N-centered radicals in self-terminating radical cyclizations: Experimental and computational studies

Article abstract of DOI:10.1021/jo702261u

(Chemical Equation Presented) Intermolecular addition of photochemically generated N-centered aminium and amidyl radicals 13a-d and 16a,b, respectively, to the cyclic alkyne 1 initiates a radical translocation/cyclization cascade, followed by an oxidative termination step that eventually leads to formation of the bicyclic ketones 7a and 8a. Computational studies were performed to gain insight into the mechanism of these reactions, which are an interesting modification of the recently discovered concept of self-terminating radical cyclizations.

Full text of DOI:10.1021/jo702261u

N-Centered Radicals in Self-Terminating Radical Cyclizations:
Experimental and Computational Studies
Uta Wille,*,†,‡ Gerold Heuger,†,‡ and Christian Jargstorff§
School of Chemistry/BIO21 Molecular Science and Biotechnology Institute, The UniVersity of Melbourne,
30 Flemington Road, ParkVille, Victoria 3010, Australia, ARC Centre of Excellence for Free Radical
Chemistry and Biotechnology, Otto-Diels Institut fu¨r Organische Chemie, Christian-Albrechts UniVersita¨t
Kiel, Olshausenstrasse 40, 24098 Kiel, Germany
uwille@unimelb.edu.au
ReceiVed October 21, 2007
Intermolecular addition of photochemically generated N-centered aminium and amidyl radicals 13a-d
and 16a,b, respectively, to the cyclic alkyne 1 initiates a radical translocation/cyclization cascade, followed
by an oxidative termination step that eventually leads to formation of the bicyclic ketones 7a and 8a.
Computational studies were performed to gain insight into the mechanism of these reactions, which are
an interesting modification of the recently discovered concept of self-terminating radical cyclizations.
SCHEME 1
Introduction
Oxidation reactions performed under mild and heavy metal
free conditions are a highly desirable goal in synthetic organic
chemistry. Self-terminating radical cyclizations are a new
concept in radical chemistry recently discovered by us: CtC
triple bonds in cyclic and open-chain alkynes are transformed
into ketones under mild conditions in a radical addition/
cyclization sequence using O-centered radicals of type XO• as
oxidant.1 The assumed mechanism is shown in Scheme 1 for
the exemplary reaction of cyclodecyne 1 (with Y ) O).
The initially formed vinyl radical 2a undergoes a transannular
1,5 or 1,6 hydrogen atom transfer (HAT), 2a f 3a/4a, followed
by a 5- or 6-exo radical cyclization, 3a/4a f 5a/6a, respectively.
Homolytic cleavage of the O-X bond in the final step leads to
the isomeric cis-fused bicyclic ketones 7a and 8a with simul-
taneous release of a radical X•. Since X• does not act as a radical
chain carrier, this sequence can be regarded as a “self-
† The University of Melbourne.
‡ ARC Centre of Excellence for Free Radical Chemistry and Biotechnology.
§ Christian-Albrechts Universita¨t Kiel.
(1) (a) Wille, U.; Plath, C. Liebigs Ann./Recl. 1997, 111. (b) Wille, U.;
Lietzau, L. Tetrahedron 1999, 55, 10119. (c) Wille, U.; Lietzau, L.
Tetrahedron 1999, 55, 11456. (d) Lietzau, L.; Wille, U. Heterocycles 2001,
55, 377. (e) Wille, U. Chem. Eur. J. 2002, 8, 340. (f) Dreessen, T.; Jargstorff,
C.; Lietzau, L.; Plath, C.; Stademann, A.; Wille, U. Molecules 2004, 9,
480. (g) Jargstorff, C.; Wille, U. Eur. J. Org. Chem. 2003, 3173. (h) Wille,
U.; Jargstorff, C. J. Chem. Soc., Perkin Trans. 1 2002, 1036. (i) Wille, U.
Tetrahedron Lett. 2002, 43, 1239. (j) Wille, U. J. Am. Chem. Soc. 2002,
124, 14. (k) Stademann, A.; Wille, U. Aust. J. Chem. 2004, 57, 1055. (l)
Sigmund, D.; Schiesser, C. H.; Wille, U. Synthesis 2005, 1437.
terminating, oxidative radical cyclization”, in which XO•
represents a synthon for O atoms in solution.
Since the initial discovery of self-terminating radical oxy-
•
genations a decade ago using the O-centered nitrate radical, NO3
(X ) NO2), we have shown that this sequence represents a
general concept that can be applied to every major class of
10.1021/jo702261u CCC: $40.75 © 2008 American Chemical Society
Published on Web 01/16/2008
J. Org. Chem. 2008, 73, 1413-1421
1413

Wille et al.
SCHEME 2
FIGURE 1. Reaction of imidyl radicals 9 with cyclodecyne leads to
imidyl-substituted bicyclic alkenes 10 and 11 through a radical addition/
cyclization sequence.
organic and inorganic O-centered radicals XO• (see Scheme 1
for the various X). Intrigued by the scope of this methodology,
we therefore decided to examine the performance of other
heteroatom-centered radicals in self-terminating radical reac-
tions.
In an earlier paper we have demonstrated that the N-centered
phthalimidyl radicals (Im•, 9) do indeed react with cycloalkyne
1 through C-N bond formation to give imidyl-substituted
bicyclic alkenes 10 and 11 (Figure 1).2 These products are likely
formed through a transannular HAT/exo radical cyclization
sequence, e.g., 2b f 3b/4b and 3b/4b f 5b/6b, respectively,
similar to that shown in Scheme 1 (with YX• ) Im•). However,
since homolytic cleavage of N-acyl bonds, in analogy to the
terminating step in radical cyclizations involving O-centered
radicals, seemed not to be an energetically favorable pathway
for the radical intermediates 5b/6b, the latter underwent
stabilization through alternative mechanisms, which were not
further explored.
Therefore, we reasoned that self-terminating radical cycliza-
tions with N-centered radicals of type RXN• should be possible,
if the substituent X would give a stable radical X• that is released
through homolytic fragmentation of the N-X bond in the final
step of the cyclization cascade shown in Scheme 1 (with Y )
NR). In the case of cycloalkyne 1, a successful self-terminating
radical sequence should then lead to formation of the CdN
double bond in the imines 7b and 8b. However, because of the
intrinsic instability of imines,3 at least partial hydrolysis of the
latter to the respective ketones 7a and 8a could also occur.
