Experimental and quantum-mechanical investigation of the vinylsilane-iminium ion cyclization

Article abstract of DOI:10.1039/b212333a

A vinylsilane-ketiminium ion cyclization involving iminium species derived from amines 6 and 7 was investigated experimentally as a possible approach to some biologically interesting 1-azaspirocycles. However, even under conditions of microwave irradiatio

Full text of DOI:10.1039/b212333a

View Article Online / Journal Homepage / Table of Contents for this issue
Experimental and quantum-mechanical investigation of the
vinylsilane-iminium ion cyclization†
Lisbet Kværnø, Per-Ola Norrby and David Tanner*
Department of Chemistry, Technical University of Denmark, Building 201, Kemitorvet,
DK-2800 Kgs. Lyngby, Denmark. E-mail: dt@kemi.dtu.dk
Received 16th December 2002, Accepted 5th February 2003
First published as an Advance Article on the web 24th February 2003
A vinylsilane-ketiminium ion cyclization involving iminium species derived from amines 6 and 7 was investigated
experimentally as a possible approach to some biologically interesting 1-azaspirocycles. However, even under
conditions of microwave irradiation at high temperatures no such cyclization was observed whereas (in line with
previous results) the corresponding vinylsilane-aldiminium ion cyclizations were more successful. Aldiminium species
substituted α to nitrogen displayed no diastereoselectivity in the cyclization of precursors derived from 6 while high
trans diastereoselectivity could be obtained for iminium species derived from 7. Quantum-mechanical investigations
of the general reaction mechanism underlined the lack of reactivity of ketiminium species and also convincingly
explained the observed diastereoselectivities of aldiminium species. The calculations further revealed that (Z)-
vinylsilanes cyclize via a silicon-stabilized β-carbocation, and that any formal aza-Cope rearrangement of the starting
material to an allylsilane-iminium species does not take place in a concerted fashion. However, the calculations show
that the aza-Cope rearrangement precedes cyclization for the corresponding (E)-vinylsilanes, the overall reaction
being energetically slightly less favoured than cyclization of the (Z)-isomers.
In contemplating possible routes to the 1-azaspirocyclic portion
of halichlorine, we considered the option of extending the
Introduction
The 1-azaspirocyclic system1 is characteristic of several bio-
vinylsilane-iminium ion cyclization introduced by Overman
logically significant natural products, exemplified in Fig. 1 by
and coworkers (Scheme 1).9–15 For a synthesis of halichlorine,
halichlorine 12,3 and histrionicotoxin 2.4–6 Our group has a
this would obviously involve reaction of a ketiminium species
long-standing interest in developing synthetic routes to such
and, in view of the complete lack of literature precedent, such
compounds,7,8 a major part of the challenge being presented by
a stratagem must be considered as bold since the vast majority
the nitrogen-bearing stereogenic center in the target molecules.
of known vinylsilane-iminium ion cyclizations utilize pre-
cursors based on formaldehyde,10–12,16–18 with a few examples
being reported for iminium ions formed from higher alde-
hydes,11,13–15,19 but none from ketones. We were also unable to
find examples of azaspirocyclization via reaction of vinyl
silanes with the more reactive N-acyliminium20,21 species. As
far as we are aware, the only reports22–24 which describe aza-
spirocyclization via organosilicon chemistry all involve the
more reactive25 allylsilane species, the most recent example22,23
having an N-o-nitrobenzoyl-activated ketiminium precursor.
Our first task was therefore to construct a model system and
to search for reaction conditions which would allow spiro-
cyclization via a vinylsilane-ketiminium ion reaction (vide
infra).
