Rule of five cyclizations in 5-hexenyl radicals and photocycloadditions of 1,5-hexadienes: Effect of 4-oxa substitution

Article abstract of DOI:10.1002/poc.1120

The effect of 4-oxa substitution on the regiochemistry and rate of 5-hexenyl radical cyclizations was investigated, as a potential model for [2 + 2] photocycloadditions of 2-acyl-4-oxa-1,5-hexadienes. Increasing the electron density in the alkene decreases the rate of cyclization in the 4-oxa-hexejiyl radicals, relative to the all carbon analogs, but has little effect on the regioselectivity of the cyclization. The radical model does not reproduce the high degree of 1,6 closure, observed in the [2 + 2] photocycloadditions for 4-oxa-1,5-hexadiene la. However, the radical model does reinforce the interpretation that ground state conformational effects, engendered by substitution remote from the reacting centers have important rate consequences for cyclization reactions. Copyright

Full text of DOI:10.1002/poc.1120

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY
J. Phys. Org. Chem. 2007; 20: 83–87
Published online 5 February 2007 in Wiley InterScience
(
www.interscience.wiley.com) DOI: 10.1002/poc.1120
Rule of five cyclizations in 5-hexenyl radicals
and photocycloadditions of 1,5-hexadienes:
effect of 4-oxa substitution
Albert R. Matlin,* Karen Feit Brinton and Belinda Tsao Nivaggioli
Department of Chemistry and Biochemistry, Oberlin College, Oberlin, Ohio 44074, USA
Received 12 June 2006; revised 2 August 2006; accepted 2 August 2006
ABSTRACT: The effect of 4-oxa substitution on the regiochemistry and rate of 5-hexenyl radical cyclizations was
investigated, as a potential model for [2 þ 2] photocycloadditions of 2-acyl-4-oxa-1,5-hexadienes. Increasing the
electron density in the alkene decreases the rate of cyclization in the 4-oxa-hexenyl radicals, relative to the all carbon
analogs, but has little effect on the regioselectivity of the cyclization. The radical model does not reproduce the high
degree of 1,6 closure, observed in the [2 þ 2] photocycloadditions for 4-oxa-1,5-hexadiene 1a. However, the radical
model does reinforce the interpretation that ground state conformational effects, engendered by substitution remote
from the reacting centers have important rate consequences for cyclization reactions. Copyright # 2007 John Wiley &
Sons, Ltd.
KEYWORDS: hexenyl radical; cyclization kinetics; conformational effects; photocycloadditions
INTRODUCTION
of excited dienones (Scheme 1). Previously, Wolff and
Agosta had shown that 2-acyl-1,5-hexadienones exclu-
sively undergo photochemical reactions to produce
products that were derived from initial closure to a
Intramolecular radical cyclizations and intramolecular
[
2 þ 2] photocycloadditions have been the focus of
1,2
7
intense mechanistic scrutiny. In addition, these reac-
tions have found broad application in the synthesis of
five-membered ring biradical. In contrast, we found that
the oxa-substituted dienones 1a and 1b gave almost equal
amounts of the 1,5 closure products 4a/4b and 1,6 closure
products 5a/5b (Table 1). This behavior was very
different from that observed with the methylene analog
1e which despite methyl substitution at C-5 in the
3,4
complex carbocyclic ring systems. It has been widely
noted that these cyclizations strongly favor, where
possible, the formation of five-membered rings. The
so-called ’rule of five’ closure appears to be another
manifestation of the kinetic preference for the formation
of five-membered rings. Some years ago, Wolff and
Agosta showed that the cyclization of substituted
7
hexadiene system, produced only 1,5 closure product 4e.
We also found that methyl substitution at C-3, remote
from the reacting double bonds, had a dramatic effect on
the regioselectivity, yield, and the rate of reaction.
Irradiation of 1c produced only 1,5 closure product 4c.
The reaction proceeded 2 to 4.5 times faster than with 1a
and 1b, and the yield was greatly improved from ꢀ20%
(for 1a and 1b) to 65 %. These results prompted us to
inquire if the radical cyclization model, described by
Wolff and Agosta could be extended to explain the
different photochemistry of the 2-acyl-4-oxa-substituted
1,5-hexadiene systems. Below, we report the results of a
study of the regiochemistry and kinetics for a series of
4-oxa-5-hexenyl radical cyclizations.
