Electrostatic repulsion by the charged tail of a radical controls the stereochemistry of coupling with anthracenide. Reversibility of benzylic fragmentation

Article abstract of DOI:10.1039/a705248k

Reaction of anthracenide A^.- with 1-pivaloyl-2,3-diphenylaziridines cis-4a and trans-4a yield the same products, namely PhCH₂CHPhNHCOCMe₃ (11) and AH-CHPhCHPhNHCOCMe₃ (6) (AH = dihydroanthryl). The steric differentiation is lost when the two ketyls 12 formed from 4a undergo homolytic ring cleavage forming the same anionic radical 13. Extremely short reactions (≤10 s) give 6 as the erythro isomer exclusively or nearly so. Coupling of 13 with A^.- does not form the precursor (threo-8) of threo-6 owing to electrostatic repulsion between A^.- and the anionic tail of 13 in the preferred conformation of the latter. Radical coupling is not completed within this short time so 11 can be formed directly from 13 via 10, the amide anion of 11. Reduction of 13 to carbanion 9 by outer-sphere electron transfer or via 8 and its benzylic fragmentation (BFR) is the other path to 11. Extending the time to 1 or 2 min has the following effects. Coupling of 13 with A^.- is completed at the expense of 11. Second, more than a trace of threo-6 is detected indicating that BFR 8→9 + A (A = anthracene) is reversible and that addition of dianion 9 to A proceeds without pronounced stereochemical preference. With even more time the erythro:threo ratio changes in favour of threo-6 and finally can even reach a value slightly less than 1. Simultaneously the amount of 11 increases slowly at the expense of the total 6 indicating that part of BFR which becomes irreversible by carbanion protonation 9 + THF→10. With much longer reaction times imidate ion 10 eliminates (E)-stilbene. Both isomers of 6 have been independently synthesized from the two isomers of 4a and anthracene hydride AH^-.

Full text of DOI:10.1039/a705248k

View Article Online / Journal Homepage / Table of Contents for this issue
Electrostatic repulsion by the charged tail of a radical controls the
stereochemistry of coupling with anthracenide. Reversibility of
benzylic fragmentation1,2
Thomas Mall and Helmut Stamm*
Faculty of Pharmacy, University of Heidelberg, Neuenheimer Feld 346, D-69120 Heidelberg,
Germany
Reaction of anthracenide A Ϫ with 1-pivaloyl-2,3-diphenylaziridines cis-4a and trans-4a yield the same
᝽
products, namely PhCH2CHPhNHCOCMe3 (11) and AH᎐CHPhCHPhNHCOCMe3 (6)
(AH = dihydroanthryl). The steric differentiation is lost when the two ketyls 12 formed from 4a undergo
homolytic ring cleavage forming the same anionic radical 13. Extremely short reactions (р10 s) give 6 as
the erythro isomer exclusively or nearly so. Coupling of 13 with A Ϫ does not form the precursor (threo-8)
᝽
of threo-6 owing to electrostatic repulsion between A Ϫ and the anionic tail of 13 in the preferred
᝽
conformation of the latter. Radical coupling is not completed within this short time so 11 can be formed
directly from 13 via 10, the amide anion of 11. Reduction of 13 to carbanion 9 by outer-sphere electron
transfer or via 8 and its benzylic fragmentation (BFR) is the other path to 11. Extending the time to 1 or 2
min has the following effects. Coupling of 13 with A Ϫ is completed at the expense of 11. Second, more
᝽
than a trace of threo-6 is detected indicating that BFR 8→9 ϩ A (A = anthracene) is reversible and that
addition of dianion 9 to A proceeds without pronounced stereochemical preference. With even more time
the erythro:threo ratio changes in favour of threo-6 and finally can even reach a value slightly less than 1.
Simultaneously the amount of 11 increases slowly at the expense of the total 6 indicating that part of BFR
which becomes irreversible by carbanion protonation 9 ϩ THF→10. With much longer reaction times
imidate ion 10 eliminates (E)-stilbene. Both isomers of 6 have been independently synthesized from the
two isomers of 4a and anthracene hydride AHϪ.
been shown to be discrete steps.4,5 The intermediate carbanion 2
Introduction
(R1 = Ph) resembles the intermediate 2 (R1, R2, R3 ≠ Ar) formed
The base induced fragmentations (FR) of 9-substituted 9,10-
dihydroanthracenes 1 (Scheme 1) are observed when the leaving
Ϫ
in reactions of anthracenide A
with alkyl halides Hal-
᝽
CR1R2R3 or similar reagents.6 While 1 was the main product in
the reactions of A Ϫ, so far 1 with R1 = Ar has been found in
᝽
low yields only twice. Beckwith and Waters7 obtained 1
2
–
•
(R1 = Ph, R2 = R3 = H) from A and benzyl chloride in ether
Ϫ
᝽
together with other products including PhCH2CH2Ph whose
formation may now be interpreted to result from BFR followed
by SN2 reaction of the benzyl anion with benzyl chloride. This
would be the first BFR although it was unrecognized as such. In
–
A•
HalCR1R2R3
the second case5 the short lifetime of 2 obtained from A Ϫ and
–H+
+H+
–
᝽
4c in THF was attributed to BFR followed by typical reactions
of the benzyl anion 3.