In this work we report on the performance of two different
classes of N-centered radicals, e.g., aminium and amidyl radicals,
in self-terminating radical cyclizations using the well-established
reaction with the cyclic alkyne 1 as the model system. The
experimental studies performed to reveal the synthetic scope
of the concept are augmented by computational studies to obtain
insights into the mechanism of the reactions under investigation.
TABLE 1. Experimental Conditions and Results for the Reaction
of Aminium Radicals 13a-d with Cycloalkyne 1a
[radical
yieldb/%
entry precursor]/mM [1]/mM solvent
acid
7a + 8a
1
2
3
4
5
6
7
8
12a: 17
12a: 6
12a: 9
12a: 13
12a: 6
12a: 8
11
11
11
9
10
11
11
9
CH2Cl2 0.1 M MAc
CH2Cl2 0.1 M MA
CH2Cl2 0.1 M TFA
CH2Cl2 0.1 M HOAc
CH2Cl2
3
31
7
2
0
6
9
3
MeCN
MeCN
MeOH
0.1 M MA
0.1 M TFA
1.25 M HCl
12a: 8
12a: 4
9
14a: 3
13
11
6
8
8
11
10
8
9
6
CH2Cl2 0.1 M MA
MeCN 0.1 M TFA
CH2Cl2 0.1 M TFA
10
31
61
57
4
37
27
16
56
28
7
10
11
12
13
14
15
16
17
18
19
14a: 24
12b: 21
12b: 32
12b: 10
12b: 20
12b: 9
12c: 25
12c: 25
12d: 17
12d: 15
MeCN
MeCN
MeOH
MeOH
0.1 M TFA
0.1 M MA
1.25 M HCl
1.25 M HCl
CH2Cl2 0.1 M TFA
MeCN 0.1 M TFA
CH2Cl2 0.1 M TFA
MeCN 0.1 M TFA
8
a Mercury lamp (TQ150), 60 min, 10 mL of solvent. b Combined yield
(based on minor compound) determined by GC using n-hexadecane as
internal standard. c MA ) malonic acid.
radicals, amidyl radicals react at nitrogen only.6 The aminium
radicals 13a-d were generated through photolysis of the
respective N-nitrosoamines 12a-d under acidic conditions. The
dibenzylaminium radical 12a was also photochemically gener-
ated from N-chloro amine 14a (Scheme 2). The benzyl-
substituted amidyl radicals 16a,b were obtained through pho-
tolysis of the corresponding N-nitrosoamides 15a,b. No indication
for interfering reactions by nitric oxide or chlorine, the byprod-
ucts in the radical generation process, was found. All photo-
chemical experiments were performed at least in duplicate.
Reaction of Aminium Radicals 13a-d with 1. In Table 1
are compiled the experimental conditions and results of the
reaction of aminium radicals 13a-d with 1. The experiments
were carried out on analytical scales by irradiating the radical
precursor 12 in the presence of cycloalkyne 1 in the respective
solvent/acid system (10 mL) for 60 min.7 After neutralization
and workup, the reaction mixtures were analyzed by quantitative
GC. For details see the Experimental Section.
Results and Discussion
Experimental Studies. Whereas neutral aminyl radicals
preferably react with alkenes by HAT and are therefore not
suitable for our purposes, N-protonation increases the electro-
philicity of the radical center, and successful addition of
aminium radicals to π systems has been reported in the
literature.4,5 Compared with aminium radicals, amidyl radicals
are less electrophilic. Although they are delocalized π-allyl
(2) Wille, U.; Kru¨ger, O.; Kirsch, A.; Lu¨ning, U. Eur. J. Org. Chem.
1999, 3185.
(3) Neale, R. S. J. Org. Chem. 1967, 32, 3263.
(5) (a) Chow, Y. L. Can. J. Chem. 1965, 43, 2711. (b) Chow, Y. L.
Tetrahedron Lett. 1964, 2333. (c) Chow, Y. L.; Colon, C. Can. J. Chem.
1967, 45, 2559.
(6) Fossey, J.; Lefort, D.; Sorba, J. Free Radicals in Organic Chemistry;
John Wiley & Sons/Masson: Chichester, UK, 1995.
(7) It was verified that 1 was completely consumed after 60 min of
irradiation time.
(4) Reviews on N-centered radicals: (a) Wolf, M. E. Chem. ReV. 1963,
63, 55. (b) Neale, R. S. Synthesis 1971, 1. (c) Minisci, F. Synthesis 1973,
1. (d) Chow, Y. L. Acc. Chem. Res. 1973, 6, 354. (e) Minisci, F. Acc. Chem.
Res. 1975, 8, 165. (f) Chow. Y. L.; Danen, W. C.; Nelsen, S. F.; Rosenblatt,
D. H. Chem. ReV. 1978, 78, 243. (g) Stella, L. Angew. Chem., Int. Ed.
Engl. 1983, 22, 337.
1414 J. Org. Chem., Vol. 73, No. 4, 2008

N-Centered Radicals in Self-Terminating Radical Cyclizations
TABLE 2. Experimental Conditions and Results for the Reaction
of Amidyl Radicals 16a,b with Cycloalkyne 1
The first experiments were performed using the dibenzyl-
aminium radical 13a, which was believed to perfectly satisfy
the prerequisites for self-terminating radical cyclizations, since
a stable benzyl radical should be released during the terminating
homolytic bond scission (see Scheme 1; Y ) NBn, X• ) Bn•).
Indeed, irradiation of a 60% excess of radical precursor 12a in
the presence of 1 in 0.1 M malonic acid (MA) in dichlo-
romethane led to formation of the ketones 7a and 8a in,
however, only very low yield (entry 1).8 Imines of types 7b
and 8b were not found in any experiment in this study.9 Since
there is no obvious alternative pathway for formation of the
ketones 7a/8a, other than through hydrolysis of imines 7b/8b
(or related species; see later), we consider ketone formation as
indirect evidence for a successful radical addition/cyclization
sequence. Using an excess of 1 resulted in an increase of the
combined yield of 7a/8a to about 31% (entry 2). Other solvents,
e.g., acetonitrile or methanol, and other acids, e.g., trifluoroacetic
acid (TFA), acetic acid, or hydrochloric acid, did not lead to a
further improvement (entries 3, 4, and 6-8). No clear influence
of the presence of oxygen on the yield was found (data not
shown).