5-hexenyl radicals could be used as a model to rationalize
the regiochemistry of the mode of closure (1,5 vs. 1,6) in
intramolecular [2 þ 2] photocycloadditions of 1,5-
5
hexadien-3-ones. In these systems, the excited enone
triplet was operationally equivalent to a localized
mono-radical at the b-enone carbon.
Several years ago, we reported a study on the
intramolecular photocycloaddition of a series of
2
-acyl-4-oxa-1,5-hexadienes 1a,b,c, as part of a program
designed to explore the effects of oxa substitution on the
6
reactivity of radical and diradical intermediates. We
were interested to explore if oxa substitution would affect
the partitioning between 1,5- and 1,6-cyclization modes
RESULTS AND DISCUSSION
*
Correspondence to: A. R. Matlin, Department of Chemistry and
Bromo-vinyl ethers 6 were prepared by mercuric acetate
catalyzed alcohol exchange of 1-bromo-3-propanol or
Biochemistry, Oberlin College, Oberlin, OH 44074, USA.
E-mail: albert.matlin@oberlin.edu
Copyright # 2007 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2007; 20: 83–87

8
4
A. R. MATLIN ET AL.
O
R1
O
R1
Table 2. Regiochemistry of 5-hexenyl radicals cyclization:
,5 (7) versus 1,6 closure (8)
1
1,5
X
X
a
7:8
Radical
R1
O
2
4
R2
R2
X
hν
R2
O
R1
R1
9a
>100:1
O
O
1
X
X
1,6
O
R2
R2
3
5
d: R = R = H, X = CH
2
9b
1.6:1
CH3
CH3
a: R = R = H, X = O
1
2
1
2
b: R = H, R = CH , X = O
e: R = H, R = CH , X = CH
1
2
3
1 2 3 2
c: R = CH , R = H, X = O
f: R = CH , R = H, X = CH
1 3 2 2
1
3
2
9
c
>100:1
Scheme 1
O
1
2
-bromo-3-butanol with either ethyl vinyl ether or
-methoxypropene. Radicals 9a,b,c were generated by
b
treating degassed decalin solutions of the appropriate
bromide 6 with tri-n-butyltin hydride, in the presence of
AIBN as an initiator. The products (7, 8, 12) were
determined by comparison of GC and GC–MS data with
authentic samples. The regiochemical selectivity (1,5 vs.
9d
9e
49:1
b
0.67:1
CH3
1
,6 closure) of radicals 9a, 9b, and 9c are summarized in
Table 2, along with comparative data for other 5-hexenyl
radical cyclizations. These data show that the regio-
selectivity of 4-oxa-substituted radicals 9a, 9b, and 9c do
not significantly deviate from the trends reported for the
all carbon analogs 9d, 9e, and 9f. In all these cases, the
CH3
c
9
f
>24:1
a
Cyclizations performed at 40 8C except for 9f (80 8C).
See ref. 8.
See ref. 18.
b
c
Table 1. Regiochemistry of photocycloaddition of dienones
1: 1,5 (4) versus 1,6 closure (5)
Dienone
4:5
oxa-substituted systems show a slightly greater prefer-
ence for 1,5 closure than the carbon systems. Radicals 9a
and 9c undergo 1,5 cyclization to give, respectively,
tetrahydrofuran products 7a and 7c, with only trace
amounts of the 1,6 closure products tetrahydropyrans 8a
and 8c. Methyl substitution at C-5 promotes 1,6 closure in
both radicals 9b and 9e. The oxa system 9b gives 40% 1,6
cyclization (8b) and the methylene analog 9e gives 60%
O
a
a
1
1
a
O
1.6:1
O
b
1.1:1
O
1,6 closure (8e). The generally accepted mechanism for
tri-n-butyl radical generated hexenyl radical cyclizations
CH3
8
is shown in Scheme 2. Previous studies have shown that
CH3
the observed product composition is determined by the
initial mode of cyclization of the radical 9, since the
intermediate cyclized radicals 10 and 11 do not revert to
O
a
1
c
e
Only 4c
O
9
–11
the acyclic radicals under the reaction conditions.