CR1R2R3
CR1R2R3
The classic papers on reactions of aromatic radical anions
have not considered the possibility of BFR although they
revealed a different behaviour of alkyl and benzyl halides in
reactions with naphthalenide.8 Benzylic halides never yielded
substituted naphthalenes or dihydronaphthalenes. This may
also be explained by BFR of a naphthalene analogue of 2. A
recent paper9 described a homolytic BFR in the reaction of
naphthalenide with a benzoylaziridine: benzoylnaphthalenes
were obtained after a very short reaction time only. Thus, BFR
is probably a general phenomenon, but is most important in the
anthracene series for obvious reasons.
The benzyl anion 3 arises in the presence of anthracene A
and may attack it resulting in a reverse BFR. This has not yet
been observed but may be expected since addition of alkyl-
lithium species to A has been described.10 Search for direct
evidence of this reversibility and the possibility of diastereo-
selectivity (see below) stimulated a study of reactions with the
cis–trans pair of 4a. The pivaloylaziridines 4a were selected
rather than the benzoylaziridines 4b in order to avoid compli-
cating carbonyl reactions and to shorten the lifetime of the
ketyl. In spite of the very unfavourable reduction potentials of
A (Ϫ1.98 V)11 and of a pivaloylaziridine (about Ϫ2.7 V),12,13
single election transfer (SET) to a pivaloylaziridine has previ-
1
2
R1 = Ph
+
–CR1R2R3
3
A
AH2
Ph
–
NCOR2
R1
AH–
4a R1 = Ph; R1 = CMe3
4b R1 = R2 = Ph
4c R1 = H; R2 = Ph
Scheme 1
carbanion 3 (heterolytic FR) or ketyl 3 (R2 = OϪ, homolytic
FR) is stabilized. Stabilization is usually due to the phenyl
group (R1 = Ph) giving the leaving species a benzyl structure
and providing the practical term ‘benzylic FR’ (BFR) when this
reaction was detected and studied at Heidelberg. Deproton-
ation at position 10 and elimination of the benzyl species have
J. Chem. Soc., Perkin Trans. 2, 1997
2135

View Article Online
Table 1 Reactionsa of trans-4a with AHϪ
Reagents (mmol/cm3 THF)
Yieldsb of products
Run
AH2
BuLi
trans-4a
Time
erythro-6
11
1
2
3
4
6.25/70
7.5/70
7.5/70
30/150
5
5
5
4.8/40
1/20
1.26/25
4.96/50
5 min
30 min
17 h
82
(53)
(28)
100
(31)
25
5 min
(57)
a The solution of AH2 in THF was cooled to Ϫ160 ЊC and BuLi (in hexane) was added; stirring was begun as soon as possible and continued until
quenching; the solution of 4a in THF was added dropwise within 3 min (4 min in run 1, 5 min in run 4) at room temperature. The reactions were
quenched with acetic acid. b Yields in parentheses were calculated from 1H NMR spectra of product mixtures.
ously been observed.13 The general problem of unfavourable
potentials in SET reactions has already been discussed by Bank
and Juckett,11 but with acylaziridines an essential aspect may be
the interaction of the counter ion with the acylaziridine before,
during and after SET.
The possibility of diastereoselectivity was considered since 1
with an enantiomeric excess has been obtained from optically
active alkylating agents.6
cmϪ1). The SN2-reactivity of acylaziridines decreases when the
steepness of the nitrogen pyramid increases.15 This is borne out
by the reactions of cis-4a with AHϪ (Table 2). No run in Table 2
produced exclusively threo-6 although this was the main prod-
uct in runs 1–3. Keeping the excess of AHϪ small (run 1 and 2)
did not completely prevent the formation of 11 via BFR and
also provided the ketone 7. The attack on the carbonyl group of
4a, despite severe steric hindrance, was possible because the
steep pyramid in cis-4a not only slows down the competing
ring-opening, it also makes the carbonyl group susceptible to
nucleophilic attack. Such carbonyl attack is reversible and this
explains the lack of ketone 7 in the other runs where 4a was
consumed by the competing reaction. The equilibrium concen-
tration of the carbonyl adduct vanishes when 4a disappears.
Run 5 illustrates the time-dependence of BFR.
Results and discussion
SN2-like ring-opening of cis-4a and trans-4a by AHϪ appeared
to be a simple way to prepare pure samples of the two diastereo-
meric dihydroanthracenes 6 required as authentic material
BFR generates the carbanion-imidate ion 9. Protonation of
the carbanionic site by AH2 was confirmed by experiments with
–
2
[2H4]AH2 (87–88% H in each non-aromatic position). From
AH–
both isomers of 4a, 11 was obtained containing (1H NMR) 82%
of [2H]11 (PhCHDCH2NHCOCMe3).
erythro-6a and threo-6a can easily be differentiated by H
+
Ph
Ph
1
Ph
Ph
O
Ph
NMR spectroscopy. Strong ring current effects in the crowded
molecules generate remarkable shift differences for all non-
aromatic signals except for the pseudo-equatorial H in position
10 of the dihydroanthracene. In particular, the strong singlets
for But are valuable: 0.71 vs. 1.33 ppm (1.09 ppm for 11). Even
the aromatic ortho-signals of NCCPh were in the olefin region
and differed by 0.3 ppm. Thus, expected unseparable mixtures
N
O–
NH
Ph
erythro-5
erythro-6
N
O
trans-4a
cis-4a
of products in reactions with A Ϫ can be analyzed by 1H NMR
᝽
7
O
spectroscopy.
