3b shown in
Figure 5, they show that the main transitions are of similar
nature as those of 8C+, that is, they also involve mainly MOs
that are centered on the allylic azomethine ylide moiety. In
agreement with the experimental spectra of ionized 3a and
3b, TD-B3LYP predicts those spectra to be dominated by
two excitations around 2 and 3 eV, respectively, with the
latter being about four times more intense than the former.
Perhaps the shoulder at 510 nm is in part due to the two
weak transitions that are predicted around 2.6 eV and that
correspond to charge transfer from the benzene to the allyl
moieties of 9C+. In any event, the calculations of the poten-
tial energy surface and the results in Table 2 leave no doubt
that the spectra in Figure 5 are those of 9C+, although we
can make no prediction with regard to the conformations
creases from 49 to 258, while
the inversion barrier almost dis-
appears). In particular, the
ꢀ
distal C C bond in the aziridine
ring is hardly affected by ioniza-
tion.
Figure 13. Shape of the singly
occupied MO of the radical
cation obtained from 1,2-di-
phenylaziridine and structural
changes on oxidation.
Nevertheless the C-phenyl
ring does not remain without an
effect on the reactivity of 4a:
due to the stabilization of the
product, the ring-opening reac-
tion (which had been nearly thermoneutral in 1C+) becomes
exothermic by about 10 kcalmolꢀ1 while the barrier decreas-
es to 19.4 kcalmolꢀ1 in 4aC+, that is, it lies almost midway
between those in 1C+ (33 kcalmolꢀ1) and in 2bC+ and close
+
to that in 3aC (16 kcalmolꢀ1[28]). Thus, at the transition state
ꢀ
around the allylic N C bonds in this case.
for ring opening, the effects of the two phenyl rings appear
to cancel almost completely.
Due to the great similarity of the spectra observed after
ionization of 4a and of 1, we refrained from carrying out
electronic structure calculations on 4aC+. Suffice it to say
that the “reactive” state where the unpaired electron resides
in the symmetric Walsh MO of the aziridine lies almost
1 eV above the state where the SOMO corresponds to that
shown in Figure 9, and the transition to that state is very
weak, so it does not interfere with the electronic structure
of 4aC+.
+
Table 2. Calculated electronic transitions of endo–exo-9C by using the
TD-B3LYP method.
States
Experiment[a]
TD-B3LYP
Configurations[b]
[nm]
[eV]
[nm]
[eV]
f[a]
1
2
–
720
–
1.72
–
616
–
2.01
–
(56)1[c]
0.91 ꢄ 56 ! 57
ꢀ0.38 ꢄ53 ! 36
0.96 ꢄ 55 ! 56
+0.23 ꢄ55 ! 57
0.96 ꢄ 54 ! 56
ꢀ0.23 ꢄ 54 ! 57
0.87 ꢄ 53 ! 56
+0.25 ꢄ 56 ! 57
0.134
1,2,3-Triphenylaziridine (5): The HOMO of 5 from which
ionization occurs is once again centered on the N-phenyl
moiety, as in 1 and 4a. Thus, it is not surprising that the
structure of the aziridine ring changes very little on ioniza-
tion. However, the phenyl rings that are attached to the C
atoms of the aziridine begin to show their influence on the
way to the product in that the activation energy for ring
opening is only 5.43 kcalmolꢀ1 (up 4.5 kcalmolꢀ1 from 2,3-
diphenylaziridine radical cation 3cC+, due to the influence of
the N-phenyl ring) and the process is exothermic by
17.9 kcalmolꢀ1,[33] halfway between the 1,2-diphenyl- and
3
4
5
(510 sh)
(2.43)
474
470
396
2.62
2.64
3.13
0.003
0.002
0.509
445
2.79
[a] From the spectrum of ionized 3b shown in Figure 5. [b] MO 56 is the
singly occupied HOMO of 9C+ which corresponds to the allylic NBMO of
the azomethine ylide; MOs 53 and 57 are also allylic MOs, whereas the
nearly degenerate MOs 53 and 54 are centered on one or the other of
the two benzene rings. [c] See Table 1.
+
+
2,3-diphenylaziridine radical cations, 4aC and 3cC .
1,2-Diphenylaziridine (4a): We have seen above that a
phenyl ring attached to the N atom of an aziridine effective-
ly prevents ring opening on ionization, whereas a phenyl
ring attached to a C atom promotes it. Therefore we were
anxious to see what theory would tell us about the fate of
1,2-diphenylaziridine (4a).