Thus, despite the fact that 4-oxa substitution greatly
perturbs the electronic nature of the double bond being
attacked, there is no significant effect on the partitioning
of radical 9 to cyclize to either radical 10 or 11. This result
reinforces the computational analysis which shows that
radical addition reactions have early transition states and
that the regioselectivity can be largely attributed to steric
O
b
1
Only 4e
CH3
a
See ref. 6.
See ref. 7.
b
Copyright # 2007 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2007; 20: 83–87
DOI: 10.1002/poc

RULE OF FIVE CYCLIZATIONS
Table 3. Relative cyclization rate constants
85
a
R1
X
Radical
Temperature (8C)
k1/k2
k1/k3
R2
9
O
9
a
40
697
0.030
k , Bu SnH
k1
k2
3
3
6
9
5
0
546
327
0.047
0.062
R1
R1
R1
CH3
X
X
X
R2
9
c
O
40
1214
0.183
R2
Bu SnH
R2
1
0
11
12
Bu SnH
3
3
6
3
954
679
0.216
0.276
9
0
R1
R1
a
Rate constants are ꢁ10%.
X
X
R2
CH3
R2
7
8
capture of 9 by Bu SnH is not significantly different than
3
8
,10
the value determined for the n-hexyl radical.
values are listed in Table 4. 4-Oxa substitution retards the
The
a: R = R = H, X = O
d: R = R = H, X = CH
1 2 2
1
2
rate of cyclization of the unsubstituted system 9a by
3
b: R = H, R = CH , X = O
e: R = H, R = CH , X = CH
8
.7-fold, relative to the carbon analog 9d. The reduced
1
2
3
1
2
3
2
c: R = CH , R = H, X = O
f: R = CH , R = H, X = CH
1
3
2
1 3 2 2
rate of the radical addition to an electron-rich vinyl ether
has also been seen in intermolecular reactions of alkyl
radicals to vinyl ethers, and is consistent with attributing
Scheme 2
1
5
some nucleophilic character to simple alkyl radicals.
1
2
factors.
Houk and Beckwith have both reported
This effect may also explain the slight increased
preference for 1,5- cyclization seen with 4-oxa substi-
tution, where the nucleophilic radical prefers to add to the
less electron-rich end of the double bond. Methyl
substitution at C-3 (9c) enhances the rate of cyclization
six-fold at 40 8C and 4.5-fold at 90 8C, relative to 9a. The
effect of substitution, remote from the reacting centers is
most likely the result of conformational factors, where
computational models which attribute the change in the
regiochemistry of the cyclization with 5-methyl substi-
tution to steric factors which destabilize the chair-like
conformation, leading to 1,5 cyclization and permit the
,6-cyclization mode to become competitive
Kinetic analyses of the radical cyclizations provide
16
13,14
1
.
some insight to the effect of remote methyl substitution at
C-3. Kinetic studies on the cyclization of bromides 6a and
6
6
c (Fig. 1) were performed at three temperatures (40 8,
5 8, and 90 8C). The data analysis was based on the
Table 4. Estimated absolute rate constants k1
accepted mechanistic scheme (Scheme 2). The relative
rate constants are listed in Table 3 and were determined
by fitting the data to the integrated rate equation reported
by Smith and Butler for the 3-oxa-5-hexenyl radical.
Following previous studies, estimates of the absolute rates
were made by assuming that the rate of bimolecular
ꢂ1 a
k1 (s )
Radical
10
O
4
5
9
9
a
c
3.0 ꢃ 10
1.8 ꢃ 10
CH3
R1
R1
R1
O
O
AIBN (cat)
O
O
+
R2
Bu SnH,
3
Br
R2
R2
CH3
6
7
8
5 b
9
d
1.1 ꢃ 10
a: R = R = H
1
2
b: R = H, R = CH
1
2
3
c: R = CH , R = H
1
3
2
a
408C.
See ref. 8.
b
Figure 1
Copyright # 2007 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2007; 20: 83–87
DOI: 10.1002/poc

8
6
A. R. MATLIN ET AL.
C-3 substitution increases the energy of the ground state
extended conformation but has little effect on the
transition state. This results in a decrease in the activation
energy of the cyclization reaction (’gem-dialkyl
electivity of the cyclization (1,5 vs. 1,6) is largely
unchanged compared to the all carbon system, the
electronic perturbation caused by oxa substitution
decreases the rate of cyclization. These results are
consistent with the view that radical cyclizations have
17
effect’ ). A similar effect was noted by Beckwith
where 9f cyclized 3.2-fold faster at 80 8C than the parent
system 9d.