Ϫ
A reaction between A and 4a can only occur by single
᝽
electron transfer. For an example6 of SET with simultaneous
+
A–
threo-5
threo-6
seemingly SN2-like reactions of A Ϫ, see below. SET generates
᝽
the two ketyls cis-12 and trans-12 (Scheme 3) which rapidly
form the same amidatoalkyl radical 13. This could be anti-
–
cipated. The reaction of 13 with excess A Ϫ as well as follow-up
–H+
(AH–)
erythro-5
threo-5
᝽
+
reactions have been investigated.
Ph
–
Ph
Ph
The first experiments were performed under conditions that
leave as little time as possible for the expected BFR. The sur-
prising results of these extremely short reactions are shown in
Table 3 and Scheme 3. No difference in products from either
aziridine had been expected and was supported experimentally.
The main product was always 11 but 6 was not obtained as a
diastereomeric mixture since only (or nearly only)† erythro-6
was found. It is unlikely that about 65% of 11 in these very
short reactions had been formed by BFR of an intermediate
coupling product erythro-8 or threo-8. Further results described
below generally rule out carbanion 8 as the main precursor of
11 in the runs of Table 3. Considering the strong tendency of
Ph
N
O–
N
O–
erythro-8
threo-8
9
Ph
O–
Ph
O
H+
(AH2)
H+
(RCO2H)
9
N
NH
Ph
Ph
10
11
Scheme 2
excess A Ϫ to couple with a carbon radical rather than to reduce
(Scheme 2). Synthesis of erythro-6 from trans-4a (Table 1) was
easy with a small excess of AHϪ and a short reaction time (run
1). Runs 2–4, carried out with a large excess of AHϪ, viz. 5:1,
show the expected BFR, yielding 11, and its dependence on
time and on the concentration of AHϪ. Both effects counteract
in runs 2 and 4 to give nearly the same result.
᝽
it to a carbanion,6,13 it appears reasonable to assume that a
† It is likely that traces of threo-6 arose from a small impurity in 4a. In
the last step of its synthesis 4a may be ring-opened by chloride ion
causing contamination by PhCHClCHPhNHCOCMe3. SET cleavage
of the C᎐Cl bond without deprotonation would provide an uncharged
radical, i.e. 13 protonated on nitrogen.
The IR carbonyl bands14 confirm the expectation that cis-4a
(1683 cmϪ1) has a steeper nitrogen pyramid than trans-4a (1664
2136
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Table 2 Reactionsa of cis-4a with AHϪ
Reagents (mmol/cm3 THF)
Yieldsb of products
Run
AH2
7.5/70
7.5/70
7.5/70
7.5/70
30/150
BuLi
cis-4a
Time
threo-6
11
(3)
(7)
(33)
(42)
(51)
(63)
98
7
cis-4a
1
2
6.25
6.25
5
5
25
4.73/20
4.9/30
1/20
1.37/20
4.93/50
5 min
5 h
45 min
1 d
30 min
1 h
27 h
(49)
(55)
60
(26)
(22)
(14)
20
22
21
3
4 c
5 d
a The solution of AH2 in THF was cooled to Ϫ160 ЊC and BuLi (in hexane) was added; stirring was begun as soon as possible and continued until
quenching; the solution of 4a in THF was added dropwise within 5 min (1 min in run 1, 3 min in run 2) at room temperature. The reactions were
quenched with acetic acid. b Yields in parentheses were calculated from 1H NMR spectra of product mixtures. c About 15% of products were lost
during workup. d 20 cm3 samples were withdrawn after the time given and worked up.
Ϫ
Table 3 Short-term reactionsa of cis-4a and trans-4a with A
᝽
Reagents (mmol/cm3 THF)
Yields (1H NMR) of products
Time of
Run
A
Na
4a
addition/s
erythro-6
threo-6
11
1
2
7/100
7/100
6/100
6/100
5.0
5.0
4.4
4.7
trans 1.0/20
cis 1.0/20
cis 1.87/20
cis 1.65/20
7
5
7
26
18
0
65
68
Trace
0
Trace
3
19
66
4 b
10
22c
49d
a A, Na and THF were stirred for about 1 day. The solution of 4 in THF was added within the time given. The reactions were immediately quenched
with excess acetic acid before air was allowed to enter. b Reaction was quenched with CF3CO2D. c 67% deuterium incorporated in pseudo-axial
position 10 of the dihydroanthracene. d 63% of 11 contained deuterium as indicated by NCCHDPh.
H
Ph
Ph
Ph
Ph
Ph
Ph
•–
O–
O–
–
•
–
O
•
N
Ph
N
N
O–
•
A•–
•
•
C
Ph
N
4a
12
13
cis/trans
cis/trans
H
–
2 H+
Ph
A•–
Ph
13
α
•
O–
β
N
C
Ph
Ph
Ph
Ph
13
N
N
Scheme 4
O–
O
lobe of the singly occupied π orbital can only proceed from
either side of the plane given by Ph᎐αC᎐βC᎐H. Only a little
steric disparity of these two sides may arise from the ‘radical
tail’ βCHPh (imidate). However, this tail must create a very
erythro-6
erythro-8
A + 9
pronounced electrostatic disparity. Formation of the threo
threo-8
threo-6
erythro-8
Ϫ
product from coupling of 13 with A will be hindered by
᝽
charge repulsion. The observed stereoselectivity results clearly
from a SET reaction of A Ϫ. In contrast, the reported6 small to
erythro-6
᝽
2 H+
moderate stereoselectivity probably does not come from a reac-
11
9
tion of A Ϫ itself. An SN2 reaction of its dimer or rather of one
᝽
Scheme 3
carbanion site of this dimer followed by BFR would easily
explain the excess of Walden inversion in reactions with optic-
ally active alkylating agents. A similar explanation is not pos-
sible for the present results.