As it turned out, the influence of the N-phenyl ring domi-
nates over that of the C-phenyl ring, that is, ionization
occurs from an MO that is similar to the HOMO of 1
(Figure 13, see Figure 9). Consequently, the geometry
changes on ionization of 4a are also very similar to those
Conclusions
In summary we successfully generated and identified the
radical cations that are formed on oxidation of different
phenyl-substituted aziridines by pulse radiolysis in aqueous
solution containing different oxidants or in n-butyl chloride
at room temperature, and by g radiolysis in a Freon matrix
at 77 K. As it turns out, the position of the phenyl substitu-
ent(s) determines the fate of the incipient aziridine radical
cations: if the N atom carries a phenyl ring, this effectively
protects the aziridine radical cation from decaying to an
ꢀ
suffered by 1, that is, a shortening of the N phenyl bond ac-
Chem. Eur. J. 2005, 11, 1294 – 1304
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1301

J. Mattay, T. Bally et al.
azomethine ylide radical cation, due to the benzylic reso-
nance stabilization that prevails in N-phenylaziridine radical
cations. The benzylic Ph-NC+ chromophore is also responsi-
ble for the intense D0!D2 transition in the 440–480 nm
range that is characteristic for ring-closed aziridine radical
cations.
Upon pulse radiolysis in aqueous solution the radical
cation of 2,3-diphenylaziridine was found to undergo sponta-
neous deprotonation to form an allylic radical, while in n-
butyl chloride hydrogen abstraction from the aziridine gen-
erates the same species. We also provided an additional
access to aziridine radical cations upon photolysis of aque-
ous solutions of aziridines with an excimer laser for 2a, 3b,
4a and 4b.[29]
By contrast, phenyl rings attached to the C atoms of the
aziridine lower the barrier for ring opening, to the extent
that the process becomes nearly activationless in 2,3-diphe-
nylaziridine radical cation. This is mainly due to the prefer-
ential stabilization of the resulting allylic azomethine ylide
radical cations by phenyl rings attached to their terminal C-
atoms. This chromophore gives also rise to intense near-UV
transitions, but in addition it distinguishes itself by a weak
broad band at 500–800 nm. In addition, the two types of rad-
ical cations can be distinguished by their reactivity towards
oxygen: In aqueous solution, the ring-opened azomethine
ylide radical cations are quenched with a higher rate than
the ring-closed aziridine radical cations, and in n-butyl chlo-
ride the former are not quenchable by O2 (in contrast to the
latter).
Quantum chemical calculations support the above conclu-
sions and helped to assign the observed spectra. They also
allow to understand the influence of phenyl rings attached
to different positions of aziridine radical cations on their re-
activity. In Figure 14 we summarize the results that pertain
to the ring-opening reactions of the various types of aziri-
dine radical cations that were investigated in the present
study.
Experimental Section
Starting materials: 1-Phenylaziridine (1),[34] 1,2-diphenylaziridine (4a),[35]
1-(p-methoxyphenyl)-2-phenylaziridine (4b)[35] and trans-2,3-diphenylazir-
idine (3a)[36] were prepared by reported procedures.
1-Butyl-2-phenylaziridine (2a):[11, 37] Starting with commercially available
styrol oxide the corresponding b-amino alcohol was prepared according
to the procedure of Chapman and Triggle.[38] The formation of 1-butyl-2-
phenylaziridine occurred as described by Okada et al.[37]
1-Butyl-trans-2,3-diphenylaziridine (3b):[11] By using trans-stilbene as
starting material epoxidation with MCPBA formed the corresponding
trans-stilbene oxide in yields of 75%. Following the procedure of Deyrup
and Moyer[39] the epoxide was opened to the b-amino alcohol. The cycli-
zation as described above[37] formed 1-butyl-trans-2,3-diphenylaziridine in
yields of 58%.
All spectroscopical data of the aziridines have been published else-
where.[11]
Pulse radiolysis: For pulse radiolysis, 0.4–2 ms electron pulses from a
3 MeV van de Graaff accelerator were used at dose rates of 30–3000 rad
per pulse. N2O-, argon- or oxygen-saturated solutions containing 0.1 up
to 9 mm aziridine were flowed through a quartz cell (20 mm path length)
while being irradiated. As precursor of the oxidants we used thallium(i)
sulfate, potassium peroxodisulfate and sodium azide in an excess of 10
times of the concentration of the aziridine.