Fig. 1
Notwithstanding the specific goal outlined above, our inter-
† Electronic supplementary information (ESI) available: experimental
est in the vinylsilane-iminium ion reaction was of a more
procedures, spectroscopic data and copies of 13C NMR spectra for
general nature since cyclizations based on this process are
compounds 4–16. Selected bond lengths in geometry optimized
mechanistically intriguing. Two different reaction pathways
structures of α-substituted (Z)-vinylsilanes. Cartesian coordinates,
have been suggested,9–15 with the first, more direct, mechanism
(Scheme 1, upper half ) involving iminium ion formation
raw energies and lower frequencies of computationally characterized
species 17A–32H. See http://www.rsc.org/suppdata/ob/b2/b212333a/
Scheme 1
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1041

View Article Online
followed by cyclization to give the β-carbocation which is stabil-
ized by overlap between the vacant p-orbital and the correctly
aligned C–Si bond. This is followed by facile loss of the silyl
group promoted by an external nucleophile. Alternatively, the
iminium ion has been suggested to participate in a [3,3]-sigma-
tropic aza-Cope rearrangement (Scheme 1, lower half ) to give
the more reactive allylsilane followed by cyclization and con-
comitant loss of the silyl group. The latter mechanism becomes
of particular interest when the iminium species shown in
Scheme 1 has R1 ≠ H, since the aza-Cope process coupled with
iminium ion equilibration could lead to complete racemization
of enantiomerically pure starting materials. Overman has pre-
sented convincing evidence for the intervention of the aza-Cope
rearrangement: isolation of some Eschweiler–Clarke methyl-
ated starting material, as an isomeric mixture of (E)- and
(Z)-vinylsilanes, from the endocyclic reaction of iminium ions
derived from secondary amines and formaldehyde, while no
olefin isomerization was observed in the absence of form-
aldehyde.9–12 The result shown in Scheme 2 is also interesting in
this respect since in contrast to earlier reports13 such substrates
could be transformed into iminium ions with formaldehyde and
directly cyclized with only minor amounts of racemization.14
only the (Z)-vinylsilane led to product.15,19 This result
emphasizes the importance of stabilization of the developing
β-carbocation; easy access to correct alignment of the C–Si
bond with the vacant p-orbital is possible only for the
(Z)-vinylsilane.9,15,19 Thus, when the aza-Cope rearrangement
is precluded, the cyclization still takes place and it is quite
possible that both suggested mechanisms coexist even when the
iminium ion precursor can rearrange.
Returning to the projected halichlorine synthesis, we decided
to start by examining the model spirocyclization shown in
Scheme 3. For this purpose, the vinyl silanes 6 and 7 were
prepared in racemic form as shown in Scheme 4, the (Z)-vinyl-
silanes being chosen in order to favour cyclization via the
β-carbocation intermediate if this were the dominant mechan-
istic pathway.
Scheme 3
Results and discussion
1. Cyclization studies: aldimines vs. ketimines
Vinylsilane-iminium ion cyclizations of preformed imines have
so far only been reported to take place in moderate yields with
acyclic imines derived from non-enolizable aldehydes11,13,14
in the presence of Brønsted acids. To investigate the corre-
sponding ketimine cyclization, amine 6 was treated with cyclo-
pentanone to afford 8 in 92% yield (Scheme 5). Unfortunately,
no cyclization to spirocycle 9 occurred, either by treatment of
8 with TFA in refluxing CH3CN or by in situ imine formation
and subsequent treatment with TFA at reflux. Attempts to
promote the cyclization of 8 with the Lewis acids TMSOTf,
BF3ؒOEt2, TiCl4 or Ti(iPrO)4 at either Ϫ78 ЊC or room temper-
ature or alternatively activation of 8 by in situ acylation with
Scheme 2
These results indicate that even if the aza-Cope rearrange-
ment occurs, the successive cyclization process must be faster
than the iminium ion equilibration. In accordance with these
results, analogous cyclizations involving vinylsilanes and some
chiral N-acyliminium ions have also been reported to take place
with only partial loss of enantiopurity.26
Generally, the stereochemistry of the vinylsilane moiety is
not critical since both cis- and trans-vinylsilanes have been
reported to cyclize.10 However, in the case of cyclization of a
substrate unable to participate in the aza-Cope rearrangement,
Scheme 4
Scheme 5
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1042

View Article Online
Scheme 6
p-NO2C6H4COCl were also unsuccessful. In a control experi-
when equimolar amounts of 7 and benzaldehyde were adsorbed
ment, amine 6 was transformed into imine 10 and subjected to
TFA in CH3CN at 80 ЊC. This gave cyclized products 11A and
11B as a 1 : 1 diastereomeric mixture in a moderate 37% com-
bined yield. Lack of diastereoselectivity in such cyclizations has
been noted earlier for analogous imines.13,14
on solid supports (zeolites, neutral Al2O3, silica gel or
montmorillonite K10) followed by heating in a commercial
microwave oven.
The yield of 14 could be enhanced somewhat by conversion28
of 7 into the α-aminonitrile 16 in 84% yield (Scheme 7) followed
by AgBF4-promoted cyanide abstraction and cyclization.11,13,14
This gave 14 in 46% yield, exclusively as the trans-diastereomer,
as noted for the previous cyclization (Scheme 6). The stereo-
selective formation of the trans-isomer is also in accordance
with previously reported results for silver-promoted cycliz-
ation of some other α-aminonitriles.13,14 Disappointingly, the
analogous conversion of 7 into the α-aminonitrile by treatment
with cyclopentanone did not proceed even at 120 ЊC (for
comparison, quantitative conversion of cyclopentanone with
benzylamine took place at 60 ЊC). At this point we abandoned
our attempts to find reaction conditions conducive to cycliz-
ation of ketiminium ions with vinyl silanes.