substantial part of the non-coupling amidatoalkyl radical 13
survived until quenching or until workup. The deuterium
incorporation (63%) into 11 (49%) in run 4 by quenching with
CF3CO2D points to dianion 9 as the precursor for 31% of 11
Thus, coupling of a charged radical‡ with a radical anion can
be controlled by an electrostatic effect exerted by the ‘tail’ of
the charged radical. This novel stereoselectivity principle
should be more effective and more useful when the reacting
and hence to the partial reduction of radical 13 by A Ϫ or, more
᝽
likely,13 by ketyls 12. Tentatively, one may also consider deuter-
ium incorporation by an internal deuterium atom transfer with
a cyclic six-membered transition state after the imidate part of
13 has accepted a deuteron on the oxygen. Protonation of
amide anions occurs first on oxygen.16 The drawback of this
anionic radical is not a benzylic radical whose coupling with
Ϫ
A
is followed by BFR. Another drawback of this particular
᝽
case may be an easy reduction of 13 as expected from its
alternative is clear: a rather stable benzyl radical would have to
ؒ
‡ It appears advisable to differentiate between anionic radicals as 13
generate a radical N᎐C᎐O with the unpaired electron in the
᎐
and radical anions as A Ϫ. Chemists usually associate the term
᝽
wrong orbital for resonance stabilization.
radical anion with a species whose charge comes from the singly
occupied MO (cf. e.g. Radical Ions, ed. E. T. Kaiser and L. R.
Kevan, Interscience, New York, 1968, part of the series:
Reactive Intermediates in Organic Chemistry).
How can the very high diastereoselectivity of coupling be
explained? The preferred conformation (Scheme 4) of 13 has
α-phenyl and β-hydrogen eclipsed. Radical coupling with either
J. Chem. Soc., Perkin Trans. 2, 1997
2137

View Article Online
Ϫ
Table 4 Product ratios erythro-6:threo-6:11:(E)-stilbene from reactions of cis-4a and trans-4a with A
᝽
Reagents (mmol/cm3 THF)a
Molar ratios of products
(see text)
Run
1
A
Na
4a
Time
17.5/100
15.0
cis 3.0/40
2 min
5 min
30 min
1 h
54:9:37:—
46:17:38:—
36:21:42:—
34:20:46:—
7:6:40:43
6.5 h
2
3
4
5
15.0/150
12.0/150
9.0/100
7.0/100
13.5
8.8
6.0
3.5
cis 4.99/50
cis 3.96/50
trans 3.11/40
cis 2.92/40
2 min
30 min
3 h
6 h
10 h
1 min
5 min
10 min
20 min
30 min
2 min
15 min
30 min
1 h
66:8:26:—
50:20:30:—
31:17:35:17
17:10:37:36
7:5:43:45
39:7:54:—
34:8:58:—
19:9:72:—
6:10:84:—
6:8:87:—
51:5:44:—
40:8:52:—
30:10:60:—
28:12:60:—
0:0:0:100 (93%)
15:6:78:—
7:10:83:—
4:8:88:—
10 h
1 min
5 min
30 min
a A in THF and Na were stirred for 1 d. Stirring was continued. The solution of 4a in THF was added from a dropping funnel within 10–15 s. The
reactions were quenched with acetic acid after sampling.
structure and the special electron source 12. A third unfavour-
able factor is possibly a relatively slow formation of erythro-8.
Coupling of an aromatic radical anion with an alkyl radical
is known to be a very fast reaction.11 In particular, coupling
was faster than any other reaction of the intermediate
amidatoalkyl radical, a substituted But radical, in the reaction
run 1 of Table 3, provided erythro-6 as the main product after 2
min, in accord with an incomplete coupling of 13 in the runs
given in Table 3. Detection of р9% of threo-6 after 2 min is
evidence for a reversible BFR of erythro-8. This reverse BFR
(Scheme 3, bottom) can be discussed with the help of Scheme 4
when one simply replaces A Ϫ by A and radical 13 by carbanion
᝽
of 2,2-dimethyl-1-pivaloylaziridine with A Ϫ where no counter-
9. An electrostatic repulsion that hinders formation of threo-8
does not exist between carbanion 9 and neutral A. One even
may suspect that 9 and A for steric reasons prefer to form threo-
8 since internal electrostatic repulsion will push away the
charged oxygen from the carbanion site (cf. the conformation in
Scheme 4) and thus create or increase the steric disparity that
favours the threo product. Summing up, the reverse BFR should
produce both diastereomers but perhaps more threo than the
erythro product. Indeed, with increasing reaction time, the
erythro–threo ratio for 6 gradually changed from 54:9 to 7:6 in
run 1.
᝽
part of 11 was found under conditions (run 2 of Table 1 in
ref. 13) comparable to those of run 3, the run with the smallest
Ϫ
excess of A in Table 3. As compared to this previous case,
᝽
coupling of 13 may be slowed by steric hindrance arising from
β-phenyl and by the decrease from 2 to 1 in the statistical factor.
Ϫ
Coupling of unreacted anionic radical 13 with available A
,
᝽
as well as BFR of 8, proceeds with reaction time. Thus, several
Ϫ
experiments with varying excesses of A
were conducted
᝽
(Table 4) in a manner that would show the change in products
and/or their relative yields with time. Samples, sufficient in size
for routine workup, of the reaction solution were withdrawn
with a pipette at the times given for each run and worked up.