On the other hand, 7a/8a were not formed in the absence of
acid (entry 5). When 13a was generated from the respective
N-chloro amine 14a, a similar outcome with regard to the yield
of 7a/8a was found, although the best solvent/acid system for
this reaction was acetonitrile/0.1 M TFA (entries 9 and 10).
During the irradiation of precursor 14a, formation of significant
amounts of N,N-dibenzylamine was observed. The different
behavior of dibenzylaminium radicals 13a in dependence of their
precursor can be explained by the fact that radical generation
through irradiation of N-nitrosoamine 12a under acidic condi-
tions occurs out of the protonated state leading directly to the
aminium radical 13a, whereas photolysis of N-chloro amine 14a
leads initially to aminyl radicals, which are protonated only in
a subsequent step to yield 13a.10 Obviously, under our conditions
protonation of the aminyl radicals was competing with hydrogen
abstraction (presumably from the solvent) to give N,N-diben-
zylamine. Because of this side reaction, N-chloroamines were
not further explored as a source for aminium radicals in this
study.
[radical
yielda/%
entry
precursor]/mM
[1]/mM
solvent
7a + 8a
b,c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 20
15a: 80
15a: 20
15a: 40
15a: 10
15b: 20
15b: 20
15b: 20
15b: 20
15b: 20
15b: 20
15b: 20
10
10
10
10
10
10
10
10
10
40
10
10
10
10
10
10
10
10
10
10
CH2Cl2
CHCl3
7
6
b,c
cyclohexaneb,c
benzeneb,c
MeOHb,c
MeCNb,c
acetoneb,c
MeCNb,c,d
MeCNb,e
MeCNb,f
MeCNg
traces
11
traces
48
52
65
35
17
47
42
17
38
25
26
17
15
26
22
MeCNg
MeCNg
MeCNb,c
acetoneb,e
MeCNb,c,d
MeCNc,h
MeCNc,d,h
MeCNc,i
MeCNg
a Combined yield (based on minor compound) determined by GC using
n-hexadecane as internal standard. b Rayonet photoreactor at λ ) 350 nm.
c Reaction time 24 h. d Under O2. e Reaction time 96 h. f Reaction time 48
h. g Mercury lamp (TQ150), reaction time 5 h. h Rayonet photoreactor at λ
) 300 nm. i Rayonet photoreactor at λ ) 254 nm.
To conclude, the reactions of aminium radicals 13a-d with
1 were highly inconclusive. Although the findings suggest that
oxidative radical cyclizations with N-centered radicals should
principally work, comparison of the various entries in Table 1
shows that optimized experimental conditions for the reaction
of one radical, e.g., concentration ratio between alkyne and
radical precursor, as well as the solvent and acid system, were
not applicable to the reaction of a different radical (see, for
example, entry 11 vs 16, entry 17 vs 19, etc.); in other words,
the reaction conditions had to be re-optimized for each of the
different radicals. Thus, in the case of radicals 13b-d the
highest yields of 7a/8a were obtained with an excess of the
respective radical precursor (entries 11, 17, and 18), but the
opposite appeared to be the case with radical 13a (entry 2). We
cannot exclude, however, that the aqueous workup procedure,
which was required to neutralize the reaction mixtures prior to
analysis, led to secondary reactions, e.g., hydrolysis, etc., so
that the observed GC data may not reflect the results of the
radical reaction itself.11
Reaction of Amidyl Radicals 16a,b with 1. In contrast to
aminium radicals, self-terminating radical cyclizations with the
amidyl radicals 16a and 16b could be performed under neutral
conditions that allowed direct GC analysis of the reaction
mixture without prior workup. The experimental conditions and
results are compiled in Table 2. The reactions were also
performed on analytical scales by irradiating the radical precur-
sor 15 in the presence of cycloalkyne 1 in various solvents (5
mL) using either a mercury lamp or a Rayonet photoreactor at
different wavelengths (e.g., λ ) 350, 300, or 254 nm). The
reaction mixtures were analyzed by quantitative GC and GC/
MS. For details see the Experimental Section.
To our surprise, the reaction of 1 with the aminium radicals
13b and 13c resulted in significantly higher combined yields
of 7a/8a of up to 61% (entry 11), or 56% (entry 17),
respectively, although highly unstable alkyl radicals (methyl in
the former, ethyl in the latter case) should be released in the
final â-fragmentation step. Even the reaction of the cyclic
piperidinium radical 13d with 1 led to formation of 7a/8a in an
albeit moderate combined yield of 28% (entry 18), despite the
fact that 13d was deliberately chosen as representative for an
N-centered radical possessing substituents with, at least what
we believed, poor leaving group ability. Our unexpected result
that no correlation between the stability of the leaving radical
and the yield of oxidized products 7a/8a was found leads to
the question whether the assumed mechanism of the cyclization
cascade is actually according to that shown in Scheme 1. We
will address this in the second part of this paper.
(8) The ketones 7a and 8a were unambiguously identified by comparison
of their GC retention times with those of authentic samples, in addition to
GC/MS experiments.
(9) Attempts to trap N-containing intermediates, for example through in
situ reduction of imines with sodium borohydride, were not successful.
(10) Cessna, A. J.; Sugamori, S. E.; Yip, R. W.; Lau, M. P.; Snyder, R.
S.; Chow, Y. L. J. Am. Chem. Soc. 1977, 99, 4044.
As before, also in the reactions of the amidyl radicals 16a,b
with 1, the imines 7b/8b were not formed, and the bicyclic
ketones 7a and 8a were identified as exclusive products.9 We
(11) GC analysis of the crude, acidic reaction mixture was not possible.
J. Org. Chem, Vol. 73, No. 4, 2008 1415

Wille et al.
SCHEME 3
intermediately formed imines 7b/8b (or related intermediates;
see below). Thus, although the released radical 19 from the first
cyclization sequence does not act as a radical chain carrier, it
undergoes transformation into a species that initiates a second
cyclization cascade. We will be investigating further radicals
exhibiting similar multistep activities in the future.