Most vinylsilane-iminium ion cyclizations in the literature
have been reported to take place by in situ treatment of a
secondary amine with excess paraformaldehyde in CH3CN or
another polar solvent at reflux in the presence of a strong
Brønsted acid,10–12 while cyclization with heptanal as the alde-
hyde required reaction temperatures of 120 ЊC.11 Accordingly,
application of standard cyclization conditions (paraform-
aldehyde, CSA, refluxing CH3CN) to amine 7 also promoted a
smooth cyclization to give 12 in 72% yield (Scheme 6). The
analogous reaction with heptanal at 120 ЊC gave the cyclized
product 13 as a single diastereomer although in only 29% yield,
while the less reactive benzaldehyde afforded diastereomerically
pure 14 in 10% yield. Compound 14 was shown by NOE differ-
ence experiments to be the trans-2,6-isomer with the phenyl
group pseudoequatorial (Fig. 2). Attempted cyclization by
treatment of 7 with cyclopentanone and CSA at 140 ЊC led only
to a slow protodesilylation while use of the stronger acid TfOH
led to complete and rapid protodesilylation to give 15B in 75%
isolated yield.
Fig. 2 Relevant NOEs for 14.
In an attempt to raise the yields of these cyclizations and to
promote the desired ketiminium reaction, experiments were
carried out with microwave irradiation as the source of heat.27
Unfortunately, microwave irradiation reactions of 7 with hep-
tanal, benzaldehyde and cyclopentanone, respectively, with
CSA in acetonitrile at 160 ЊC gave results similar to those
obtained under thermal conditions, while reaction temper-
atures of 200 ЊC effected only decomposition of both starting
material and products. The microwave irradiation experiments
were also performed with the milder acid pyridinium p-toluene-
sulfonate (PPTS) or totally without addition of acid in either
acetonitrile, methanol or water, as the solvents, but no cycliz-
ation was detectable. The same lack of reactivity was observed
Scheme 7
2. Quantum mechanical investigations of reaction mechanism
Having established that ketiminium species are too unreactive
toward vinylsilanes, we now return to the general question of
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1043

View Article Online
Table 1 (Z)-Vinylsilanes. Calculated free energies relative to starting iminium ions 17, in kJ molϪ1
R = H
R = Me
A
B
C
D
E
F
G
H
18
19
20
21
22
23
27.4
2.5
33.4
78.9
72.8
121.8
97.2b
136.2
38.2
18.0
38.4
Ϫ15.9
23.6
60.8
94.9
81.4
62.6
83.1
42.2
66.9
125.0
116.0
133.4
91.5
120.3
142.3
56.7b
91.8
75.2
51.7
83.0
48.8
87.2
103.2
49.6
92.7
113.7
Ϫ5.8
99.0
a
a
a
a
54.2
105.6
100.8
146.1
108.6
a Not determined. b The trimethylsilyl group had migrated to C4 in the geometry optimization.
Fig. 3 Calculated mechanism for cyclization of (Z)-vinylsilanes.
reaction mechanism (Scheme 1) and the diastereoselectivity
observed in the present investigation (Schemes 6 and 7) and in
others.13,14 For mechanistic insight we decided to investigate the
vinylsilane-iminium ion cyclization by quantum-mechanical
methods at the B3LYP level,29–36 which to the best of our
knowledge has not been reported previously. Both reactions of
iminium species derived from primary (Fig. 3, A–D) and
secondary (Fig. 3, E–H) amines with formaldehyde, acetalde-
hyde and acetone, respectively, were studied (Fig. 3, 17). Since
both the (E)- and the (Z)-iminium ions in principle can cyclize,
two sets of calculations were carried out for aldiminium ions
with the methyl substituent either pseudoequatorial (Fig. 3, B
and F) or -axial (Fig. 3, C and G), respectively, in chair-like
transition states and intermediates. All calculations were con-
ducted in vacuo since solvation effects are believed to be largely
independent of the methyl substitution pattern and thus to be
of equal size for similar structures, regardless of the number of
methyl groups.
together with the free trimethylsilyl cation (23) were also calcu-
lated but due to the high energy of the unstabilized silyl cation
in the gas phase, these energies do not reflect the presumed
stability of the product amines.