The thermochemistry of the ring opening of the different
aziridine radical cations is compared with that of the parent
azirdine radical cation and the N-methyl derivative, calculat-
ed also at the B3LYP/6-31G* level.[28]
Upon radiolysis in water OH radicals and solvated electrons were gener-
ated.[40–42] By using a N2O-saturated solution the amount of the OH radi-
cals was increased through the reaction: N2O + H2O ! N2 + OHꢀ
+
OHC.[43] The OH radical was important for the formation of the oxidants.
In the case of thallium(i) sulfate the oxidizing species was generated ac-
cording to the reaction: Tl+ + OHC ! TlOHC+.[21] By using sodium azide,
the N3C was formed via N3ꢀ + OHC ! N3C + OHꢀ.[44] In the measurements
with peroxodisulfate, the produced solvated electrons create the sulfate
2ꢀ
2ꢀ
radical anion: S2O8 + eꢀ ! SO4 + SO4Cꢀ.[26,27] In this case the OH
radicals could disturb, therefore tert-butanol was added in order to trap
the OH radicals.[23,45] Pulsing in n-butyl chloride, solvent radical cations
were formed on direct ionization.[25,46] All these reactive species can oxi-
dize the substrates in a one electron transfer reaction. The corresponding
substrate radical cations were formed.
g Radiolysis: g Radiolysis of 10ꢀ2–10ꢀ3 m solutions of organic samples in
Freons at 77 K constitute a very clean way of generating radical cations
under conditions where they persist for hours.[46] For the present experi-
ments we employed a 1:1 mixture of CFCl3 and CF2BrCF2Br[47] which
formed a transparent glass at 77 K.[48] Samples were exposed to a total
dose of about 5 kSv of 60Co radiation (1.173 and 1.332 MeV) in a Gam-
macell 220 charged with about 450 TBq. Electronic absorption spectra
were recorded on a Perkin–Elmer Lambda 900 instrument.
Quantum chemical calculations: The geometries of all species except the
singlet nitrenes were optimized by the B3LYP density functional
method[49, 50] as implemented in the Gaussian program package,[51,52] using
the 6-31G* basis set. All stationary points were characterized by second
derivative calculations.
For the ylide radical cation, 8C+ that resulted from oxidative ring opening
of 1-methyl-2-phenylaziridine 3c, excitation energies and oscillator
strengths were calculated by the CASSCF/CASPT2 procedure[53] by using
the ANO/S basis set,[54] at the B3LYP/6-31G* optimized geometries of
Figure 14. Thermochemistry of the ring-opening reactions of different
phenylaziridines from B3LYP/6-31G* calculations.
1302
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1294 – 1304

Aziridines
FULL PAPER
the radical cations. The active space that was employed in the CASSCF
calculations is indicated in Table 1 where the results are listed. Level
shifts[55] had to be applied to eliminate intruder states in the CASPT2
runs for all excited states under consideration, but we checked carefully
that no artefacts were introduced by this technique. Under this condition,
the weight of the zero-order CASSCF wavefunction in the PT2 expansion
was between 0.7 and 0.72 for all states. All CASSCF/CASPT2 calcula-
tions were performed with the Molcas program.[56]
[24] P. OꢆNeill, S. Steenken, D. Schulte-Frohlinde, J. Phys. Chem. 1975,
79, 2773.
[25] T. Shida, W. H. Hamill, J. Chem. Phys. 1966, 44, 2369.
[26] E. Heckel, A. Henglein, Ber. Bunsenges. Phys. Chem. 1966, 70, 149.
[27] L. Dogliotti, E. Hayon, J. Phys. Chem. 1967, 71, 2511.
[28] B. Mꢁller, PhD Thesis, University of Fribourg, Fribourg (Switzer-
land), 2000.
[29] Interestingly, laser flash photolyis of 3b in a 1:1 mixture of acetoni-
trile and water gave also rise to a band at 430 nm (in contrast to the
spectra observed previously in pure acetonitrile or methanol), next
to those of the azomethine ylide at 500 nm and a 280 nm band that
was assigned to an iminium ion obtained by protonation of the azo-
methine ylide. Thus, it appears that in the presence of water aziri-
dine radical cations and their ring-opening products can also be
formed by flash photolysis (see C. Gaebert, C. Siegner, J. Mattay,
M. Toubarz, S. Steenken, Photochem. Photobiol. Sci. 2004, 3, 990).
[30] The bands at 480 and 500 nm persisted on radiolysis of oxygen-satu-
rated BuCl solutions, where the radical cations were not formed.
The species responsible for these bands were therefore not very sen-
sitive to molecular oxygen, in accord with their assignment to neu-
tral azomethine ylides.