Computational Studies
Our experiments have clearly revealed that N-centered
radicals can undergo addition to alkynes, where they initiate a
radical cyclization sequence. However, since no apparent
correlation between the stability of the leaving radicals and the
yield of cyclized products, 7a/8a, was found in the reaction of
either aminium or amidyl radicals with 1, the question arises,
whether the mechanism of self-terminating radical cyclizations,
as shown in Scheme 1 (which we believe is correct for
O-centered radicals),13 is applicable to N-centered radicals. We
have therefore performed computational studies to obtain a
detailed mechanistic insight into the key steps in self-terminating
radical cyclizations, particularly the initial addition of N-centered
radicals to alkynes (1 f 2b) and the terminating bond scission
(5b/6b f 7b/8b).
The calculations were carried out using the Gaussian 03
program.14 Geometry optimizations were performed using the
BHandHLYP/6-311G** and BHandHLYP/6-311++G** meth-
ods.15 The ground and transition states were verified by
vibrational frequency analysis at the same levels of theory, and
all identified transition states showed only one imaginary
frequency. The spin expectation value, s2 , was very close to
0.75 after spin annihilation. The one-electron reduction potentials
E°, absolute and relative to nonaqueous SCE, were calculated
according to the procedure recently described by Coote et al.16
For this, the Gibbs free energy of each species was calculated
at the BHandHLYP/6-311++G** level of theory (except for
the half reaction in entry 5 of Table 6, for which the Gibbs free
energy was computed at the BHandHLYP/6-31++G**) level,
and the solvation energy was calculated using the conductor-
like polarizable continuum model (CPCM) at the B3LYP/6-
31+G* level of theory. All geometries of the studied species
have been fully optimized in the gas phase at the BHandHLYP/
6-311++G** level of theory and in the presence of solvent,
i.e., acetonitrile, at the B3LYP/6-31+G* level of theory. The
believe that the latter must have been generated from the
respective imines 7b/8b or imine-type intermediates upon
injection onto the GC column (we will discuss this possibility
below).
Initial experiments were performed in which 2:1 mixtures of
radical precursor 15a and 1 in various solvents were irradiated
for 24 h at λ ) 350 nm in a Rayonet photoreactor. Generally,
in contrast to the experiments with aminium radicals, the
reactions with amidyl radicals appeared to be easier to control.
Both acetonitrile and acetone were the best solvents, which often
gave similar outcomes (because of this, Table 2 shows mostly
the results of the experiments performed in acetonitrile), and
combined yields of around 50% of 7a/8a were obtained in the
reaction of the acetamidyl radical 16a with 1 (entries 6 and 7).
Halogenated solvents, hydrocarbons, benzene, or methanol
performed significantly poorer (entries 1-5). Elongation of the
reaction time to 96 h (entry 9) and increasing the total
concentration (entry 10) resulted in reduced yields. Using the
stronger mercury lamp as a light source, an outcome similar to
that in experiment 6 was obtained with an irradiation time of
only 5 h (entry 11). A much larger excess of radical precursor
15a (entry 12) does not affect the yield significantly, whereas
an equimolar ratio of 15a and 1 (entry 13) clearly resulted in a
poorer outcome. Interestingly, the reaction performed in the
presence of oxygen led to a significant increase of 7a/8a
formation (entry 8 vs 6).
Compared with 16a, the tosylamidyl radical 16b, which
contained two suitable leaving groups with the tosyl radical
supposedly being even more stable than the benzyl radical, gave
generally lower yields of 7a/8a. A maximum yield of 38% was
found when the irradiation was performed in acetonitrile for
24 h at λ ) 350 nm (entry 14). Longer reaction times (entry
15), shorter irradiation wavelengths (entries 17-19), or a
different light source (entry 20) as well as the presence of
oxygen (entries 16 and 18) only resulted in a reduction of the
yield of 7a/8a.
(13) Dreessen, T. Ph.D. Thesis, Universita¨t Kiel, 2004.
(14) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K.
N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.;
Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li,
X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.;
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.;
Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.;
Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels,
A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03,
Revision B.04; Gaussian, Inc.: Pittsburgh, PA, 2003.
In a seemingly unrelated project, we became interested in
studying the potential of carbamoyloxyl radicals in self-
terminating radical oxygenations. Thus, when the O-centered
radical 18, which was generated through photolysis of the
corresponding Barton ester 17 in acetonitrile,1g was reacted with
excess cycloalkyne 1, the bicyclic ketones 7a/8a were obtained
in a surprisingly high combined GC yield of 155% with respect
to the radical precursor 17 (Scheme 3).
We suggest that after the first successful radical cyclization
initiated by addition of 18 to 1 (see Scheme 1), the released
carbamoyl radical 19 undergoes decarbonylation12 to give the
N-centered acetamidyl radical 20. The latter subsequently reacts
with 1 to yield further 7a/8a after decomposition of the
(15) Single-point energy calculations performed with the QCISD/cc-
pVDZ and CCSDT/cc-pVDZ methods on BHandHLYP/6-311G** and
BHandHLYP/aug-cc-pVTZ optimized geometries showed excellent agree-
ment with results obtained from BHandHLYP/6-311G** and BHandHLYP/
6-311++G** computations.
(12) The rate constants for decarbonylations are in the order 106-107
s-1; see for example: (a) Turro, N. J.; Gould, I. R.; Baretz, B. H. J. Phys.
Chem. 1983, 87, 531. (b) Kurnysheva, O. A.; Gritsan, N. P.; Tsentalovich,
Y. P. Phys. Chem. Chem. Phys. 2001, 3, 3677.
(16) Namazian, M.; Coote, M. L. J. Phys. Chem. A 2007, 111, 7227.