Surprisingly, the β-carbocations 19 are stable only with a
pseudoaxial N-substituent. The result of forcing R into pseudo-
equatorial positions followed by energy minimizations was
also partial loss of the silyl substituent to give intermediary
π-silyl complexes similar in structure to transition states 22 but
with R pseudoequatorial.
As can be seen from the calculated free energies relative to
reactant iminium ions 17 for all transition states and intermedi-
ates (Table 1, depicted in Fig. 4), the two largest energy barriers
of similar heights in the reaction profile are for the initial
formation of the β-carbocations 19 via transition states 18
and the rearrangement to allylsilanes 21 via transition states 20.
The three possible pathways available to 19 [reversal to 17,
rearrangement to 21 and the facile irreversible loss of the silyl
substituent to give 23 (vide supra)] are all energetically so feas-
ible that they presumably coexist. In the following discussion
formation of the initial cyclization transition states 18 has been
emphasized.
The calculations also show that the energies to form transi-
tion states 18 are very dependent on the degree of substitution
of the iminium carbon, with the iminium ions 17A and 17E
from formaldehyde being the most reactive and the ketiminium
ions 17D and 17H being the least reactive. A combination of
steric bulk and alkyl group stabilization of the positive charge
would account for this. This is in line with the experimental
results, where even microwave irradiation did not provide suffi-
cient energy for the cyclization of ketiminium ions, and it does
not seem likely that suitable reaction conditions for this type of
cyclization can be found.
2.1 (Z )-Vinylsilanes. For all (Z)-vinylsilanes, both with
protonated imines (17A–D) and in situ formed iminium ions
(17E–H), the first steps of the reaction mechanism can be
described as depicted in Fig. 3. The well-known hyperconju-
gative stabilization of the β-carbocation by the silyl substituent is
great enough to allow for short-lived intermediates 19 and thus
any aza-Cope rearrangement to form 21 (via transition states
20) does not take place concertedly for (Z)-vinylsilanes.
Since all calculations were carried out in the absence of
external nucleophiles, the transition states 22 for direct loss of
the silyl substituent37 actually proved to be transition states for
facile migration of the trimethylsilyl substituent from C3 to C4
(only calculated for E–H). From 19A as the simplest example,
an intermediary π-silyl complex similar in structure to transi-
tion states 22 was localized and proved to be only 0.6 kJ molϪ1
higher in energy than 19A which clearly demonstrates the soft-
ness of the energy curve. No further calculations were carried
out in vacuo for the unaided elimination of the silyl substituent
since the irreversible loss of the silyl substituent is presumed to
be energetically much more favourable in the presence of
external nucleophiles. The relative energies of product amines
The reaction profiles for cyclization of the (E)- and
(Z)-aldiminium ions (17B/F and 17C/G, respectively) are gen-
erally very similar with the latter being slightly favoured, so
there can only be smaller steric clashes between a pseudo-
equatorial methyl group (B/F) and the pseudoaxial trimethyl-
silyl substituent in the various chair-like transition states and
intermediates. When N-alkyl substituents are present, this
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1044

View Article Online
Table 2 (Z)-Vinylsilanes. Selected bond lengths in geometry optimized structures, in Å
R = H
R = Me
A
B
C
D
E
F
G
H
17 C2–C3
18 C2–C3
19 C2–C3
20 C2–C3
17 C5–C6
18 C5–C6
19 C5–C6
20 C5–C6
17 C3–Si
18 C3–Si
19 C3–Si
20 C3–Si
4.539
2.007
1.548
1.627
1.543
1.617
1.560
2.108
1.915
1.955
2.175
1.955
4.650
1.791
1.498a
1.652
1.540
1.638
1.547a
2.072
1.912
1.989
2.193a
1.969
4.540
1.918
1.554
1.640
1.543
1.621
1.557
2.102
1.914
1.965
2.174
1.963
4.646
1.779
1.501a
1.671
1.541
1.631
1.545a
2.070
1.912
1.998
2.184a
1.970
3.518
1.944
1.568
1.612
1.551
1.627
1.589
2.067
1.914
1.965
2.104
1.981
3.668
1.852
1.589
1.647
1.545
1.638
1.586
2.020
1.909
1.986
2.134
1.984
4.652
1.906
1.572
1.605
1.544
1.628
1.587
2.042
1.902
1.970
2.103
1.988
4.744
1.849
1.586
1.664
1.544
1.639
1.575
2.007
1.905
1.989
2.170
1.984
a The trimethylsilyl group had migrated to C4 in the geometry optimization, 19 C4–Si is given.