The products 6 and 11 were chromatographically obtained as
ternary mixtures that were analyzed by 1H NMR spectroscopy
giving the molar ratios of products and approximate values of
yields. When the sum of the approximate yields showed a large
deficit, the hydrocarbon fraction, mainly anthracene, was ana-
lyzed for trans-stilbene by 1H NMR spectroscopy giving a crude
yield or rather an upper yield limit that made the sum of
approximate yields exceed 100%. The amount of stilbene was
then adapted to give a total of 100% making the product ratios
include stilbene when necessary after long-term sampling.
Formation of stilbene should produce the same amount of
pivaloylamide that, however, was obviously lost during workup
owing to volatility and solubility in water. The same problem is
already known from pivaloylamide carrying short alkyl
groups.13 Most runs of Table 4 were performed with cis-4a since
cis-4a is obtained in a pure state much easier than trans-4a.
Summarizing, the experimental precision is not so high as in
the runs of Tables 1–3 but the product ratios listed in Table
4 depend on time and other details in a manner that allows
reasonable interpretations and mechanistic conclusions.
The yield of 11 dropped from about 65% (Table 3) to р37%
in the 2 min sample of run 1. This indicates that in the very
short runs of Table 3 about half of 11 arose directly from rad-
ical 13 without intermediacy of carbanion 9. After 2 min,
Ϫ
radical 13 should have completed its coupling with excess A
᝽
leaving only the carbanion 9 as a precursor of 11. The other
samples of run 1 show that carbanion 9 reacts faster with A
than with THF. This follows from a comparison of erythro–
threo isomerization with the increase in amount of 11 and fits
nicely with the previous finding5 that the benzylic anion formed
from 4c reacted faster with the carbonyl group of available 4c
than with the solvent THF. Thus, the BFR path leading to 9
really contributes little to the formation of 11 in the short-term
samples of run 1. Direct reduction of radical 13 to carbanion 9
by A Ϫ or 12 is probably the main path leading to 11 in run 1 of
᝽
Table 4. The competition between coupling and reduction of 13
as measured by the product quantities deviates from the corre-
Ϫ
sponding competition known13,6 for A and alkyl radicals.
᝽
Reasons for this deviation are found in the above discussions
but a different behaviour of alkyl and benzyl radicals cannot be
excluded. A very slow process, recognizable after 6 h and pos-
sibly at work already after one hour, is the elimination of trans-
stilbene from amide anion 10. This will be discussed below.
Run 1, conducted with the same ratio of reactants (5:1) as in
2138
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Run 2 performed with a reactant ratio of 2.7:1 resembled
run 1 in all ways. In both runs the colour of the solution after
addition of 4a remained dark-green. In run 3, with a reactant
Reactions with anthracene hydride AH؊
These reactions were performed using the previously19
described technique. Necessary details are given in Tables 1 and
2. The residue obtained by evaporation was taken up in dichlo-
romethane and washed with water. Evaporation left a residue,
which was subjected to chromatography (silica gel Merck,
0.063–0.200 mm, thickness × length column in cm and other
details are given with each run). Preparative layer chromato-
graphy (PLC) was performed on plates 20 × 20 cm, 2 mm thick
(Merck 5717, silica gel 60F254).
ratio of 2.2:1, this dark-green colour changed slowly to light
Ϫ
yellow–green indicating an insufficient excess of A which
᝽
explains the low extent of coupling. The molar amount of 11
was always high, higher than that of both isomers of 6 taken
together, whereby the reaction time was obviously not long
enough to observe conversion of 10 to stilbene. Perhaps
favoured by the small amount of erythro-6 formed after 1 min,
later on slightly more threo-6 than erythro-6 was found. This
correlates better with the erythro–threo rate difference of BFR
than with the rate difference in reverse BFR. Generally, there is
good reason to assume that the rate of BFR depends on the
stereochemistry of 8. The steric situation for BFR is different
for erythro- and threo-8. Steric crowding in 2 had even sup-
pressed BFR in another case.17
Run 1, Table 1. Chromatography (3 × 50 cm) with CH2Cl2
removed hydrocarbons. Elution with CH2Cl2–ethyl acetate
(10:1) provided erythro-6 (1.813 g, 82%), mp 193–194 ЊC
(Found: C, 86.0; H, 7.3; N, 3.0. C33H33NO requires C, 86.2; H,
7.2; N, 3.1%); νmax/cmϪ1 3400 (NH), 1648 (amide I) and 1522
(amide II); δ 0.71 (s, But), 2.09 (d, J 19.0, 10-H pseudo ax), 3.18
(d, J 19.0, 10-H pseudo eq), 3.60 (dd, J 2.8 and 11.5, NCCH),
4.18 (s br, 9-H pseudo eq), 5.41 (d, J 8.0, NH), 5.53 (dd, J 8.0
and 11.5, NCH), 6.33–6.36 (m, 2 × o᎐H of NCCPh), 6.94–6.96
(m, 4 × ArH), 7.09–7.28 (m, 6 × ArH), 7.37–7.42 (m, 2 × ArH),
7.51–7.57 (m, 2 × ArH) and 7.70–7.73 (m, 2 × ArH).
Run 4 with trans-4a and a reactant ratio of 1.9:1 showed also
the trends already described but a comparison with run 3
reveals some differences in quantities of products that may be
related to cis–trans isomerism. The rates of steps 4→12 and
12→13 may be different for cis and trans. The first step should
be faster for the cis isomer owing to the steep nitrogen pyramid
that makes cis-4a resemble a ketone while trans-4a has more
amide character. A kind of stereoelectronic effect should
accelerate the second step for trans-12 since generation of a
stabilized benzylic radical demands coplanarity of PhC᎐C,
which is possible only for trans-12. A shorter lifetime of trans-
12 means less time for reduction of 13 explaining the difference
of ratios in runs 4 and 3.