12
early transition states and that the regioselectivity is
largely determined by geometric constraints, imposed by
18
13,14
The 4-oxa-5-hexenyl radical does not provide a direct
model to explain the regiochemistry observed in the
intramolecular photocycloadditions of 2-acyl-4-oxa-1,5-
hexadienes 1a,b,c. The significant amount of 1,6-closure
product observed for 1a does not correlate with the radical
system (9a). However, the radical model may give some
insight into the course of the photocycloadditions.
Mechanistically, these two reactions are significantly
different. In the radical reactions, the product distri-
butions are governed by the irreversible kinetically
controlled cyclizations to radicals 10 and 11. The
regiochemistry of the cyclization displays little sensitivity
to electronic effects, but is highly sensitive to non-bonded
interactions at C-3 and C-5. In contrast, the formation of
products in the photocycloaddition reactions are much
more complex and depend on the partitioning of the initial
mode of cyclization of the dienone excited state to
biradicals 2 and 3, coupled with the relative rates of
reversion to starting material versus closure/fragmenta-
tion to products by the transient biradicals. In addition,
the photochemical reaction is further complicated by the
various possible spin multiplicities (singlet and triplet) of
the excited dienone and biradical intermediates, along the
reaction pathway. If the radical system can be used as a
guide, we would expect the regioselectivity of the initial
bond formation in the photocycloaddition of 1 would be
largely unaffected by oxa substitution. The dramatic
increase in 1,6-closure products observed for 1a and 1b,
relative to the analogous carbon systems, would then arise
by an increase in the rate of fragmentation of 1,4-biradical
the competing transition states.
Although the radical model does not offer a direct
parallel to the regioselectivity observed for the photo-
cyclizations of the 4-oxa-1,5-hexadienes, the radical
model does offer mechanistic insights to the more
complex photochemical reaction and suggests that the
regioselectivity in the photocyclization may be governed
by the reversion rates of the biradical intermediates. In
addition, the radical model reinforces the interpretation
that ground state conformational effects, engendered by
modest substitutions, remote from the reacting centers
have important consequences on the rates of cyclization
reactions.
EXPERIMENTAL
General
NMR spectra were obtained on a Bruker NR200 MHz
spectrometer in CDCl . Analytical gas chromatography
3
(GC) was carried out on a Hewlett-Packard 5890A
chromatograph, equipped with a 12 m ꢃ 0.2 mm OV-101
capillary column. Mass spectra were recorded on an
Hewlett-Packard 5970 MS coupled to an Hewlett-Packard
5890 GC.
Azoisobutyronitrile (AIBN) was recrystallized from
methanol and stored under argon. Anhydrous decalin (cis,
trans mixture) was used without further purification.
Tri-n-butyltin hydride was distilled at reduced pressure,
prior to use. Ethyl vinyl ether and 2-methoxypropene
were distilled over sodium metal, prior to use.
—
intermediate 3 (X—O) to the photo-Claisen product 5.
Formation of a carbonyl group (vs. an alkene) promotes
this fragmentation pathway. The radical model shows that
C-3 methyl substitution increases the overall rate of
cyclization and promotes 1,5 over 1,6 closure. This
suggests that the effect of C-3 substitution in the
photochemical cyclization of 1c is to facilitate the
preferential formation of biradical intermediate 2c over
General synthesis of bromo-vinyl ethers (6). The
appropriate bromo-alcohol was mixed with a large excess
(ꢀ30ꢃ) of either ethyl vinyl ether or 2-methoxypropene
in an oven-dried 100 ml round bottom flask. A catalytic
amount of mercuric acetate (ꢀ0.1 equivalents) was added
and the reaction was heated at reflux for 24 h in a nitrogen
atmosphere. The reaction mixture was diluted with ether,
washed with sodium bicarbonate and brine solutions, and
dried over sodium sulfate. The ether was removed by
simple distillation. The product was purified by flash
chromatography (silica gel, 2% ether, pentane), followed
by bulb-to-bulb distillation at reduced pressure. The pure
bromo-vinyl ethers were stored over K CO under N and
3
c. The less likely alternative explanations require that
C-3 substitution increases the relative rates of reversion of
biradical 3c, compared with fragmentation to photo-
product 5c and/or C-3 substitution increases the relative
rates of collapse of biradical 2c to photoproduct 4c,
compared with reversion to starting material.