Fig. 4 Free energy reaction profiles for (Z)-vinylsilanes.
Fig. 5 Calculated mechanism for cyclization of (E)-vinylsilanes.
reactivity difference (G compared to F) is somewhat more
pronounced, however (see Fig. 4, R = Me).
between the C–Si σ-orbital and the empty p-orbital since
these cannot be coplanar (Fig. 6). In accordance with this, our
calculations revealed no β-carbocations analogous to 19 as
intermediates in the reactions of (E)-vinylsilanes41 and these
substrates thus participate in the aza-Cope rearrangement in a
concerted fashion prior to cyclization (Fig. 5). As can be seen
from the calculated free energies (Table 3, depicted in Fig. 7),
the energy barriers for the rearrangement follow the same
pattern as for the (Z)-vinylsilanes, i.e. they are very dependent
on the degree of substitution of the iminium carbon. Thus, the
iminium ions 24A/E from formaldehyde are the least hindered
and most reactive while rearrangement of the ketiminium ions
24D/H is much higher in energy.
The stabilizing hyperconjugation when the C–Si bond and
the vacant p-orbital of the β-carbocations 19 are coplanar is
demonstrated in particular by the elongated C–Si bonds (Table
2, 19 C3–Si) where this bond length is in the range of 2.103–
2.193 Å compared to C–Si bond lengths of 1.902–1.915 Å
observed for 17 and 1.880–1.890 Å observed for the three
methyl groups on silicon. This is in accordance with the work of
White et. al.38 who have shown by X-ray crystallography that
the Si–C bond lengthens with increasing electron demand.
Another interesting feature is that the C2–C3 distance in transi-
tion states 18 is generally shorter when pseudoequatorial
methyl groups are present (B,D,F,H). This could be interpreted
as a later transition state in accordance with Hammond’s postu-
late and the Marcus equation,39,40 since unfavourable sterical
clashes of the pseudoequatorial methyl groups (R2=Me) and
the pseudoaxial trimethylsilyl substituent in the carbocations
19B,D,F and H raise their energies and therefore delay the
transition states for their formation.
2.2 (E )-Vinylsilanes. In contrast to the (Z)-vinylsilanes, all
the analogous (E)-vinylsilanes 24A–H have only a poor overlap
Fig. 6 Orientation of C–Si bond and empty p-orbital in (Z)- and
(E)-vinylsilane carbocations.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1045

View Article Online
Table 3 (E)-Vinylsilanes. Calculated free energies relative to starting iminium ions 24, in kJ molϪ1
R = H
R = Me
A
B
C
D
E
F
G
H
25
26
23
35.1
9.0
68.8
82.9
60.4
112.4
92.5
60.5
115.3
140.6
115.4
160.7
48.4
9.3
86.2
91.4
57.5
123.3
89.1
49.3
122.2
133.1
108.2
157.7
Table 4 (E)-Vinylsilanes. Selected bond lengths in geometry optimized structures, in Å
R = H
R = Me
A
B
C
D
E
F
G
H
24 C2–C3
25 C2–C3
24 C5–C6
25 C5–C6
24 C3–Si
25 C3–Si
4.408
1.634
1.544
1.959
1.906
1.966
4.491
1.650
1.542
1.958
1.904
1.974
4.430
1.656
1.545
1.963
1.904
1.967
4.479
1.676
1.544
1.964
1.902
1.979
3.223
1.654
1.557
1.913
1.905
1.960
4.608
1.674
1.543
1.915
1.900
1.967
4.358
1.673
1.544
1.922
1.899
1.961
3.791
1.697
1.547
1.930
1.898
1.973
Table 5 α-Substituted (Z)-vinylsilanes. Calculated free energies relative to starting iminium ions 27C (R = H) and 27G (R = Me), respectively, in kJ
molϪ1
R = H
R = Me
A
B
C
D
E
F
G
H
27
28
29
30
31
32
Ϫ6.3
66.4
Ϫ6.5
0
0.1
Ϫ18.3
76.7
70.5
76.1
Ϫ6.5
140.4
Ϫ14.6
78.3
69.9
81.7
Ϫ2.3
143.3
0
68.6
51.5
60.3
Ϫ8.3
143.3
Ϫ7.7
81.9
53.9
70.5
Ϫ2.7
140.5
76.1
60.0a
82.0
4.1
102.5
65.2
45.8
64.4
1.7
84.8
59.1
83.5
11.7
98.3
49.3a
69.1
Ϫ3.3
98.2
102.4
a The trimethylsilyl group had partly migrated to give a π-silyl complex in the geometry optimization.