Run 2, Table 1. Chromatography (3 × 17 cm) with toluene
removed hydrocarbons. Elution with ethyl acetate provided a
mixture (322 mg) consisting (1H NMR spectroscopy) of
erythro-6 (242 mg, 53%) and 11 (80 mg, 28%).
Run 3, Table 1. Chromatography (3 × 17 cm) with toluene
removed hydrocarbons. Elution with ethyl acetate yielded 11
(356 mg, 100%), mp 141–142 ЊC (Found: C, 81.2; H, 8.1; N, 5.1.
C19H23NO requires C, 81.1; H, 8.2; N, 5.0%); νmax/cmϪ1 3350
(NH), 1631 (amide I) and 1531 (amide II); δ 1.09 (s, But), 3.07
(dd, J 7.8 and 13.7, 1 H of CH2), 3.12 (dd, J 6.5 and 13.7, 1 H of
CH2), 5.26 (mc, NCH), 5.95 (d, J 7.2, NH), 7.04–7.07 (m,
2 × ArH) and 7.15–7.32 (m, 8 × ArH).
In order to verify the proposed non-carbanion path for 13 to
11, run 5 was performed with a reactant ratio of 1.2:1, i.e.
with practically equimolar quantities. In comparison to the
other runs of Table 4, the formation of both 6 was a minimum
with all samples in accord with the requirement of two equiv-
Run 4, Table 1. Only 20 cm3 (withdrawn with a pipette) of the
solution were quenched and worked up. Chromatography
(3 × 17 cm) with toluene removed hydrocarbons. Elution with
ethyl acetate yielded a mixture (172 mg) consisting (1H NMR
spectroscopy) of erythro-6 (129 mg, 57%), 11 (43 mg, 31%) and
anthraquinone (12 mg).
alents of A Ϫ. The ratio of coupling and non-coupling
᝽
resembles the respective ratios in Table 3. An alternative path
to 6 can be excluded although Beckwith and Waters have des-
cribed the addition of benzyl radical to A, but the observed main
reactions of the arising dihydroanthryl radical were dimeriz-
ation and coupling with a second benzyl radical.7 Neither of
these products were found in run 5. The high amounts of 11 in
run 5 can only have arisen by hydrogen abstraction 13→10,
certainly from THF in these relatively long reaction times.
Elimination of olefines in SET reactions of acylaziridines
have previously been described and possible mechanisms dis-
cussed.18 Direct elimination from a precursor of type 10 was
verified leaving some doubts about whether this is the whole
explanation. The counter ion may have some influence. It was
recently shown4 that the sulfonyl counterpart of 9 with counter
ion Liϩ immediately provided stilbene without any intermediate
proton abstraction from AH2. The results shown in Table 4
exclude cleavage of 9 and 13 and leave the internal proton trans-
fer in 10 as the route to the formation of stilbene under the
present conditions.
Run 1, Table 2. Chromatography (1.5 × 90 cm) with toluene
removed hydrocarbons and then 7 (246 mg, 20%), mp 128–
130 ЊC (Found: C, 86.3; H, 7.6. C19H20O requires C, 86.3; H,
7.6%); νmax/cmϪ1 1703 (C᎐O); δ 1.28 (s, But), 3.81 (d, J 18.0, 10-
᎐
H pseudo eq), 4.75 (d, J 17.9, 10-H pseudo ax), 5.53 (s, 9-H
pseudo eq) and 7.12–7.35 (m, 8 × ArH). Elution with CH2Cl2–
ethyl acetate (10:1) provided cis-4a (287 mg, 21%) and a mix-
ture (1.092 mg) consisting (1H NMR spectroscopy) of threo-6
(1.055 g, 49%) and 11 (37 mg, 3%).
Run 2, Table 2. Chromatography (1.5 × 90 cm) with toluene
removed hydrocarbons and then 7 (290 mg, 22%). Elution with
CH2Cl2–ethyl acetate (10:1) yielded a mixture (1.338 g) consist-
ing (1H NMR spectroscopy) of threo-6 (1.240 g, 55%) and 11
(98 mg, 7%).
Run 3, Table 2. Chromatography (1.5 × 90 cm) with CH2Cl2
removed hydrocarbons and then threo-6 (347 mg, 60%), mp
216–217 ЊC (Found: C, 86.1; H, 7.2; N, 3.0. C33H33NO requires
C, 86.2; H, 7.2; N, 3.1%); νmax/cmϪ1 3400 (NH), 1646 (amide I)
and 1522 (amide II); δH 1.33 (s, But), 1.93 (d, J 18.8, 10-H
pseudo ax), 3.18 (d, J 19.0, 10-H pseudo eq), 3.45 (dd, J 2.2 and
11.7, NCCH), 4.66 (s br, 9-H pseudo eq), 5.66 (dd, J 9.6 and
11.7, NCH), 6.19 (d, J 9.4, NH), 6.04–6.07 (m, 2 o᎐H of
NCCPh), 6.65–6.71 (m, 2 × ArH), 6.86–7.41 (m, 13 × ArH)
and 7.68–7.71 (m, 1 × ArH). Elution with CH2Cl2–ethyl acetate
(1:1) yielded a mixture (141 mg) consisting (1H NMR spec-
troscopy) of 11 (116 mg, 33%) and anthraquinone (25 mg).