2
3
2
CONCLUSIONS
stored in the freezer.
4-Oxa-substituted 5-hexenyl radicals cyclize to give
tetrahydrofurans and tetrahydropyrans. While the regios-
1-Bromo-4-oxa-5-hexene (6a). 3-Bromo-1-propanol
(2.20 g, 15.8 mmol) was allowed to react with 50 ml of
Copyright # 2007 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2007; 20: 83–87
DOI: 10.1002/poc

RULE OF FIVE CYCLIZATIONS
87
ethyl vinyl ether and mercuric acetate (0.51 g, 1.6 mmol).
The bromo-vinyl ether product (1.94 g, 11.8 mmol, 75%
yield) distilled at 60–80 8C at 28 torr. H NMR: 6.45 (1 H,
equation (the integrated rate equation was essentially
taken from ref. 10 after correcting for the typographical
error that omitted the minus sign).
1
dd, J ¼ 14.3, 6.9 Hz), 4.19 (1 H, dd, J ¼ 14.3, 2.2 Hz), 4.01
k þ k
k þ k þ k ½Bu SnHꢄ
1
2
1
2
3
3
I
(
(
1 H, dd, J ¼ 6.8, 2.2 Hz), 3.81 (2 H, t, J ¼ 5.9 Hz), 3.50
2 H, t, J ¼ 6.5 Hz), 2.18 (2 H, quintet, J ¼ 6.2 Hz).
½
12ꢄ ¼ ½Bu SnHꢄ ꢂ
ln
F
3
I
k
k þ k
3
1
2
1
3
C NMR: 151.6, 86.8, 65.2, 32.1, 30.0. GC–MS: 164
Mþ), 166 (M þ 2).
where [12] is the final concentration and [Bu SnH] is
F 3 I
(
the initial concentration. This analysis allowed determi-
nation of the values of k /k and k /k or k /k by using the
substituted equation:
1
2
1
3
2 3
1
-Bromo-5-methyl-4-oxa-5-hexene (6b). 3-Bromo-
1
5
1
-propanol (2.52 g, 18 mmol) was allowed to react with
0 ml of 2-methoxypropene and mercuric acetate (0.60 g,
.8 mmol). The bromo-vinyl ether product (0.86 g,
.8 mmol, 27% yield) distilled at 70–80 8C at 30 torr.
1
þ ½Bu SnHꢄ
3
I
½
12ꢄ ¼ ½Bu SnHꢄ ꢂ xðr þ 1Þ ln
F
3
I
xðr þ 1Þ
4
where x ¼ k /k and r ¼ k /k ¼ [7]/[8]. For an exper-
2
3
1 2
1
H NMR: 3.84 (2 H, brs), 3.77 (2 H, t, J ¼ 5.8 Hz), 3.51
imentally determined value of r, x was varied until the
value for [12] calculated matched the value of [12]
observed.
(
(
2 H, t, J ¼ 6.5 Hz), 2.2 (2 H, quintet, J ¼ 6.2 Hz), 1.8
F
F
13
3 H, s). C: NMR: 159.5, 81.4, 64.5, 32.1, 30.1, 20.8.
GC–MS: 178 (Mþ), 180 (M þ 2).
Acknowledgements
1
-Bromo-3-methyl-4-oxa-5-hexene (6c). 1-Bromo-
3
6
3
-butanol (4.83 g, 31.6 mmol) was allowed to react with
0 ml of ethyl vinyl ether and mercuric acetate (1.2 g,
.8 mmol). The bromo-vinyl ether product (1.53 g,
.6 mmol, 27% yield) distilled at 65–75 8C at 30 torr.
We thank M. A. Mehta for helpful discussions. We
acknowledge is made to the donors of the Petroleum
Research Fund (grant 20612-B4), administered by the
American Chemical Society, and Oberlin College for
financial support.