2.3 ꢀ-Substituted (Z )-vinylsilanes. In an attempt to explain
the observed diastereoselectivity in the vinylsilane-iminium ion
cyclization when substituents α to nitrogen of a secondary
amine were present (13 and 14, Schemes 6 and 7) together with
the complete lack of diastereoselectivity for preformed imines
from primary amines (11A/B, Scheme 5), the calculations were
repeated for aldiminium ions with a methyl group either
pseudoequatorial (R4 = Me, A,B,E and F) or -axial (R3 = Me,
C,D,G and H) at C6 as a model study of the experimentally
investigated benzyloxyethyl substituent (Fig. 8). Once more,
silicon’s hyperconjugative stabilization of the β-carbocation
allowed for a short-lived intermediate 29, implying that any
aza-Cope rearrangement of 27 to 31 takes place in a stepwise
fashion (cf. Section 2.1).
As can be seen (Table 5, depicted in Fig. 9), the energy
barriers for formation of the initial transition states 28A–H and
for the aza-Cope rearrangements 30A–H are of similar heights
in the reaction sequence. For protonated imines (R=H, A–D)
two favoured initial transition states of comparable energies,
28A (66.4 kJ molϪ1) and 28C (65.2 kJ molϪ1), both with a
pseudoequatorial α-substituent (R4 = Me) are found. Since 28A
and 28C lead to diasteomerically different products, i.e. 32A
and 32C which are cis and trans with respect to the ring substi-
tuents, respectively, this lack of energy difference also corre-
sponds well with the observed lack of diastereoselectivity in the
cyclization of preformed imines (e.g. 11A/B, Scheme 5). In con-
trast to these results, one transition state 28G (68.6 kJ molϪ1) is
formed more readily than the remaining three for in situ formed
iminium ions (R = Me, E–H). This corresponds well with the
observed diastereoselectivity for iminium ions from secondary
amines (e.g. 13 and 14, Schemes 6 and 7). In analogy with the
unsubstituted (Z)-vinylsilanes from secondary amines (Table 1,
Fig. 7 Free energy reaction profiles for (E)-vinylsilanes.
The lack of effective hyperconjugation of the C–Si bond for
(E)-vinylsilanes is also demonstrated by C–Si bonds in the
range of 1.960–1.979 Å (Table 4) in the transition states 25
compared to the elongated C–Si bonds in the range of 2.103–
2.193 Å observed for the β-carbocations 19 from the (Z)-vinyl-
silanes. The bond lengths in the transition states 25 are gener-
ally slightly longer for the ketiminium ions (25D/H) which
might be interpreted as a “looser” transition state caused by
unfavourable steric interactions of the additional methyl
groups.
The activation energies are generally slightly higher than for
the corresponding (Z)-vinylsilanes but probably not enough to
preclude the reaction to occur with iminium ions derived from
formaldehyde42 or higher aldehydes. For the (E)-vinylsilanes,
the small reactivity difference between (E)- and (Z)-iminium
ions (24B/F and 24C/G, respectively) is in favour of the former,
presumably due to smaller steric clashes between a pseudo-
equatorial methyl group (B/F) and the pseudoequatorial
trimethylsilyl substituent.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1046

View Article Online
Fig. 8 Calculated mechanism for cyclization of α-substituted (Z)-vinylsilanes.
spirocycles but even under conditions of microwave irradiation
at high temperatures, no cyclization was observed. In line with
previous results11,13–15,19 vinylsilane-iminium ion cyclizations
employing electrophilic species derived from aldehydes were
more successful. High trans diastereoselectivity could be
obtained for the cyclization of iminium species from second-
ary amines substituted α to nitrogen while no diastereo-
selectivity was observed for analogous cyclizations of iminium
species from primary amines. Quantum-mechanical investi-
gations of the reaction mechanism further underlined the
difficulties of finding reaction conditions appropriate for
cyclization of ketiminium species. The calculations also
revealed that (Z)-vinylsilanes cyclize via a silicon-stabilized
β-carbocation, and that any rearrangement of the starting
material to an allylsilane-iminium species does not take
place via a concerted aza-Cope rearrangement. For (E)-vinyl-
silanes, however, the aza-Cope rearrangement precedes cycliz-
ation and the overall reaction is energetically slightly less
favoured than cyclization of the (Z)-isomers. Our calculations
also provide the first cogent explanation for the observed trans
diastereoselectivity for (Z)-vinylsilane iminium ions from
secondary amines with substituents α to nitrogen and the lack
of diastereoselectivity for (Z)-vinylsilane iminium ions from
the corresponding primary amines.