Run 4, Table 2. Chromatography (3 × 17 cm) with toluene
removed hydrocarbons. Elution with CH2Cl2–ethyl acetate
(10:1) provided a mixture (395 mg) that was separated by PLC
with CH2Cl2–ethyl acetate (50:1) into threo-6 and a mixture.
Scratching out and eluting with hot ethyl acetate yielded (i)
Experimental
Characterization of products was accomplished by H NMR
1
spectroscopy (Bruker W250 instrument, CDCl3 solution, signal
multiplicity given, mc = multiplet centred at; J values in Hz) and
IR (Perkin-Elmer 283 instrument; KBr tablets unless otherwise
stated).
All reactions were performed in dry, continuously stirred
THF under nitrogen whose quality was secured with a THF
solution of sodium naphthalenide.
Starting materials
Aziridines cis-4a and trans-4a are known.14 [2H4]AH2 was pre-
pared from AH2 and [2H6]DMSO6 as described in ref. 19.
J. Chem. Soc., Perkin Trans. 2, 1997
2139

View Article Online
pure threo-6 (130 mg) and (ii) a mixture (197 mg) consisting (1H
NMR spectroscopy) of threo-6 (34 mg, total 164 mg, 26%) and
11 (163 mg, 42%).
Run 4, Table 3. Chromatography (3 × 17 cm) with toluene
removed hydrocarbons. Elution with CH2Cl2 provided
anthraquinone (61 mg). Elution with ethyl acetate yielded a
mixture (400 mg) consisting (1H NMR spectroscopy) of
erythro-6 (171 mg, 22%, deuterated at position 10-H pseudo-
axially to an extent of 67%), [2H]11 (229 mg, 49%, degree of
deuteration, 63%) and a trace of threo-6.
Run 5, Table 2. At the times given in Table 2, 20 cm3 were
withdrawn from the solution with a pipette, quenched and
worked up. Chromatography (3 × 17 cm) with toluene removed
hydrocarbons. Elution with ethyl acetate provided the following
results (1H NMR spectroscopy). The 30 min sample gave a mix-
ture (136 mg) consisting of threo-6 (50 mg, 22%), 11 (71 mg,
51%) and anthraquinone (15 mg). The 1 h sample gave a mix-
ture (135 mg) consisting of threo-6 (32 mg, 14%), 11 (87 mg,
63%) and anthraquinone (16 mg). The 27 h sample gave pure 11
(136 mg, 98%).
Runs of Table 4. The samples (20 cm3) of the solution were
withdrawn by means of a pipette. Keeping the sample and the
remaining solution free of air whilst removing samples was dif-
ficult and required quick action. The precision of sampling was
consequently not high. Each sample was quenched with acetic
acid. Evaporation provided a residue that was taken up in
CH2Cl2 and washed with water. Evaporation gave a residue that
was subjected to chromatography (3 × 17 or 3 × 20 cm) with
toluene which removed the hydrocarbons. Further elution was
performed with ethyl acetate, providing the product mixture, or
was performed with CH2Cl2–ethyl acetate (2:3), providing
anthraquinone, followed by elution with ethyl acetate to give
the product mixture. Two samples of run 1 are described with
full details. The other samples of all runs were analysed in the
same manner.
Reactions of 9,10,10-trideutero-AHϪ with trans-4a and cis-4a
[2H4]AH2 (5.5 mmol) in THF (70 cm3), BuLi (5 mmol, in hex-
ane) and trans-4a in THF (20 cm3) reacted for 22 h and was
worked up as described above for reactions given in Table 1.
Chromatography (3 × 15 cm) with toluene removed hydro-
carbons. Elution with ethyl acetate provided a mixture (335 mg)
consisting (1H NMR spectroscopy) of erythro-[2H3]6 (203 mg,
43%) and [2H]11 (132 mg, 46%). Separation by PLC (CH2Cl2–
ethyl acetate 50:1) provided the pure products: erythro-[2H3]6
(deuteration р90%), mp 193–195 ЊC; the 1H NMR data match
with those of erythro-6 except for very weak (р10% of the
required integral) signals for 9-H and both 10-H atoms while
the double doublet (dd) at 3.60 ppm had changed to a doublet
Run 1, Table 4, sample after 5 min. The product mixture (156
mg) consisted (1H NMR spectroscopy) of erythro-6 (98 mg,
50%), threo-6 (17 mg, 8%) and 11 (41 mg, 34%).
Run 1, Table 4, sample after 6.5 h (volume of sample 60 cm3).
The product mixture (224 mg) consisted (1H NMR spec-
troscopy) of erythro-6 (42 mg, 7%), threo-6 (36 mg, 7%) and 11
(146 mg, 40%). The hydrocarbon fraction (1.288 g) contained
(1H NMR spectroscopy) (E)-stilbene (116 mg, р50%) identi-
fied by comparison (1H NMR spectroscopy, tlc) with authentic
material.
1
(d) (J 11.5); [2H]11 (deuteration 82%), mp 140–141 ЊC; the H
NMR data match with those of 11 except for very weak signals
(р10% of the required integral) at 3.07 ppm (1 H of CH2) and
the change of the dd at 3.12 ppm (1 H of CH2) to d (J 6.5).