8
1
H NMR: 6.63 (1 H, dd, J ¼ 14.2, 6.6 Hz), 4.28 (1 H, dd,
J ¼ 14.2, 1.2 Hz), 4.07 (1 H, m), 4.0 (1 H, dd, J ¼ 6.6,
1
.6 Hz), 3.45 (2 H, t, J ¼ 6.1 Hz), 2.0 (2 H, m), 1.2 (3 H, d,
REFERENCES
13
J ¼ 6.2 Hz). C NMR: 150.7, 88.6, 73.4, 39.5, 29.7, 19.6.
GC–MS: 178 (Mþ), 180 (M þ 2).
1. Schuster DI. In CRC Handbook of Organic Photochemistry and
Photobiology, (2nd edn), Horspool W, Lenci F (eds). CRC Press:
USA 2004.
KINETICS AND PRODUCT STUDIES
2
. Newcomb M, Choi S, Horner JH. J. Org. Chem. 1999; 61:
225–1231.
. Crimmins MT. Chem. Rev. 1988; 88: 1453–1473.
1
3
The kinetics studies were modeled after the studies by
8,10
4. Della EW, Kostakos C, Smith PA. Org. Lett. 1999; 1: 363–365.
. Wolff S, Agosta WC. J. Org. Chem. 1980; 45: 3139–3141.
6. Matlin AR, Lectka TC, McGarvey DJ, Jacob P, Picken HA.
Butler and Walling.
vinyl ethers 6 (80–350 mM), tri-n-butyl tin hydride
30–100 mM), AIBN (0.6–3.2 mM), and an internal
Decalin solutions of the bromo-
5
Tetrahedron Lett. 1987; 28: 5083–5086.
. Wolff S, Agosta WC. J. Am. Chem. Soc. 1983; 105: 1299–1304.
. Walling C, Cioffari A. J. Am. Chem. Soc. 1972; 94: 6059–6064.
(
7
8
standard (octane) were prepared and sealed in Pyrex
tubes, following degassing by three freeze-pump-thaw
cycles. The ratio of bromo-vinyl ether to tin hydride was
9. Lamb RC, Ayers PW, Toney MK. J. Am. Chem. Soc. 1963; 85:
483–3486.
3
1
1
0. Smith TW, Butler GB. J. Org. Chem. 1978; 43: 6–13.
1. However radicals with electron withdrawing substituents on the
radical center in 9 undergo reversible cyclization. See: Julia M,
Acc. Chem. Res. 1971; 4: 386–392.
ꢀ
3:1 in each sample. Three concentrations were
examined at each temperature studied and duplicate
tubes of each concentration were prepared. The samples
were heated at the desired temperatures until the reactions
consumed all of the hydride and then analyzed by
capillary GC. Each tube was analyzed at least twice and
the results were averaged. Products were identified by
comparison to authentic samples and GC–MS. The
combined chemical yields (GC) of products 7, 8, and 12
1
2. Houk KN, Paddon-Row MN, Spellmeyer DC, Rondan N, Nagase
S. J. Org. Chem. 1986; 51: 2874–2879.
13. Spellmeyer DC, Houk KN. J. Org. Chem. 1987; 52: 959–973.
1
4. Beckwith ALJ, Schiesser CH. Tetrahedron Lett. 1985; 26:
73–376.
5. Geise B. Angew. Chem. Int. Ed. Engl. 1983; 22: 753–764.
3
1
16. However, it should be noted that bimolecular additions of radicals
to substituted alkenes occurs at the less substituted end of the
alkene independent of the electronic nature of the substituent on
the alkene. See: Tedder JM, Walton JC, Tetrahedron 1980; 36:
701–707.
(
based on the initial hydride concentration), where
8
9
4–91% for radical 9a, 80–90% for radical 9b, and
5–100% for radical 9c. Response factors were deter-
1
7. Eliel EL, Allinger NL, Angyal SJ, Morrison GA. Conformational
Analysis. Wiley: New York, 1965; 191–192.
8. Beckwith ALJ, Lawrence T, Serelis AK. Chem. Commun. 1980:
11: 484–485.
mined for all products with reference to the internal
standard. The data were fitted to the integrated rate
1
Copyright # 2007 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2007; 20: 83–87
DOI: 10.1002/poc