1416 J. Org. Chem., Vol. 73, No. 4, 2008

N-Centered Radicals in Self-Terminating Radical Cyclizations
SCHEME 4
TABLE 3. Calculated Activation Energies, Eq, and Reaction
Enthalpies, ∆E, for the Reaction of Acetamidyl Radicals 16a with 1
(in kJ mol-1; BHandHLYP/6-311G**)a,b
TABLE 4. Calculated Activation Energies, Eq1-4, and Reaction
Enthalpies, ∆E1-3, for the Addition of Aminium and Amidyl
Radicals 13a and 16a/c, Respectively, to 2-Butyne in the Absence
and Presence of Oxygen (in kJ mol-1; BHandHLYP/6-311G**)a
step
1
reaction
Eq
∆E
entry
1
R1
Bnb
E1
∆E1
E2
∆E2
E3
∆E3
E4
q
q
q
q
1 + 16a f 2b
35.2
40.8
29.6
15.3
35.1
32.0
129.9
119.7
-80.7
-66.3
-54.9
-57.5
-121.1
-163.9
24.5
28.0c -87.1 78.0
56.5 23.1 -39.8
72.3 24.4 -32.0
8.9
3.4
2.8
2
3
4
2b f 4b (1,6-HAT)
4b f 6b (6-exo)
6b f 8b + Bn•
32.7c -69.1 86.6
2
3
Ac
42.0
49.1
-91.1 54.3 -5.8 59.1 -24.0
-74.4 61.1
-63.7 65.8
-51.9 69.6
9.4 63.5 -15.5 -2.0
35.7 59.4 -20.1
45.8 61.1 -11.4
PhSO2 55.1
11.0
57.6
5.0
8.2
a Numbers in italics contain zero-point vibrational energy correction
(ZPC). b Protonated N. c Via an alkyne/radical association complex (see
text).
a Numbers in italics contain zero-point vibrational energy correction
(ZPC). b Numbering of the reaction steps and intermediates is according to
Scheme 1 (with R ) Ac and X ) Bn).
barrier and reaction enthalpy for the initial radical addition to
the CtC triple bond as well as for the terminating â-fragmenta-
tion are different for the various N-centered radicals used in
this study. To get insight into the initial radical attack,
calculations were performed on the addition of the aminium
radical 13a and amidyl radicals 16a and 16c (the latter mimics
tosylamidyl radical 16b) to alkynes in the absence and presence
of oxygen, using 2-butyne as a simplified model for the
cycloalkyne 1. Scheme 4 shows the computed potential surface
of these reactions, and the results of the calculations are listed
in Table 4. At this point, we will focus the discussion on the
radical addition in the absence of O2.
archive entries of the Gaussian output files for all optimized
ground and transition state structures in this work are available
as electronic Supporting Information.
To explore whether a self-terminating radical cyclization
cascade is principally feasible with N-centered radicals, we
investigated at first the potential surface of the reaction between
1 and the acetamidyl radical 16a leading to the bicyclo[4.4.0]-
decane derivative 8b (see Scheme 1 with R ) Ac and X )
Bn). In Table 3 are compiled the calculated activation energies,
Eq, and reaction enthalpies, ∆E, for each of the four reaction
steps in the proposed mechanism. The data show that the entire
reaction sequence is thermodynamically favorable with a total
reaction energy ∆Etotal of ca. -225 kJ mol-1. Initial radical
addition (step 1), transannular HAT (step 2), and transannular
radical cyclization (step 3) have low Eq, and all three steps are
strongly exothermic.17 In this sequence, the final homolytic bond
cleavage is the only step associated with a significantly high
Eq of some 130 kJ mol-1 (or ca. 120 kJ mol-1 with zero-point
vibrational energy correction, ZPC, included), which is also
slightly endothermic. However, compared with the ring opening
6b f 4b in reversal of the 6-exo cyclization, which requires
not only around 150 kJ mol-1 but is also strongly endothermic,
â-fragmentation 6b f 8b with release of a benzyl radical should
principally be both the kinetically and thermodynamically more
preferable pathway for intermediate 6b.
Attack of the positively charged radical 13a to 2-butyne
proceeds via initial formation of a radical-alkyne association
q
complex, with a low E1 of ca. 33 kJ mol-1 (entry 1, ZPC
included),18 whereas E1q for the addition of amidyl radicals 16a
and 16c to 2-butyne is 16-25 kJ mol-1 higher (entries 2 and
3).19 The reaction enthalpy ∆E1 is strongly negative in all cases.
Transition state 21 has Z geometry and leads initially to a
Z-configured vinyl radical. In Table 4, ∆E1 refers to an
E-configured vinyl radical 22. However, since the energy
difference between Z and E vinyl radicals is generally not only
minute but their interconversion requires also only few kJ mol-1
,
(18) The addition of 13a to 2-butyne was found to have a negative
q
activation barrier (E1 ) -11.5 kJ mol-1, or -4.3 kJ mol-1 with ZPC
Whereas the energetics of both transannular HAT and exo
cyclization should not vary significantly with the nature of the
N-substituent at the π system, it is very likely that activation
included), which strongly suggests formation of an association complex
between 13a and 2-butyne prior to the actual radical addition. This complex
was located on the potential surface and computed to be ca. 38 kJ mol-1
lower in energy than the free reactants.
(19) The lower Eq for the addition of 16a to 1, compared to the value
calculated for addition of 16a to 2-butyne (see Table 3), is believed to be
due to release of ring strain in the ten-membered ring as a result of change
in hybridization at the π bond from sp to sp2 during the radical addition.
(17) Orienting computations have revealed that the energetics of the
parallel pathway leading to the bicyclo[5.3.0]decane derivative 7b via a
1,5-HAT and 5-exo cyclization are in the same order of magnitude.
J. Org. Chem, Vol. 73, No. 4, 2008 1417

Wille et al.
q
TABLE 5. Calculated Activation Energies, E5 , and Reaction
SCHEME 5
Enthalpies, ∆E5-6, for the Homolytic and Heterolytic Bond Scission
in 28 and 31, Respectively (in kJ mol-1
)
a
q
entry
1b
R1
Me
R2
Bn
E5
∆E5
5.2
∆E6
92.4
82.4
90.7
80.6
51.6
41.1
51.5
40.7
-10.9
2c
3b,d
4c,d
5c
Me
Me
Me
Me
Me
Me
Me
Me
Me
Bn
Bn
Bn
Bn
Me
Me
Me
Me
Me
0.6
-15.5
-17.0
-35.4
-21.0
-39.1
248.4
229.2
SCHEME 6
6b
124.4
110.8
128.8
115.4
104.9
90.6
64.7
40.4
66.3
42.3
36.7
10.0
33.4
6.9
7c
8b,d
9c,d
10c
11b
12b
13b
14b
15b
16b
17b
18b
104.3
90.0
548.3
518.2
-(CH2)5-
-(CH2)4-
Bn
122.9
108.6
114.9
102.4
119.5
108.0
161.5
154.2
158.3
143.2
167.5
155.3
32.9
64.2
45.1
54.6
38.7
38.0
Ac
21.2
Bn
Ac
114.9
100.2
95.5
68.2
110.9
97.5
-20.6
-29.5
2.8
Ac
Me
the error in ∆E1 is negligible.6,20 Interestingly, the lower
electrophilicity of the uncharged dibenzylaminyl radical, com-
pared with its protonated counterpart 13a, is reflected by the
finding that addition of the former to 2-butyne requires an
activation energy that is higher by 60 kJ mol-1 and less
exothermic by ca. 55 kJ mol-1 than 13a (data not shown). It
should be noted that an association complex between the
uncharged amidyl radical 16a and 2-butyne prior to addition
was also located, but found to be only lower in energy by ca.