For the diphenylaziridines, it was impossible to carry out CASSCF/
CASPT2 calculations so we resorted to the more economical time-de-
pendent response theory,[57] where the poles and the residues of the fre-
quency-dependent polarizability are calculated, whereby the former cor-
respond to vertical excitation energies and the latter to oscillator
strengths. We employed the density-functional based implementation of
this method (TD-DFT)[58] using again the B3LYP functional and the 6-
31G* basis set as described above.
[31] R. J. Lichter, K. Crimaldi, D. A. Baker, J. Org. Chem. 1982, 47,
3524.
Acknowledgement
[32] K. N. Houk, S. Searles, M. D. Rozeboom, S. E. Seyedrezai, J. Am.
Chem. Soc. 1982, 104, 3448.
[33] An intrinsic reaction coordinate calculation from the transition state
for ring opening of 5 leads to the exo–exo isomer of the triphenyla-
zomethine ylide radical cation which was, however easily converted
into the more stable exo–endo isomer.
Support provided by the Deutsche Forschungsgemeinschaft and the
Fonds der Chemischen Industrie is gratefully acknowledged. T.B. and
B.M are grateful for support by the Swiss National Science Foundation
(project No. 2000–067881.02). C.G. thanks the Studienstiftung des Deut-
schen Volkes for a scholarship.
[34] H. W. Heine, B. L. Kapur, C. S. Mitch, J. Am. Chem. Soc. 1954, 76,
1173.
[35] V. Franzen, H. E. Driesen, Chem. Ber. 1963, 96, 1881.
[36] D. Tanner, C. Birgersson, A. Gogoll, Tetrahedron 1994, 50, 9797.
[37] I. Okada, K. Ichimura, R. Sudo, Bull. Chem. Soc. Jpn. 1970, 43,
1185.
[38] N. B. Chapman, D. J. Triggle, J. Chem. Soc. 1963, 1385.
[39] J. A. Deyrup, C. L. Moyer, J. Org. Chem. 1969, 34, 175.
[40] A. Henglein, W. Schnabel, J. Wendenburg, Einfꢀhrung in die Strah-
lenchemie, Verlag Chemie, Weinheim 1969.
[41] I. G. Draganic, Z. D. Draganic, The Radiation Chemistry of Water,
Academic Press, New York, 1971.
[1] J. Backes in Methoden der Organischen Chemie Houben-Weyl,
Vol. E 16c (Ed.: D. Klamann), Thieme, Stuttgart, 1992, pp. 370–677.
[2] J. A. Deyrup in: Small Ring Heterocycles, Vol. Part 1 (Ed.: A. Hass-
ner), Wiley, New York, 1983, pp. 1–214.
[3] R. Huisgen, Angew. Chem. 1963, 75, 604; Angew. Chem. Engl. Int.
Ed. 1963, 2, 656.
[4] J. W. Lown, Rec. Chem. Prog. 1971, 32, 51.
[5] R. Huisgen in 1,3-Dipolar Cycloaddition Chemistry, Vol. 1 (Ed.: A.
Padwa), Wiley, New York, 1984, pp. 1–176.
[6] V. Caer, A. Laurent, E. Laurent, R. Tardevil, Z. Cebulska, R. Bart-
nik, Nouv. J. Chim. 1987, 11, 351.
[42] R. W. T. Spinks, R. J. Woods, An Introduction to Radiation Chemis-
try, Wiley, New York, 1990.
[7] T. Brigaud, E. Laurent, R. Tardevil, Z. Cebulska, R. Bartnik, J.
Chem. Res. Synop. 1994, 8, 330.
[43] J. H. Baxendale, M. A. Rodgers, Chem. Soc. Rev. 1978, 7, 235.
[44] Z. B. Alfassi, R. H. Schuler, J. Phys. Chem. 1985, 89, 3359.
[45] E. Bothe, D. J. Deeble, D. G. E. Lemaire, R. Rashid, M. N. Schuch-
mann, H.-P. Schuchmann, D. Schulte-Frohlinde, S. Steenken, C.
von Sonntag, Radiat. Phys. Chem. 1990, 36, 149.
[46] T. Shida, Electronic Absorption Spectra of Radical Ions, Elsevier,
Amsterdam, 1988.
[47] A. Grimison, G. A. Simpson, J. Phys. Chem. 1968, 72, 1776.
[48] C. Sandorfy, Can. J. Spectrosc. 1965, 10, 85.