Acknowledgements
We thank the Technical University of Denmark, H. Lundbeck
A/S, The Danish Natural Sciences Research Council and The
Carlsberg Foundation for financial support. Thomas Høyer
and Leo Pharmaceuticals are gratefully acknowledged for
providing access to the microwave equipment.
Fig. 9 Free energy reaction profiles for α-substituted (Z)-vinylsilanes.
18G compared to 18F), the (Z)-iminium ion with its pseudo-
axial methyl group (R1 = Me) suffers from the least steric
hindrance in the transition state 28G while in all cases (28A,
28C and 28G), a pseudoequatorial α-substituent (R4 = Me)
is favoured. Quite interestingly, when the energies of the
rearranged intermediates 31 relative to reacting iminium ions
27 are compared with the results for the rearranged aldiminium
ions in Table 1 (21B/C/F/G, E = 42.2 Ϫ 75.2 kJ molϪ1), it can be
seen that when alkyl substituents are present at C6, the energy
difference between 31 and 27 is much smaller (Table 4, E = Ϫ6.5
Ϫ 11.7 kJ molϪ1). Altogether, the results of the quantum-
mechanical calculations are in excellent agreement with the
observed reactivity/diastereoselectivity trends in the vinylsilane-
iminium ion cyclization.
References
1 Other recent synthetic methods for 1-azaspirocycles: (a) M. J.
McLaughlin, R. P. Hsung, K. P. Cole, J. M. Hahn and J. S. Wang,
Org. Lett., 2002, 4, 2017–2020; (b) M. D. B. Fenster, B. O. Patrick
and G. R. Dake, Org. Lett., 2001, 3, 2109–2112; (c) S. Ciblat,
J. L. Canet and Y. Troin, Tetrahedron Lett., 2001, 42, 4815–4817;
(d ) T. Ito, N. Yamazaki and C. Kibayashi, Synlett, 2001, 1506–1510;
(e) P. Huang, K. Isayan, A. Sarkissian and T. Oh, J. Org. Chem.,
1998, 63, 4500–4502.
2 M. Kuramoto, C. Tong, K. Yamada, T. Chiba, Y. Hayashi and
D. Uemura, Tetrahedron Lett., 1996, 37, 3867–3870.
3 D. Trauner, J. B. Schwarz and S. J. Danishefsky, Angew. Chem.,
Int. Ed., 1999, 38, 3542–3545.
4 Isolation of natural product: J. W. Daly, I. Karle, C. W. Myers,
T. Tokuyama, J. A. Waters and B. Witkop, Proc. Natl. Acad. Sci. U.
S. A., 1971, 68, 1870–1875.
5 First total synthesis of racemic material: S. C. Carey, M. Aratani
and Y. Kishi, Tetrahedron Lett., 1985, 26, 5887–5890.
6 First enantioselective total synthesis: G. Stork and K. Zhao,
J. Am. Chem. Soc., 1990, 112, 5875–5876.
Conclusions
In summary, a vinylsilane-ketiminium ion cyclization was
investigated experimentally as a possible approach to 1-aza-
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1047

View Article Online
7 D. Tanner and L. Hagberg, Tetrahedron, 1998, 54, 7907–7918.
8 D. Tanner, L. Hagberg and A. Poulsen, Tetrahedron, 1999, 55, 1427–
1440.
9 T. A. Blumenkopf and L. E. Overman, Chem. Rev., 1986, 86, 857–
873.
10 L. E. Overman, T. C. Malone and G. P. Meier, J. Am. Chem. Soc.,
1983, 105, 6993–6994.
11 C. J. Flann, T. C. Malone and L. E. Overman, J. Am. Chem. Soc.,
1987, 109, 6097–6107.
12 S. F. McCann and L. E. Overman, J. Am. Chem. Soc., 1987, 109,
6107–6114.