8 kJ mol-1 compared to the free reactants. Because of this, we
did not further pursue locating similar association complexes
in the reactions of other uncharged radicals with alkynes.
After having established that addition of both aminium and
amidyl radicals to CtC triple bonds should be kinetically
feasible reactions, we turned our attention to the terminating
â-fragmentation. Using the simplified model system 28 for the
actual radicals 5b/6b (see Scheme 1), the transition state 29 for
the homolytic cleavage of N-C bonds as well as of the product
imine 30 and the released radical R2• was calculated. The
potential surface is shown in Scheme 5, and the results are
compiled in Table 5.
Ac
Ac
Bn
PhSO2
Bn
27.1
97.3
PhSO2
88.2
-11.8
a Numbers in italics contain zero-point vibrational energy correction
(ZPC). b BHandHLYP/6-311G**. c BHandHLYP/6-311++G**. d Proto-
nated N.
a more exothermic (or less endothermic, respectively) reaction
by ca. 4 kJ mol-1, and we therefore used this method for
computing ∆E6 for the heterolytic bond scission (see below).
Interestingly, homolytic cleavage of an N-C bond at a
quarternary, protonated nitrogen, a very likely scenario under
the acidic conditions in the experiments using aminium radicals,
not only lowers E5q for release of a benzyl radical by ca. 40 kJ
mol-1 (entries 3 vs 1 or 4 vs 2) and for release of a methyl
radical by ca. 25 kJ mol-1 (entries 8 vs 6 or 9 vs 7), but also
makes this cleavage process thermodynamically significantly
more favorable. To conclude, of the various aminium radicals
studied in this work, a successful termination of the radical
cyclization sequence by homolytic â-fragmentation should only
be possible with the dibenzylaminium radical 13a.
The required energy for homolytic bond scission at an amide
or imide nitrogen depends on the nature of the remaining
substituent R1. Generally, release of an acetyl radical is
kinetically and thermodynamically even more unfavorable
(entries 14 and 16) than release of a methyl radical (entry 15).
Cleavage of a benzyl radical through fragmentation of an
acetamide N-C bond (entry 13) requires some 20 kJ mol-1
more energy than fragmentation of a sulfonamide N-C bond
(entry 18). Of all the various â-fragmentations studied, the
kinetically and thermodynamically most superior leaving group
should be the phenylsulfonyl radical (entry 17), for which an
As expected, the computations revealed that release of a
benzyl radical from an amine-derived precursor 28 (entries 1-4)
is both kinetically and thermodynamically significantly more
favorable than release of an alkyl radical (entries 6-9, 11, and
12). The influence of basis sets augmented with diffuse functions
q
on the outcome is not very large. It appears that E5 is usually
reduced by 1-2 kJ mol-1 when the BHandHLYP/6-311++G**
method is used (entries 1 vs 2, 3 vs 4, and 8 vs 9), except for
the cleavage of the methyl radical from the neutral 28 (entry 6
vs 7), where the augmented basis set resulted in an ca. 4 kJ
mol-1 higher value for E5 , compared to the 6-311G** basis.
q
Generally, the BHandHLYP/6-311++G** method resulted in
(20) Wille, U.; Dreessen, T. J. Phys. Chem. A 2006, 110, 2195.
1418 J. Org. Chem., Vol. 73, No. 4, 2008

N-Centered Radicals in Self-Terminating Radical Cyclizations
TABLE 6. Calculated Reduction Potentials E° (in V), Using BHandHLYP/6-311++G** Gas-Phase Energies with Solvation Energies Obtained
via CPCM Models, at the B3LYP/6-31+G* Level
a Versus nonaqueous SCE. b BHandHLYP/6-31++G** gas-phase energies.
E5 of only ca. 27 kJ mol-1 and ∆E5 of -30 kJ mol-1 (ZPC
included) were calculated at the BHandHLYP/6-311G** level
of theory.
respective N-centered radical is assumed). To explore this
possibility, we calculated the reduction potentials, E°, both
absolute and relative to nonaqueous SCE, for selected aminyl,
aminium, and amidyl radicals as well as for simplified model
systems of the R-nitrogen radical intermediates 5b/6b, which
are listed in Table 6.
q
Although these computations unanimously support our per-
ception of good or poor leaving groups in self-terminating
radical reactions, they clearly contradict the experimental
findings. This leads to the conclusion that the radical cyclizations
investigated in this study cannot be terminated by a homolytic
bond scission. As a mechanistic alternative, one could imagine
that the radical intermediate 28 is oxidized by some means to
the iminium ion 31 (we will discuss the nature of the oxidant
below), which then undergoes heterolytic bond scission to give
imine 30 with release of the substituent R2 as a cation (see
Scheme 5). However, computations of this alternative pathway
showed that this process is highly endothermic for cleavage of
both a benzyl and a methyl cation (Table 5, entries 5 and 10).
A transition state could not be located. Calculations performed
using a solvation model did not lower ∆E6 significantly (data
not shown). Therefore, such a heterolytic termination step can
clearly be ruled out.
Because both radical and cationic terminations are an unlikely
mechanism, how are the observed ketones 7a and 8a formed in
our reactions? A possible pathway could involve the iminium
ions 33 and 34, respectively, which are hydrolyzed during
aqueous workup, in the case of aminium radicals, or undergo
thermal decomposition upon injection onto the GC column, in
the case of amidyl radicals, to give 7a/8a (Scheme 6). Formation
of 33 and 34 could occur through oxidation of the R-nitrogen
radical intermediates 5b and 6b by the respective electrophilic
N-centered radical itself, which was usually present in excess
(quantitative conversion of the radical precursor into the
The data show that the aminium radicals 13a and 13b are
both good oxidizing agents (entries 1 and 2), whereas aminyl
radicals are not (entries 9 and 10). Compared with aminium
radicals, the amidyl radicals 16a and 16c (entries 3 and 4; the
phenylsulfone residue was used as a simplified model for the
tosyl residue in 16b) have a slightly weaker oxidation capability.