[8] C. Gaebert, Diploma Thesis, Westfꢀlische Wilhelms-Universitꢀt,
Mꢁnster (Germany), 1994.
[9] C. Gaebert, PhD Thesis, Westfꢀlische Wilhelms-Universitꢀt, Mꢁn-
ster (Germany), 1997.
[10] C. Gaebert, J. Mattay, J. Inf. Record. 1996, 23, 3.
[11] C. Gaebert, J. Mattay, Tetrahedron 1997, 53, 14297.
[12] C. Gaebert, C. Siegner, J. Mattay, M. Toubartz, S. Steenken, J.
Chem. Soc. Perkin Trans. 2 1998, 2735.
[13] C. Siegner, C. Gaebert, J. Mattay, S. Steenken, J. Inf. Record. 1998,
24, 253.
[49] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[50] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785.
[51] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, R. E.
Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Daniels,
K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi,
R. Cammi, B. Mennucci, C. Pommelli, C. Adamo, S. Clifford, J.
Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K.
Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslow-
ski, J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I.
Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A.
Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W.
Gill, B. G. Johnson, W. Chen, M. W. Wong, J. L. Andres, C. Gon-
zales, M. Head-Gordon, E. S. Repogle, J. A. Pople, Gaussian, Inc.,
Pittsburgh, PA, 1998.
[14] J. L. Holmes, J. K. Terlouw, Can. J. Chem. 1976, 54, 1007.
[15] A. Maquestiau, Y. van Haverbeke, R. Flammang, A. Menu, Bull.
Soc. Chim. Belg. 1979, 88, 53.
[16] M. H. Lien, A. C. Hopkinson, Can. J. Chem. 1984, 62, 922.
[17] A. P. Schaap, G. Prasad, S. D. Gagnon, Tetrahedron Lett. 1983, 24,
3047.
[18] A. P. Schaap, S. Siddiqui, G. Prasad, E. Palomino, L. Lopez, J. Pho-
tochem. 1984, 25, 167.
[19] A. P. Schaap, G. Prasad, S. Siddiqui, Tetrahedron Lett. 1984, 25,
3035.
[20] X.-Z. Qin, F. Williams, J. Phys. Chem. 1986, 90, 2292.
[21] P. OꢆNeill, D. Schulte-Frohlinde, J. Chem. Soc. Chem. Commun.
1975, 387.
[22] H. Christensen, Int. J. Radiat. Phys. Chem. 1972, 4, 311–333.
[23] H. Christensen, K. Sehested, E. J. Hart, J. Phys. Chem. 1973, 77,
983.
[52] B. G. Johnson, P. M. W. Gill, J. A. Pople, J. Chem. Phys. 1993, 98,
5612.
Chem. Eur. J. 2005, 11, 1294 – 1304
www.chemeurj.org
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1303

J. Mattay, T. Bally et al.
[53] K. Andersson, B. O. Roos in Modern Electronic Structure Theory,
Vol. Part 1, Vol. 2 (Ed.: D. R. Yarkony), World Scientific Publ., Sin-
gapore, 1995, pp. 55.
[54] K. Pierloot, B. Dumez, P.-O. Widmark, B. O. Roos, Theor. Chim.
Acta 1995, 90, 87.
[55] B. O. Roos, K. Andersson, M. P. Fꢁlscher, L. Serrano-Andrꢂs, K.
Pierloot, M. Merchan, V. Molina, J. Mol. Struct. 1996, 388, 257.
[56] K. Andersson, M. Barysz, A. Bernhardsson, M. R. A. Blomberg,
D. L. Cooper, T. Fleig, M. P. Fꢁlscher, C. d. Graaf, B. A. Hess, G.
Karlstrçm, R. Lindh, P.-ꢅ. Malmqvist, P. Neogrꢇdy, J. Olsen, B. O.
Roos, A. J. Sadlej, M. Schꢁtz, B. Schimmelpfennig, L. Seijo, L. Ser-
rano-Andrꢂs, P. E. M. Siegbahn, J. Stꢈlring, T. Thorsteinsson, V. Ver-
yazov, P.-O. Widmark, University of Lund (Sweden), 2000.
[57] M. E. Casida in Recent Advances in Density Functional Methods,
Part I (Ed.: D. P. Chong), World Scientific Publ., Singapore, 1995,
pp. 155.
[58] R. E. Stratmann, G. E. Scuseria, M. J. Frisch, J. Chem. Phys. 1998,
109, 8218.
Received: June 4, 2004
Published online: January 5, 2005
1304
ꢃ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1294 – 1304