[2H4]AH2 (17.5 mmol) in THF (120 cm3), BuLi (15 mmol, in
hexane) and cis-4a (3.15 mmol) in THF (40 cm3) were reacted
as described above. A sample (20 cm3) was withdrawn after 3 h
and a second sample (80 cm3) after 4 d. Workup and chrom-
atography (3 × 17 cm) of the second sample with toluene
removed hydrocarbons. Elution with ethyl acetate provided
[2H1]11 (325 mg, 73%), identical with the product described
above. Workup of the first sample and chromatography (3 × 17
cm) with toluene removed the hydrocarbons followed (ethyl
acetate) by mixture I (75 mg) and mixture II (93 mg). Mixture I
consisted (1H NMR spectroscopy) of threo-[2H3]6 (63 mg) and
anthraquinone (12 mg). Mixture II consisted (1H NMR spec-
troscopy) of threo-[2H3]6 (54 mg, total 117 mg, 64%) and [2H]11
(40 mg, 36%). Recrystallization of mixture I from CCl4 pro-
vided pure threo-[2H3]6, mp 215–216 ЊC; the 1H NMR data
matched with those of threo-6 except for very weak (р10% of
the required integral) signals for 9-H and both 10-H, while the
dd at 3.45 ppm (NCCH) had changed to a doublet (J 11.8).
Acknowledgements
Financial support by the Fonds der Chemischen Industrie is
gratefully acknowledged.
References
1 Aziridines, Part 72; For Parts 70 and 71, see refs. 3 and 4.
2 Arene Hydrides. Part 15; for Parts 13 and 14, see refs. 3 and 4.
3 T. Mall and H. Stamm, J. Chem. Res., (S), 1996, 375.
4 T. Mall and H. Stamm, Pharmazie, 1996, 51, 831.
5 H. Stamm and R. Falkenstein, Chem. Ber., 1990, 123, 2227.
6 M. Malissard, J.-P. Mazaleyrat and Z. Welvart, J. Am. Chem. Soc.,
1977, 99, 6933; E. Hebert, J.-P. Mazaleyrat, Z. Welvart, L. Nadjo
and J.-M. Savéant, Nouv. J. Chim., 1985, 9, 75.
7 A. L. J. Beckwith and W. A. Waters, J. Chem. Soc., 1957, 1001.
8 H. E. Zieger, I. Angres and L. Marasco, J. Am. Chem. Soc., 1973,
95, 8201; S. Bank and J. F. Bank, Tetrahedron Lett., 1971, 4581;
Y.-J. Lee and W. D. Closson, Tetrahedron Lett., 1974, 1395.
9 P.-Y. Lin, K. Bellos, J. Werry, P. Assithianakis, R. Weiß, T. Mall,
G. Bentz and H. Stamm, J. Prakt.Chem., 1996, 338, 270.
10 A. W. Brinkmann, M. Gordon, R. G. Harvey, R. W. Rabideau and
L. Stothers, J. Am. Chem. Soc., 1970, 92, 5912; E. J. Panek and
T. J. Rogers, J. Am. Chem. Soc., 1974, 96, 6921.
11 S. Bank and D. A. Juckett, J. Am. Chem. Soc., 1976, 98, 7742.
12 D. Archier-Jay, N. Besbes, A. Laurent, E. Laurent, S. Lesniak and
R. Tardivel, Bull. Soc. Chim. Fr., 1993, 537.
13 P.-Y. Lin, J. Werry, G. Bentz and H. Stamm, J. Chem. Soc., Perkin
Trans. 1, 1993, 1139.
14 T. Mall, B. Buchholz and H. Stamm, Arch. Pharm., 1994, 327, 377.
15 R. Falkenstein, T. Mall, D. Speth and H. Stamm, J. Org. Chem.,
1993, 58, 7377.
16 R. Huisgen and H. Brade, Chem. Ber., 1957, 90, 1432; D. W. Farlow
and R. B. Moodie, J. Chem. Soc. (B), 1970, 334.
Ϫ
Reactions with anthracenide A
᝽
These reactions were performed following the previously13
described techniques. Necessary details are given in Tables 3
and 4. Workup was carried out as described under reactions
with AHϪ.
Run 1, Table 3. Chromatography (3 × 20 cm) with toluene
removed hydrocarbons. Elution with CH2Cl2 provided
anthraquinone (75 mg). Elution with ethyl acetate yielded a
mixture (303 mg) consisting (1H NMR spectroscopy) of
erythro-6 (120 mg, 26%) and 11 (183 mg, 65%).
Run 2, Table 3. Chromatography (3 × 20 cm) with toluene
removed hydrocarbons. Elution with CH2Cl2–ethyl acetate (2:3)
yielded a mixture (276 mg) consisting (1H NMR spectroscopy)
of erythro-6 (85 mg, 18%), 11 (191 mg, 68%) and a trace of
threo-6.
Run 3, Table 3. Chromatography (3 × 17 cm) with toluene
removed hydrocarbons. Elution with CH2Cl2 provided
anthraquinone (34 mg). Elution with ethyl acetate yielded a
mixture (507 mg) consisting (1H NMR spectroscopy) of
erythro-6 (162 mg, 19%) and 11 (345 mg, 66%).
17 A. Sommer and H. Stamm, Liebigs Ann. Chem., 1992, 99.
18 P.-Y. Lin, K. Bellos, J. Werry, P. Assithianakis, R. Weiß, T. Mall,
G. Bentz and H. Stamm, J. Prakt. Chem., 1996, 338, 270.
19 H. Stamm, A. Sommer, A. Woderer, W. Wiesert, T. Mall and
P. Assithianakis, J. Org. Chem., 1985, 50, 4946.
Paper 7/05248K
Received 9th June 1997
Accepted 22nd July 1997
2140
J. Chem. Soc., Perkin Trans. 2, 1997