Most importantly, however, all aminium and amidyl radicals
used in this study are thermodynamically capable of oxidizing
their corresponding R-nitrogen radical intermediates to iminium
ions (the respective redox systems are the half reactions 1 + 8,
2 + 7, 3 + 5, and 4 + 6 in Table 6). On the basis of these data,
we conclude that the termination step in radical cyclization
cascades initiated by addition of N-centered radicals to alkynes
proceeds as proposed in Scheme 6 through oxidation of the
radical intermediates 5b/6b by the N-centered radicals them-
selves. Subsequent hydrolysis or thermal decomposition of the
resulting iminium ions 33 and 34, respectively, leads to the
bicyclic ketones 7a/8a. We believe that this redox pathway must
be much faster than the homolytic N-C bond scission, and this
would explain why a correlation between stability of the leaving
group and yield was not found in our experiments.
Finally, we explored the role of oxygen in these reactions,
especially the significant increase in yield of 7a/8a in the
reaction of amidyl radicals 16a with 1 (see Table 2, entry 8).
Because reaction of C-centered radicals with oxygen is a
J. Org. Chem, Vol. 73, No. 4, 2008 1419

Wille et al.
convenient way to produce peroxyl radicals,6,21 we wondered
whether peroxyl radical formation could also occur with
N-centered radicals (see Table 4 and Scheme 4). Interestingly,
oxidant in these reactions is the N-centered radical itself. The
reason for the different behavior of O- and N-centered radicals
with regards to the termination step may be a combination of
the relative ease to oxidize the R-nitrogen intermediates 5b/6b
and the difficulty to cleave N-C bonds in a homolytic fashion.
Thus, the reaction of N-centered radicals with alkynes represents
a highly interesting variation of self-terminating radical cy-
clizations, in which the radical addition/cyclization cascade is
terminated by a redox process. The entire cascade allows
oxidative transformation of alkynes into ketones under very mild
conditions, and we will study the application of this methodol-
ogy using other cyclic and open-chain alkynes in the future.
Whereas the conditions of the reactions involving amidyl
radicals appeared to be easily controllable, the behavior of
aminium radicals was very unpredictable and no general reaction
conditions were found. In the case of the acetamidyl radical
16a, a significant increase of the yield of 7a/8a was found in
the presence of oxygen. Computational studies lead to the
suggestion that at least partial trapping of the N-centered radicals
by oxygen is possible. The resulting peroxyl radicals may then
undergo addition to the CtC triple bond to give a very labile
vinyl radical that undergoes rapid γ-fragmentation to an R-oxo
carbene. Subsequent transannular C-H insertion of the carbene
may lead to formation of additional 7a/8a, parallel to the
pathway involving direct addition of the amidyl radical to the
alkyne. Recently, we have reported on the first example of
formation of R-oxo carbenes through γ-fragmentation in vinyl
radicals,23 and we are currently intensely exploring the scope
of these reactions in our laboratory.
q
the calculations show for amidyl radical 16a that E2 required
for formation of peroxyl radical 24 (with R1 ) Ac) is higher
by only ca. 12 kJ mol-1 than E1q for addition of 16a to 2-butyne
(entry 2). Thus, although the former reaction appears to be
kinetically less favorable than the latter, formation of peroxyl
radicals is a possible pathway, when one reaction partner, e.g.,
O2, is present in large excess, which was the case in our
experiments. The subsequent addition of 24 (with R1 ) Ac) to
2-butyne leads to a peroxylvinyl radical 26. This radical addition
q
requires an activation energy (E3 entry 2) similar to that
required for the addition of sulfonamidyl radical 16c to 2-butyne
(E1q entry 3), which we know is feasible under our experimental
conditions. The peroxyl bond in vinyl radical 26 (with R1 )
Ac) is extremely labile, and γ-fragmentation to an R-oxo carbene
q
(not shown) is a virtually barrierless process (E4 ) ( 0 kJ
mol-1). R-Oxo carbenes have been suggested as intermediates
in the peroxide, dimethyldioxirane, and HOF‚CH3CN mediated
epoxidation of 1, which leads to formation of 7a/8a through a
stereoselective transannular 1,5 or 1,6 carbene C-H insertion
reaction.22 Therefore, if under our experimental conditions a
fraction of the amidyl radicals 16a reacts with oxygen to produce
ultimately R-oxo carbenes, two parallel pathways leading to
formation of 7a/7b are possible. In contrast to this, formation
of peroxyl radicals 23 through reaction of oxygen with diben-
zylaminium radicals 13a or sulfonamidyl radicals 16c, respec-
tively, is associated with a significantly higher activation barrier
(E2q in entries 1 and 3) so that an R-oxo carbene pathway parallel
to the direct addition of these radicals to the CtC triple bond
in 1 seems less likely.
Experimental Section
1. Synthesis of the Radical Precursors. The compounds 12b-d
were commercially available. 12a,24 14a,25 and 15b26 were prepared
according to literature procedures.
Conclusions
N-Benzyl-N-nitrosoacetamide (15a). (a) N-Benzylacetamide:
Triethylamine (76.6 mL, 0.55 mol) was added to a solution of
This work was performed in order to explore whether the
concept of self-terminating radical cyclizations of alkynes, which
was discovered by us using O-centered radicals XO•, and which
enables oxidation of alkynes to ketones under very mild
conditions, could be extended to N-centered radicals RXN•,
using the reaction with the cyclic alkyne 1 as a model system.
Through a combination of experimental and computational
studies, it was revealed that both aminium as well as amidyl
radicals readily undergo intermolecular addition to the CtC
triple bond in alkynes, which triggers a series of intramolecular
radical translocation and cyclization processes. However, in
contrast to the homolytic fragmentation of the weak O-X bond
with release of a radical X•, which occurs in the reaction of
O-centered radicals with alkynes, computational studies revealed
that the termination step in the case of N-centered radicals is
an oxidative process, by which the isomeric iminium ions 33/
34 are formed, which subsequently undergo hydrolytic or
thermal decomposition to the corresponding ketones 7a/8a. The
27
benzylamine (54.6 mL, 0.5 mol) in THF (150 mL), the mixture
was cooled to 0 °C, and acetylchloride (37.3 mL, 0.52 mol) in THF
(50 mL) was slowly added. The reaction mixture was stirred for 3
h, warmed to rt, and quenched with saturated sodium bicarbonate
solution (30 mL). The phases were separated and the aqueous layer
was washed with dichloromethane (3 × 20 mL). The combined
organic layers were dried over magnesium sulfate and concentrated.