13 G. W. Daub, D. A. Heerding and L. E. Overman, Tetrahedron, 1988,
44, 3919–3930.
mechanics using MacroModel [pseudosystematical Monte Carlo
conformational search32 (100 steps, limit 30 kJ molϪ1) with CHCl3 as
solvent, MMFFs force field33] followed by a quantum-mechanical
geometry optimization using Jaguar (B3LYP34,35/6-31G*). Initial
transition state geometries were located by coordinate driving
followed by a full refinement (B3LYP/6-31G*, eigenvector following,
quantum-mechanical initial Hessian, ultrafine accuracy level, fine
grid density). Transition states and minima were verified to have
exactly one and zero negative eigenvalues, respectively. Negative
eigenvalues were animated using Molekel v. 4.136 and were proved in
all cases to be transition states for formation of the expected C–C
bond (in three cases of intermediary cations, 17E, 24B and Me3Si؉,
one negative eigenvalue was found and shown to be a transition
state for rotation of a methyl group). All quantum-mechanical
calculations were performed in vacuo since the software does not
allow for the important frequency analyses to verify located
transition states when a solvent is included.
14 P. Castro, L. E. Overman, X. M. Zhang and P. S. Mariano,
Tetrahedron Lett., 1993, 34, 5243–5246.
15 R. M. Burk and L. E. Overman, Heterocycles, 1993, 35, 205–225.
16 L. E. Overman and A. J. Robichaud, J. Am. Chem. Soc., 1989, 111,
300–308.
30 F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liskamp,
M. Lipton, C. Caufield, G. Chang, T. Hendrickson and W. C. Still,
J. Comput. Chem., 1990, 11, 440–467.
17 L. E. Overman, C. J. Flann and T. C. Malone, Org. Synth., 1990, 68,
188–195.
18 P. Compain, J. Gore and J. M. Vatele, Tetrahedron, 1996, 52, 6647–
6664.
31 http://www.schrodinger.com/. Schrödinger (Portland, OR, USA),
2003.
19 L. E. Overman and R. M. Burk, Tetrahedron Lett., 1984, 25, 5739–
32 J. M. Goodman and W. C. Still, J. Comput. Chem., 1991, 12, 1110–
5742.
1117.
20 W. N. Speckamp and M. J. Moolenaar, Tetrahedron, 2000, 56,
3817–3856.
33 T. A. Halgren, J. Comput. Chem., 1996, 17, 490–519.
34 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
35 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785–789.
36 36 http://www.cscs.ch/molekel/. Swiss Center for Scientific
Computing, ETH (Zürich, Switzerland), 2003.
37 Located by coordinate driving of the carbon–silicon bond followed
by a full refinement, see ref. 29.
38 (a) A. J. Green, Y.-L. Kuan and J. M. White, J. Org. Chem., 1995, 60,
2734–2738; (b) V. Y. Chan, C. I. Clark, J. Giordano, A. J. Green,
A. Karalis and J. M. White, J. Org. Chem., 1996, 61, 5227–5233.
39 F. Jensen, in Introduction to Computational Chemistry, Wiley, 1999,
chapter 15.5.
21 W. N. Speckamp and H. Hiemstra, Tetrahedron, 1985, 41, 4367–
4416.
22 P. Sun, C. X. Sun and S. M. Weinreb, Org. Lett., 2001, 3, 3507–
3510.
23 P. Sun, C. X. Sun and S. M. Weinreb, J. Org. Chem., 2002, 67,
4337–4345.
24 R. Ahmed-Schofield and P. S. Mariano, J. Org. Chem., 1985, 50,
5667–5677.
25 I. Fleming, J. Dunoguès and R. Smithers, Org. React., 1989, 37,
57–575.
26 F. P. J. T. Rutjes, J. J. N. Veerman, W. J. N. Meester, H. Hiemstra and
H. E. Schoemaker, Eur. J. Org. Chem., 1999, 1127–1135.
27 The reactions were performed in a multi-mode microwave oven,
ETHOS 1600 from Milestone, Italy.
28 M. P. Georgiadis and S. A. Haroutounian, Synthesis, 1989, 616–618.
29 Methods for calculations. Calculations were performed using Jaguar
v. 4.2 and Macromodel v. 7.030 in Maestro v. 4.1.31 Acyclic reactants
and intermediates were initially energy-minimized by molecular
40 G. S. Hammond, J. Am. Chem. Soc., 1955, 77, 334–338.
41 For all transition states 25A–H, elongation of either C2–C3 or C5–
C6 by 0.05 Å followed by an energy minimization (trust radius 0.1)
led to structures very similar to reactants 24A–H or rearranged
intermediates 26A–H, respectively, and thus no carbocationic
intermediates were verified.
42 Acyclic (E)-vinylsilanes have also been shown experimentally to
cyclize readily with formaldehyde, see ref. 10.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 0 4 1 – 1 0 4 8
1048