The crude product was purified by vacuum distillation (0.8 mbar,
138 °C) to give N-benzylacetamide as a yellowish solid (49.0 g,
1
66%). Mp 59 °C. H NMR (CDCl3, 400 MHz) δ 7.32-7.24 (m,
5H), 6.13 (s, broad, 1H), 4.38 (d, J ) 5.6 Hz, 2H), 1.98 ppm (s,
3H). 13C NMR (CDCl3, 100 MHz) δ 170.0, 138.1, 128.6, 127.7,
127.4, 43.6, 23.0 ppm.
(b) N-Benzyl-N-nitrosoacetamide (15a):28 A solution of N-
benzylacetamide (3.0 g, 0.02 mol) in a mixture of acetic acid (10
mL) and acetic anhydride (50 mL) was cooled to 0 °C, and sodium
nitrite (3.04 g, 0.044 mol) was added over 1 h. The mixture was
stirred at 0 °C until TLC indicated compete conversion (6 h), then
the mixture was warmed to 10-15 °C, poured into ice/water, and
extracted with diethyl ether (3 × 20 mL). The combined organic
(21) Alfassi, Z. P. Chemistry of Free Radicals: Peroxyl Radicals; John
Wiley & Sons: Chichester, UK, 1997.
(22) (a) Ciabattoni, J.; Campbell, R. A.; Renner, C. A.; Concannon, P.
W. J. Am. Chem. Soc. 1970, 92, 3826. (b) Dryuk, V. G. Tetrahedron 1976,
32, 2855. (c) Zeller, K.-P.; Kowallik, M.; Haiss, P. Org. Biomol. Chem.
2005, 3, 2310. (d) Curci, R.; Fiorentino, M.; Fusco, C.; Mello, R.; Ballistreri,
F. B.; Failla, S.; Tomaselli, G. A. Tetrahedron Lett. 1992, 33, 7929. (e)
Sun, S.; Edwards, J. O.; Sweigart, D. A.; D’Accolti, L.; Curci, R.
Organometallics 1995, 14, 1545. (f) Murray, R. W.; Singh, M. J. Org. Chem.
1993, 58, 5076. (g) Espenson, J. H.; Zhu, Z. J. Org. Chem. 1995, 60, 7728.
(h) Dayan, S.; Ben-David, I.; Rozen, S. J. Org. Chem. 2000, 65, 8816.
(23) Wille, U.; Andropof, J. Aust. J. Chem. 2007, 60, 420.
(24) Nakajima, M.; Warner, J. C.; Anselme, J.-P. Tetrahedron Lett. 1984,
25, 2619.
(25) Heuger, G.; Kalsow, S.; Go¨ttlich, R. Eur. J. Org. Chem. 2002, 1848.
(26) Overberger, C. G.; Anselme, J.-P. J. Org. Chem. 1963, 28, 592.
(27) Khan, A. T.; Choudhury, L. H.; Ghosh, S. Eur. J. Org. Chem. 2005,
2782.
(28) Itoh, T.; Nagata, K.; Matsuya, Y.; Miyazaki, M.; Ohsawa, A.
Tetrahedron Lett. 1997, 38, 5017.
1420 J. Org. Chem., Vol. 73, No. 4, 2008

N-Centered Radicals in Self-Terminating Radical Cyclizations
phases were washed with water (30 mL), aqueous sodium carbonate
(5%), and water (30 mL) and then dried over magnesium sulfate.
The solvent was removed in vacuo, and the product was purified
by flash chromatography (SiO2, petroleum spirits/diethyl ether 3:1)
to give 15a as an orange liquid (2.70 g, 76%), which was stored
(b) General procedure for the reaction of amidyl radicals
with 1: 1 was dissolved in the respective solvent, and oxygen was
removed by ultrasound treatment for 30 min under a steady flow
of nitrogen. The radical precursor (15a,b) was then added and the
reaction mixture was irradiated using a mercury lamp or in a
photoreactor for the given time under a steady flow of nitrogen. A
defined amount of n-hexadecane was then added to the reaction
mixture as internal standard, which was subsequently analyzed by
quantitative GC.
1
under nitrogen at -30 °C. H NMR (CDCl3, 500 MHz) δ 7.26-
7.18 (m, 5H), 4.92 (s, 2H), 2.79 (s, 3H). 13C NMR (CDCl3, 100
MHz) δ 174.6, 134.6, 128.8, 128.0, 42.1, 22.8 ppm (1 overlapped
arom. C).
2. Photochemical Radical Reactions. The irradiations were
performed at 25-30 °C. A temperature associated with the reaction
could not be observed under the experimental conditions. A dark
reaction between the radical precursors and 1 could be strictly
excluded in all cases.
(a) General procedure for the reaction of aminium radicals
13a-d with 1: 1 and the radical precursor (12a-d or 14a) were
dissolved in the respective solvent, and oxygen was removed by
ultrasound treatment for 15 min under a steady flow of argon. The
reaction mixture was irradiated using a mercury lamp for the given
time under a steady flow of argon. After neutralization with
saturated sodium bicarbonate solution, the phases were separated
and the aqueous layer was washed with dichloromethane. The
combined organic phases were dried over magnesium sulfate and
concentrated. After the residue was dissolved in dichloromethane,
a defined amount of n-hexadecane was added as internal standard
for the subsequent analysis of the mixture by quantitative GC.
Acknowledgment. This work was supported by the Aus-
tralian Research Council under the Centre of Excellence
program, Deutsche Forschungsgemeinschaft, Universita¨t Kiel,
Victorian Institute for Chemical Sciences High Performance
Computing Facility, High Performance Computing Facility of
the University of Melbourne, and the Australian Partnership for
Advanced Computing.
Supporting Information Available: General experimental
information and the archive entries of the Gaussian output files for
all optimized geometries in this work. This material is available
free of charge via the Internet at http://pubs.acs.org.
JO702261U
J. Org. Chem, Vol. 73, No. 4, 2008 1421